diff --git "a/result.json" "b/result.json"
new file mode 100644--- /dev/null
+++ "b/result.json"
@@ -0,0 +1 @@
+[{"meeting_id": "TS3005c", "summary": "Project Manager began a conceptual design meeting with the group, they discussed the components design, interface design at first. In the presentation, material, colour, type and supplement were referred. Then marketing shared the watch trend over the past and future, and thought young and old people's different tastes should be taken into consideration when designed. After that, the group made a discussion about the remote control concept. The final decision about remote control was made by a round shape and green colour with speech recognition and LCD screen to appeal to the target population.", "dialogue": "Speaker C: Oh, how do you find?\nSpeaker C: What's the doer?\nNone: Doo doo doo doo Transmitter transmitter Transmitter\nNone: No! Finche, eh?\nNone: If you don't see here Like a little box?\nNone: So I can just shoot off, take...\nNone: I also stood What?\nSpeaker E: What?\nSpeaker B: Please.\nSpeaker E: JBL on the wall\nNone: JBL on the wall Blah, no!\nNone: I don't know, is it okay?\nSpeaker B: Do I am?\nSpeaker C: Mm, I'm wearing that cute\nNone: No, no, no Right, sideways Move to a meeting room and break a laptop.\nNone: Laptop in hand.\nNone: Suck the internet drive.\nNone: I'm not sure.\nNone: I'm not sure.\nNone: I'm not sure.\nNone: I'm not sure.\nNone: I'm not sure if I'm not sure.\nNone: I'm not sure.\nNone: I'm not sure.\nNone: You should speak in with me.\nNone: You look fine.\nNone: Can I tell you what this looks like?\nNone: No?\nNone: Oh.\nNone: Hello.\nNone: Hello.\nNone: Hello there little friend.\nNone: Hello.\nNone: Halle.\nNone: Hello.\nNone: Hello.\nNone: Sorry.\nNone: Taw.\nNone: Artistic, artistic.\nSpeaker A: A tree.\nNone: Happy tree friends.\nNone: Tea.\nSpeaker D: Okay welcome everyone to our next meeting.\nSpeaker D: I'm busy writing and busy leading the meeting.\nSpeaker D: I'm not sure if I'm going to write the meeting, but I've prepared a little presentation once again.\nSpeaker D: Or at least an agenda.\nSpeaker D: I think the biggest part of the presentation will be on your side.\nSpeaker D: We are here at the conceptual design meeting, which is hereby opened.\nSpeaker D: Once again I will try to write some minutes, which I just, from a previous meeting, placed inside our project folder.\nSpeaker D: Which was quite some typing.\nSpeaker D: Today we once again have three presentations, if I'm right.\nSpeaker D: And after that we will take a decision on the remote control concept.\nSpeaker D: And just as the last time we have 40 minutes to accomplish that.\nSpeaker D: Okay.\nSpeaker D: Well, I'd say let's start with the first presentation in the same order as last time.\nSpeaker D: Okay.\nSpeaker D: I'll take it away.\nSpeaker C: Bling.\nSpeaker C: Okay.\nSpeaker C: And welcome.\nSpeaker C: You all.\nSpeaker C: Components design.\nSpeaker C: First of all, I'd like to comment on some of those things.\nSpeaker C: I'll elaborate some of the things I did.\nSpeaker C: I'll elaborate it on the concept.\nSpeaker C: What should be said about the components?\nSpeaker C: Its properties and what kind of material we used to make one of those remote controls.\nSpeaker C: First of all, I've subtracted some of the components that are used from what the remote control has formed.\nSpeaker C: First of all, the case, the case, the surrounding of the remote control.\nSpeaker C: I would like to give you an idea of how I thought about doing the screw which was gone there.\nSpeaker C: Okay.\nSpeaker C: The case is made from rubber, I suppose.\nSpeaker C: That's one of the, because when you use a remote control, a lot of people will drop their remote control in the break because the titanium was also an option.\nSpeaker C: But it's a very expensive material.\nSpeaker C: Rubber is, I think, the best suitable material to use for remote control.\nSpeaker C: It's also possible to create fancy colors with rubber.\nSpeaker C: Rubber makes it easy to let it self-color.\nSpeaker C: Titanium, you have to paint it.\nSpeaker C: And with that, it's possible to scratch it or make it ugly.\nSpeaker C: The total piece of rubber that's used to make the case is the same color.\nSpeaker C: So if you scratch it, it's still the same color, perhaps it's a little bit damaged.\nSpeaker C: But it's a very strong material.\nSpeaker C: I had an idea, it's single curved.\nSpeaker C: So it's two-dimensional.\nSpeaker C: I think it's best to draw a green.\nSpeaker A: It's a color.\nSpeaker C: It's a color.\nSpeaker C: New.\nSpeaker C: Forward blank.\nSpeaker E: Okay.\nSpeaker C: That's my, it's black.\nSpeaker C: Okay.\nSpeaker C: I thought of an idea like this.\nSpeaker C: Or not that.\nSpeaker C: Blank.\nSpeaker C: Okay.\nSpeaker C: So it also looks nice when it's on your table.\nSpeaker C: So you get, it doesn't lay flat down on the table, but it stands.\nSpeaker C: It's a size few.\nSpeaker C: Size few, yes, a size few.\nSpeaker C: So I'm not technically good at 3D modeling.\nSpeaker C: It's just an idea I had, so it's very, so it also looks nice when it's on the table.\nSpeaker C: The graphical user interface and the buttons.\nSpeaker C: We also thought about that already.\nSpeaker C: I thought about the LCD touch cream, which is the easiest to clean too.\nSpeaker C: One of the great advantages of the LCD screen is just use some ClosX or another cleaning stuff.\nSpeaker C: And it should be made of strong plastic and it should be bright.\nSpeaker C: Well, I already explained some properties of that material.\nSpeaker C: And I think, oh, you also almost concluded about that.\nSpeaker C: This should be our two button component.\nSpeaker C: So that's all about the buttons.\nSpeaker C: Vectories, we also thought about that already, will be chargeable with an option for a mount station.\nSpeaker C: So you can put the remote control in a mount station so it charges itself up instead of plugging it in or something like that.\nSpeaker C: Questions?\nSpeaker C: No, no, no, no.\nSpeaker C: Just looking.\nSpeaker C: And they should be long lasting.\nSpeaker C: Not be empty in about two minutes or 30 minutes or 40 minutes of use.\nSpeaker C: And next step is the chip, the component that makes or transmits the signal to the television.\nSpeaker C: There was an option to use a rather simple chip that I think because of our highly requiring requirements.\nSpeaker C: There should be an advanced chip in it with also the ability to facilitate speech recognition, which unfortunately is still in a test phase.\nSpeaker C: So there should be some more investigation on that side.\nSpeaker C: My personal preference is I also overheard in the last meeting that we should use our own business colors.\nSpeaker C: That was correct, wasn't it?\nSpeaker C: I think they are rather boring for use with rubber.\nSpeaker D: Well, business colors, I thought it was the slogan and the corporate image.\nSpeaker C: They should be in mind, wasn't it?\nSpeaker D: But I don't think you have to make the entire thing in there.\nSpeaker C: Okay, so it doesn't says to have the slogan.\nSpeaker C: It's very recognizable.\nSpeaker C: Okay, okay, well, that's possible, of course.\nSpeaker C: You could make a little R&R on the top of the machine.\nSpeaker C: So they are pretty boring, I suggest because of the availability of rubber to make fantastic colors.\nSpeaker C: And also in a lot of possible colors, so it's possible to make very fancy remote controls, which people, in which people will find their interesting.\nSpeaker C: And we'll buy faster when they look at the same old gray or black colors.\nSpeaker C: As I said before, rubber is impossible to damage severely.\nSpeaker C: Instead of, of course, you can break it when you break it.\nSpeaker C: For example, with a pair of scissors or something like that.\nSpeaker C: But if you drop it, it's not broken right away instead of using plastic, hard plastic or titanium.\nSpeaker C: I personally liked the single curved remote control because it makes it fancy.\nSpeaker C: Well, that's all about my findings.\nSpeaker D: Okay, thank you.\nSpeaker A: Next.\nSpeaker A: I saw a little bit about the interface.\nSpeaker A: I should look.\nSpeaker A: We determined that there will be no buttons, but only on the LCD screen, so I had to look on that.\nSpeaker A: And the design is therefore based on what we just thought of.\nSpeaker A: First, there are some new findings and new technology for speech recognition.\nSpeaker A: And this is that you give a question through the device and it answers you.\nSpeaker A: And they already put this in a coffee maker.\nSpeaker A: And so that you say, good morning coffee maker and it says to you back, good morning, Joe, or what you need.\nSpeaker A: There's an easy way to program that.\nSpeaker A: You say, record into the device, into the speaker, and then you say the question and three seconds later, you say the answer and then when you say the question, it gives you the answer.\nSpeaker A: Perhaps it's useful because people lose the remote, they can yell remote.\nSpeaker A: Remote, where are you?\nSpeaker A: And that's cold or something.\nSpeaker A: Perhaps we could implement that.\nSpeaker A: And then I have to go out of the presentation because I tried to make some kind of idea of how to do plaque, but I can draw.\nSpeaker A: Don't make too much of it.\nSpeaker A: I tried to the LCD screen, I tried to draw.\nSpeaker A: I thought at least the icon for the volume.\nSpeaker A: I don't know if there's an icon for the program.\nSpeaker A: No, just P.\nSpeaker A: So, and then the buttons above and below.\nSpeaker A: The mute button also recognized as an icon.\nSpeaker F: Where's the button for two digits?\nSpeaker A: I forgot that one.\nSpeaker A: I thought I forgot something.\nSpeaker A: And the numbers, it should be a bit larger, I think.\nSpeaker A: It's not really on scale and so forth.\nSpeaker A: An options button.\nSpeaker A: And I thought that the button for the tail text apart, because it's not really options, I think.\nSpeaker A: The options is the settings of the remote and of the TV and that kind of thing.\nSpeaker A: You could call it settings or something.\nSpeaker A: But this is a bit how I thought it.\nSpeaker A: And the LCD somewhere on your remote, perhaps you could be more curvy to the remote.\nSpeaker A: So that it's better in your hand or something.\nSpeaker A: And microphone for the speech recognition if you want to implement that.\nSpeaker A: And then if you press the options button, then we have an example of...\nSpeaker A: You could get the other options with what you could do.\nSpeaker A: And you could do with something like this.\nSpeaker A: And it's also, I think we discussed earlier that other people don't really want to use these extra settings.\nSpeaker A: And other people also don't really want to use this kind of option we use.\nSpeaker A: So they want to use one button and then something happens.\nSpeaker A: And not choose with this kind of...\nSpeaker A: And you could put a little scrolling device on the side of your remote so you could scroll across these things.\nSpeaker A: That's an option.\nSpeaker A: That was my finding thing.\nSpeaker D: Thanks.\nSpeaker E: No, no Danny.\nSpeaker D: Do you remember this presentation?\nSpeaker F: Okay.\nSpeaker F: I'm going to tell you something about the trend washing.\nSpeaker F: The trends form the past years, what people like, what the youngsters like, what the only people like about shapes, colors, material and stuff.\nSpeaker F: The method I use was like I told, I watched the threads from past years about color shapes, material they wanted from elderly and young people.\nSpeaker F: So you can keep that in mind for designing the device itself.\nSpeaker F: The first thing I made, the most important thing people liked last year was that the remote control should look fancy.\nSpeaker F: The second important thing that should be innovative, like the LCD screen is quite innovative so that should be a great meeting for this.\nSpeaker F: And the third thing is it should be easy to use.\nSpeaker F: I think with only one menu for button channel volume it should also be enough for easy to use.\nSpeaker F: Personal preferences for the young people, they liked fruity colors like banana yellow, strawberry red and stuff.\nSpeaker F: Grass green.\nSpeaker F: Yeah, like that.\nSpeaker F: The round shapes and soft materials, materials like the rubber, it should be soft, it should be feeling spongy or a spongebob like this.\nSpeaker A: Let's build it into a sponge.\nSpeaker F: Elderly people like colors which are being seen in autumn like wood, dark brown, deep red and stuff.\nSpeaker F: They like square shapes with round edges and hot materials like wood, titanium, those kind of materials they like.\nSpeaker F: This is a bit like the young people's like, the fruity colors, innovative, all the colors you see the blue, red, white, yellow, that stuff.\nSpeaker F: And then I personally thought a front side of the shape should be something for the young people's like this or something.\nSpeaker F: It's a bit like banana and the color should be yellow or something.\nSpeaker F: And for elderly people just plain old because we decided to have two kind of remotes to designs or was it two colors?\nSpeaker F: Different colors, yeah.\nSpeaker F: We should decide whether it's going to be with round shapes.\nSpeaker F: I think like my colleague, you said, is that better or for the elderly people something like the iPod or something with round squares.\nSpeaker F: Simple but easy to use.\nSpeaker F: So that's it.\nSpeaker F: All the colors I can maybe change the colors that are also yellow.\nSpeaker A: I don't know which shape you should take.\nSpeaker D: The color of the color will be easier than changing.\nSpeaker C: Change the shape of the image.\nSpeaker A: Perhaps you could find something in the middle.\nSpeaker A: Round squares.\nSpeaker C: Round corners, but square, yeah.\nSpeaker D: Maybe then both groups won't buy it.\nSpeaker A: If you do it square with round corners, but in the middle of it.\nSpeaker A: Yeah, I know what you mean.\nSpeaker C: Kind of like a...\nSpeaker A: Yeah, whatever.\nSpeaker A: Like this.\nSpeaker A: A bit.\nSpeaker C: Yeah.\nSpeaker A: So it's a bit square, but it's also a bit...\nSpeaker C: Kind of like a beer glass.\nSpeaker A: So, but then...\nSpeaker A: I know what you mean.\nSpeaker A: Same size.\nSpeaker C: It's also easy to put in your hand.\nSpeaker A: Yeah, but it's also how other remotes are shaped.\nSpeaker A: So that's a good thing.\nSpeaker A: So that's easy to use.\nSpeaker A: People know that.\nSpeaker C: We'll recognize that as a remote control.\nSpeaker C: What I saw in your...\nSpeaker F: Look, something like that.\nSpeaker F: Out of colors, like red, brown.\nSpeaker F: They like the wood a lot.\nSpeaker F: Maybe we could give it like a wooden look or something.\nSpeaker E: Kind of old colors.\nSpeaker F: A bit old school style.\nSpeaker F: Red, sun, medieval colors.\nSpeaker F: It's all together.\nSpeaker F: Those kind of things.\nSpeaker F: So you see the big difference between the young people.\nSpeaker F: Fresh, exciting, and old people.\nSpeaker F: So that's easy to do with color, have you?\nSpeaker F: So, sorry?\nSpeaker A: That's easy to do with the color.\nSpeaker F: Yeah, I think it's easier to do in color than in shape.\nSpeaker F: Otherwise, we have to get different shapes.\nSpeaker F: And color is way easier than shapes.\nSpeaker F: And material, yeah, rubber.\nSpeaker F: Rubber is, like I said, young people like more soft materials, and spongy ones, and the old people like plain wood.\nSpeaker F: So we have to decide if we're going to use really hard rubber or soft rubber.\nSpeaker F: Or something.\nSpeaker F: Something we can see that.\nSpeaker F: Also in between.\nSpeaker F: Yeah, soft rubber, which you can...\nSpeaker F: Feel it.\nSpeaker F: I know.\nSpeaker C: I don't think you should be promoted.\nSpeaker F: No, or what should not be on the next question.\nSpeaker F: No, no, but you have to like erase or something.\nSpeaker F: That's a bit you can press it in or something harder.\nSpeaker A: But like this kind of rubber.\nSpeaker A: Yeah, something like this, yeah.\nSpeaker A: This is quite hard.\nSpeaker F: Yeah, it's quite hard, but you can press it in.\nSpeaker F: It feels kind of spongy.\nSpeaker F: Spongy.\nSpeaker E: Spongy.\nSpeaker E: Something like this.\nSpeaker A: I don't think it's rubber.\nSpeaker D: So we need a spongey feeling.\nSpeaker A: If you have something.\nSpeaker E: You're going to invite Spongebob.\nSpeaker E: Maybe he can...\nSpeaker A: So we should first decide about shape, I think.\nSpeaker A: Yeah, I think it's the better thing to do.\nSpeaker A: Then you can fit the LCD screen in it.\nSpeaker A: Oh, Moses.\nSpeaker A: I'm going to start with...\nSpeaker C: I also can't help but notice.\nSpeaker C: As you used, you had a remote control, but the LCD screen was rather small.\nSpeaker C: Yeah, I think the LCD screen should be...\nSpeaker C: Like your size.\nSpeaker C: But I think it should be larger.\nSpeaker C: Yeah, three quarters.\nSpeaker C: So you don't have to put your...\nSpeaker F: But it won't get that small.\nSpeaker F: What LCD screen is.\nSpeaker C: So when you get this...\nSpeaker C: Kind of like this.\nSpeaker C: Or should it be larger?\nSpeaker C: Larger, I think.\nSpeaker C: Larger?\nSpeaker C: Because you want to put your hand...\nSpeaker A: Yeah, because you can't put your hand there.\nSpeaker A: And then you won't touch the screen.\nSpeaker A: Okay.\nSpeaker E: True, true, true, true.\nSpeaker E: You're thumb here.\nSpeaker A: But not on the screen.\nSpeaker A: Yeah.\nSpeaker A: That's what I mean.\nSpeaker A: Because when you put your...\nSpeaker C: Yeah, but if you make the LCD screen as large as the remote control itself, you always get some...\nSpeaker C: You always touch it.\nSpeaker F: You're going to be very greasy and stuff.\nSpeaker F: But it won't get that small because you have how much?\nSpeaker F: 9, 10, 11, 12, 13, 14, 15 buttons on the screen.\nSpeaker F: 1, 2, 0.\nSpeaker F: You don't want it to take it too small.\nSpeaker F: You have to...\nSpeaker F: Because some fat people would think fingers.\nSpeaker F: I don't think you're doing it for the first three buttons this time.\nSpeaker C: We've got...\nSpeaker C: Well, if you want to make it international, Japanese people get it rather small hands.\nSpeaker C: Yeah, true.\nSpeaker C: But we would love...\nSpeaker C: Yeah, we have the zoom option.\nSpeaker C: Right?\nSpeaker C: The zoom option?\nSpeaker C: Of course.\nSpeaker C: Yeah.\nSpeaker C: You can make it larger with accompanying a greater field to push the button.\nSpeaker C: Yeah.\nSpeaker D: And we won't include a pen or something to point.\nSpeaker C: I think we won't do her fingers.\nSpeaker C: Yeah.\nSpeaker C: Your fingers.\nSpeaker C: Yeah.\nSpeaker C: You can lose a pen.\nSpeaker C: You lose a pen.\nSpeaker C: You can't use...\nSpeaker D: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: They think it's handling the use pen.\nSpeaker A: Yeah.\nSpeaker A: I think this is a good size for the screen.\nSpeaker A: I don't know how large the actual remote should be.\nSpeaker A: But...\nSpeaker D: Maybe we have to add a tissue to remove the grease from all that.\nSpeaker D: Yeah.\nSpeaker D: That can be used for any soft tissue.\nSpeaker D: Yeah.\nSpeaker D: You can buy those.\nSpeaker D: Okay.\nSpeaker D: So maybe I should show some points on which we should take a decision.\nSpeaker D: Sure.\nSpeaker D: So we can discuss these points.\nSpeaker D: There are those points.\nSpeaker D: Energy question mark.\nSpeaker D: How should we supply the thing with energy and how are we going to do a separate docking station and then put a cues in it.\nSpeaker D: Chip on print and case.\nSpeaker D: Those are points my coach advised me to discuss here.\nSpeaker D: But I hope you have ideas about them.\nSpeaker A: I think you were batteries and then...\nSpeaker A: The docking station.\nSpeaker A: Yeah.\nSpeaker D: So that's the first point we'll include rechargeable batteries.\nSpeaker F: Yeah.\nSpeaker F: Which you can recharge through the docking station.\nSpeaker F: Yeah.\nSpeaker F: So if the batteries are dead.\nSpeaker F: I kind of like your...\nSpeaker C: Change them.\nSpeaker C: Hey, you got some of those wireless mice.\nSpeaker C: Yeah.\nSpeaker B: Computer mice.\nSpeaker C: Yeah.\nSpeaker C: Kind of like those kind of batteries.\nSpeaker A: Yeah.\nSpeaker A: But it should be, I think, normal batteries.\nSpeaker A: Not like to...\nSpeaker F: Yeah.\nSpeaker F: Normal plane batteries.\nSpeaker F: Yeah.\nSpeaker F: Normal plane batteries.\nSpeaker D: Yeah.\nSpeaker D: You can buy the supermarket or rechargeable.\nSpeaker D: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nSpeaker A: Yeah.\nSpeaker A: And what was with the chip on print?\nSpeaker C: The chip on print.\nSpeaker C: You got...\nSpeaker C: Yeah.\nSpeaker C: I think so.\nSpeaker C: Chip on print with a...\nSpeaker C: But simple.\nSpeaker C: Yeah.\nSpeaker C: Simple, but we also discard that.\nSpeaker C: Didn't we?\nSpeaker C: Yeah.\nSpeaker D: How does that work with LCD screen?\nSpeaker D: You still have a...\nNone: Print.\nSpeaker C: Yeah.\nSpeaker E: You always have a print plate.\nSpeaker C: Thank you, Juan.\nSpeaker E: You always have a print plate, right?\nSpeaker E: Yeah, sure.\nSpeaker E: Of course.\nSpeaker E: Always.\nSpeaker E: So...\nSpeaker E: I don't know what we have to decide about that.\nSpeaker E: Yeah.\nSpeaker D: Well, it's a good question.\nSpeaker D: It was in there and I didn't know any other...\nSpeaker C: Chip on print, I think what I mean with the regular rubber buttons that you got, it's always clear for the remote control when you press a certain button.\nSpeaker C: But when you got an LCD screen...\nSpeaker C: Yeah, okay.\nSpeaker C: With no...\nSpeaker C: With no...\nSpeaker C: The buttons are not always on the same place.\nSpeaker C: For example, if you enlarge a button, or if you got several options appearing on your screen, the co-ordinations aren't always the same.\nSpeaker C: When you got a regular button, the button of standby is always on the same place.\nSpeaker C: Yeah, but...\nSpeaker F: And you get on a print plate of LCD screen that's more advanced than a normal print plate.\nSpeaker F: Yeah.\nSpeaker F: So that's not of any discussion, I think.\nSpeaker D: No, you need some kind of CPU, I guess.\nSpeaker D: I suppose so.\nSpeaker C: I don't know.\nSpeaker F: I don't think...\nSpeaker C: It is a simple CPU, but it is...\nSpeaker F: Yeah, okay.\nSpeaker F: But I don't know if...\nSpeaker F: Yeah, because it has to...\nSpeaker F: Yeah, it's quite a simple LCD screen.\nSpeaker F: I think the remote control has to know where you're in the settings menu or just...\nSpeaker C: You just want to turn on the volume.\nSpeaker C: Yeah, okay, true.\nSpeaker F: True.\nSpeaker F: True.\nNone: True.\nSpeaker C: So it has to have some kind of...\nSpeaker C: Calculation unit or CPU.\nSpeaker C: To know in which state you are and which button you are pressing it at the right moment.\nSpeaker D: Because we're projecting the buttons on the LCD screen.\nSpeaker C: And the touchscreen makes it possible to get coordination of your finger on the screen.\nSpeaker C: In the case, yeah, we already discussed the case, so we wanted to make it from rubber.\nSpeaker A: But the hard rubber like this or software rubber?\nSpeaker A: The hard rubber, I think.\nSpeaker C: That's easy to add to that.\nSpeaker C: Yeah.\nSpeaker A: It bounces back from the floor when you...\nSpeaker A: We have different colors.\nSpeaker A: Yeah, this is...\nSpeaker F: So the shapes is something between square shapes with round corners and round shapes.\nSpeaker F: No, I don't think.\nSpeaker F: I think it's more routes than square.\nSpeaker A: Yeah, it's a bit...\nSpeaker A: So it meets...\nSpeaker A: It's a bit longer.\nSpeaker F: I think it meets more the young people than the older people.\nSpeaker C: Yeah, but that's what we wanted.\nSpeaker C: Yeah, that's how it started.\nSpeaker C: It's our main target.\nSpeaker F: Yeah.\nSpeaker F: Load it for the years, I think.\nSpeaker C: And how about my idea of making it with one single curve?\nSpeaker F: Oh, yeah, that's cool.\nSpeaker F: I think I thought that was quite good.\nSpeaker C: Because it's a gadget and you want to show it off, of course.\nSpeaker D: Yeah, yeah, you have to fancy design it right away.\nSpeaker F: You can put it on your table with LCD screen.\nSpeaker F: You don't have to get it in your head.\nSpeaker F: You can put it next to you and then...\nSpeaker F: Yeah, that's cool.\nSpeaker F: It's not easier.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker C: So you can write that down and get a single curved rubber.\nSpeaker C: Fancy colored.\nSpeaker C: You can't remote control.\nSpeaker F: You can't...\nSpeaker F: Oh, you can draw it on your paper and then load it on the...\nSpeaker A: Yeah, you can...\nSpeaker D: Yeah, but I don't understand.\nSpeaker E: And load it on the user, the server.\nSpeaker D: And about the user interface, there are also some questions.\nSpeaker D: But the concepts, I think, you have some ideas.\nSpeaker D: Yeah, I...\nSpeaker A: And then you showed your drawing.\nSpeaker A: I should go...\nSpeaker A: Go...\nSpeaker A: Go...\nNone: Go...\nNone: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nNone: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nNone: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker B: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker A: Go...\nSpeaker C: Can you hear me?\nSpeaker C: screen vertical it doesn't matter if you make it in a rectangle\nSpeaker A: if we make it like this so it says put it like this square\nSpeaker F: I think it's easier to have it something like but the minus here plus over here minus here plus over here and on here the other buttons and on here the top the options and then you have something like the P over here and the sounds\nSpeaker A: I think that's not what you're used to I think I'll put a plus in the middle here and then the P in the middle and the sound\nSpeaker F: something like you\nSpeaker C: take your time plus minus\nSpeaker E: plus plus almost\nSpeaker A: minus I think when you're holding it you could press the minus and the plus and with the other finger the minus and the plus\nSpeaker F: yeah I think you're going to you're going to use it with one thumb\nSpeaker A: but I think perhaps I have some examples in the room\nSpeaker C: we'll need that to be used to build the engineering\nSpeaker E: who's used to build the engineering?\nSpeaker A: that's me I've got some examples because perhaps you should choose what's most often used because we can use that better\nSpeaker F: consistency I have those numbers\nSpeaker E: I don't really know\nSpeaker D: everybody's searching in it\nSpeaker F: channel selection per hour, 168 volume selection four times an hour\nSpeaker A: but I mean if it's usually plus or above each other or next to each other I don't know how much\nSpeaker C: much I often use but\nSpeaker A: yeah that depends on the remote here's it's below here also here's next to each other I think it's\nSpeaker F: I think I think because I have two televisions in the home one is horizontal, one is vertical so it doesn't really matter yeah it depends\nSpeaker B: I think the volume is usually\nNone: above each other because you go higher and lower and the the program is next to each other because you go further and back that's how it's usually okay let's go to chase\nSpeaker C: okay let's go to chase yeah well I think we don't have to decide about that now\nSpeaker D: where we will put the button it is user interface\nSpeaker F: yeah it's complicated\nSpeaker A: there's in the actual design you should know where you play yeah interface shall we implement that?\nSpeaker D: yeah well we just heard about the new I think it's easy for it\nSpeaker F: where are you that this is?\nSpeaker A: I'm here but then you should also find a place if you do it like this you could put it in a corner or something you can talk into the corner yeah microphone\nSpeaker C: not even necessary because a lot of you can make a very sensitive microphone to make it possible to just put it around underneath it or on the bottom of it yeah\nSpeaker D: maybe at the bottom where you can can hold it with your hand there's also a microphone over there\nSpeaker C: I think that's done I'm very important just it doesn't matter where the microphone is\nSpeaker A: I would you should decide\nSpeaker C: okay I should okay\nSpeaker E: I think where it isn't seen the most\nSpeaker C: it shouldn't be\nNone: it's unsightly put it in a logo of the company between the\nSpeaker F: around the art because I think it should be in an\nSpeaker D: important position where people can see it because it is the unique idea of our remote the speech control I don't want to put it\nSpeaker C: I'm going to make it any more fancy because you can see it\nSpeaker A: actually it does because you can find it better but it's a way for you to so it makes more fancy because other remotes don't have that\nSpeaker D: okay but are we talking about the button or about the microphone?\nSpeaker D: microphone is very small thing but you can make it look like it's big so it's really an important function of the remote\nSpeaker F: I think the left\nSpeaker E: other corner should be the best\nSpeaker F: where it isn't the most is sight\nSpeaker C: okay I think I'll put it there I don't mind it doesn't really matter\nSpeaker E: actually does it\nSpeaker C: okay\nSpeaker A: so\nSpeaker D: well type supplements interface type yeah well the LCD\nSpeaker F: I think like you said like scroll next to the remote isn't that handy?\nSpeaker F: I think it's better to just type what you'd like to do on the screen if you want to go back if you want to go back if you want to choose audio settings you press audio settings and it goes to\nSpeaker C: that submenu\nNone: that's my name to the fan yeah the young people do like the scroll\nSpeaker C: so why not on side?\nSpeaker A: or at least I don't know if you just scroll but the menu aren't either there like most and I think you can navigate through a menu yeah I think\nSpeaker F: the settings faster I think the scroll is easier if you have a lot of options but if you don't have a lot of options because when you\nSpeaker C: when you use the settings menu for example to look up some some channels on your television you should scroll down on a menu which probably does not fit on your screen yeah okay so then it's very handy to scroll down okay you make just a rubber just like on your mouse or just put it on a side and it's\nSpeaker D: very easy to use and I think it would make it even more fancy because you have another interesting thing on the side\nSpeaker A: it's also different\nSpeaker D: yeah okay well there's a five minutes warning any more questions about about the design?\nSpeaker D: or do you need to face a\nSpeaker E: component? or color sword?\nSpeaker E: Clear?\nSpeaker C: okay what's the standard color?\nSpeaker C: standard? no we have a different color how many colors? you have different colors but you should have a standard color I think black, black, I think\nSpeaker F: with a yellow, darker or something like this color or something\nSpeaker C: just a regular, a large color and then have different covers I think it's better to have silver now\nSpeaker F: I think it's more silver\nSpeaker C: silver, silver, black, silver, black\nSpeaker F: silver, black, silver, black, silver, black, silver\nSpeaker D: it's sort of like our own pink colors\nSpeaker F: I think we have silver, black, and between those like, I don't know, five colors between them, silver, green, yellow\nSpeaker D: what about a yellow thing? I mean, it's better to have a yellow, you have?\nSpeaker C: definitely, fancy. red? yeah, that's right, but if you use silver, silver, silver, silver isn't fancy true? when you use titanium, silver is fancy, but when you use silver rubber, this is fancy silver has to shine and rubber doesn't shine especially when you just made a soft rubber like this you understand?\nSpeaker D: I would think about colors like red, yellow, blue, blue\nSpeaker F: of course, the fruity colors and the other colors like red and brown, I add dark red and brown\nSpeaker D: okay okay, so\nSpeaker A: do I have to have a normal black one or\nSpeaker D: who's singing? you're singing okay, so that wraps it up everybody knows what to do, well I know what to do, I wrote it down here, what to do the next meeting is once again in 30 minutes, here are the individual actions and especially notice that the interfaces on our are going to work together on a prototype drawing on a smart board and of course to all four of us instructions will be emailed by\nSpeaker A: personal coach. we'll work together, but do we have to stay here or do we?\nSpeaker D: well, I just wait for email and find out I don't know okay, well that was what I had to say final thoughts from anyone?\nSpeaker D: no, thank you very much\nSpeaker F: thank you very much\nSpeaker A: keep talking English now I'll make a new one for you\nNone: thanks and I'll face out bye okay, well I'll try to put minutes in the project folder as soon as possible okay hopefully okay good luck good luck this is a smart board so yeah, this is smart board so\nNone: okay\n"}, {"meeting_id": "TS3008d", "summary": "Firstly, User Interface and Industrial Designer introduced the prototype of remote control based on the previous discussion of its form, material and the colour. The prototype was a pretty simple design with an obvious company logo on it and two buttons in the middle. Secondly, Marketing designed an evaluation test under the guidance of the advantages of its main features discussed before and the team gave one to seven points to each feature of the product to see if the prototype would meet the original requirements and goals. Lastly, the team calculated the cost of the prototype and found that it was not easy to land on the exact budget. After trying to incorporate as many functions as the team could for a while, the team decided to use regular chips, normal batteries and a combination of cheap materials and fancy ones, which made the budget under control.", "dialogue": "None: or I\nNone: Okay. You questioner.\nNone: Oh, you questioner.\nNone: I mean, you questioner.\nNone: I mean, you know.\nNone: No.\nNone: Power by that.\nNone: They will know it.\nNone: It's a little sign on it.\nSpeaker D: Little round with a line.\nNone: Okay.\nSpeaker B: Is it nice?\nNone: Okay.\nSpeaker A: Welcome to the details design meeting.\nSpeaker A: Again, I'm going to take minutes.\nSpeaker A: Oh, we're going to have a prototype presentation first.\nSpeaker A: Who's going to give the prototype presentation?\nSpeaker A: You two guys?\nSpeaker A: Okay.\nSpeaker A: Go ahead.\nSpeaker B: I'm going to help you.\nNone: You have a common family.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: We've made a prototype.\nSpeaker D: We've got our aspects from the last meeting, especially we looked at the form material and the color.\nSpeaker D: We've drawn here the prototype.\nSpeaker D: The logo is pretty obvious to see on the remote control, but it is necessary.\nSpeaker D: When you want to build your company to a level higher.\nSpeaker D: Our interface elements are shown in the drawing.\nSpeaker D: Maybe you can point them.\nSpeaker B: Well, all functions are discussed.\nSpeaker B: I think the most of the functions are obvious.\nSpeaker B: It's a little bit power button.\nSpeaker B: Then the nine channels.\nSpeaker B: The volume at the side and the other side is the programs.\nSpeaker B: Then we have just two buttons.\nSpeaker B: We place them in the middle, the menu, and for the teletext.\nSpeaker A: No, the mute button.\nSpeaker D: The mute button is here.\nSpeaker C: You've got to point the out which is the volume button.\nSpeaker B: Most of them are right handed.\nSpeaker C: You've got to make it clear on that.\nSpeaker C: I don't have time at any point.\nSpeaker C: Next to that, I kind of miss the zero.\nSpeaker C: This was one thing I wanted to ask.\nSpeaker A: There are different ways for remote controls to do like, I call it teens and twenties.\nSpeaker A: Yes.\nSpeaker A: Two numbers.\nSpeaker C: It's a television.\nSpeaker A: It's a television.\nSpeaker A: Do we need extra buttons?\nSpeaker D: Yes, you have a lot of standard buttons.\nSpeaker D: With the one and the double cross.\nSpeaker A: You don't need them.\nSpeaker A: A lot of remote controls work when you push the one first.\nSpeaker A: I don't think so.\nSpeaker A: Yes, you have televisions.\nSpeaker A: I don't think so, really.\nSpeaker A: You have to have remote controls that don't have these buttons, but you still can select the number in the twenties.\nSpeaker C: Remote control nowadays, they come with the television or the other way around.\nSpeaker A: No, I really think it's because when you put a button on it with one and then a dash, it's the same thing as when you just push the one.\nSpeaker A: Because it gives you the functionality of that separate button.\nSpeaker A: You also have to apply.\nSpeaker D: Some televisions don't accept that.\nSpeaker A: Yes, because it's for television, it's exactly the same thing.\nSpeaker C: No, no, no.\nSpeaker C: Some television respond differently.\nSpeaker C: No, listen, listen.\nSpeaker A: When you push the button, the remote control gives a signal.\nSpeaker A: In the first place, it gives a signal which it would also send when you put the separate button on it.\nSpeaker A: That the one with dash, that signal gives, and when you don't push another button on the remote control within five seconds, then the remote control gives the signal for channel one.\nSpeaker A: No, I think it works that way.\nSpeaker C: No, it works.\nSpeaker C: If you haven't got a special button for it, if you push the one, then on your television, there will appear one and a line.\nSpeaker A: Yeah, but it's exactly the same name that would appear when you put the separate button.\nSpeaker D: When you put all the televisions, you have to click on a special button, then you go to a next level.\nSpeaker D: You can push two buttons.\nSpeaker A: Yeah, but you don't understand my point.\nSpeaker A: I think it's exactly the same thing.\nSpeaker D: Yes, but some televisions don't support it.\nSpeaker A: No, but then they will also support that button because it's the same thing.\nSpeaker A: Listen, that special button, the way you're talking about that, it's just a signal to the TV that they have to put a one on your screen.\nSpeaker A: And a dash, so you can still put another number on it.\nSpeaker A: When you don't have the separate button and you push the one, it's exactly the same thing.\nSpeaker A: The remote control gives that same signal as it would give when you only have to...\nSpeaker C: So, how can some televisions need the input first?\nSpeaker C: Yeah, but do you give the input?\nSpeaker A: No, they need the one.\nSpeaker A: That's the same thing as the button.\nSpeaker A: No, that's not true.\nSpeaker A: Yes, it is.\nSpeaker A: Think about it.\nSpeaker B: It's simply not true.\nSpeaker B: No, but when you push the one, you can show on the telly one and just...\nSpeaker B: And it's the same thing where it happens.\nSpeaker A: And the remote control gives another signal after five seconds, that is just one.\nSpeaker C: Now, the remote control doesn't give signal after five seconds.\nSpeaker C: The remote control is a stupid thing.\nSpeaker C: If you push a button, it sends it immediately to the telly one.\nSpeaker A: Yeah, but I know for sure that some televisions that the remote control supplies only as the buttons with the one and the dash and the two and the dash.\nSpeaker A: But when you use a remote control that doesn't support these buttons, it still works.\nSpeaker A: But okay, we'll...\nSpeaker A: Definitely not.\nSpeaker B: Definitely not.\nSpeaker B: And usability lab.\nSpeaker B: No, we'll apply them then for now.\nSpeaker B: I don't know.\nSpeaker B: I don't know if it's necessary.\nSpeaker A: Yeah, up next to the zero, the one and the two.\nSpeaker A: Yeah, I think so.\nSpeaker A: Yeah, for now, if we don't know for sure whether...\nSpeaker B: Yeah, okay.\nSpeaker D: And the button for the SCART audio video external input.\nSpeaker C: You can access that file server on the right side.\nSpeaker A: When I said about it, the remote control sending another signal, that might not be true.\nSpeaker A: But I still think the old TVs in some way support it.\nSpeaker A: I don't know.\nSpeaker A: No, no.\nSpeaker A: I think it's more...\nSpeaker A: Maybe we both don't really understand how it is.\nSpeaker A: I would really work.\nSpeaker A: But I think there's more to it than remote control sends one signal at one button.\nSpeaker D: I do think that...\nSpeaker A: Some...\nNone:...\nNone:...\nSpeaker A:...\nSpeaker A:...\nSpeaker D:...\nSpeaker D:...\nSpeaker D:...\nSpeaker A:...\nSpeaker A:...\nSpeaker D:...\nSpeaker D: Okay, I'll always see.\nSpeaker D: When you make the technology that it will give more signals, it could work.\nSpeaker D: It's just the basic idea of the most common and simple operations on the remote.\nSpeaker C: I kind of missed the docking station.\nSpeaker B: Yeah.\nSpeaker D: Yes.\nSpeaker D: It's here.\nSpeaker B: And there's nothing.\nSpeaker B: I think it's pretty basic.\nSpeaker B: There's nothing really funny about it.\nSpeaker B: There's one function and that's the one function.\nSpeaker D: But maybe we can make the docking station a bit standard for the other products we sell.\nSpeaker D: Because real reaction sells more products than only remote controls.\nSpeaker D: So maybe we can use the docking station for example, MPD players or hearing devices.\nSpeaker C: I think that's very difficult because of different shapes of devices.\nSpeaker D: Yes, but when you put the same photoises on it, you can put the lowest part of it.\nSpeaker D: When it's the same as the other products, you can put it all on the same...\nSpeaker C: Well, it has got to fit into the shape of course.\nSpeaker C: Yes, but we can make...\nSpeaker C: The technology and the voltage can be the same.\nSpeaker C: That's true.\nSpeaker C: But if you all make them having a bottom like this...\nSpeaker D: No, we can make the most lowest part all the same.\nSpeaker D: Yeah, that's true.\nSpeaker D: When the recharger has a bit that points out, we can place all on top of it.\nSpeaker D: Yes, but...\nSpeaker C: Which has to be big enough for the big...\nSpeaker C: If it's full then, it isn't going to fall down.\nSpeaker C: That's a bit...\nSpeaker D: No, when you make it large enough, it will not.\nSpeaker D: But then it's a little bit...\nSpeaker C: If, like this, I'll point it out.\nSpeaker C: If you've got...\nSpeaker C: But it's just an idea.\nSpeaker C: If it's like this, I won't draw it.\nSpeaker C: If you've got a base, which is...\nSpeaker C: It's as big as this.\nSpeaker D: It's as big as this.\nSpeaker D: So we can make all the products and flattens this.\nSpeaker C: But if you've got a tiny player, it can...\nSpeaker D: Yes, but when you make a bit of a big, low leaning...\nSpeaker B: It's leaning backwards, I think.\nSpeaker C: What you could do if you're from the bottom...\nSpeaker C: Oh, right.\nSpeaker C: But you could make it like a hole in it.\nSpeaker D: A little hole or...\nSpeaker D: A little project, go deeper in it.\nSpeaker A: Let's talk about the docking station later.\nSpeaker A: Maybe we have to consider the docking station anyway.\nSpeaker A: Because we have some...\nSpeaker A: Yes.\nSpeaker A: Cost issues.\nSpeaker A: It's still to come.\nSpeaker A: We have to look at that.\nSpeaker D: The look and feel would be great on this remote control.\nSpeaker D: I'm glad to call this.\nSpeaker D: You always will pick up the remote control in the smallest area.\nSpeaker D: Then your left thumb, your right thumb, is near the program button, which is the most common used function.\nSpeaker D: And all the other buttons are available for your thumb.\nSpeaker D: So it's a really good design.\nSpeaker D: Yes.\nSpeaker D: But that's it?\nSpeaker D: Yes, on the side, there will be a strip of rubber.\nSpeaker D: And in the middle, there's a hard material, a hard plastic, with light behind it.\nSpeaker D: Okay.\nSpeaker C: And other lights.\nSpeaker C: I think you have the lights.\nSpeaker D: Yes, you can make also neon lights on it.\nSpeaker C: Or the buttons that can make lights on it.\nSpeaker C: Maybe the logo.\nSpeaker D: Yes.\nSpeaker D: But it will also use batteries.\nSpeaker A: And we want to.\nSpeaker A: Okay.\nSpeaker A: For now, this is good enough.\nSpeaker C: Okay.\nSpeaker A: Yeah.\nSpeaker A: What was on the...\nSpeaker D: Did all the aspects of the interface buttons were...\nNone: Okay.\nSpeaker A: But the color, because we're going to use one color for the plastic, and closer one color for the rubber.\nSpeaker A: Yes.\nSpeaker A: And we're going to do the buttons in the...\nSpeaker A: Or we're going to have rubber buttons.\nSpeaker A: And there will be the same color as the rubber on the side.\nSpeaker A: Yeah.\nSpeaker A: I think that'll be good enough.\nSpeaker A: I think we should use a darker color for the plastic, and maybe some more brighter and flashier.\nSpeaker A: Yes.\nSpeaker D: Maybe we can use on the lights on the side, we can use multiple lights.\nSpeaker D: So...\nSpeaker D: It will...\nSpeaker A: Yeah.\nSpeaker A: We'll talk about the lights later.\nSpeaker A: Yes.\nSpeaker A: Because also, yeah, it depends on the costs and such.\nSpeaker A: Yeah.\nSpeaker A: But...\nSpeaker A: And we have to agree upon the exact colors, but maybe I don't know if that's important, but we'll talk about it later.\nSpeaker A: Okay.\nSpeaker A: For now, this is okay.\nSpeaker A: The next...\nSpeaker A: You're going to give a presentation too?\nSpeaker A: I have to see the...\nSpeaker A: Yeah.\nSpeaker C: I'm going to do something right there.\nSpeaker A: Detail design.\nSpeaker C: We've got to do that on the right.\nSpeaker C: On the right-most screen, because...\nSpeaker C: Evaluation criteria.\nSpeaker A: Yeah.\nSpeaker C: That's me.\nSpeaker C: Okay.\nSpeaker A: Okay.\nSpeaker C: All right.\nSpeaker C: I will be needing that image, so leave it, please.\nSpeaker C: Go away.\nSpeaker C: Right, we're going to evaluate that design.\nSpeaker C: According to a few points.\nSpeaker C: The four of us are going to do that together.\nSpeaker C: I want to have a color over here.\nSpeaker C: Come on.\nSpeaker C: Right.\nSpeaker C: The remote is not ugly, a bit weird sentence, but the positive thing has to be on the left, so I said not ugly instead of ugly.\nSpeaker C: What would you say?\nSpeaker C: We've got to give points to all of these to evaluate that design, and please forget the drawing skills of these guys.\nSpeaker C: Okay.\nSpeaker C: The remote controls are not ugly.\nSpeaker C: How do you feel?\nSpeaker A: I think four maybe would be appropriate because it really depends on taste.\nSpeaker A: I mean, it's kind of dairy art design.\nSpeaker A: Yes.\nSpeaker A: So maybe a lot of people find it really ugly, and other people find it really cool.\nSpeaker A: I don't know how you think.\nSpeaker A: How do you guys feel?\nSpeaker D: I think the front will give it a more less earlier side, because you can make it in your own...\nSpeaker D: Yes, you can make it in your own...\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: More to your own personality or house style.\nSpeaker A: We didn't...\nSpeaker A: We're not planning to use fronts, I believe.\nSpeaker A: Not fronts, but they're not designed to fit their thing.\nSpeaker C: And that's still...\nSpeaker C: Yeah, it's a little personal touch, I guess.\nSpeaker A: Maybe we should do three or something in our...\nSpeaker C: What would you guys think?\nSpeaker C: Personally.\nSpeaker B: Personally.\nSpeaker B: We can make it a one.\nSpeaker C: Yes, but what is it?\nSpeaker D: I think two or three...\nSpeaker D: No.\nSpeaker B: I agree.\nSpeaker B: Two or three?\nSpeaker B: I go for the politics, so I go for two.\nSpeaker C: I was thinking about three, so I...\nSpeaker A: I was thinking about four, so I think three is...\nSpeaker C: Three is a bit...\nSpeaker C: What am I doing?\nSpeaker C: I'll mark it.\nSpeaker C: Remote controls and then makes stepping easy.\nSpeaker C: Yeah.\nSpeaker A: Let's make it a one.\nSpeaker A: One.\nSpeaker B: One, you agree?\nSpeaker B: Okay, I agree.\nSpeaker B: That's one thing for sure.\nSpeaker B: I'm not having usability, so...\nSpeaker C: I totally agree.\nSpeaker C: Remote controls, the remote controls, relevant buttons, are prominently visible.\nSpeaker C: Yes.\nSpeaker C: Yeah.\nSpeaker A: Two are aware, I guess.\nSpeaker A: It's something we really work into.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker A: It's all about buttons.\nSpeaker C: Every button is relevant.\nSpeaker C: And they're all...\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: All right.\nSpeaker C: That's one.\nSpeaker E: You agree?\nSpeaker C: Yeah.\nSpeaker C: Remote control hasn't got too much redundant or unneeded buttons, I think.\nSpeaker C: We totally succeeded there.\nSpeaker A: Well, maybe two because of the menu part or something.\nSpeaker D: Yeah, the buttons of the one, the two, the digits.\nSpeaker D: Yeah, we don't know if the data...\nSpeaker C: That's true, that's true.\nSpeaker C: They're definitely needed.\nSpeaker A: It can also always be more simplistic.\nSpeaker C: You put it on a tool?\nSpeaker C: Yes.\nSpeaker C: The remote control has got a really trendy look.\nSpeaker D: Yes.\nSpeaker D: One.\nNone: Martin.\nSpeaker A: Well...\nSpeaker A: I think two.\nSpeaker A: Yeah, yeah, yeah.\nSpeaker A: It's hard to say from a speech.\nSpeaker D: Yeah.\nSpeaker D: We've tried to make it the best training look ever.\nSpeaker D: Ever.\nSpeaker C: Yeah.\nNone: You know.\nSpeaker A: But I do think it's more...\nSpeaker B: I don't make it a tree because...\nSpeaker B: Yeah.\nSpeaker A: I think that it's more trendy than beautiful.\nSpeaker C: Yeah, I agree.\nSpeaker A: So I think maybe it has to score higher on this than on the...\nSpeaker C: I was planning to give it a two where I give the not ugly...\nSpeaker C: The three.\nSpeaker C: Oh, yeah.\nSpeaker C: That's true.\nSpeaker C: You agree or not too?\nSpeaker A: Yeah, when you compare it to the...\nSpeaker C: Great.\nSpeaker C: Remote control hasn't got too much redundant or unneeded buttons.\nSpeaker A: What's the difference with...\nSpeaker A: I copied that one.\nSpeaker C: Well, forget that.\nSpeaker C: Okay.\nSpeaker C: Um...\nSpeaker C: No way.\nSpeaker C: Remote control has got innovative technology implanted.\nSpeaker C: No.\nSpeaker B: No.\nSpeaker B: We're not.\nSpeaker B: Well, maybe the LCD.\nSpeaker B: LCD.\nSpeaker A: Yeah, but we...\nSpeaker A: You mean the rubber stuff?\nSpeaker C: Yeah, and the lights.\nSpeaker C: Yeah, but you have to...\nSpeaker A: That's not innovative.\nSpeaker A: That's not innovative.\nSpeaker A: And I don't know if it's also really not innovative.\nSpeaker A: Light's up.\nSpeaker C: It's not...\nSpeaker A: Seven?\nSpeaker A: No, six.\nSpeaker A: Oh, seven maybe.\nSpeaker A: No, six.\nSpeaker C: Why not seven?\nSpeaker A: Yeah, why not seven?\nSpeaker D: Because we've tried to make it a little bit innovative.\nSpeaker D: Yeah.\nSpeaker A: But how?\nSpeaker A: It depends on the...\nSpeaker D: With the lights, it's kind of future.\nSpeaker A: No, I think...\nSpeaker A: I think actually it's seven maybe, but there's nothing innovative about it.\nSpeaker C: It's true.\nSpeaker C: I agree.\nSpeaker D: But still, you can retrieve it when it's gone.\nSpeaker D: Innovate it in general.\nNone: Yeah, you didn't draw a But you said...\nSpeaker C: No, no.\nSpeaker C: The docking station is...\nSpeaker A: I mean, the docking station, but I think...\nSpeaker A: It's a part of the remote docking station.\nSpeaker A: It's going to be a kind of a problem.\nSpeaker A: Okay.\nSpeaker A: And when the speaker...\nSpeaker A: Well, let's leave it open for us later to see.\nSpeaker A: But because we have to reevaluate anyway.\nSpeaker A: Well, yeah.\nSpeaker C: Well, the agenda says evaluate now, so I think...\nSpeaker A: Okay, for now it's a six or a seven, maybe...\nSpeaker A: But retrieval or...\nSpeaker A: Yeah, but I don't know if it's very...\nSpeaker C: I think that's very innovative for our control.\nSpeaker D: How would you innovate a remote control more?\nSpeaker A: Yeah, more through like TV functionality.\nSpeaker A: To put it on your head.\nSpeaker A: No, no, no, no.\nSpeaker A: You know it when I...\nSpeaker A: You must be innovative technology for remote controls, but more...\nSpeaker A: Yes, sir.\nSpeaker A: In how you control stuff, not in how you find your...\nSpeaker A: Yeah.\nSpeaker A: I definitely don't think it's a five.\nSpeaker A: Let's think about it later.\nSpeaker C: The remote control is easy to use.\nSpeaker A: Yeah, as a one or two, at least.\nSpeaker C: I think a two.\nSpeaker A: Yeah.\nSpeaker A: I think a two.\nSpeaker C: More on.\nSpeaker C: The remote control hasn't got...\nSpeaker A: No, I want to see that one before.\nSpeaker A: I just get one...\nSpeaker C: I've just filled...\nSpeaker C: Yeah.\nSpeaker C: That way.\nSpeaker D: You like the buttons?\nSpeaker C: I found 12 questions so much, but it's still a stop.\nSpeaker A: No control will be bought by...\nSpeaker C: It will be bought by people under the age of 14.\nSpeaker A: Yeah, definitely.\nSpeaker D: In comparing with people of...\nSpeaker D: No, no, no, no.\nSpeaker C: Just in general.\nSpeaker C: If they buy it.\nSpeaker A: Yeah, two.\nSpeaker D: We don't know.\nSpeaker B: Yeah, what do you think?\nSpeaker B: Yeah.\nSpeaker B: I think two.\nSpeaker D: I agree.\nSpeaker D: Contact two.\nSpeaker D: Yes, two, but only when you compare it with elderly.\nSpeaker A: There's no question.\nSpeaker A: It's just...\nSpeaker A: It will be bought by people under 40.\nSpeaker A: Yeah, it can be very...\nSpeaker A: And I don't mean to be...\nSpeaker A: Just guessing.\nSpeaker A: I'll just make it a two.\nSpeaker D: Make it a two.\nSpeaker D: When it succeeds, it can get a two...\nSpeaker D: Right.\nSpeaker C: The remote control has recognizable corporate image color, logo or slogan.\nSpeaker C: Yeah, thank you.\nSpeaker A: Slogan is quite obvious.\nSpeaker A: No, the slogan.\nSpeaker C: Oh, sorry.\nSpeaker A: No, no, the slogan.\nSpeaker A: You can put that on the side if we would like to.\nSpeaker D: I'm a mess.\nSpeaker D: Encrypt it with...\nSpeaker D: Yeah, and I will...\nSpeaker A: I'm gonna do that.\nSpeaker A: I still think it's gonna be a two or three, maybe three this time.\nSpeaker C: Yes, three.\nSpeaker C: I agree.\nSpeaker C: Because of the slogan in the...\nSpeaker C: Remote control has got a basic design intended for no...\nSpeaker C: So one or two?\nSpeaker C: Yeah, two.\nSpeaker C: Yes.\nSpeaker C: Two?\nSpeaker A: Yeah, make it a two.\nSpeaker C: Two.\nNone: All right.\nSpeaker C: We've got to add up the scores now to see our total average four, five, seven, nine.\nSpeaker C: Look at that.\nSpeaker C: Fifteen, seventeen, twenty-one, twenty-four, twenty-six.\nSpeaker C: Twenty-six.\nSpeaker C: It's a two point six.\nSpeaker C: All right.\nSpeaker A: Yeah.\nSpeaker A: Yeah, that's mostly the...\nSpeaker A: When we score higher on innovative technology, we would score it too.\nSpeaker A: Yeah.\nSpeaker A: Which is quite a great score.\nSpeaker A: Okay.\nSpeaker A: This was the evaluation.\nSpeaker C: This was my evaluation.\nSpeaker A: Because I still think that the most important part of this meeting is...\nSpeaker C: The pretty nice job until now.\nSpeaker C: Yeah.\nSpeaker C: Is this your...\nSpeaker C: Is there something after this meeting?\nSpeaker C: Well, I think we've got to fill out another question now.\nSpeaker B: Okay.\nNone: Okay.\nSpeaker A: So open door.\nSpeaker C: Yeah.\nSpeaker A: Okay. Finance.\nSpeaker A: Because I received a spreadsheet.\nSpeaker A: Yeah, but I actually don't need this presentation, I guess.\nSpeaker C: That's not a matter.\nSpeaker A: I'm going to open the spreadsheet and we're going to work this out together because I didn't really finish it.\nSpeaker A: I...\nSpeaker A: Didn't really finish it.\nSpeaker A: But we will see we'll stumble upon some problems.\nSpeaker A: Probably won't look for any...\nSpeaker A: Probably have already opened it here.\nSpeaker A: I'll try it again.\nSpeaker A: First of all, the...\nSpeaker A: All the docking station and...\nSpeaker A: Costs and such are not included in this list.\nSpeaker A: But let's start with the meeting.\nSpeaker A: Let's conclude one battery.\nSpeaker A: I'll explain it.\nSpeaker A: The components are listed over here.\nSpeaker A: Price is given.\nSpeaker A: We indicate the amount of components, specific component, how much we need over them.\nSpeaker A: And then...\nSpeaker A: Don't watch the number yet.\nSpeaker A: I don't know if it's filled in properly.\nSpeaker A: Okay. We need one battery and one battery.\nSpeaker A: I think we're going to use enough.\nSpeaker A: We don't need kinetic, solar, solar, hand-dun, ammo.\nSpeaker A: Okay. This is the first problem.\nSpeaker A: I think we should know how many simple chips, regular chips.\nSpeaker D: It's one chip, but you have to choose one format.\nSpeaker A: Okay.\nSpeaker D: But simple chip is enough, I think.\nSpeaker D: With the lights and the retrieval, can we find this information?\nSpeaker C: I haven't got an idea in which we need to use.\nSpeaker A: I think it was your job in the first meeting, for your first presentation to make this clear.\nSpeaker A: But then you had some time problems.\nSpeaker A: Do you know what chip we need?\nSpeaker D: The email I got said, simple chip.\nSpeaker D: But when we put in the speaker and the retrieval device, it will cost a bit more like, I think, the advanced chip.\nSpeaker D: Maybe.\nSpeaker C: And how do you know?\nSpeaker C: I mean, you got that email.\nSpeaker A: Maybe you can look it up right now.\nSpeaker D: They didn't know about a retriever or a speaker.\nSpeaker A: Okay. When we leave the retriever in such a site, then it's a simple chip.\nSpeaker A: Then it would be a simple chip.\nSpeaker A: And with the retriever, it would be an advanced chip.\nSpeaker A: Yes.\nSpeaker C: All right.\nSpeaker C: I point out the advanced chip for now, I guess.\nSpeaker C: Okay.\nSpeaker D: That would be enough for future recommendations.\nSpeaker A: It would be because of a lot of problems.\nSpeaker A: The sample sensor, sample speaker.\nSpeaker A: Is that the speaker we were to?\nSpeaker A: I don't know what it is.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: Okay. We went for the double-curved case.\nSpeaker A: Made out of plastic and rubber.\nSpeaker A: And with a special collar.\nSpeaker A: I guess that's what we were.\nSpeaker B: Well, special collar.\nSpeaker A: Yeah, I don't know about the special collar.\nSpeaker D: Otherwise you can't think of standard plastic collar.\nSpeaker A: Yeah, I think we have a special collar.\nSpeaker C: We have a standard rubber.\nSpeaker C: All right.\nSpeaker A: Okay. Then the push button.\nSpeaker A: I was just counting them.\nSpeaker A: I think you have to indicate the amount of push buttons we want to use, isn't it?\nSpeaker A: Oh.\nSpeaker A: Yes.\nSpeaker A: Well, that's a bit of a problem because...\nSpeaker A: But I really don't understand that because I can imagine a remote control with far more push buttons, and it wouldn't be possible according to this sheet.\nSpeaker B: The same post buttons.\nSpeaker D: No, it's only when you use push buttons, it will cost that much.\nSpeaker D: I don't think so because it's a scroll wheel.\nSpeaker C: If you use a scroll wheel, the yellow row is the amount of...\nSpeaker A: So in the number of components you plan to use in the files and it tells the cost, I don't know.\nSpeaker D: Maybe it's the kind of push buttons.\nSpeaker D: You can have four kind of push buttons.\nSpeaker B: One, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, and thirteen.\nSpeaker A: This is one.\nSpeaker A: Okay. 12. Okay. Then it will be 18.\nSpeaker A: Because I rated them as...\nSpeaker A: Oh, for four months.\nSpeaker A: Yes.\nSpeaker A: I think that... Plus the mute button, and it will be 18.\nSpeaker A: 18.\nSpeaker C: One, two, three, four, five.\nSpeaker D: Why is that so expensive?\nSpeaker A: Yeah, I don't understand.\nSpeaker A: I don't get the point because it would be relatively so expensive, just small buttons.\nSpeaker B: Is it sense? The 50 cents? 50 cents for one single stupid button.\nSpeaker C: So where are you...\nSpeaker D: Okay, let's make it just one.\nSpeaker D: It's 80% of the price of the...\nSpeaker A: Yeah, now it's already...\nSpeaker A: Yeah, exactly.\nSpeaker A: Shall we just give our own interpretation to it?\nSpeaker A: Because else we would really have a problem.\nSpeaker A: It would be impossible to make it...\nSpeaker C: I couldn't understand it if it was 50 cents per button.\nSpeaker D: When you have the same amount of button, you have to put it in a carton.\nSpeaker D: And less buttons than this.\nSpeaker A: Isn't possible.\nSpeaker A: This is the most simple... Yeah, it is possible, but I've never seen one before.\nSpeaker D: I've seen one remote control.\nSpeaker D: Yeah, without the numbers.\nSpeaker D: Yeah, only with the page up, page down.\nSpeaker D: And all of you...\nSpeaker A: Keep the numbers.\nSpeaker C: But I wouldn't want to own that.\nSpeaker C: That's still four.\nSpeaker C: It's still for little children.\nSpeaker D: They can handle that remote control.\nSpeaker A: Yeah, then the text would also be impossible.\nSpeaker D: Yes, it's just for a little option.\nSpeaker A: Okay, well, just...\nSpeaker A: Okay, but then still, when we...\nSpeaker A: There is no room for a docking station or something.\nSpeaker A: Let's see.\nSpeaker A: We have a button supplement.\nSpeaker A: We'll give the buttons a special color.\nSpeaker A: We'll give them a special form.\nSpeaker A: I think we should mark the special form thing because this will be some special form.\nSpeaker A: So, incorporated in these big buttons, I guess.\nSpeaker B: Special color, why is special color?\nSpeaker A: Because the buttons will be matching color between the buttons and the rubber surroundings.\nSpeaker A: I think that's what they mean by a special color.\nSpeaker A: I think all the special color things have to be marked over here because that's what we were planning to do, making it...\nSpeaker D: Special form also.\nSpeaker A: Yeah, special material also because there's rubber and the buttons have to be rubber.\nSpeaker D: What is the normal material?\nSpeaker C: Yeah, I don't know.\nSpeaker C: Classic, I think.\nSpeaker B: Classic.\nSpeaker D: Classic.\nSpeaker A: Okay, but the problem now is that...\nSpeaker A: There's no such thing as a docking station in this list, but we can all imagine that it would be impossible to make a docking station for 30 cents.\nSpeaker D: But we can sell the remote control and sell the docking station separately.\nSpeaker D: Yes, and we don't have to tell it, but we can say we can almost make it impossible to buy a remote control without a docking station.\nSpeaker A: Yeah, I do like the idea, but the docking station isn't relevant for this project anymore.\nSpeaker A: No, but then we have to find out what chip we need.\nSpeaker C: Yeah, I really don't get it. I mean, if it's a simple chip, I think we can all be honest.\nSpeaker A: I think the special color thing has to be marked because I think that's what they...\nSpeaker D: But for two euros and 30 cents, we don't get a docking station.\nSpeaker A: I don't know. Can we find out about the chips? Because when we don't need a docking station, then probably we also need a simple chip.\nSpeaker B: And then we can get...\nSpeaker A: Maybe then we can do something extra.\nSpeaker A: But it's still always going to get more expensive with the Simona.\nSpeaker A: Then we have some money left. We can put...\nSpeaker A: Then for two euros, we can put a scroll wheel on it.\nSpeaker A: Yeah, we know.\nSpeaker C: Ah, why?\nSpeaker C: I mean, if it would cost two euros, the kind of total thing, it would be nice, too.\nSpeaker C: I mean, we're not going to add...\nSpeaker C: But what can we do?\nSpeaker A: How are we going to do the kinetic cells? That's also maybe an idea.\nSpeaker B: But what can we do with the simple chip? What's the difference with the regular chip?\nSpeaker A: Yeah, that's what we can find out. Maybe we can find it in your email right now.\nSpeaker A: And we know exactly what it will cost us. Maybe it's nice to know.\nSpeaker B: Regular chip. Because we don't have special functions to use.\nSpeaker B: Yeah, but when we...\nSpeaker C: We have a farm chip, for example, I'm like the hand in ammo.\nSpeaker A: But when we skip...\nSpeaker A: When we don't use the...\nSpeaker A: I'm going to make the docking station.\nSpeaker A: Then we still...\nSpeaker A: We need something else maybe to make it kind of special, because that was our special future.\nSpeaker C: We can make a plane docking station for two euros.\nSpeaker C: We'll go back tomorrow.\nSpeaker A: Yeah, okay. You can also do the...\nSpeaker A: Without the recharge...\nSpeaker A: It still is a special remote control, you know.\nSpeaker A: It's for a special one.\nSpeaker C: But we can make a docking station for two euros if you don't put the recharge function in it.\nSpeaker C: I mean...\nSpeaker C: It has a shape.\nSpeaker C: Yeah, but for two euros.\nSpeaker A: And we have to use the advanced chip.\nSpeaker A: And two euros. Is it even possible?\nSpeaker C: Why should that not be possible?\nSpeaker A: Because then we...\nSpeaker A: Three cents left.\nSpeaker C: Now, for the docking station, if you choose the simple chip.\nSpeaker A: Yeah, but...\nSpeaker A: Yeah, I don't know, because maybe...\nSpeaker A: Yeah, we'll find out what the same thing is.\nSpeaker B: That's the question.\nSpeaker B: If we do need an advanced chip.\nSpeaker B: Yes, and we...\nSpeaker A: And we listen to my world's vision.\nSpeaker A: Sample, sample, sample, sample, sample, sample.\nSpeaker D: I don't know either.\nSpeaker D: It isn't in my information.\nSpeaker D: I've got a...\nSpeaker D: A schematic view of the remote control, but nothing about advanced chips or...\nSpeaker A: Let's take a look at it for presentation.\nSpeaker A: Technical functions?\nSpeaker B: No.\nSpeaker B: I've got here on my...\nSpeaker B: I will put a...\nSpeaker D: I will put a page on it.\nSpeaker D: When my mouse works again.\nSpeaker E: Oh, oh, oh.\nSpeaker E: Hey.\nSpeaker D: My mouse is...\nSpeaker C: Dead.\nSpeaker D: Yes.\nSpeaker C: Reanimated.\nSpeaker C: Oh, died.\nSpeaker D: Ah, I've got it.\nSpeaker D: I will put my email on the...\nSpeaker D: On the network.\nNone: What the hell are these?\nSpeaker C: It's on it.\nSpeaker B: Oh, what a...\nNone: Yeah, it's open.\nSpeaker A: I don't think it's in here already.\nSpeaker B: There's only just...\nSpeaker B: There's nothing about...\nSpeaker A: In the presentation of yours, there was also something about different components.\nSpeaker A: Which one was it?\nSpeaker A: Component.\nSpeaker A: Fictional requirements?\nSpeaker C: No, that was my personal...\nSpeaker B: Component design, maybe.\nSpeaker A: Yes, that was mine.\nSpeaker A: Yes, that was the second.\nSpeaker D: Yeah, that was my second.\nSpeaker A: Your first presentation.\nSpeaker C: It's at the bottom.\nSpeaker C: It's a taskbar.\nSpeaker B: Yeah, that's the other one.\nSpeaker B: Oh.\nSpeaker B: This is not the right one.\nSpeaker A: I don't...\nSpeaker A: Okay, sorry.\nSpeaker A: This is the other one.\nSpeaker A: Maybe something about...\nSpeaker A: In these chips.\nSpeaker D: This is the same...\nSpeaker D: This is only the possibilities.\nSpeaker B: Yeah.\nSpeaker D: Yeah, we can use a simple regular or advanced chip.\nSpeaker D: Yeah, nice.\nSpeaker C: That's the same.\nSpeaker A: You know that a push button requires a simple chip, but a scroll wheel requires...\nSpeaker A: Oh, okay.\nSpeaker A:...a ring tube is not open.\nSpeaker A: Okay, so we only need a simple chip.\nSpeaker D: With a light.\nSpeaker B: A little light.\nSpeaker B: Yeah, but that's just the same as the...\nSpeaker A: No, no, this is a simple chip.\nSpeaker C: That's that neat.\nSpeaker A: The scroll wheel...\nSpeaker A: This only states that the scroll wheel requires a regular chip, and that a display requires an advanced chip.\nSpeaker A: So we don't need anything.\nSpeaker C: This display is, of course, for some...\nSpeaker C: For some text.\nSpeaker C: Yeah.\nSpeaker C: I don't think that...\nSpeaker C: No.\nSpeaker C: Just a little light.\nSpeaker B: I think a normal simple chip will be okay.\nSpeaker B: Yeah, I agree.\nSpeaker A: And what's the sample sensor slash sample speaker?\nSpeaker D: Maybe you can say against the remote page.\nSpeaker D: Yeah.\nSpeaker D: Page up, page run.\nSpeaker C: Yeah, I guess so too.\nSpeaker B: Yeah.\nSpeaker B: That's not what we want.\nSpeaker C: No.\nSpeaker C: Well, we might want it, but...\nSpeaker A: Okay.\nSpeaker D: All in 12 euros.\nSpeaker A: In the back two costs.\nSpeaker D: Yeah.\nSpeaker D: 12 euros and 50 cents.\nSpeaker A: So we're going to use a simple chip.\nSpeaker D: Great.\nSpeaker C: Delete.\nSpeaker C: Yeah.\nSpeaker D: In the light.\nSpeaker D: Where are the lights?\nSpeaker D: Yeah, the lights.\nSpeaker B: There's no...\nSpeaker C: Oh, they're free, I guess.\nSpeaker C: Category.\nSpeaker A: No, there is someone left to be spent.\nSpeaker A: Can we do it?\nSpeaker B: I think we can make a darker station.\nSpeaker A: Okay, but what we have to think about now is that...\nSpeaker A: Is this still a special remote control?\nSpeaker A: I mean, it hasn't got any innovative technology.\nSpeaker A: We aren't going to apply any innovative technology anyway, I think.\nSpeaker A: I don't see any possibility to do so because it wouldn't fit our design philosophy.\nSpeaker A: Is there some extra...\nSpeaker A: Maybe, I think, maybe the kinetic thing is something instead of the rechargeable.\nSpeaker A: The rechargeable thing was something to...\nSpeaker A: So people wouldn't have to worry about their batteries anymore.\nSpeaker A: Maybe if we put the kinetic thing in it.\nSpeaker A: But sometimes you put it...\nSpeaker A: Yeah, you leave...\nSpeaker A: Yeah, I know, but still, they will think about that.\nSpeaker A: I mean, if you...\nSpeaker D: Kinetics are nowadays only used in watches.\nSpeaker A: It's made for people, but they don't...\nSpeaker A: If it was useless technology, they wouldn't put it as a possibility.\nSpeaker A: A solar sensor used for the hand of the knife.\nSpeaker A: The target group are people who zap regularly and throw with their remote control.\nSpeaker A: Because I think when there was a remote control where it was useful to have a kinetic power source, then it would be this one because it's one that gets thrown around a lot and gets used a lot.\nSpeaker A: Maybe that's a cool thing about it, you know?\nSpeaker A: You don't use batteries.\nSpeaker A: I've never seen it before in a remote control.\nSpeaker B: Then we can make a docking station five minutes.\nSpeaker A: No, we can make a docking station anyway.\nSpeaker C: That's not true.\nSpeaker A: Yeah, we can also...\nSpeaker A: We can make a docking station.\nSpeaker A: We can still make...\nSpeaker B: Look at now, we got two...\nSpeaker B: Two-thirty left.\nSpeaker B: We can make a docking station.\nSpeaker D: A docking station like that?\nSpeaker D: With a cable, with buttons on it, with retrieval, a device in it.\nSpeaker C: The power device is very cheap.\nSpeaker C: That's just a regular power cable.\nSpeaker A: It would be serious.\nSpeaker A: The docking station would be...\nSpeaker A: With a button.\nSpeaker D: With a button to retrieve it, so it will beep.\nSpeaker D: So it's wireless technology.\nSpeaker A: We haven't looked at the discosts of the speaker and other stuff.\nSpeaker A: I don't think it's realistically to do so.\nSpeaker C: Well, then it's useless project.\nNone: Look at the case.\nSpeaker D: The case of...\nSpeaker B: Then we don't have...\nSpeaker A: Look at all these special stuff.\nSpeaker A: The colors are special, the form is special.\nSpeaker A: This is a whole concept.\nSpeaker A: Maybe with the kinetic thing.\nSpeaker A: I think we could do a compromise with the kinetic thing.\nSpeaker C: Can we say 15 euros?\nSpeaker C: No, it's the...\nSpeaker D: No.\nSpeaker D: Then we have to sell it for 30 euros.\nSpeaker C: No.\nSpeaker C: We only make less profit over it.\nSpeaker C: You can sell for 27.5.\nSpeaker C: Then you make as much profit as you would with 12.5 production costs.\nSpeaker A: My suggestion is just forget about the whole docking station thing and make it like...\nSpeaker A: I also feel this concept of making it kinetic.\nSpeaker A: Because of the...\nSpeaker D: I don't think we can do it with both.\nSpeaker D: Maybe we can do it both in the remote.\nSpeaker D: Battery and kinetic.\nSpeaker D: No.\nSpeaker A: No.\nSpeaker A: That wouldn't...\nSpeaker C: No.\nNone: Thirteen, twenty.\nSpeaker A: Yeah, and it's also not a good...\nSpeaker A: It's not a good...\nSpeaker A: It's really only kinetic.\nSpeaker A: You don't want to think about batteries anymore.\nSpeaker A: Yes, but when it's kinetic...\nSpeaker D: No, you have to shake it.\nSpeaker D: And when it's empty.\nSpeaker A: It's great.\nSpeaker C: You have to throw it through the room like 20 times an hour.\nSpeaker A: No, no, no.\nSpeaker A: This is very sophisticated technology.\nSpeaker A: When you use it for...\nSpeaker A: You're remote like once today or maybe even less.\nSpeaker D: No, that's not true.\nSpeaker D: A watch is kinetic.\nSpeaker D: Because you walk all the time.\nSpeaker A: We can make it...\nSpeaker A: No.\nSpeaker A: Be able to remote control gets...\nSpeaker A: Why do they state that this technology can be used to be...\nSpeaker D: Yeah, solar cells are also stated.\nSpeaker A: Yeah, but a calculator also works on solar cells.\nSpeaker D: Yeah, you solar cells.\nSpeaker A: Because I think the whole dynamic part, you know, appeals to me thinking of our design philosophy, you know, with the rubber parts and sturdiness of the thing.\nSpeaker A: And when you move it around a lot, then people find the idea funny than when I move my remote control around.\nSpeaker C: That's true.\nSpeaker C: It's funny for a week.\nSpeaker C: I guess something like that where you have to move it around very frequently.\nSpeaker C: It's the most...\nSpeaker A: Yeah, you don't have to.\nSpeaker A: Trust me, the idea of this technology is that you don't think about it.\nSpeaker A: It just happens.\nSpeaker C: No, I don't move.\nSpeaker C: My remote control...\nSpeaker A: Okay, we don't have to do it, but we just have a lack of key futures.\nSpeaker A: You know, you have to put something on your box to make people buy it.\nSpeaker A: And we can really do the docking thing.\nSpeaker A: It's not...\nSpeaker A: Yeah, we can do it, but it would be an easy way out.\nSpeaker D: You can do it for 50 cents.\nSpeaker C: We've got more than 50 cents.\nSpeaker A: Okay, but we have to do...\nSpeaker A: The case alone is...\nSpeaker D: Because we only have a minute left or so.\nSpeaker D: The case alone for a remote control is at least one euro.\nSpeaker D: Then we have one euro 30 for the whole documentation.\nSpeaker A: Okay, we can leave it on bed.\nSpeaker A: And then our concept is ready.\nSpeaker B: Keep remote control.\nSpeaker A: Yeah, we...\nSpeaker A: Make some extra profit on it.\nSpeaker C: No, we won't.\nSpeaker C: Any ideas?\nSpeaker C: Of course not.\nSpeaker D: It's great.\nSpeaker B: We just have to go all what we did today again.\nSpeaker B: We have to do it over.\nSpeaker C: We'll come back tomorrow.\nSpeaker A: There's still some concept and some special left.\nSpeaker A: It's going to excel in the field of design and the looks and feel.\nSpeaker A: That's what makes it special.\nSpeaker A: Yeah, and I would have liked the kinetic part as well to give it some extra special future.\nSpeaker A: I know it will work.\nSpeaker A: They're not putting technologies on this list if it was impossible to make it happen.\nSpeaker D: Why not a hand dynamic then?\nSpeaker A: Okay, but we leave like this.\nSpeaker A: Then it's nowhere.\nSpeaker C: We can do anything else.\nSpeaker C: Warning.\nSpeaker C: Finish meeting now.\nSpeaker D: We're done.\nSpeaker A: Okay.\nSpeaker A: Is this it?\nSpeaker A: Sure.\nSpeaker A: We're going to do the last cheat.\nSpeaker B: We have to do this.\nSpeaker C: Yes, yes.\nSpeaker C: Celebration.\nSpeaker C: I don't see why, but where's the champagne?\nSpeaker C: I think we've got to fill out another question to be honest.\nSpeaker D: I don't hear a bell.\nSpeaker C: No, not yet.\nSpeaker E: All right.\nSpeaker C: We can do it here.\nSpeaker C: I'll see you guys in a minute.\nSpeaker C: Bye.\nSpeaker C: I don't think so.\nSpeaker B: Can we do it here?\nSpeaker C: I don't know.\nSpeaker C: I don't think so.\nSpeaker D: Just fill that one in.\nSpeaker B: Yeah, we're doing now.\nSpeaker B: Oh, okay.\nSpeaker B: Nice.\nSpeaker B: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "IS1001c", "summary": "The meeting focused on the cost and design implication of including the wheel, the best way to include the light, ensuring that buttons are well-placed, and choosing a fashionable fruit as the remote's design. Project Manager reiterated that several decisions, like not having an LCD, having a wheel, and including a lighting feature, have been made in the previous meeting. Then, Industrial Designer highlighted that the wheel was expensive, though, likely still within the budget. User Interface led some discussion about placing the buttons in a way that minimizes strain for the user. Finally, Marketing informed the group that fruits and vegetables were the upcoming fashion trend. The group decided that a banana would be a good option for the design and discussed how existing priorities could be adapted to a curved design.", "dialogue": "None: Oh yeah. Excuse me.\nNone: Of course I can stay.\nNone: Okay.\nSpeaker E: A nice picture. Okay.\nSpeaker E: So we are here for the concept design meeting.\nSpeaker E: So we will first start by summarizing the previous meeting and the decision we've taken.\nSpeaker E: I will take notes during this meeting so that you can look at my folder to see the summary of this meeting afterwards.\nSpeaker E: So each of you will show us the various investigations they've done during.\nSpeaker E: Previous hours.\nSpeaker E: We'll then take the decision and concept and then we will define the next task to be done before the next meeting.\nSpeaker E: So last time we decided to have a simple interface.\nSpeaker E: We also decided to have a will to change channel, preview channel button, channel digits.\nSpeaker E: This button should be protected by plastic cover or something for the remote control to look very simple.\nSpeaker E: We have also buttoned for volume and to switch on of the TV.\nSpeaker E: We have also the lightning feature for the remote control to be easy to find.\nSpeaker E: So for the first development and low cost we have decided to have no LCD nor voice features.\nSpeaker E: So now we will have three presentations.\nSpeaker E: So the concept you have specification by industrial designer.\nSpeaker E: The specification of the UI by...\nSpeaker E: The last point is train watching by market expert.\nSpeaker E: So maybe we can start with industrial design.\nSpeaker E: So this is the presentation idea.\nSpeaker E: Maybe I can switch slide on your request.\nSpeaker C: I only have three slides.\nSpeaker C: So I just look at the...\nSpeaker C: Just on some web pages to find some documentation.\nSpeaker C: And I think a remote control is, as I mentioned previously, you just have a very simple chip.\nSpeaker C: And the user interface is just done usually by push button.\nSpeaker C: And in our case we are using a wheel control.\nSpeaker C: So I was looking basically for that chip, which is a very, very standard.\nSpeaker C: And I just look for the wheel sensor and the standard push button.\nSpeaker C: And we can change directly.\nSpeaker C: In fact, I have the number of that element, which is very standard for remote control.\nSpeaker C: The push button are usually extremely cheap.\nSpeaker C: But I just have one problem and this is related with the wheel sensor, which seems to be quite expensive.\nSpeaker C: Okay. And I think if we could just talk about that, if we really need a wheel sensor, or if we can combine something with a push button.\nSpeaker C: Okay. A wheel sensor is 50 times the price of a push button.\nSpeaker E: But is it a significant price on the world remote control?\nSpeaker E: Because we can afford up to 12 euros for the price of the remote control.\nSpeaker E: Yeah. So will this, including all possible things, so buttons will, and the chip, be lower than 12 euros to produce?\nSpeaker C: But I don't think that we should, we should talk about the design of the box.\nSpeaker C: So we need some money.\nSpeaker B: I also have to say, did you receive the mail about voice recognition?\nSpeaker C: Oh, that's all.\nSpeaker C: No.\nSpeaker B: You received something?\nSpeaker B: Yeah. We have an email from the Wine Factory division that they have basically a voice recognition chip already developed.\nSpeaker B: Hmm.\nSpeaker B: It says right now we just use it to record the answers to particular questions.\nSpeaker B: But I guess we could be adapted, I guess it's possible.\nSpeaker B: Instead of recording the answers, you can just record something simpler.\nSpeaker E: Okay. And there can recognize some commands.\nSpeaker B: Yeah. You can recognize commands. You can record new commands and stuff.\nSpeaker B: So if you already have the other tip.\nSpeaker E: Yeah. You could use it.\nSpeaker E: Okay. Maybe we can just listen to this presentation and then take the decision later.\nSpeaker E: Okay.\nSpeaker C: And according to those news.\nSpeaker C: But I think it's, yeah. Sorry. I haven't written my personal references.\nSpeaker C: I just want to mention the problem of the wheels and so, which is much more expensive than any push button.\nSpeaker C: Okay.\nSpeaker C: So we could reduce that. We have already some good things with the back lights of the push button.\nSpeaker B: I have a question about that actually.\nSpeaker B: What is the purpose of the light?\nSpeaker C: Just to make something which is slightly more designed that is a square root box.\nSpeaker E: You can use the button.\nSpeaker B: But the dark also. But in the dark.\nSpeaker B: Yeah. But it's going to go all the way down the light.\nSpeaker E: It will be turned on when the user moves the remote control. I think no.\nSpeaker B: What if you move it and you have it? You don't need to find it.\nSpeaker B: You can see the buttons better of course. Yeah. Yeah. Yeah. Yeah.\nSpeaker B: True.\nSpeaker B: But if you move it, then you have to have some sensor to move it to detect that you're moving it.\nSpeaker C: As soon as you start to move the remote control you have the right.\nSpeaker B: Yeah. But you need another sensor for that. Right. Yeah. Again.\nSpeaker B: No, it's too expensive.\nSpeaker C: I don't think that this is really expensive. But at the end this is plenty of the rich devices.\nSpeaker B: But the build installs to be. Yeah. But the experience will be too expensive.\nSpeaker C: First of all I was thinking to have a continuous light.\nSpeaker C: And you press the button. You have the light on your remote control.\nSpeaker C: Or once you want to turn off. But it can be battery consuming now.\nSpeaker C: To have the light always on. A little bit.\nSpeaker E: Well we will discuss that after maybe. Okay.\nSpeaker B: The other presentation. No, my one should be in the shared folder. Yeah.\nSpeaker B: It was last time I saw it.\nSpeaker E: And it is.\nSpeaker F: Okay.\nSpeaker B: So, okay. Let's move the next slide.\nSpeaker B: So basically I want very simple. Right.\nSpeaker B: That's a major idea. I suppose. Yeah.\nSpeaker B: So I just looked at some current designs on the web of usually more complicated remote controls.\nSpeaker B: And I selected two of them. Because even though they have many buttons they look quite simple.\nSpeaker B: And in our case we just rejected the buttons that we don't need. Yeah.\nSpeaker E: And also does it fit well in. And because it was the risk problem.\nSpeaker B: Well these remote are quite big. So go to the next page.\nSpeaker B: So the hub works buttons as you can see. But most of them we just need the ones in the middle.\nSpeaker B: Yeah. So from the bottom where there's the numbers and then the top until the channel.\nSpeaker B: So this middle part. And the left one is exactly the same.\nSpeaker B: So it's basically more or less how we would like it with big volume control, big channel control,\nSpeaker E: and mutant power. Yeah. Yeah. It's the basic thing. So it's only the central part.\nSpeaker B: So basically something that we would look more or less the same as these two.\nSpeaker E: Yeah. With maybe a more economic design on the... Yeah.\nSpeaker B: If you look for example, I think that the volume and the buttons that are there on the top part are not very easy to reach your thumb. Yeah. It could be on the right side for example.\nSpeaker B: Okay. Yeah. Because we don't have these input buttons and all the stuff that they have.\nSpeaker B: And I think that the plastic cover is not very good idea because you open it, it can break.\nSpeaker B: You can do various things. Okay.\nSpeaker B: You just need to put the channel numbers somewhere a bit out of the way.\nSpeaker B: Okay. So that's a bit better. Yeah.\nSpeaker B: Okay. And just to press the other big buttons. But it's not be difficult to press the channel button side there.\nSpeaker B: I think that if you put the cover, it would be even more difficult for the user. Okay.\nSpeaker B: Okay. Yeah.\nSpeaker B: Usually what I have noticed that people put the plastic cover on things that you normally don't mess with, like buttons for tuning the channels and stuff like that.\nSpeaker B: Yeah. That you want to protect it with.\nSpeaker B: I think it's an intelligent algorithm. Okay.\nSpeaker B: So I don't think, yeah. This is just the wheel.\nSpeaker B: Some wheels can be pushed down, could use a push down of the wheel for their call if you want.\nSpeaker B: So you could just basically use one, just wheel and a user could use just the wheel to everything.\nSpeaker B: Yeah. At the time. In that case.\nSpeaker E: Maybe the wheel will be a good advantage over our competitors.\nSpeaker E: Otherwise it's pretty standard about the fact that it's very simple.\nSpeaker E: So maybe this was to have more expense on that aspect.\nSpeaker B: Yeah. I guess the market itself would tell us about that.\nSpeaker E: Okay. So we can move to the, is there any question for designer of user interface?\nSpeaker E: We can move to the next part. Maybe and discuss afterwards.\nSpeaker A: Okay. Okay. You can go. Yeah.\nSpeaker A: So now the recent investigation we have done for the remote control.\nSpeaker A: So the most important aspect for remote controls is to be fancy look and feel and not current functional look and feel.\nSpeaker A: And the second aspect is that the remote control should be technological innovative.\nSpeaker A: Okay. And the third most important aspect is to, is that the remote control should be easy to use. So things we have, we have speak about before. Yeah.\nSpeaker A: And so you can go after.\nSpeaker A: And there is a fashion which is in Paris and Milan that I've detected the following trends.\nSpeaker A: Fruits and vegetables will be the most important time for closes shoes and furniture.\nSpeaker A: So maybe if our remote control has to be a food form or vegetable forms, something like that.\nSpeaker C: What's important, April?\nSpeaker A: And the material is expected to be spongy.\nSpeaker A: So I don't know which material can be spongy and if you...\nSpeaker B: This is good also for, I think, just the electronics and the things that work, right?\nSpeaker E: Yeah. I think it is good also to have a spongy control metal television.\nSpeaker A: Yeah, because it's progress. It's progress, yeah.\nSpeaker B: Hey, that's a cool one.\nSpeaker B: We could say that if you throw it, you have a sensor and you throw it and you can tell the vision and it's a ten-hit-off.\nSpeaker B: When it takes a shock.\nSpeaker C: Yeah. I know that they do that for around the clock.\nSpeaker A: And this, you can say that.\nSpeaker A: No.\nSpeaker A: You can go before?\nSpeaker A: Before? Yes.\nSpeaker A: And you know, yeah, the more important aspect is the fancy look and feel.\nSpeaker A: After is the technological innovative and after the easy twig.\nSpeaker E: Yes. I think it's innovative to use the wheel because I think the one is in the side.\nSpeaker A: Yeah, that's why I think we have to keep that if possible.\nSpeaker E: Yeah. I think it makes it both easy and both innovative.\nSpeaker E: So I think it's a good aspect and it's a good look.\nSpeaker E: Cool. Fancy.\nSpeaker E: Yeah. We have to make it look like a fruit or a vegetable.\nSpeaker E: Yeah. What about it?\nSpeaker E: A color that remembers something.\nSpeaker E: Things like that.\nSpeaker B: The obvious thing is a banana, I guess.\nSpeaker A: Maybe.\nSpeaker A: I thought about a P, for example. You know, a P is like that.\nSpeaker A: And it's easy to have in a hundred times.\nSpeaker E: Yeah. A pure.\nSpeaker E: A banana is also going to look.\nSpeaker E: Maybe pure, yeah, something like that.\nSpeaker A: Oh, a fruit like that, I don't know.\nSpeaker E: Yeah. We can discuss that.\nSpeaker E: Is there anything you want to add?\nSpeaker B: Is there any fruit that is spongy?\nSpeaker E: I don't think so.\nSpeaker E: I think we can have like a pure is good, fit well or banana, you too.\nSpeaker C: So maybe look and feel what about a piece of ice with blue LED inside.\nSpeaker C: But that's not in the train.\nSpeaker E: The train is spongy and vegetable fruit.\nSpeaker E: It's not in the toilet.\nSpeaker E: I think spongy is good because it will be robust as well.\nSpeaker E: So I think we can keep the wheel because it's easy, it's innovative, even if the cost is a bit higher.\nSpeaker E: And we also have to find a fruit like pure or banana with any other idea you have.\nSpeaker E: What kind of fruit would you like to control your TV with?\nSpeaker C: Banana is a nice idea.\nSpeaker E: Banana is also yellow, so you can't lose your remote control.\nSpeaker C: Don't use the banana.\nSpeaker C: The banana is curving like that, but when the banana is curving like that, with the wheel on the top, you can control.\nSpeaker C: But you don't have the first button.\nSpeaker B: You can just have the skill and you move your hand here to push the button and then you move on the other side.\nSpeaker B: You can have it on two sides and it will be cool.\nSpeaker E: I think it's a good design and it's innovative as well.\nSpeaker E: Maybe we can keep the banana and it will be very easy to find.\nSpeaker E: Everybody knows what is a banana.\nSpeaker C: Basically, if you start with fancy fruits, you can also take the banana and feed it\nSpeaker B: with the color scheme of our company.\nSpeaker D: Oh, yeah.\nSpeaker E: It's really a good point.\nSpeaker B: I hope this video does not manage to die.\nSpeaker E: I think so. We have to take some decisions on those aspects.\nSpeaker E: So for components, we have to think about those aspects.\nSpeaker C: So we will just use a standard battery.\nSpeaker C: We know exactly which one we are going to use.\nSpeaker C: What do you mean by case?\nSpeaker E: I think it's the box that should be a spongy banana shape.\nSpeaker C: For me, if we use a spongy banana case, it doesn't matter.\nSpeaker C: I just want to have something to print to fix my components on two of that box.\nSpeaker B: It can't be inside the inner structure.\nSpeaker B: The thing is that the buttons and the wheel have to be a spongy.\nSpeaker B: Then the buttons and the wheel have to move.\nSpeaker B: It's going to bend a lot.\nSpeaker B: I think it's possible.\nSpeaker E: I think it's not. The button would be infected.\nSpeaker E: It should be something out-shaped with a spongy cover.\nSpeaker C: This is like a...\nSpeaker E: Okay.\nSpeaker E: Out-shaped with spongy cover.\nSpeaker E: And standard battery.\nSpeaker E: A cheap one print.\nSpeaker E: There's no specific problem.\nSpeaker E: So we agreed to put the wheel.\nSpeaker C: The wheel on the top button.\nSpeaker C: Why do we want some buttons?\nSpeaker B: Well, usually hold...\nSpeaker B: It won't be good also for the left hand user, right?\nSpeaker D: Yeah.\nSpeaker E: So it has to be symmetrical.\nSpeaker B: Yeah, but it's okay.\nSpeaker B: So let's say that...\nSpeaker B: Basically, you can only take two sides.\nSpeaker B: One side where there's a thumb, and the other side where there's...\nSpeaker B: Yeah, also the thumb.\nSpeaker B: Or you could use this one, but it won't.\nSpeaker B: It's very comfortable.\nSpeaker B: Yeah, maybe this thumb is more comfortable.\nSpeaker B: This for the wheel and then this for the buttons.\nSpeaker E: Yeah, I think it's okay for ghosts right and left.\nSpeaker B: So if we get left, we have...\nSpeaker E: I think you can turn it this way or so.\nSpeaker E: You can do those with...\nSpeaker E: With buttons.\nSpeaker B: Yeah, the buttons and wheels, then we'll need 10th around.\nSpeaker B: The buttons are on the other side.\nSpeaker B: You can't see them.\nSpeaker E: Well, you will get used to it.\nSpeaker E: And more buttons have to be here and the wheels are here.\nSpeaker E: The buttons are lighted, so you immediately identify the right side.\nSpeaker E: Because you have light on buttons.\nSpeaker B: Yeah, I know, but if you hold the left hand and the wheels here, and the buttons are here, they only turn around.\nSpeaker B: The buttons will be on the other side.\nSpeaker E: No, you... I think you will use it on the right or left hand, whether you're right here or left here.\nSpeaker E: I think for left it's okay.\nSpeaker E: I can do this movement and for right here as well.\nSpeaker E: I think this doesn't change that much.\nSpeaker B: Okay.\nSpeaker D: Yeah?\nSpeaker B: Okay, maybe.\nSpeaker D: Yeah.\nSpeaker E: So, for interface we said that we have this banana shape with buttons on the side.\nSpeaker E: And the wheel on the top.\nSpeaker E: So the color is yellow.\nSpeaker E: I think it's... we define everything according to what we should...\nSpeaker E: but the decision we should take, yeah.\nSpeaker E: So maybe we can... we can work on due respect until next meeting.\nSpeaker E: So, have the final look and feel design according to the decision.\nSpeaker E: And have the user interface design and then evaluate the product.\nSpeaker E: That is to say, check if it fits the requirement given by the users.\nSpeaker E: But according to your presentation, it seems to be okay.\nSpeaker E: It seems to be fancy, innovative and easy to use.\nSpeaker E: So, to prepare the prototype, I would suggest that the industrial designer and the user interface designer work together.\nSpeaker E: That would be better, I think.\nSpeaker E: And so you will receive further instruction by emails as usual.\nSpeaker E: Yes, master.\nSpeaker E: Do you need to add anything?\nSpeaker E: No.\nSpeaker E: You feel okay?\nSpeaker E: Yeah.\nSpeaker E: You feel free to express what you want to say.\nSpeaker E: You don't feel too constrained.\nSpeaker E: You don't feel free to answer this.\nSpeaker E: No, you can't.\nSpeaker E: Okay.\nSpeaker E: So, see you.\n"}, {"meeting_id": "Bro023", "summary": "The meeting began with a short introduction to Hans Guenter, who was arriving soon and would be visiting for a few weeks. The team then discussed integrating mean log magnitude spectral subtraction into the SmartKom system. It seemed, however, that lengthening the time for this task was not very effective. The team also learned more about Wiener filtering which worked well when mixed with other approaches. It was almost as effective as the best systems. The team then delved into spectral subtraction and discussed the various ways in which it could be refined. The meeting ended with comments on the additional latency that was caused by this method.", "dialogue": "Speaker E: And Hans Gunter will be here with the mechanics x2's day or so.\nSpeaker E: Oh, okay.\nSpeaker E: So he's going to be here for about three weeks.\nSpeaker D: Oh, that's right.\nSpeaker E: Just for a visit.\nSpeaker E: We might end up with some longer collaboration or something.\nSpeaker E: So he's going to look in on everything we're doing and give us his thoughts.\nSpeaker E: And so be another good person looking at things.\nSpeaker A: Oh.\nSpeaker A: Yeah.\nSpeaker A: Oh, yeah.\nSpeaker A: Oh, yeah.\nSpeaker A: Yeah.\nSpeaker A: Oh, yeah.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Now, here we are after three weeks.\nSpeaker E: He's very, very easy going.\nSpeaker E: Easy to talk to.\nSpeaker E: Very interesting.\nSpeaker E: Everything.\nSpeaker E: Yeah.\nSpeaker B: Yeah, we're married and we're not sure.\nSpeaker E: Yeah.\nSpeaker E: Yeah, he's been here before.\nSpeaker E: I mean, he's been here.\nSpeaker G: He's been here.\nSpeaker G: He's been here for a year or six months, something like that.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: He's done a couple of stays here.\nSpeaker G: So I guess we got lots to catch up on.\nSpeaker G: We haven't met for a couple of weeks.\nSpeaker G: We didn't meet last week morning.\nSpeaker G: I went around and walked everybody and seemed like they had some new results.\nSpeaker G: Other than them coming up and telling me I figured we should just wait a week and they can tell both, you know, all of us.\nSpeaker G: So why don't we start with you, Dave?\nSpeaker G: Oh, okay.\nSpeaker G: Then we can go on.\nSpeaker A: So since we're looking at putting this on, we need a lot of things.\nSpeaker A: I think you expect those factors to make this our conscious.\nSpeaker A: I just ask you if it would work using the path only and possibly present the pathway for me.\nSpeaker A: And so I just ask who I use 12 seconds from the path and the path and the path for me and 12 seconds.\nSpeaker G: 12 seconds from the current.\nSpeaker G: 12 seconds from the current.\nSpeaker A: 12 seconds.\nSpeaker A: 12 seconds from the end of the current frame.\nSpeaker A: So we've had to do it using a 12 seconds sandwich window.\nSpeaker A: I think it was a drop in performance, but it was slightly dropped.\nSpeaker A: Is that right?\nSpeaker E: Yeah, I mean, it was pretty tiny.\nSpeaker A: So that was encouraging.\nSpeaker A: That's encouraging for the idea of using it in hydroaccus.\nSpeaker A: It's like smart home.\nSpeaker A: And another issue I'm thinking about is in the smart home system.\nSpeaker A: So it's like 12 seconds in the early test.\nSpeaker A: It's like a good length of time.\nSpeaker A: But what happens if you have left in 12 seconds?\nSpeaker A: So I would be before back in May I did some experiments using say two seconds or four seconds or six seconds.\nSpeaker A: In those I trained the models using mean subtraction with the mean calculated over two seconds or four seconds or six seconds.\nSpeaker A: And here I was curious what if I trained the models using 12 seconds.\nSpeaker A: But I gave it a situation where the test I was subtracting using two seconds or four seconds or six seconds.\nSpeaker A: So I did that for about three different conditions.\nSpeaker A: And I think it was four seconds and six seconds and eight seconds, something like that.\nSpeaker A: And it seems like it hurts compared to if you actually trained the models using that same length of time.\nSpeaker A: But it doesn't hurt that much.\nSpeaker A: You usually left in 25% although I could see one where it was a point eight percent or so rise in word error rate.\nSpeaker A: But this is where even if I train on the model means subtractive with the same length of time as in the test, the word error rate is around 10% or 9%.\nSpeaker A: So it doesn't seem like that.\nSpeaker E: But looking at the other way, isn't it what you're saying that it didn't help you to have the longer time for training if you were going to have a short time for?\nSpeaker A: That's true.\nSpeaker E: I mean why would you do it if you knew that you were going to have short windows?\nSpeaker G: It seems like you're in normal situations you would never get 12 seconds of speech.\nSpeaker B: You need 12 seconds in the past two weeks to make it right?\nSpeaker B: Are you looking at six seconds in future and six in?\nSpeaker G: No, it's all 12 seconds in the past.\nSpeaker G: Is this 12 seconds of regardless of speech or sound?\nSpeaker E: The other thing which may be a little bit of something else we've talked about in terms of windowing and so on is that I wonder if you trained with 12 seconds.\nSpeaker E: And then when you were two seconds in, you used two seconds and when you were four seconds in, you used four seconds and you basically build up to the 12 seconds so that if you have very long utterances, you have the best.\nSpeaker E: But if you have shorter utterances, you use which you can.\nSpeaker A: Right. And that's actually what we're applying to do in small calm.\nSpeaker A: Yeah.\nSpeaker A: So the question I was trying to get out with those experiments is does it matter what models you use? Does it matter how much time you use to calculate the mean when you were doing the training data?\nSpeaker E: Right. But I mean the other thing is that that's, I mean the other way of looking at this going back to mean capital subtraction versus rasta kind of things is that you could look at mean capital subtraction, especially the way you're doing it as being kind of filter.\nSpeaker E: So the other thing is just to design a filter.\nSpeaker E: You know, basically you're doing a high pass filter or a band pass filter or some sort and just design a filter.\nSpeaker E: And then you know, filter will have certain behavior and you can look at the startup behavior.\nSpeaker E: You start up with nothing and you know, you will.\nSpeaker E: If you have an IAR filter, for instance, it will not behave in the study state way that you would like it to behave until you get along enough period.\nSpeaker E: But by just constraining yourself to have your filter be only subtraction of the mean, you're kind of tying your hands behind your back because there's filters have all sorts of temporal and spectral behaviors.\nSpeaker E: And the only thing consistent we know about is that you want to get rid of the very low frequency component.\nSpeaker B: But do you really want to calculate the mean you neglect all the silenced regions or you just use everything that's 12 seconds?\nSpeaker A: You mean in my task so far?\nSpeaker A: Yeah. Most of the silences have been cut out.\nSpeaker A: Okay.\nSpeaker A: Just just introwards silences.\nSpeaker B: And they are like pretty short.\nSpeaker B: So you really need a lot of speech to estimate the mean in it?\nSpeaker A: Well, if I only use six seconds, it still works pretty well.\nSpeaker A: I saw my task before.\nSpeaker A: I was trying 12 seconds, but that was the best.\nSpeaker A: Okay.\nSpeaker A: And increasing past 12 seconds didn't seem to help.\nSpeaker A: Yeah, I guess it's something I need to play with more of this.\nSpeaker A: I had to set that up for the smart answers that made maybe if I trained on six seconds, it would work better when I only had two seconds or four seconds.\nSpeaker D: Yeah.\nSpeaker E: And again, if you take this filtering perspective and if you essentially have it build up over time, I mean, if you computed means over two and then over four and over six, then it's literally getting into the kind of ramp up of a filter anyway.\nSpeaker E: And so you may just want to think of it as a filter.\nSpeaker E: But if you do that, then in practice, somebody using a smart-com system wouldn't think they're using it for a while.\nSpeaker E: It means that their first utterance instead of getting a 40% error rate reduction, they'll get what you get without this policy, you get 30%.\nSpeaker E: And then the second utterance that you give, they get the full benefit of it.\nSpeaker G: If it's ongoing, you catch the utterances.\nSpeaker G: That's how you get your...\nSpeaker E: Well, I'm saying in practice, yeah.\nSpeaker E: That's somebody's using a system to ask for directions or something.\nSpeaker E: Okay.\nSpeaker E: You know, they'll say something first and to begin with, if it doesn't get them quite right, maybe they'll come back and say, excuse me.\nSpeaker E: Or I mean, you should have some policy like that anyway. And in any event, they might ask a second question.\nSpeaker E: It's not like what he's doing doesn't improve things.\nSpeaker E: It does improve things just not as much as he would like.\nSpeaker E: And so there's a higher probability of making an error for a utterance.\nSpeaker G: It would be really cool if you could have...\nSpeaker G: There's probably users who would never like this, but if you had... could have a system where before they began to use it, they had to introduce themselves verbally.\nSpeaker G: Yeah.\nSpeaker G: I have my name and so on and so on.\nSpeaker G: And you could use that initial speech to do all these adaptations.\nSpeaker E: Right.\nSpeaker E: Well, the other thing, I guess, which I don't know much about... as much as I should about the rest of the system, but couldn't you...\nSpeaker E: If you sort of did a first pass, I don't know what kind of capability we have at the moment for doing second passes on some kind of little small lannister, a graph, or fusion network or something.\nSpeaker E: But if you did first pass with either without the means of subtraction or with a very short time one, and then once you actually had the whole utterance in, if you did the longer time version then based on everything that you had, and then at that point only used it to distinguish between, you know, top-end possible utterances or something.\nSpeaker E: You might... it might not take very much time. I mean, I know in the large vocabulary systems people were evaluating on the past, some people really pushed everything in to make it in one pass, but other people didn't and had multiple passes.\nSpeaker E: And the argument against multiple passes has often been, but we want this to be, you know, had nice interactive response.\nSpeaker E: And the counter argument to that, which is a BBN, I think, had was, yeah, but our second response is, second passes and third passes are really, really fast.\nSpeaker E: So if your second pass takes a millisecond, who cares?\nSpeaker A: So the idea of the second pass would be waiting until you have more recorded speech or...\nSpeaker E: If it turned out to be a problem, that you didn't have enough speech because you need longer window to do this processing, then one tactic is looking at the larger system, and not just at the front end stuff, is to take in the speech with some simpler mechanism or shorter time mechanism, do the best you can and come up with some possible alternates of what might have been said, and either in the form of an end-best list or in the form of a lattice or confusion network or whatever.\nSpeaker E: And then the decoding of that is much, much faster, or can be much, much faster if it isn't a big bushing network.\nSpeaker E: And you can decode that with speech that you've actually processed using this longer time subtraction.\nSpeaker E: So I mean, it's common that people do this sort of thing where they do more things that are more complex, a require looking over more time, whatever in some kind of second pass.\nSpeaker E: And again, if the second pass is really, really fast, another one I've heard of is in connected digit stuff, going back and through backtrace and finding regions that are considered to be a digit, but which have very low energy.\nSpeaker E: So, I mean, there's lots of things you can do in second pass, as in all sorts of levels.\nSpeaker E: Anyway, I'm throwing too many things out.\nSpeaker G: So is that...\nSpeaker G: Edit?\nSpeaker G: I got that.\nSpeaker B: Do you want to go for it?\nSpeaker B: So last two weeks, I've been working on that Vener filtering.\nSpeaker B: And I found that single, like I just do a normal Vener filtering, like the standard method of Vener filtering.\nSpeaker B: That doesn't actually give me any improvement over, like...\nSpeaker B: It actually improves over the baseline, but it doesn't need something like 50% or something.\nSpeaker B: So I've been playing with the baseline MFCC.\nSpeaker G: Yeah.\nSpeaker B: So that's the improvement is around 30% over the baseline.\nSpeaker E: Is that using contamination with something else?\nSpeaker B: No, just one stage Vener filter, which is a standard Vener filter.\nSpeaker E: No, but I mean, contamination with our non-normalization or with the...\nSpeaker B: Yeah, yeah, yeah.\nSpeaker B: So I just plug in the Vener filtering.\nSpeaker B: In our system where... Does it mean it gets worse?\nSpeaker B: No, it actually improves over the baseline of not having a Vener filter in the whole system.\nSpeaker B: Like I have an LDF plus online normalization.\nSpeaker B: And then I plug in the Vener filter in that.\nSpeaker B: So it improves over not having the Vener filter.\nSpeaker B: So it improves, but it doesn't take it like beyond like 30% over the baseline.\nSpeaker B: So...\nSpeaker E: That's when I'm confused about it, because I think I thought that our system was more like 40% without the Vener filter.\nSpeaker E: No, it's like...\nSpeaker G: What is the new VAD?\nSpeaker B: No, it's old VAD.\nSpeaker B: So my baseline was...\nSpeaker B: This is like the baseline is 95.68, 89.\nSpeaker E: So I mean, if you do all these word errors, it's a lot easier.\nSpeaker E: What is that?\nSpeaker E: If you do all these word error rates, it's a lot easier.\nSpeaker B: Oh, okay, okay, okay.\nSpeaker B: I don't have it. It's all accurate.\nSpeaker C: Yeah, the baseline is something similar to...\nSpeaker C: I mean, the baseline that you were talking about is the MFCC baseline.\nSpeaker B: Yeah, there are two baselines.\nSpeaker B: Okay, so the baseline...\nSpeaker B: One baseline is the MFCC baseline.\nSpeaker B: When I said 30% improvement is like the MFCC baseline.\nSpeaker E: So as a startup, the MFCC baseline is what?\nSpeaker E: Is it what level?\nSpeaker B: It's just the male frequency and...\nSpeaker B: That's what's the number.\nSpeaker B: So I don't have that number here.\nSpeaker B: Okay, okay, I have it here.\nSpeaker B: It's the VAD plus the baseline, actually.\nSpeaker B: I'm talking about the MFCC plus I do a frame dropping on it.\nSpeaker B: So that's like...\nSpeaker B: The word error rate is like 4.3.\nSpeaker B: 4.3.\nSpeaker B: 4.3.\nSpeaker B: And 0.7.\nSpeaker B: What's 10.7?\nSpeaker B: It's the medium mismatch.\nSpeaker B: Okay, sorry.\nSpeaker B: It's the well-match medium mismatch in the high mismatch.\nSpeaker B: So I don't have...\nSpeaker B: 4.3, 10.7.\nSpeaker B: And 40.\nSpeaker B: 40%.\nSpeaker B: It's the high mismatch.\nSpeaker B: Okay.\nSpeaker B: And it becomes like 4.3.\nSpeaker E: Not changed.\nSpeaker B: Yeah, it's like 10.1.\nSpeaker B: Still the same.\nSpeaker B: And the high mismatch is like 18.5.\nSpeaker E: 18.5.\nSpeaker E: And what were you just describing?\nSpeaker B: The one is...\nSpeaker B: This one is just the baseline plus the...\nSpeaker B: We don't filter plugged into it.\nSpeaker E: But where's the online normalization and so on?\nSpeaker B: Okay, so...\nSpeaker B: Sorry.\nSpeaker B: So with the online normalization, the performance was...\nSpeaker B: And...\nSpeaker B: Okay, so it's like 4.3.\nSpeaker B: And again, that's the 10.4 and 20.1.\nSpeaker B: That was with online normalization and LDA.\nSpeaker B: So the well matched as like literally not changed by adding online or LDA on it.\nSpeaker B: But the...\nSpeaker B: I mean, even the medium mismatch is pretty much the same.\nSpeaker B: And the high mismatch is improved by 20% absolute.\nSpeaker E: Okay.\nSpeaker E: And what kind of number...\nSpeaker E: What are we talking about here?\nSpeaker E: Is this...\nSpeaker B: It's Italian.\nSpeaker B: Italian.\nSpeaker E: Yeah.\nSpeaker E: And what...\nSpeaker E: So what was the corresponding numbers say for the Alcatel system?\nSpeaker E: Yeah, 3.3.\nSpeaker C: Yeah, 3.3.\nSpeaker C: Yeah, 3.4.\nSpeaker C: Yeah, 3.4.\nSpeaker C: 8.7.\nSpeaker C: And 13.7.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker D: Okay.\nSpeaker D: Thanks.\nSpeaker F: Okay.\nSpeaker B: So this is the single stage winner filter with the noise estimation was based on first 10 frames.\nSpeaker B: Actually, I started with using the VAD to estimate the noise.\nSpeaker B: And then I found that it works.\nSpeaker B: It doesn't work for finish and Spanish because the VAD endpoints are not good to estimate the noise.\nSpeaker B: Because it cuts into the speed sometime.\nSpeaker B: So I end up overestimating the noise and getting a worse result.\nSpeaker B: So it works only for Italian by using a VAD to estimate noise.\nSpeaker B: Work for Italian because the VAD was trained on Italian.\nSpeaker B: So this was giving...\nSpeaker B: This was not improving a lot on this baseline of not having the winner filter on it.\nSpeaker B: And so I ran this stuff with one more stage of winner filtering on it.\nSpeaker B: But the second time what I did was I estimated the new winner filter based on the cleaned up speech.\nSpeaker B: And did a smoothing in the frequency to reduce the variance.\nSpeaker B: I mean, I observed that a lot of bumps in the frequency when I do this winner filtering, which is more like a musical noise or something.\nSpeaker B: And so by adding another stage of winner filtering, the results on the speech that car was like...\nSpeaker B: So I still don't have the word error rate.\nSpeaker B: I'm sorry about it.\nSpeaker B: But the overall improvement was like 56.46.\nSpeaker B: This was again using 10 frames of noise estimate and two stage of winner filtering.\nSpeaker B: And rest is like the LDAP and online organization all remaining the same.\nSpeaker B: So this was like compared to 57 is what you got by using the French telecom system, right?\nSpeaker C: No, I don't think so. Is it on Italian?\nSpeaker B: No, this is all of the whole speech that car.\nSpeaker B: 57 point.\nSpeaker B: Right.\nSpeaker B: Yeah, so the new new winner filtering scheme has like some 56.46, which is like one person still less than what you got using the French telecom system.\nSpeaker E: But it's a pretty similar number in any event.\nSpeaker E: It's very similar.\nSpeaker E: Yeah. But again, you're more or less doing what they were doing, right?\nSpeaker B: It's different in a sense.\nSpeaker B: Like, I'm actually cleaning up the cleaned up spectrum, which they're not doing.\nSpeaker B: They're what they're doing is they have two stage stages of estimating the winner filter.\nSpeaker B: Yeah, but the final filter what they do is they take it to the time domain by doing an inverse Fourier transform.\nSpeaker B: And they filter the original signal using that filter, which is like final filter is acting on the input noise speech rather than on the clean up.\nSpeaker B: So this is more like I'm doing winner filter twice, but the only thing is that the second time I'm actually smoothing the filter and then cleaning up the cleaned up spectrum first level.\nSpeaker B: Okay.\nSpeaker B: And so that that's that's what the difference is. And actually I tried it on the original clean.\nSpeaker B: I mean, the original spectrum where like I second time estimate the filter, but actually clean up the noisy speech rather than the first output of the first stage.\nSpeaker B: And that doesn't seem to be giving me that machine improvement.\nSpeaker B: I just didn't run it for the whole case.\nSpeaker B: And what I what I tried was by using the same thing, but so we actually found that the word is very like crucial.\nSpeaker B: I mean, just by changing the word itself gives you the lot of improvement by instead of using the current word, if you just take up the bad output from the channel zero.\nSpeaker B: When instead of using channel zero and channel one, because that was the reason why I was not getting a lot of improvement for estimating the noise.\nSpeaker B: So I just use the channel zero watt to estimate the noise so that it gives me some reliable markers for this noise estimation.\nSpeaker E: What's the channel zero then?\nSpeaker D: So it's like close talking.\nSpeaker B: Yeah, the close talking.\nSpeaker B: Because the channel zero and channel one are like the same speech only when the same endpoints, but only thing that the speech is very noisy for the channel one.\nSpeaker B: So can I also use the output of the channel zero for channel one for bad?\nSpeaker B: I mean, that's like a cheating.\nSpeaker E: Right. I mean, so what are they going to do?\nSpeaker E: Do we know yet about the source what the rules are going to be?\nSpeaker C: Yeah, so actually you received a new document.\nSpeaker C: Yeah, that's describing this.\nSpeaker C: And what they did find out is to not to align the utterances, but to perform recognition only on the close talking microphone.\nSpeaker C: Did you get the recognition to get the boundaries of speech?\nSpeaker E: So it's not like that's being done in one place or one time that's just a role and we you were permitted to do that.\nSpeaker C: I think they will send files, but we don't want.\nSpeaker E: Also, they will send files so everybody will have the same boundaries to work with.\nSpeaker B: Yeah, but actually the alignment actually is not seems to be improving in like on all cases.\nSpeaker C: Yeah, so what happened here is that the overall improvement that they have with this method.\nSpeaker C: Well, to be more precise what they have is they have these alignments and then they drop the beginning silence and the end silence, but they keep 200 milliseconds before speech and 200 after speech.\nSpeaker C: And they keep the speech post is also.\nSpeaker C: And the overall improvement over the MFCC baseline.\nSpeaker C: So when they just at this friend wrapping in addition is 40% right 14% I mean, which is which is the overall improvement.\nSpeaker C: But in some cases it doesn't improve at all like.\nSpeaker B: It gives like negative in some time like some Italian and the IDGETs right.\nSpeaker B: So by using the end point that speech actually it's worse than the baseline in some instances which could be due to the.\nSpeaker C: The other thing also is that 14% is less than what you obtain using a real VAD.\nSpeaker C: Yeah, our new result cheating like this. So I think this shows that there's still work.\nSpeaker C: But working on the VAD is still still important.\nSpeaker G: Can I ask just a high level question.\nSpeaker G: Can you just say like one or two sentences about Wiener filtering and why are people doing that? What's the deal?\nSpeaker B: So the Wiener filter it's like you try to minimize.\nSpeaker B: So the basic principle of Wiener filter is like you try to minimize the difference between the noise signal and the clean signal.\nSpeaker B: If you have two channels like let's say you have a clean signal and you have an additional channel where you know what is the noise signal.\nSpeaker B: And then you try to minimize error between these two.\nSpeaker B: So that's the basic principle and you can do that. If you have only a noise signal available with you, you try to estimate the noise from the assuming that the first few frames are noise or if you have voice activity director you estimate the noise spectrum.\nSpeaker B: And then you assume the noise is same.\nSpeaker B: Yeah, after the speech starts.\nSpeaker B: But that's not the case in many of our cases but it works reasonably well. And then you what you do is you.\nSpeaker B: So again, I can try down some of this.\nSpeaker B: Yeah, and then you do this. This is the transfer function of the Wiener filter.\nSpeaker B: So SF is the clean speech spectrum, and N is the noisy, and so this is the transfer function.\nSpeaker B: And then you multiply your noisy, and you get an estimate of the clean, and so.\nSpeaker B: But the thing is that you have to estimate the SF from the noisy spectrum what you have.\nSpeaker B: So you estimate the NF from the initial noise portions and then you subtract that from the current noise spectrum to get an estimate of the SF.\nSpeaker B: So sometimes that becomes zero because you don't have a true estimate of the noise.\nSpeaker B: So the filter will have like sometimes zeros, and it because some frequency values will be zeroed out because of that. And that creates a lot of discontinuities across the spectrum with the filter.\nSpeaker B: So that's what that was just the first stage of Wiener filtering that I tried.\nSpeaker G: So is this basically similar to just regular spectrocentrum?\nSpeaker E: It's all pretty related. There's a whole class of techniques where you try in some sense to minimize the noise.\nSpeaker E: And it's typically a mean square sense in some way.\nSpeaker E: And spectral subtraction is one approach to it.\nSpeaker G: So people use the Wiener filtering in combination with the spectral subtraction typically or are they sort of non-seating techniques?\nSpeaker B: They are very similar techniques. So it's like I've been seeing Wiener filter with spectral subtraction.\nSpeaker E: I mean in the long run you're doing the same thing, but you make different approximations. In spectral subtraction for instance there's an estimation factor.\nSpeaker E: So you figure out what the noise is and you multiply that noise spectrum times some constant and subtract that rather than sometimes people even though this really should be in the power domain, sometimes people work in the magnitude to mean because it works better.\nSpeaker G: So why did you choose Wiener filtering over some other one of these other techniques?\nSpeaker B: The reason was we had this choice of using spectral subtraction Wiener filtering and there was one more thing which I'm trying to do is the subspace approach.\nSpeaker B: Stefan is working on spectral subtraction. So I picked up a few sort of trying to be...\nSpeaker B: We just wanted to have a few noise production, composition techniques and then pick some from that.\nSpeaker E: Yeah, I mean there's comments working on other than the other activity series. So they were just trying to cover a bunch of different things with this task and see what are the issues for each of them.\nSpeaker B: So one of the things that I tried, like I said, was to remove those zeros in the filter by doing some smoothing of the filter.\nSpeaker B: Like you estimate the HF square and then you do smoothing across the frequency so that those zeros get like flattened out and that doesn't seem to be improving by trying it on the first time.\nSpeaker B: So what I did was like I did this and then you applied in the one more, the same thing but with the smooth filter, second time. That seems to be working.\nSpeaker B: So that's what I got like 56.5% improvement on speech.car with that. And so the other thing what I tried was I used till the 10 frames of noise estimate but I used this channel 0 watt to drop the frames.\nSpeaker B: So I'm not still on estimating and that has taken the performance to like 67% in speech.car which is which like sort of shows that by using a proper watt you can just take it to further better levels.\nSpeaker B: So that's sort of like you know, best case performance. Yeah so far I've seen 67% and I haven't seen like 67% and using the channel 0 watt to estimate the noise also seems to be improving but I don't have the results for all the cases with that.\nSpeaker B: So I used channel 0 watt to estimate noise as a lesser drop frame which is like everywhere I use the channel 0 watt and that seems to be the best combination than using a few frames to estimate and then drop channel.\nSpeaker E: So I'm still a little confused is that channel zero information going to be accessible.\nSpeaker B: Now this is just to test whether we can really improve by using a better watt. So I mean so this is like the noise compensation is fixed but you make a better decision on the end points that's like seems to be.\nSpeaker B: So which means which means like by using this technical it just the way we can just take the performance by another 10% of better.\nSpeaker B: So that that was just the reason for doing that experiment and yeah but this all these things have to still try it on the TI digits which is like I'm just running and that seems to be not improving a lot on the TI digits.\nSpeaker B: So the other thing is like I'm doing all this stuff on the power spectrum so try this stuff on the Mel as well Mel and the magnitude and Mel magnitude and all those things but seems to be the power spectrum seems to be giving the best result.\nSpeaker B: So one of the reasons I thought like do the averaging after the filtering using the Mel filter bank that seems to be maybe helping whether than trying it on the Mel filter filter out goods.\nSpeaker B: Yeah that's the only thing is I could think of why it's giving improvement on the Mel and yeah so that's about the subspace stuff.\nSpeaker B: So it's like going parallely but not much of improvement I'm just have some skeleton ready need some work.\nSpeaker C: Yeah so I've been working still on the spectral subtraction so to remind you a little bit of what I did before is just to apply some spectral subtraction with an over estimation factor also to get an estimate of the noise spectrum and subtract this estimation of the noise spectrum from the signal spectrum but subtracting more when the SNR is low which is the technique that it's subtracting from the meaning.\nSpeaker C: So you over estimate the noise spectrum you multiplied no spectrum by a factor which depends on the SNR so above 20 dB it's one so you just subtract the noise and then it's generally what I use actually on linear function of the SNR which is bounded to like two or three when the SNR is below zero dB.\nSpeaker C: Doing just this either on the FFT bands or on the Mel bands doesn't yield any improvement.\nSpeaker D: What are you doing with negative?\nSpeaker C: There is also threshold of course because after subtraction you can have negative energies and so what I just do is to add to put a threshold first and then to add a small amount of noise which right now is speed shaped.\nSpeaker C: So it has the overall energy as the overall power spectrum of speech so with a bump around one key.\nSpeaker G: When you talk about there being something less than zero after subtracting the noise is that a particular frequency band?\nSpeaker G: Yes there can be frequency bands with zero energy. So you're adding some it has overall.\nSpeaker C: For each frequency I'm adding some noise but the amount of noise I add is not the same for all the frequency bands.\nSpeaker C: Right now I don't think if it makes sense to add something that's speed shaped because then you have silence portion but if some spectra are similar to the overall speech spectrum.\nSpeaker C: So this is something I can still work on.\nSpeaker G: What does that mean? I'm trying to understand what it means when you do the spectral subtraction and you get a negative.\nSpeaker G: That means that that means that the frequency range you subtracted more energy than there was actually.\nSpeaker C: You have an estimation of the noise spectrum but sometimes of course as the noise is not perfectly stationary.\nSpeaker C: Sometimes this estimation can be too small so you don't subtract enough but sometimes it can be too large also.\nSpeaker C: If the noise energy in this particular frequency band drops for some reason.\nSpeaker G: So in an ideal world if the noise were always the same then when you subtracted it the worst that you would get would be a zero.\nSpeaker G: I mean the lowest you would get would be a zero because if there was no other energy there you're just subtracting exactly the noise.\nSpeaker E: There's all sorts of deviations from ideal here.\nSpeaker E: For instance you're talking about the signal noise at a particular point even if something is sort of stationary and it's third terms of statistics there's no guarantee that any particular instantiation or piece of it is exactly a particular number or bounded by a particular range.\nSpeaker E: So you're figuring out from some chunk of the signal what you think the noise is then you're subtracting that from another chunk.\nSpeaker E: And there's absolutely no reason to think that you'd know that it wouldn't be negative in some places.\nSpeaker E: On the other hand that just means that some sense you've made a mistake because you certainly have subtracted a bigger number than is due to the noise.\nSpeaker E: Also we speak where all this stuff comes from is from an assumption that signal noise are uncorrelated and that certainly makes sense in statistical interpretation that over all possible realizations that they're uncorrelated.\nSpeaker E: So that's why we're talking about the signal noise that is there uncorrelated or assuming your goodicity that across time it's uncorrelated.\nSpeaker E: But if you just look at quarter second and you cross multiply the two things you could very well end up with something that sums to something that's not zero.\nSpeaker E: So if you do signals could have some relation to one another and so there's all sorts of deviations from ideal in this and given all that you can definitely end up with something that's negative.\nSpeaker E: But if down the road you're making use of something as if it is a power spectrum then it can be bad to have something negative.\nSpeaker E: The other thing I wonder about actually is what if you left it negative what happens? I mean because the log are you taking the log before you add them up to the mill?\nSpeaker E: No, after.\nSpeaker E: Right. So the thing is I wonder how if you put your thresholds after that I wonder how often you would end up with negative values.\nSpeaker B: Would you end up reducing the neighboring frequency when the average right when you add the negative to the positive value which is the true estimate?\nSpeaker E: Yeah.\nSpeaker E: But nonetheless you know these are it's another kind of smoothing right that you're doing.\nSpeaker E: So you've done your best shot at figuring out what the noise should be and then you subtract it off and then after that instead of instead of leaving it as is and adding things adding up to neighbors you artificially push it up which is you know it's there's no particular reason that that's the right thing to do either right.\nSpeaker E: So in fact what you'd be doing is saying well we're we're we're going to definitely diminish the effect of this frequency and this is a little frequency bin in the in the overall mill summation.\nSpeaker E: This is the thought.\nSpeaker G: I don't know if you're going to get a negative number you don't do the subtraction for that.\nSpeaker G: Yeah, although almost the opposite right.\nSpeaker G: Instead of living it negative you don't do it.\nSpeaker G: If you're subtracting is going to result in a negative number you you don't do subtraction.\nSpeaker E: Yeah, but that means that in a situation where you thought that the bin was almost entirely noise you left it.\nSpeaker D: I'm just saying that's like yeah.\nSpeaker C: And some people also it's a negative value they recompute it using interpolation from the adjacent.\nSpeaker C: Yeah, from frequency bin things that you can do.\nSpeaker E: People can also reflect it back up and essentially do a four-way rectification instead of a set of half-wave.\nSpeaker E: But it's just a thought that that might be something to try.\nSpeaker C: Yeah, well actually I tried something else based on this.\nSpeaker C: It's to put some smoothing because it seems to help.\nSpeaker C: It seems to help the winner filtering.\nSpeaker C: So what I did is some kind of non-linear smoothing actually have a recursion that computes.\nSpeaker C: Yeah, let me go back a little bit.\nSpeaker C: Actually when you do spectral subtraction you can find this equivalent in the spectral domain you can compute.\nSpeaker C: You can see that the spectral subtraction is a filter and the gain of this filter is the signal energy minus what you subtract divided by the signal energy.\nSpeaker C: And this is the gain that varies over time.\nSpeaker C: Of course depending on the noise spectrum and on the speech spectrum.\nSpeaker C: And what happened actually is that during low SNR values the gain is close to zero but it varies a lot.\nSpeaker C: And this is the cause of musical noise and all these.\nSpeaker C: The fact that we go below zero on one frame and then you can have an energy that's above zero.\nSpeaker C: So the smoothing is, I did a smoothing actually on this gain trajectory but the smoothing is low in error in the sense that I try to not smooth if the gain is I.\nSpeaker C: Because in this case we know that the estimate of the gain is correct because we are not close to zero.\nSpeaker C: And to do more smoothing if the gain is low.\nSpeaker C: Yeah so basically that's this idea and it seems to give pretty good results.\nSpeaker C: Although I just tested on Italian and Finnish.\nSpeaker C: And on Italian it seems my result seems to be a little bit better than the winner filtering.\nSpeaker B: Yeah the one you showed yesterday.\nSpeaker C: I don't know if you have these improvements, the 10 improvements for Italian finishes.\nSpeaker B: No I don't have any.\nSpeaker B: I just just have the final number here.\nSpeaker E: So these numbers he was given before with the 4.3 and the 10 by 1.\nSpeaker E: And so those were Italian right?\nSpeaker B: Yeah so no I actually didn't give the number which is the final one which is after two stages of winner filtering.\nSpeaker B: I mean that was I just told like the overall improvement is like 56.5.\nSpeaker B: So his number is still better than what I got in that two stages of winner filtering.\nNone: Right.\nSpeaker C: On Italian but on Finnish it's a little bit worse apparently.\nSpeaker E: But you have numbers in terms of water rates and...\nSpeaker E: Yeah.\nSpeaker C: That's what I just said.\nSpeaker C: You have some sensor reference.\nSpeaker C: 3.8.\nSpeaker D: Oh okay.\nSpeaker C: And then 9.1.\nSpeaker C: And finally 16.5.\nSpeaker E: And this is spectrosotraction plus what?\nSpeaker C: Blood, plus non-linear smooth thing.\nSpeaker C: Well it's the system.\nSpeaker C: It's exactly the same system.\nSpeaker C: Oh my normalization.\nSpeaker C: But LDA.\nSpeaker C: Yeah.\nSpeaker C: But instead of the both stage winner filtering it's this smooth spectrosotraction.\nSpeaker G: Right. What is it? The France telecom system is it?\nSpeaker G: Did they use spectrosotraction or when you're filtering?\nSpeaker C: They use spectrosotraction right?\nSpeaker C: Forward?\nSpeaker C: French relic.\nSpeaker C: It's a winner filtering.\nSpeaker B: Oh it's a winner filtering.\nSpeaker C: Well it's some kind of winner filtering.\nSpeaker B: It's not exactly winner filtering but some variant of winner filtering.\nSpeaker E: Yeah.\nSpeaker E: Plus I guess they have some sort of capstone normalization.\nSpeaker E: They have like...\nSpeaker B: Yeah.\nSpeaker B: They're just noise compensation technique is a variant of winner filtering.\nSpeaker B: Because they do some smoothing techniques on the final filter.\nSpeaker B: They actually do the filtering in the time domain.\nSpeaker B: So they take this HF square back taking an inverse Fourier transform.\nSpeaker B: And they convolve that time domain signal with that.\nSpeaker B: And they do some smoothing on that final filter in pulse response.\nSpeaker C: But they also have to do different smoothing.\nSpeaker C: One in the time domain and one in the frequency domain by just taking the first.\nSpeaker C: Coefficient of the impulse response.\nSpeaker C: So basically similar...\nSpeaker C: I mean what you did it's...\nSpeaker C: It's similar in the smoothing.\nSpeaker C: They also have to kind of smoothing.\nSpeaker C: One in the time domain and...\nSpeaker C: Yeah.\nSpeaker C: One in the frequency domain.\nSpeaker G: Does the smoothing in the time domain help?\nSpeaker G: Well do you get this musical noise stuff with winner filtering?\nSpeaker G: Or is that only with the spectrosotraction?\nSpeaker G: Oh you get it with...\nSpeaker B: Yeah.\nSpeaker G: Winner filtering also?\nSpeaker G: Is the smoothing in the time domain help with that?\nSpeaker B: No, you still end up with zeros in the spectrum sometimes.\nSpeaker E: I mean it's not clear that these musical noises hurt us in recognition.\nSpeaker E: We don't know if they sound bad.\nSpeaker D: Yeah.\nSpeaker D: We're not listening to it usually.\nSpeaker C: Actually the smoothing that I did, the air reduced the music and noise.\nSpeaker C: Well I cannot...\nSpeaker C: You cannot hear...\nSpeaker C: Well actually what I did not say is that this is not in the FFT bands.\nSpeaker C: This is in the male frequency bands.\nSpeaker C: So it could be seen as a smoothing in the frequency domain because I use the male bands in addition.\nSpeaker C: And then the other phase of smoothing in the time domain.\nSpeaker C: But when you look at the spectrum, if you don't have any smoothing, you clearly see like in silence portions and at the beginning and end of speech, you see spots of high energy randomly distributed over the spectrum.\nSpeaker C: That's the musical.\nSpeaker C: Which is musical noisy.\nSpeaker C: If you listen to it, if you do this in the FFT bands, then you have spots of energy randomly distributed.\nSpeaker C: And if you recentize, this spot sounds as like sounds.\nSpeaker E: None of these systems, by the way, you both are working with our system that does not have the neural net.\nSpeaker E: Right?\nSpeaker F: Yeah.\nSpeaker D: So one would hope.\nSpeaker D: I assume we have the neural net part of it with improved things further as they did before.\nSpeaker C: Yeah.\nSpeaker C: Although if we look at the result from the proposals, one of the reasons the system with the neural net was more than well, around 5% better, is that it was much better on highly mismatched condition.\nSpeaker C: I'm thinking, for instance, on the TIDGITs trained on clean speech and tested noisy speech.\nSpeaker C: For this case, the system with the neural net was much better.\nSpeaker C: But not much on the other cases.\nSpeaker C: And if we have no spectral subtraction or winner filtering, the system is, we thought neural net work is much better than before, even in these cases of mismatch.\nSpeaker C: So maybe the neural net will help less.\nSpeaker C: Maybe.\nSpeaker G: Could you do any neural net with spectral subtraction?\nSpeaker E: Yeah, could do nonlinear spectral subtraction.\nSpeaker E: But I don't know if it, I mean, just figure out what your targets are.\nSpeaker G: Yeah, I was thinking if you had a clean version of the signal one.\nSpeaker G: A noisy version.\nSpeaker G: Right.\nSpeaker G: And targets for the, you know, whatever frequency.\nSpeaker E: Yeah, well, that's not so much spectral subtraction then.\nSpeaker E: But anyway, yeah, people, people do that.\nSpeaker E: Yeah, in fact, we had visitors here who did that, I think, when you were, a little bit way back when people had been lots of experimentation over the years with training neural nets.\nSpeaker E: It's not a bad thing to do.\nSpeaker E: It's another approach.\nSpeaker E: I mean, it's, it, the objection everyone always raises, which has some truth to it, is that it's good for mapping from a particular noise to clean, then you get a different noise.\nSpeaker E: And the experiments we saw that visitors did here showed that there was at least some gentleness to the degradation when you switched to different noises.\nSpeaker E: It did seem to help.\nSpeaker E: So that, you're right, that's another way to go.\nSpeaker G: You did care on, I mean, for good cases where it, it, it, it, stuff that it was trained on.\nSpeaker G: Did it do pretty well?\nSpeaker G: Oh, yeah, it did very well.\nSpeaker E: Yeah.\nSpeaker E: But to some extent, that's kind of what we're doing.\nSpeaker E: I mean, we're not doing exactly that.\nSpeaker E: We're not trying to generate good examples.\nSpeaker E: But by trying to do the best classifier you possibly can for these little fanatic categories.\nSpeaker E: It's sort of built in.\nSpeaker E: It's, yeah, it's kind of built into that.\nSpeaker E: And that's why we have found that it, it does help.\nSpeaker E: So, yeah, I mean, we'll just have to try it.\nSpeaker E: But I would, I would, I would imagine that it will help some.\nSpeaker E: I mean, it, this have to see whether it helps more or less the same.\nSpeaker E: But I would imagine it would help some.\nSpeaker E: So I didn't even, all of this, I was just confirming that all of this was with the simplices.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker C: So this is the, well, actually this was kind of the first try with this spectral subtraction plus smoothing.\nSpeaker C: And I was kind of excited by the result.\nSpeaker C: Then I started to optimize the different parameters.\nSpeaker C: And the first thing I tried to optimize is the time constant of the smoothing.\nSpeaker C: And it seems that the one that I choose for the first experiment was the optimal one.\nSpeaker C: It's amazing how often I have.\nSpeaker C: This is the first thing.\nSpeaker C: Yeah, another thing that I, it's important to mention is that this as this as some additional latency, because when I do the smoothing, it's a recursion that estimates the means of the gain curve.\nSpeaker C: And this is a filter that has some latency.\nSpeaker C: And I noticed that it's better if we take into account this latency.\nSpeaker C: So instead of using the current estimated mean to subtract the current frame, it's better to use an estimate that's somewhere in the future.\nSpeaker G: And that's what causes the latency.\nSpeaker B: Yeah.\nSpeaker B: I mean, the mean is computed based on some frames in the future also.\nSpeaker C: No.\nSpeaker C: It's the recursion.\nSpeaker C: So it's the standard recursion, right?\nSpeaker C: And the latency of this recursion is around 50 milliseconds.\nSpeaker D: One five.\nSpeaker D: It's the one five five zero five zero.\nSpeaker D: Five zero.\nSpeaker D: Yeah.\nSpeaker B: Sorry, why is that delay coming like you estimate the mean?\nSpeaker C: Yeah, the mean estimation has some delay, right?\nSpeaker C: I mean, the filter has that estimated the mean as a distance.\nSpeaker B: Okay, so it's like it looks into the future also.\nSpeaker E: Yeah, okay.\nSpeaker E: What if you just look into the past?\nSpeaker C: It's not as good.\nSpeaker C: It's not bad.\nSpeaker C: How much?\nSpeaker C: It helps a lot of the baseline, but how much?\nSpeaker C: It's around 3% relative.\nSpeaker C: Where's?\nSpeaker C: Yeah.\nSpeaker E: So depending on all this stuff comes out, we may or may not be able to add any latency.\nSpeaker C: Yeah, but yeah.\nSpeaker C: Sorry, it's a different way.\nSpeaker C: Yeah, it's three percent.\nSpeaker C: Yeah, but I don't think we have to worry too much on that right now.\nSpeaker E: Yeah, I mean, I think the only thing is that I would worry about little, because if we completely ignore latency, then we discover that we really have to do something about it.\nSpeaker E: We're going to be finding ourselves in the bind.\nSpeaker E: So, you know, maybe you could make it 25.\nSpeaker E: Yeah, you know what I mean?\nSpeaker E: Yeah, just be a little conservative, because we may end up with this crunch where we have to cut the latency in half or something.\nSpeaker C: So, yeah, there are other things in the algorithm that I didn't play a lot yet,\nSpeaker G: which, sorry, quick question just about the latency thing. If there's another part of the system that causes the latency of 100 milliseconds, is this an additive thing or is yours hidden in that?\nSpeaker G: No, it's added.\nSpeaker G: It's added.\nSpeaker B: We can do something in parallel also.\nSpeaker B: In some cases, like if you wanted to do a voice activity detection, and you can do that in parallel with some other filtering you can do.\nSpeaker B: So, you can make a decision on that voice activity detection, and then you decide whether you want to filter or not.\nSpeaker B: But by then, you already have sufficient samples to do the filtering.\nSpeaker G: So, sometimes you can do it in.\nSpeaker G: I mean, couldn't you just also, I mean, if you know that the largest latency in the system is 200 milliseconds, couldn't you just buffer up that number of frames, and then everything uses that buffer in that way.\nSpeaker G: It's not additive.\nSpeaker E: One fact, everything is sent over and buffers kisses.\nSpeaker E: That TCP buffer, something.\nSpeaker B: I mean, the data, the superframe or something?\nSpeaker B: Yeah.\nSpeaker B: Yeah, but that has a variable latency, because the last frame doesn't have any latency, and the first frame has a 20 frame latency.\nSpeaker B: So, can't rely on that latency all the time.\nSpeaker F: Yeah.\nSpeaker B: Because, I mean, the transmission over the air interface is like a buffer, 20 frames.\nSpeaker B: Yeah.\nSpeaker B: So, but only thing is that the first frame in that 24 frame buffer has a 24 frame latency, and the last frame doesn't have any latency, because it just goes as...\nSpeaker G: Yeah, I wasn't thinking of that one in particular, but more of, you know, if there is some part of your system that has the buffer 20 frames, can't be other parts of the system draw out of that buffer and therefore not add to the latency.\nSpeaker E: Yeah, and that's sort of one of the, all of that sort of stuff is things they're debating in their standards committee.\nSpeaker F: Yeah, so, there is...\nSpeaker C: These parameters that I still have to look at, like, I played a little bit with this overestimation factor, but I still have to look more at this.\nSpeaker C: At the level of noise I add after, I know that adding noise at the system just using spellcrime subtraction without smoothing, but I don't know right now if it's still important or not, and if the level I choose before is still the right one, same thing for the shape of the noise.\nSpeaker C: And, you know, the level I choose before is still the right one, same thing for the shape of the noise.\nSpeaker C: Maybe it would be better to add just white noise instead of speech-shaped noise.\nSpeaker E: That'd be more like the aroust, I think, in a sense.\nSpeaker C: Yeah.\nSpeaker C: And another thing is to...\nSpeaker C: Yeah, for this I just use as noise estimates, the mean spectrum of the first 20 frames of each utterance.\nSpeaker C: I don't remember for this experiment, but did you use 10 frames?\nSpeaker B: I used 10 frames.\nSpeaker B: Yeah, because...\nSpeaker B: I mean, the reason was, like, in TI, did you say, I don't have a lot of 10 frames most of the time?\nSpeaker C: But so, what's this result you told me about the fact that if you use more than 10 frames, you can...\nSpeaker B: Oh, that's using the channel 0.\nSpeaker B: I use the channel 0 to estimate the noise.\nSpeaker C: But this is 10 frames plus...\nSpeaker C: channel 0.\nSpeaker C: Channel...\nSpeaker C: I know, these results...\nSpeaker C: Oh, two stage, winner filtering is 10 frames, but possibly more.\nSpeaker C: I mean, if channel 1, VAD gives you...\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Yeah, but in this experiment, I didn't use any VAD.\nSpeaker C: I just used the 21st frame to estimate the noise.\nSpeaker C: So I expected it to be a little bit better if I use more frames.\nSpeaker C: Okay, that's it for spectroscopy.\nSpeaker C: The second thing I was working on is to try to look at noise estimation and using some technique that doesn't need voice activity detection.\nSpeaker C: And for this, I simply used some code that I add from...\nSpeaker C: I add from Belgium, which is technique that takes a bunch of frames.\nSpeaker C: And for each frequency bands of this frame takes a look at the minima of the energy.\nSpeaker C: And then average this minima and take this as an energy estimate of the noise for this particular frequency band.\nSpeaker C: And there is something more to this, actually, what is done is that these minima are computed based on high-resolution spectra.\nSpeaker C: So I compute a FFT based on the long single frame, which is 64 milliseconds.\nSpeaker G: So you have one minimum per each frequency.\nSpeaker C: What I do actually is to take a bunch of...\nSpeaker C: to take a tile on the spectrogram and this tile is 500 milliseconds long and 200 hertz wide.\nSpeaker C: And this tile...\nSpeaker C: In this tile appears like the harmonics if you have a voice sound, because it's the FFT bands.\nSpeaker C: And when you take the minima of this tile, when you don't have speech, this minima will give you some noise level estimate.\nSpeaker C: If you have voiced speech, this minima will still give you some noise estimate because the minima are between the harmonics.\nSpeaker C: And if you have an or kind of speech sound, then it's not the case.\nSpeaker C: But if the time frame is long enough, like 500 milliseconds seems to be long enough, you still have portions which are very close...\nSpeaker C: which minima are very close to the noise energy.\nSpeaker E: I'm confused. You said 500 milliseconds, but you said 64 milliseconds.\nSpeaker E: What is that?\nSpeaker C: 64 milliseconds is to compute the FFT bands, the FFT.\nSpeaker C: Actually, it's better to use 64 milliseconds because if you use 30 milliseconds, then because of this shortwindowing and at low pitch sounds, the harmonics are not correctly separated.\nSpeaker C: So if you take this minima, they will overestimate the noise a lot.\nSpeaker E: So you take 64 milliseconds and then you average them over 500 or what do you do over 500?\nSpeaker C: I take a bunch of these 64 milliseconds frame to cover 500 milliseconds.\nSpeaker C: And then I look for the minima on a bunch of 50 frames.\nSpeaker C: So the interest of this is that with this technique, you can estimate some reasonable noise spectra with only 500 milliseconds of signal.\nSpeaker C: So if the noise varies a lot, you can track better track the noise, which is not the case if you rely on a voice activity detector.\nSpeaker C: So even if there are no speech poses, you can track the noise level.\nSpeaker C: The only requirement is that you must have in this 500 milliseconds segment, you must have voiced sound at least.\nSpeaker C: Because this will add you to track the noise level.\nSpeaker C: So what I did is just to simply replace the VAD based noise estimate by this estimate.\nSpeaker C: First on speech.car. Well, only on speech.car actually.\nSpeaker C: And it's slightly worse, like one percent relative compared to the VAD based estimates.\nSpeaker C: I think the reason why it's not better is that speech.car. noise is our stationary.\nSpeaker C: There is no need to have something that's active. Well, there are mainly stationary.\nSpeaker C: But I expect maybe some improvement on TI digits because in this case the noises are sometimes very variable.\nSpeaker C: So I have to test it.\nSpeaker E: But are you comparing with something a little confused again?\nSpeaker C: When you compare it with the VAD based. It's the French Silicon based spectra winner filtering and VAD.\nSpeaker C: So it's their system, but just a replace. There are noise estimate by this one.\nSpeaker C: Oh, you're not doing this with our system. I'm not. No.\nSpeaker C: Yeah, it's our system, but we're just the winner filtering from their system.\nSpeaker C: Actually, the best system that we still have is our system, but with their noise compensation scheme.\nSpeaker C: So I'm trying to improve on this and by replacing their noise estimate by something that might be better.\nSpeaker E: Okay, but the spectra subtraction scheme that you reported on also requires a noise estimate.\nSpeaker C: But could I try this for that?\nSpeaker C: No, because I did this in parallel. I say what's working on it.\nSpeaker C: For sure, I can try it also.\nSpeaker B: I think that new noise estimate technique on this winner filtering what I'm trying. I have some experiments running around other results.\nSpeaker B: So I don't estimate the noise from the time France, but you see a system.\nSpeaker C: Yeah, I am also implemented spectral widening idea, which is in the Ericsson proposal.\nSpeaker C: The idea is just to flatten the log spectrum.\nSpeaker C: And to flatten it more if the probability of silence is higher.\nSpeaker C: So in this way, you can also reduce so much reduce the musical noise.\nSpeaker C: And you reduce the variability if you have different noise shapes.\nSpeaker C: Because the spectrum becomes more flat in the silence portions.\nSpeaker C: Yeah, with this no improvement, but there are a lot of parameters that we can play with.\nSpeaker C: And actually, this could be seen as a soft version of the frame dropping.\nSpeaker C: Because you could just put a threshold and say that below the threshold, I will flat on completely flat on the spectrum.\nSpeaker C: And above this threshold, give the same spectrum.\nSpeaker C: So it would be like frame dropping because during the silence portions, which are below the threshold of voice activity, probability, you would have some kind of dummy frame, which is perfectly flat spectrum.\nSpeaker C: And this widening is something that's more soft because you widen, you just have a function.\nSpeaker C: The widening is a function of the speech probability, so it's not a hard decision.\nSpeaker C: So I think maybe it can be used together with frame dropping and we are not sure about the speech of silence.\nSpeaker E: I mean, in J. Rasta, we were essentially adding in white noise, depending on our estimate of the noise, when the over-awesome in the noise.\nSpeaker E: I think it never occurred to us to use a probability in there.\nSpeaker E: You can imagine that that made use of where the amount that you added in was a function of the probability of it being speech or noise.\nSpeaker C: Yeah, we're at noise a constant that just depending on the noise spectrum.\nSpeaker E: Because that brings in sort of powers of classifiers that we don't really have in the other estimates, so it could be interesting.\nSpeaker E: What point does the system stop recording?\nSpeaker E: How much?\nSpeaker G: It'll be a little long.\nSpeaker G: I just ran out of this space.\nSpeaker G: I think we're okay.\nSpeaker C: Yeah, so with this technique, there are some... I just did something exactly the same as the direction proposal, but the probability of speech is not computed the same way.\nSpeaker C: I think for a lot of things, actually, a good speech probability is important. For frame-dropping, you can improve from 10%.\nSpeaker C: As a news show, if you use the channel zero speech probabilities, for this, it might add.\nSpeaker C: So, yeah, the next thing I started to do is to try to develop a better voice activity detector.\nSpeaker C: For this, I think we can maybe try to train the neural network for voice activity detection on all the data that we have, including all these speech data.\nSpeaker C: So, I'm starting to obtain alignment on these databases. The way I do that is that I just use the SDK system, but I train only on the closed-talking microphone, and then I obtain the Viterbi alignment of the training utterances.\nSpeaker C: It seems to be... Actually, what I observe is that for Italian, it doesn't seem... No, it doesn't seem to be a problem.\nSpeaker B: No, it doesn't seem to be a problem.\nSpeaker C: When you enter the frame-dropping, right? Yeah, but actually, the VAD was trained on Italians. The current VAD that we have was trained on the spine-right, the attitudes with noise and length.\nSpeaker C: It seems to work on Italian, but not on the Finnish and Spanish data. So, maybe one reason is that Finnish and Spanish noise are different.\nSpeaker C: Actually, we listen to some of the utterances, and sometimes for Finnish, there is music in the recordings and strange things.\nSpeaker C: Yeah, so the idea was to train on all the databases and obtain an alignment to train on these databases.\nSpeaker C: Also, to try different kinds of features, as input to the VAD network, we came up with a bunch of features that we want to try.\nSpeaker C: Like the spec-roslow, the degree of voicing with the features that we started to develop with Kermann, with the correlation between bands, and then different kinds of features.\nSpeaker C: Energy, of course.\nSpeaker E: Okay, well, Hans Kuntel will be here next week, so I think he'll be interested in all of these things.\nSpeaker E: Transcript L-209, 4-6-28-893020, 4-22-08-0952, 5-075-121056, 9-37105-2768, 316-727-5311, 7-329-7204-2, 7-646-7011, 5-8152-268.\nSpeaker G: Transcript L-294, 6-02459-228, 05827-34, scratch that, 3746, 7991-6007-1418, 6-6-717-449, 094019-6213, 6-059-78-2606, 8-613-296973, 8-556-71576, 6.\nSpeaker A: Transcript L-281-364-827-3611, 5-364-549408, 1-369-5408, 4-232-35773-8, 1-996-6488-2402, 153-393-89, 9-365-78060, 6-8915-07935.\nSpeaker B: Transcript L-285, 7269-4416, 07604-5454, 1-884-385-387-09, 0535-393-266, 4-909-909109, 6-713-05271-236, 7-832-96, 3-385, 4-757-1276-4975.\nSpeaker C: Transcript L-286, 3-7545146-9, 3-5453-5309, 6-85-2, 5-8-2143-44, 8-265-5808-1, 5-396-1055338, 5-01-195-910, 5-343-117-859, 6-050-1187-391.\n"}, {"meeting_id": "Bro022", "summary": "The meeting began with a discussion on the TORRENT project completion being pushed for two years. Grad F then introduced intermediate categorization, which was his topic for his qualification exams. The team then discussed mean subtraction from SRI. Using it had led to an improvement in Meeting Recorder digits though near mic performance worsened. The professor points to pre-echoes as the culprit. The team continued to study differences between SRI and Aurora. The team thought it would be interesting to do the Aurora tests with the SRI system instead of the HTK. The team was also exploring the Wiener filter and VTS. The professor did not seem too excited about the VTS.", "dialogue": "Speaker E: I think for two years we were two months away from being done.\nSpeaker G: And what was that Morgan?\nSpeaker G: What project?\nSpeaker E: The Taurant chip.\nSpeaker E: Oh yeah.\nSpeaker E: We went through, Jim and I went through all the emails at one point and for two years there was this thing saying yeah we were two months away from being done.\nSpeaker E: It was very believable schedules.\nSpeaker E: We went through and schedule some years.\nSpeaker E: Oh yeah, it was very true.\nSpeaker G: So should we just do the same kind of deal where we go around and do status report kind of things?\nSpeaker G: Okay, and I guess when Sunil gets here he can do his last or something.\nSpeaker E: So we probably should wait for him to come before we do his.\nSpeaker G: Okay, good idea.\nSpeaker G: Any objection?\nSpeaker G: All in favor.\nSpeaker G: Do you want to start Morgan?\nSpeaker G: Do you have anything?\nSpeaker E: I don't do anything.\nSpeaker E: No, I mean I have involved in discussions with people about what they're doing but I think they're since they're here they can talk about it themselves.\nSpeaker D: Okay, so should I go so that you go ahead and talk about where we're at stuff for sake.\nSpeaker D: Okay.\nSpeaker D: Well this past week I've just been getting down and dirty into writing my proposal.\nSpeaker D: I just finished a section on talking about these to really categories that classify as a little step.\nSpeaker D: And I hope to get this full rough draft by Monday and you're tomorrow.\nSpeaker G: When is your meeting?\nSpeaker D: My meeting with oh you mean the cool ones.\nSpeaker D: Are the calls are happening in July 25th?\nSpeaker G: Oh, soon.\nSpeaker G: Yeah, D-Day.\nSpeaker G: Yeah.\nSpeaker G: So the idea you're going to do this paper and then you pass it out to everybody ahead of time.\nSpeaker D: Right, right.\nSpeaker D: So you write up a proposal and give it to people ahead of time and you have a short presentation.\nSpeaker D: And then everybody asks you questions.\nSpeaker G: Yeah, I remember now.\nSpeaker G: Yeah, so I was just going to ask you want to say any a little bit about it or a little bit about what you're going to you said you're talking about the features that you're looking\nSpeaker D: at. Right.\nSpeaker D: Well, I was I think one of the perplexing problems is for a while I was thinking that I had to come up with a complete set of intermediate features and intermediate categories to classify right away.\nSpeaker D: But what I'm thinking now is I would start with a reasonable set something like a regular phonetic features just to just to start off that way.\nSpeaker D: And do some phone recognition, build a system that classifies these these features of these intermediate categories using multi band techniques, combine them and do phone phoneme recognition.\nSpeaker D: Look at then I would look at the errors produced in the phoneme recognition and say, okay, well, I could probably reduce the errors if I included this extra feature or this extra intermediate category that would that would reduce certain confusions over other confusions.\nSpeaker D: And then and then reiterate build the intermediate classifiers and do phoning recognition, look at the errors and then postulate new or remove intermediate categories and then do it again.\nSpeaker G: So you're going to use Timit?\nSpeaker D: Or that for that part of the process here, I used Timit and then after doing Timit, right, that's just the phone recognition task.\nSpeaker D: I want to take a look at things that I can model within Word.\nSpeaker D: So I would then shift the focus to something like switchboard where I would be able to model intermediate categories that span across phonemes and just within the phonemes themselves.\nSpeaker D: And then do the same things as there on a large regular task like switchboard.\nSpeaker D: And for that part, I would use the SRI recognizer.\nSpeaker D: So it's already set up for a switchboard and I run some sort of tandem style processing with intermediate classifiers.\nSpeaker G: So that's why you were interested in getting your own features into the SRI files?\nSpeaker D: That's why I was asking about that.\nSpeaker D: I guess that's it.\nSpeaker D: Any questions?\nNone: Sounds good.\nSpeaker G: You just have a few more weeks.\nSpeaker D: It's about a month from now.\nSpeaker D: It's a month and a week.\nSpeaker G: So you want to go next, David?\nSpeaker G: Oh, OK, sure.\nSpeaker F: So last week I finally got results from the SRI system about this means subtraction approach.\nSpeaker F: And we got an improvement in Word, error rate, training on the TI digits, data set, and testing on meeting recorded digits of 6% to 4.5% on the far mic data using PCMF.\nSpeaker F: But the near mic performance worsened from 1.2% to 2.4%.\nSpeaker F: And why would that be considering that we actually got an improvement in near mic performance using HDK?\nSpeaker F: So with some input from Andreas, I have a theory in two parts.\nSpeaker F: First of all, the SRI system is doing channel adaptation.\nSpeaker F: And so, HDK wasn't.\nSpeaker F: So this means subtraction approach will do a channel normalization.\nSpeaker F: And so that might have given the HDK use of it a boost that wouldn't have applied in the SRI case.\nSpeaker F: And also, the Andreas pointed out the SRI system is using more parameters.\nSpeaker F: It's got finer grain acoustic models.\nSpeaker F: So those finer grain acoustic models could be more sensitive to the artifacts in the recenthasized audio.\nSpeaker F: And me and Barry were listening to the recenthasized audio.\nSpeaker F: And sometimes it seems like you get a bit of an echo of speech in the background.\nSpeaker F: And so it seems like it could be difficult for training because you could have different phones lined up with a different foreground phone, depending on the timing of the echoes.\nSpeaker F: So I'm going to try training on a larger data set.\nSpeaker F: And then the system will have seen more examples of these artifacts.\nSpeaker F: And hopefully, it will be more robust to them.\nSpeaker F: So I'm trying to use the microphone set of red speech.\nSpeaker E: And I have another thought just now, which is, remember we were talking before about, we were talking in our meeting about some of the other stuff that Evernano did where they were getting rid of low energy sections.\nSpeaker E: If you did a high pass filtering as Hirsch did late 80s to reduce some of the effects of reverberation, Evernano and Hermansky were arguing that perhaps one of the reasons for that working was that it may not have even been filtering so much, but the fact that when you filter an all positive power spectrum, you get some negative values and you've got to figure out what to do with them.\nSpeaker E: If you continue treating this as a power spectrum, so what Hirsch did was set them to zero, set the negative values to zero.\nSpeaker E: So if you imagine a waveform that's all positive, which is the time trajectory of energy, and shifting it downwards and then getting rid of the negative parts, that's essentially throwing away the low energy things.\nSpeaker E: That's the low energy parts of the speech where the reverberation is most audible.\nSpeaker E: You have reverberation from higher energy things showing up in.\nSpeaker E: So in this case, you have some artificially imposed reverberation like that.\nSpeaker E: I mean, you're getting rid of some of the other effects of reverberation, but because you have these non-calls or windows, you're getting these funny things coming in.\nSpeaker E: What if you did, I mean, there's nothing to say that the processing for this recynthesis has to be restricted to trying to get it back to the original according to some equation.\nSpeaker E: I mean, you also could just try to make it nicer.\nSpeaker E: And one of the things you could do is you could do some sort of VAD-like thing.\nSpeaker E: You actually could take very low energy sections and set them to some very low or near zero value.\nSpeaker E: I mean, I'm just saying if in fact it turns out that these echoes that you're hearing are pre-echos, whichever they are, are part of what's causing the problem.\nSpeaker E: You actually can get rid of them.\nSpeaker H: Uh-huh.\nSpeaker E: Be pretty simple.\nNone: Okay.\nSpeaker E: You can do it in a pretty conservative way so that if you made a mistake, you were more likely to keep in an echo than to throw out speech.\nSpeaker B: What is the reverberation time like in this room?\nSpeaker B: The one in the speech that you were using?\nSpeaker F: Yeah.\nSpeaker F: I don't know.\nSpeaker F: So is this room?\nSpeaker E: It's this room.\nSpeaker E: So is this just a microphone?\nSpeaker E: This microphone, a close microphone and a distant microphone.\nSpeaker E: He's doing these different tests on.\nSpeaker E: We should do measurement here.\nSpeaker E: I think we never have.\nSpeaker E: I think it's, I would guess, 0.7.8 seconds, RT60, something like that.\nSpeaker E: But it's, you know, it's this room.\nSpeaker E: But the other thing is he's putting in, I was using word reverberation in two ways.\nSpeaker E: He's also putting in a, he's taking out some reverberation, but he's putting in something because he has averages over multiple windows stretching out the 12 seconds, which are then being subtracted from the speech.\nSpeaker E: And since, you know, what you subtract, sometimes you'll be subtracting from some larger number and sometimes you won't.\nSpeaker E: So you can end up with some components in it that are affected by things that are seconds away.\nSpeaker E: And if it's a low energy comp, portion, you might actually hear some funny things.\nSpeaker F: One thing I noticed is that the mean subtraction seems to make the PZM signals louder after they've been recenticized.\nSpeaker F: So I was wondering, is it possible that one reason it helped with the Aurora baseline system is just as a kind of gain control because some of the PZM signals sound pretty quiet if you don't amplify them?\nSpeaker C: I don't see why you're signally louder after processing because you...\nSpeaker F: Yeah, I don't know why either.\nSpeaker E: I don't think just multiplying the signal by two would have any effect.\nSpeaker F: Oh, okay.\nSpeaker E: Yeah.\nSpeaker E: I mean, I think if you really have louder signals, what you mean is that you have better signalized ratio.\nSpeaker E: So if what you're doing is improving the signalized ratio, then it would be better, but just it being bigger with the same signalized ratio.\nSpeaker E: It would have effect.\nSpeaker E: Okay.\nSpeaker C: Well, the system is...\nSpeaker C: You use the absolute energy, so it's a little bit dependent on the signal level, but not so much.\nSpeaker E: Well, yeah, but it's trained and tested on the same thing.\nSpeaker E: So if you change in both training and test, the absolute level by factor too, it will matter no effect.\nSpeaker G: So if you add this data to the training site for the Aurora, or you just test it on this, did I...\nSpeaker F: What, sorry?\nSpeaker G: When we're going to just saying that as long as you do it in both training and testing, you can't have any effect.\nSpeaker G: But I was sort of under the impression that you just tested with this data.\nSpeaker G: You didn't train at all.\nSpeaker F: Right, I trained on clean TI digits.\nSpeaker F: I did the mean subtraction on clean TI digits, but I didn't...\nSpeaker F: I see.\nSpeaker F: But it made the clean TI digits any louder.\nSpeaker F: I only remember noticing it made the PCM signal louder.\nSpeaker E: Okay, well, I don't understand then.\nSpeaker F: I don't know.\nSpeaker F: If it's trying to find a reverberation filter, it could be that this reverberation filter is making things quieter, and then if you take it out, that...\nSpeaker F: Taking it out makes things louder.\nSpeaker E: Are you...\nSpeaker E: No.\nSpeaker E: I mean, there's nothing inherent about removing...\nSpeaker E: If you're really removing...\nSpeaker E: I mean, then I don't see how it makes it louder.\nSpeaker E: So I should maybe let this stuff get...\nSpeaker E: Yeah, it might just be some artifact of the processing that...\nSpeaker E: If you're...\nSpeaker E: Yeah.\nSpeaker E: I don't know.\nNone: Okay.\nSpeaker G: I wonder if there could be something like...\nSpeaker G: For the PCM data, if occasionally somebody hits the table or something, you could get a spike.\nSpeaker G: I'm just wondering if there's something about the...\nSpeaker G: You know, doing the mean normalization where it could cause you to have better signal of noise ratio.\nSpeaker E: Well, you know, there is this, right?\nSpeaker E: It may be...\nSpeaker E: If subtracting the mean log spectrum is like dividing by a spectrum.\nSpeaker E: So depending what you divide by, if your estimate is often sometimes you're getting a small number, you could make it bigger.\nSpeaker E: So it's just a question of...\nSpeaker E: It could be that there's some normalization that's missing or something to make it...\nSpeaker E: You think it shouldn't be larger, but maybe in practice it is.\nSpeaker E: That's something to think about.\nSpeaker H: I don't know.\nSpeaker C: I had a question about the system, the SRI system.\nSpeaker C: So you trained it on TI digits, but except this, it's exactly the same system as the one that was tested before and that was trained on microphone, right?\nSpeaker C: So on TI digits it gives you 1.2% error rate and on microphone it's still 0.8.\nSpeaker C: But is it exactly the same system?\nSpeaker F: I think so.\nSpeaker F: If you're talking with the microphone results that Andreas had about a week and a half ago, I think it's the same system.\nSpeaker C: So you use VTA vocal track length normalization and like MLLR transformation also.\nSpeaker E: I'm sorry, was his 0.8% result on testing on microphone or training?\nSpeaker C: Training on microphone and testing on meeting digits.\nSpeaker E: So that was done already.\nSpeaker E: So it's 0.8.\nNone: Okay.\nSpeaker C: Yeah, I've just been testing the new Aurora front end with...\nSpeaker C: Well, the system actually, so front end and HTK acoustic models on the meeting digits and it's a little bit better than the previous system.\nSpeaker C: We have 2.7% error rate and before with the system that was proposed, it was 3.9.\nSpeaker C: Oh, that's a lot better.\nSpeaker C: Getting better.\nSpeaker B: So with the HTK back in what we have, the Aurora?\nSpeaker C: Yeah, 2.7.\nSpeaker C: On the meeting, we have 2.7.\nSpeaker D: That's with the new IIR shelters.\nSpeaker C: Yeah, yeah.\nSpeaker C: So yeah, we have the new L.E. filters and I think maybe I didn't look, but one thing that makes different is this DC offset compensation.\nSpeaker C: Did you have a look at the meeting digits if they have DC component or...?\nSpeaker F: I didn't know.\nSpeaker B: No, the DC component could benefit from it if you have recording it through a mic coming.\nSpeaker B: All the mics have the DC remote, some capacity to receive right?\nSpeaker E: Yeah, but this...\nSpeaker E: No, because there's a sampling hold in the A to D and these periods these typically do have a DC offset and they can be surprisingly large.\nSpeaker B: It depends on the electronics.\nSpeaker E: The microphone isn't going to pass in a DC, but actually there are instrumentation mics that do pass go down to DC, but no, it's the electronics.\nSpeaker E: And then there's amplification afterwards and you can get...\nSpeaker E: I think it was in the Wall Street Journal data that I can't remember one of the darker things that was this big DC offset.\nSpeaker E: We didn't know about it for a while.\nSpeaker E: We were messing with it.\nSpeaker E: We were getting these terrible results.\nSpeaker E: We're talking to somebody and they said, oh yeah, everybody knows that.\nSpeaker E: There's all this DC offset.\nSpeaker E: So yes, you can get DC offset in the data.\nSpeaker G: Was that everything, Dave?\nSpeaker F: Oh, and I also did some experiments about normalizing the phase.\nSpeaker F: So I came up with a web page that people can take a look at.\nSpeaker F: And the interesting thing that I tried was Adam and Morgan had this idea since my original attempts to take the mean of the phase spectra over time and normalize using that by subtracting that off didn't work.\nSpeaker F: So we thought that might be due to problems with the arithmetic of phases.\nSpeaker F: They added this module to Pi way and there's reason to believe that that approach of taking the mean of the phase spectrum wasn't really mathematically correct.\nSpeaker F: So what I did instead is I took the mean of the FFT spectrum without taking the log or anything.\nSpeaker F: And I took the phase of that and I subtracted that phase off to normalize that didn't work either.\nSpeaker E: So we have a different interpretation of this.\nSpeaker E: He says it doesn't work.\nSpeaker E: I think it works magnificently, but just not for the task we intended.\nSpeaker E: It gets rid of the speech.\nSpeaker E: What does it leave?\nSpeaker E: At least, you know, it leaves the junk.\nSpeaker E: I mean, I think it's tremendous.\nSpeaker E: All he has to do is go back and reverse what he did before.\nSpeaker E: Well, could you take what was left over?\nSpeaker E: Exactly.\nSpeaker E: Yeah, you got it.\nSpeaker E: So it's a general rule.\nSpeaker E: Just listen very carefully to what I say and do the opposite.\nSpeaker E: Including what I just said.\nSpeaker H: And yeah, that's everything.\nSpeaker G: Do you want to go step on?\nSpeaker C: Yeah, maybe concerning this still is meeting digits.\nSpeaker C: I'm interested in trying to figure out what's still the difference between the SRI system and the overall system.\nSpeaker C: Yeah, so I think I will maybe train gender-dependent models because this is also one big difference between the two systems.\nSpeaker C: The other difference is where the fact that maybe the acoustic models of the SRI and the SRI system are more complex.\nSpeaker C: But Chuck, you did some experiment with this.\nSpeaker C: It didn't seem to happen.\nSpeaker C: Did you have some improvement with this?\nSpeaker E: What sounds like they also have, he's saying they have all these tumor kinds of adaptation, their channel adaptation, their speaker adaptation.\nSpeaker G: But there's also the normalization, like they do, I'm not sure how they do it when they're working with the digits.\nSpeaker G: But like in the switchboard, there's conversation side normalization for the non-C0 components and then utterance normalization for the C0 components.\nSpeaker C: Yeah, this is another difference.\nSpeaker C: The normalization works on utterance levels, but we have to do it, we have a system that does it online, so it might be better, it might be worse if the channel is constant.\nSpeaker B: And the acoustic models are like a triple model, so the whole world is a SRI.\nSpeaker C: Yeah, I guess it's tri-fold.\nSpeaker E: I think it's probably more than that.\nSpeaker E: So I think they use these genome things, so there's these kind of pooled models and they can go out to all sorts of dependencies.\nSpeaker B: It's like the tight states.\nSpeaker E: They have tight states and I think, I'm just guessing here, but I think they don't just have tri-phones, I think they have a range of dependencies.\nSpeaker C: Yeah, well, the first thing I want to do is just maybe these gender things.\nSpeaker C: Maybe see with Andreas, I don't know how much it helps.\nSpeaker G: So the stuff on the numbers you got, the 2.7 is that using the same training data that the SRI system used and got 1.2?\nSpeaker C: That's right.\nSpeaker C: So it's a clean DIDGIT training set.\nSpeaker G: So exact same training data.\nSpeaker G: Okay.\nSpeaker C: I guess you used the clean training set.\nSpeaker F: Right.\nSpeaker F: With the SRI system, the Aurora baseline is set up with this version of the clean training set that's been filtered with its D712 filter and to train the SRI system under SUs, the original TI digits under you, Dr. Speech Data, TI digits, which don't have this filter, but I don't think there's any other difference.\nSpeaker E: So are these results comparable?\nSpeaker E: So you are getting with the Aurora baseline, something like 2.4% on clean TI digits when training the SRI system with clean TRI digits, right?\nSpeaker E: And so is your 2.7 comparable where you're using the submitted system?\nSpeaker C: Yeah, I think so.\nSpeaker E: Okay.\nSpeaker E: So it's about the same.\nSpeaker F: It was 1.2 with the SRI system.\nSpeaker C: I'm sorry.\nSpeaker C: Complete SRI system is 1.\nSpeaker E: You were HDK.\nSpeaker E: Right.\nSpeaker E: Okay.\nSpeaker E: That's right.\nSpeaker E: Okay.\nSpeaker E: So the comparable number than for what you were talking about, then SRI would be the 2.5.\nSpeaker C: It was 4.0 something, right?\nSpeaker C: The HDK system with the SRI system, right?\nSpeaker F: The baseline Aurora 2 system, trained on TI digits, tested on meeting record in year.\nSpeaker F: I think we saw it today and it was about 6.6%.\nSpeaker E: Right.\nSpeaker E: Right.\nSpeaker E: So, yeah.\nSpeaker E: It's different between this and this.\nSpeaker C: Okay.\nSpeaker E: Good.\nSpeaker E: So they are helping.\nSpeaker E: That's good.\nSpeaker E: Yeah.\nSpeaker C: And another thing I maybe like to do is to just test the SRI system that's trained on microphone, tested on the NoECTI digits.\nSpeaker C: I'm still wondering where this improvement comes from when you train on microphone, it seems better on meeting digits.\nSpeaker C: But I wonder if it's just because maybe microphone is acoustically closer to the meeting digits than TI digit is, which is, TI digits are very clean recorded digits.\nSpeaker G: It would also be interesting to see, to do the regular Aurora test, but use the SRI system instead of the same.\nSpeaker C: Yeah, that's what I wanted just to, yeah.\nSpeaker C: So just using the SRI system, test, did some, and tested on Aurora TI digits, right?\nSpeaker G: When at the full Aurora test?\nSpeaker C: Yeah, there is this problem of multilinguality, so we don't, you have to train the SRI system\nSpeaker E: with all the different languages.\nSpeaker G: Yeah, that's what I mean. So you'll have to work.\nSpeaker G: Yeah.\nSpeaker G: Well, I mean, I guess the work would be into getting the files in the right formats or something, right?\nSpeaker G: Yeah, because when you train up the Aurora system, you're also training on all the data.\nSpeaker G: That's right.\nSpeaker G: Yeah, I mean, it's...\nSpeaker C: Yeah, yeah, yeah, yeah, yeah, yeah, yeah.\nSpeaker C: Right, I see what you mean.\nSpeaker E: That's true, but I think that also when we've had these meetings, we have to wake off in times people have not done the full range of things because on whatever is they're trying because it's a lot of work, even just with the HTK.\nSpeaker E: So it's a good idea, but it seems like it makes sense to do some pruning first with a test or two that makes sense for you and then take the likely candidates to go further.\nSpeaker C: But just testing on TI digit already gave us some information about what's going on.\nSpeaker C: Oh, yeah, okay.\nSpeaker C: The next thing is this VAD problem, but...\nSpeaker C: So I'm just talking about the curve that I sent.\nSpeaker C: I sent you.\nSpeaker C: So that shows that when the SNR decreases, the current VAD approach doesn't drop much frames for some particular noises, which might be the noises that are closer to speech.\nSpeaker E: Just clarify something for me.\nSpeaker E: They were supposedly in the next evaluation, they're going to be supplying us with boundaries.\nSpeaker E: So does any of this matter?\nSpeaker E: I mean, other than our interest in it.\nSpeaker C: Well, first of all, the boundaries might be like we would have 200 milliseconds before an after speech.\nSpeaker C: So removing more than that might still make a difference in the results.\nSpeaker E: Do we...\nSpeaker E: I mean, there's some reason that we think that's the case.\nSpeaker C: No, because we don't...\nSpeaker C: That much at that.\nSpeaker C: Still, I think it's an interesting problem.\nSpeaker C: Oh, yeah.\nSpeaker C: Yeah.\nSpeaker E: But maybe we'll get some insight on that when the gang gets back from CREAT because there's lots of interesting problems, of course.\nSpeaker E: And the thing is, if they really are going to have some means of giving us fairly tight boundaries, then that won't be so much the issue.\nSpeaker B: Yeah, so we were wondering whether that VAD is going to be like a realistic one or is it going to be some manual segmentation?\nSpeaker B: And they're like, if that VAD is going to be a realistic one, then we can as well use their markers to shift the point around, I mean, the way we want to find the...\nSpeaker B: Rather than keeping the 20 frames, we can actually go to the markers point, which we find more suitable for us.\nSpeaker B: But that is going to be something like a manual segmenter.\nSpeaker B: Then we couldn't use that information anymore because that's not going to be the one that is used in the final evaluation.\nSpeaker B: Right.\nSpeaker B: So we don't know what is the type of VAD, if they're going to provide.\nSpeaker C: And actually, there's...\nSpeaker C: Yeah, there's...\nSpeaker C: I think it's still for...\nSpeaker C: Even for the evaluation, it might still be interesting to work on this because the boundary is a part that they would provide is just starting off speech and end of speech at the utterance level.\nSpeaker B: With some gap, I mean, with some passes in the center, provided they meet that order at the hangover time, which is...\nSpeaker C: Yeah, but when you have like five or six frames, both...\nSpeaker B: Yeah, they'll just fill it up.\nSpeaker B: It twists, yeah.\nSpeaker B: Yeah.\nSpeaker E: So if you could get at some of that, all of that papers.\nSpeaker C: Yeah, for like, not this animation and not other things that we want to work on.\nNone: Okay.\nNone: But...\nSpeaker C: Yeah, so I did...\nSpeaker C: I just started to test putting together two VAD, which was not much work, actually.\nSpeaker C: I implemented VAD that's very close to the energy-based VAD that the other or guys use.\nSpeaker C: So, which is just putting a threshold on the noise energy, detecting the first group of four frames that have energy that's about this threshold.\nSpeaker C: From this point, tagging the frames are a speech.\nSpeaker C: So it removes the first silent portion of each utterance.\nSpeaker C: And it really removes it.\nSpeaker C: Still, on the noises where our MADVAD doesn't work a lot.\nSpeaker E: And so I would have thought that having some kind of spectral information, you know, when you'll days people would use energy in zero crossings, for instance, would give you some better performance, right?\nSpeaker E: Because you're meant of low energy, fricatives, or stop consonants or something like that.\nSpeaker C: Yeah.\nSpeaker C: So your point is way between...\nSpeaker E: Oh, that if you use purely energy and don't look at anything spectral, then you don't have a good way of distinguishing between low energy speech components and non-speech.\nSpeaker E: And just as a gross generalization, many non-speech noises have a low-pass kind of characteristic, some sort of slope.\nSpeaker E: And most low energy speech components that are unvoiced have a high-pass kind of characteristic and upward slope.\nSpeaker E: So having some kind of a, you know, the beginning of an S sound, for instance, just starting in, it might be pretty low energy, but it will tend to have this high frequency component whereas a lot of rumble and background noises and so forth will be predominantly low-fruicing.\nSpeaker E: You know, by itself, it's not enough to tell you, but it plus energy is sort of...\nSpeaker E: It plus energy plus timing information is sort of...\nSpeaker E: And if you look up in Rebeanor and Schaeff for like 25 years ago, or something, that's sort of what they were using then.\nSpeaker C: So it might be that what I did is...\nSpeaker C: So it removes like low energy speech frames because the way it is said just combined the two decisions, so the one from the MLB and the one from the energy based is with the end operator.\nSpeaker C: So I only keep the frames where to agree that it's speech.\nSpeaker C: So if the energy based dropped, dropped low energy speech, they are lost.\nSpeaker C: But still the way it's done right now, it helps on the noises where it seems to help on the noises where RVU was not very good.\nSpeaker E: Well, I guess I mean, one could imagine combining them in different ways, but I guess what you're saying is that the MLB based one has the spectral information.\nSpeaker C: Yeah, so...\nSpeaker C: But the way it's combined is maybe the...\nSpeaker C: Well, you can imagine...\nSpeaker C: The way you use the end operator is...\nSpeaker C: The frames that are dropped by the energy based system are dropped even if the MLB decides to give them.\nSpeaker E: Right, and that might not be optimal, but I mean, I guess principle what you'd want to do is have a probability estimated by each one and put them together.\nSpeaker G: Something that I've used in the past is when just looking at the energy is to look at the derivative and you make your decision when the derivative is increasing for so many frames, then you say that's beginning of speech.\nSpeaker G: But I'm trying to remember if that requires that you keep some amount of speech and a buffer.\nSpeaker G: I guess it depends on how you do it, but that's been a useful thing.\nSpeaker B: I mean, you're everywhere as a delay associated with it, you still have to keep the buffer.\nSpeaker B: There only make a decision because it's still needed to smooth the decision further.\nSpeaker B: That's always there.\nSpeaker G: Yeah, okay.\nSpeaker C: Well, actually, I don't maybe don't want to work too much on it right now.\nSpeaker C: I just wanted to see if it's what I observed as a ghost by this really problem.\nSpeaker C: It seems to be the case.\nSpeaker C: The second thing is the spectral subtraction, which I just started yesterday to launch a bunch of 25 experiments with different values for the parameters that are used.\nSpeaker C: So it's the Michael type spectral subtraction which use an overestimation factor.\nSpeaker C: So I subtract more noise than the noise spectra that is estimated on the noise portion of the utterances.\nSpeaker C: Should I try several overestimation factors?\nSpeaker C: And after subtraction, I also add a constant noise and I also try different noise values.\nSpeaker C: And we'll see what happened.\nSpeaker C: But still, when we look at the value depends on the parameters that you use.\nSpeaker C: More moderate overestimation factors and moderate noise level that you add, you have a lot of musical noise.\nSpeaker C: On the other hand, when you subtract more and when you add more noise, you get rid of this musical noise, but maybe you distort a lot of speech.\nSpeaker C: Well, until now it doesn't seem to have.\nSpeaker C: So the next thing maybe I will try to do is just to try to smooth the smooth the result of the subtraction to get rid of the musical noise using some kind of filter.\nSpeaker B: Can smooth the SNR estimate also?\nSpeaker C: Yeah, right.\nSpeaker B: You filter is a function of SNR.\nSpeaker C: Yeah.\nSpeaker C: So to get something that would be closer to what you try to do with inner filtering.\nSpeaker C: Yeah.\nSpeaker C: Actually, maybe you can, I think, let's see it for me.\nSpeaker B: So I've been playing with this inner filter like, and there were some bugs in the programs I was initially trying to clear them up.\nSpeaker B: Because one of the bugs was, I was assuming that always the bad initial frames were silence.\nSpeaker B: It always started in the silence state, but it was in for some utterances.\nSpeaker B: So it was in estimating the noise initially and then it never estimated because I assumed that it was always silence.\nSpeaker C: So this is on speech at Karyatalian?\nSpeaker B: Yeah, speech at Karyatalian.\nSpeaker B: So in some cases, there are a few cases actually, which I found later.\nSpeaker B: So that was one of the bugs that was there in estimating the noise.\nSpeaker B: And so once we were clear, the ran a few experiments with different ways of smoothing the estimated clean speech and how estimated the noise and smoothing of the SNR also.\nSpeaker B: And so the trend seems to be like smoothing the current estimate of the clean speech for deriving the SNR, which is like deriving the inner filter, seems to be helping then updating it quite fast using a very small time constant.\nSpeaker B: So I have like few results where the estimating the most smoothing is helping.\nSpeaker B: But still it's like it's still comparable to the baseline.\nSpeaker B: I haven't caught anything beyond the baseline, but that is like not using any inner filter.\nSpeaker B: And so I'm trying a few more experiments with different time constants for smoothing the noise spectrum and smoothing the clean speech and smoothing SNR.\nSpeaker B: So there are three time constant that I have.\nSpeaker B: So I'm just playing around.\nSpeaker B: So one is fixed in the line like smoothing the clean speech is helping.\nSpeaker B: So I'm not going to change that much.\nSpeaker B: But the way I'm estimating the noise and the way I'm estimating the SNR, I'm just trying a little bit.\nSpeaker B: So that and the other thing is like putting a floor on the SNR because that if some in some cases the clean speech is like when it's estimated it goes to very low value.\nSpeaker B: So the SNR is like pretty low.\nSpeaker B: So that actually creates a lot of variance in the low energy region of the speech.\nSpeaker B: So thing of like putting a floor also for the SNR so that it doesn't vary a lot in the low energy regions.\nSpeaker B: And so the results are like so far I've been testing only with the baseline which is which doesn't have any LDA filtering and online analysis that I just want to separate that the contributions out.\nSpeaker B: So it's just VAD plus, the VINOR filter plus the baseline system which is just the spectral I mean the male frequency coefficient.\nSpeaker B: And the other thing that I tried was by I just took one of those Carlos filters which he needed had to see whether it really helps or not.\nSpeaker B: It just ran to see whether it really degrades all of it helps.\nSpeaker B: It seems to be like it's not hurting a lot by just blindly picking up one filter which is nothing but a four hertz a band pass filter on the cubic root of the power spectrum.\nSpeaker B: So that was the filter that Carlos had.\nSpeaker B: And so just to see whether it really it's worth trying around.\nSpeaker B: So it doesn't seem to be degrading a lot on that.\nSpeaker B: So there must be something that I can be done with that type of noise compensation also because I have to ask Carlos about that.\nSpeaker B: I mean how he derived those filters and where he has any filters which are derived on co-GIS stories added with some type of noise which what we are using currently of something that so maybe we have.\nSpeaker E: This is cubic root of power spectrum.\nSpeaker B: Yeah, cubic root of power spectrum.\nSpeaker E: So if you have this band pass filter you probably get negative values right?\nSpeaker B: Yeah, and I'm like so it has like the spectrogram has like it actually enhances the onset and the offset of I mean the beginning and the end of the speech.\nSpeaker B: So it seems to be like deep valleys and the beginning and the end of like high energy regions because the filter has like a sort of mix again high type structure.\nSpeaker B: So those are the regions where the like when I look at the spectrogram there are those deep valleys on the beginning and the end of the speech but rest of it seems to be like pretty nice.\nSpeaker B: So that's something I observed using that filter and yeah there are few very not a lot of because the filter doesn't have a really deep negative portion so that it's not really creating a lot of negative values in the cubic root.\nSpeaker B: So I'll just continue with that for some maybe I'll ask Carlos later more about how to play with those filters and while making this we have that yeah that's it.\nSpeaker E: Last week you were also talking about building up the subspace.\nSpeaker B: Yeah I would actually didn't get enough time to work on the subspace last week was mostly about finding those bugs and yeah thanks.\nSpeaker B: I think about much on that.\nSpeaker G: How about you coming?\nSpeaker A: Well I'm still working with BTS and one of the things that last week say here that maybe the problem was with the deep because the signal have different level of energy and maybe talking with Stefan and with Sunil we decide that maybe it was interesting to apply your line normalization before applying BTS but then we decided that it doesn't work absolutely because we modified also the noise.\nSpeaker A: And well thinking about that with then we decide that maybe it's a good idea we don't know.\nSpeaker A: I didn't do the experiment yet to apply BTS in the castral domain.\nSpeaker E: The other thing is so and not and C0 would be a different so you could do different normalization for C0 than for other things anyway.\nSpeaker E: I mean the other thing I was going to suggest is that you could have two kinds of normalization with different time constants so you could do some normalization before the BTS and then do some other normalization after.\nSpeaker E: But C0 certainly acts differently than the others do so that's.\nSpeaker A: Well we decided to obtain the new expression if we work in the castral domain.\nSpeaker A: Well I am working in that now but I'm not sure if that would be useful.\nSpeaker A: I don't know it's quite a lot of work.\nSpeaker A: It's not too much but it's work and I want to know if we have some feeling that the result I would like to know if I don't have any feeling if this will work better than applying BTS in the castral domain will work better than applying in the filter packet of mine.\nSpeaker A: I'm not sure.\nSpeaker A: I don't know absolutely nothing.\nSpeaker E: Yeah well I think you're the first one here to work with BTS so we could call someone else up who has asking their opinion.\nSpeaker E: I don't have a good feeling for it.\nSpeaker C: Actually the BTS that you tested before was in the log domain and the code book is kind of dependent on the level of the speech signal.\nSpeaker C: So I expect it if you have something that's independent of this, I expect it to be a better model of speech.\nSpeaker E: You wouldn't even need to switch the capture right?\nSpeaker C: Just normalize the mean.\nSpeaker C: Remove the mean.\nSpeaker E: Yeah then you have one number which is very dependent on level because it is the level and the other which is not.\nSpeaker C: But here also we would have to be careful about removing the mean of speech and not of noise because it's like first doing general normalization and then noise removal which is.\nSpeaker A: Yeah we were thinking to estimate the noise with the first frames and then apply the VAD before the normalization.\nSpeaker A: I am thinking about that and working about that but I don't have this way.\nSpeaker E: I mean one of the things we talked about maybe might be time to start thinking about pretty soon is we look at the pros and cons of these different methods.\nSpeaker E: How do they fit in with one another because we talked about potentially doing some combination of a couple of them.\nSpeaker E: Maybe we may be pretty soon we'll have some sense of what their characteristics are so we can see what should be combined.\nSpeaker E: Is that it?\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: We read some digits.\nSpeaker E: Yeah.\nSpeaker G: Want to go ahead and mark?\nNone: Sure.\nNone: Transcript.\nSpeaker E: LDASH 212.\nSpeaker E: 248829109.\nSpeaker E: 7061782596.\nSpeaker E: 858779619.\nSpeaker E: 48441752.\nSpeaker E: 2462721821.\nSpeaker E: 35251.\nSpeaker E: 50381177.\nSpeaker E: 697496315.\nSpeaker C: Transcript LDASH 213.\nSpeaker C: 2454437657.\nSpeaker C: 318 O2O997.\nSpeaker C: 317388313.\nSpeaker C: 071724153985.\nSpeaker C: 0989062173.\nSpeaker C: 2995696699.\nSpeaker C: 859946451.\nSpeaker C: 2453759935.\nSpeaker D: Transcript LDASH 2147539591352.\nSpeaker D: 53510101067.\nSpeaker D: 170437155.\nSpeaker D: 924681.\nSpeaker D: 0710147667915.\nSpeaker D: 219220956097.\nSpeaker D: 0250858405.\nSpeaker D: 34237491.\nSpeaker G: Transcript LDASH 215.\nSpeaker G: 357165459919.\nSpeaker G: 552085225.\nSpeaker G: 137988958.\nSpeaker G: 056540589.\nSpeaker G: 34565268.\nSpeaker G: 30182868.\nSpeaker G: 8394407633.\nSpeaker G: 507932030.\nSpeaker B: Transcript LDASH 216.\nSpeaker B: 020132950.\nSpeaker B: 013809626.\nSpeaker B: 0274407591.\nSpeaker B: 2472293233.\nSpeaker B: 745114817.\nSpeaker B: 6918953964.\nSpeaker B: 66763799.\nSpeaker B: 51700707811.\nSpeaker F: Transcript LDASH 217.\nSpeaker F: 723954211.\nSpeaker F: 9238523639.\nSpeaker F: 047217119.\nSpeaker F: 315592660584.\nSpeaker F: 598848386934.\nSpeaker F: 851149017.\nSpeaker F: 03006587.\nSpeaker F: 532425788.\nSpeaker A: Transcript LDASH 218.\nSpeaker A: 8896023667.\nSpeaker A: 7807172039.\nSpeaker A: 064322851.\nSpeaker A: 51559961.\nSpeaker A: 4378766857.\nSpeaker A: 036926029.\nSpeaker A: 381558434.\nSpeaker A: 415501528.\nSpeaker A: 8378.\nNone: 8378.\nNone: 8378.\nNone: 8378.\n"}, {"meeting_id": "IS1001b", "summary": "As an extension to the previous discussion, this meeting basically dealt with the functional design of the remote control, for which User Interface, Marketing and Industrial Design respectively gave presentations on user requirement specification, subjects' main frustrations and expectations for remote controls and technical design. Subsequent to individual presentations, Project Manager then presented the new requirements from the management board, based on which the decisions on the needed functions of the remote control were made and the pre-arrangement of the next meeting was subsequently told.", "dialogue": "Speaker A: Okay, everyone's ready.\nSpeaker A: Hello.\nSpeaker A: So we are here for our functional design.\nSpeaker A: Okay.\nSpeaker A: So we are here for the functional design meeting.\nSpeaker A: So first I will show the agenda.\nSpeaker A: So we will, I will take notes during this meeting.\nSpeaker A: So I will try to summarize it and put that somewhere in the shed folder if you want to look at it afterwards.\nSpeaker A: So then each of you will lead the presentation on the task that has been required last time.\nSpeaker A: So user requirements, specification, technical function design and working design.\nSpeaker A: Then I will present you some new project requirements I received from the management board.\nSpeaker A: Then we will take the decision on the remote control needed functions.\nSpeaker A: And then I will assign you to task for the next part of the meeting, of the process.\nSpeaker A: So we want to start the presentation of what the...\nSpeaker B: Do you want to start?\nSpeaker D: I can start.\nSpeaker D: Should be my folder, no?\nSpeaker A: Yeah, maybe there.\nSpeaker D: Where are you?\nSpeaker F: I'm still here.\nSpeaker A: I'm going to go out.\nSpeaker C: Okay.\nSpeaker A: There is a technical problem.\nSpeaker D: Do you think we said it in the wrong folder?\nSpeaker D: Maybe.\nSpeaker D: Is it possible?\nSpeaker D: You put it somewhere in something like this.\nSpeaker D: I don't remember the name of something like messenger army or something.\nSpeaker B: What do you have in shortcut?\nSpeaker B: Go up.\nSpeaker B: Participant too.\nSpeaker D: Yeah, go up.\nSpeaker C: Yeah.\nSpeaker D: Again?\nSpeaker D: No.\nSpeaker D: We are back.\nSpeaker D: We have no...\nSpeaker D: Maybe messenger army.\nSpeaker D: Messenger.\nSpeaker D: Over?\nNone: Okay.\nSpeaker D: No, there is nothing.\nNone: There is no...\nSpeaker A: We have a technical problem.\nSpeaker D: Yeah.\nSpeaker B: Otherwise, could you just describe by hand with the white ball?\nSpeaker A: Yeah.\nSpeaker A: If you remember, yeah.\nSpeaker D: Yeah.\nSpeaker D: So basically, what we want here is a remote control, right?\nSpeaker D: Yeah.\nSpeaker D: So the question goes first of all, what to control?\nSpeaker D: So most people want to have a remote for the high-five and TV and stuff like that.\nSpeaker D: But other people want also remote for controlling toys like robotic pets and robots and stuff.\nSpeaker D: And other people also want to have remotes for controlling the whole house.\nSpeaker D: Yeah.\nSpeaker D: So there is a project I think called X-Houd or something like that.\nSpeaker D: That does that.\nSpeaker D: You can integrate your remote with computers and stuff.\nSpeaker D: So there is one, not one thing.\nSpeaker D: The other thing is the finder feature by whistling or whatever.\nSpeaker D: If you have the finder feature, then you can also have at the same time, general voice commands if you want.\nSpeaker D: Yeah.\nSpeaker D: So I think it would be a package in that case.\nSpeaker D: So the user in the Facebook consists of two parts.\nSpeaker D: One is the voice command part and one is the actual buttons part.\nSpeaker D: The buttons part would be a set of buttons for choosing devices, a set of buttons for spatial navigation and space, a set of buttons for linear access of medium and a set of buttons for random access.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: What do you mean by linear access?\nSpeaker D: Like a video tape goes forward backwards, fast and stuff.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker A: So spatial navigation, linear access, random access and there is a false one.\nSpeaker A: So the button offers spatial navigation?\nSpeaker D: Yeah.\nSpeaker D: For spatial navigation, for example, you might have a TV in the menu and you're going to\nSpeaker A: see it. Okay.\nSpeaker A: Then linear access, then random access.\nSpeaker D: Yeah.\nSpeaker D: And also parameter.\nSpeaker A: Ah, yeah.\nSpeaker D: For a meter, okay.\nSpeaker D: So if there are common parameters, maybe we should put spatial buttons for that.\nSpeaker D: Okay.\nSpeaker D: Or maybe we could have everything generic, but there are a lot of remotes on the market right now.\nSpeaker D: Okay.\nSpeaker D: So basically, this is most of the, almost everybody has the stuff.\nSpeaker A: Okay.\nSpeaker A: And voice command, did you?\nSpeaker D: Voice command, we could specify anything.\nSpeaker D: We could assign a command to any button.\nSpeaker D: Okay.\nSpeaker D: If we have enough processing power, I guess.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: So that's, that close your investigations?\nSpeaker A: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I think so.\nSpeaker A: Okay.\nSpeaker A: Maybe we can have a look at the user requirements with.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: I don't know if you can open.\nSpeaker E: I don't know if I can open it.\nSpeaker E: Maybe you can.\nSpeaker A: It's not here.\nSpeaker E: It's in, okay.\nSpeaker E: It's in the same room.\nNone: Yeah.\nSpeaker E: Yeah.\nSpeaker A: In which folder?\nSpeaker E: You could do.\nSpeaker E: Here.\nSpeaker E: Here.\nNone: Here.\nNone: Short cut to random.\nSpeaker E: But it's not.\nSpeaker E: Yeah.\nNone: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nNone: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nNone: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Maybe you can send it to me by email just to participant one.\nSpeaker A: At AME.\nNone: Yeah.\nSpeaker A: I can do that.\nSpeaker A: I'll try to show it to everyone.\nSpeaker A: That would be more of a whatever.\nSpeaker E: Okay.\nSpeaker A: You said it?\nSpeaker E: It's participant one.\nSpeaker A: Yeah.\nSpeaker A: This is the email.\nSpeaker D: I'm designing the, you can run my face.\nNone: Okay.\nNone: Okay.\nSpeaker E: You can.\nSpeaker C: Okay.\nSpeaker A: So maybe I can switch slides whenever you ask.\nSpeaker A: That would be more convenient.\nSpeaker A: So okay.\nSpeaker A: Okay.\nSpeaker E: So you can go.\nSpeaker E: Okay.\nSpeaker E: So in our usability lab, we observed the remote control use among 100 subjects.\nSpeaker E: And this subject also filled a questionnaire.\nSpeaker E: Okay.\nNone: Yeah.\nSpeaker E: And yeah, here are the results.\nSpeaker E: So you can see that 75% of users find most remote controls ugly.\nSpeaker E: So we have to find something to make them more nice, more kind.\nSpeaker E: 80% of users would spend more money when a remote control would look fancy.\nSpeaker E: 80% of users would spend more money when a remote control would look up.\nSpeaker E: It's not good.\nSpeaker E: So okay.\nSpeaker A: So it's not really good.\nSpeaker E: But I can say, yeah, 50, 75% of users say they zap a lot.\nSpeaker E: So we have to have a remote control very hard for that.\nSpeaker E: The buttons have to be like to say, resisting to shocks.\nSpeaker E: And 50% of users say they only use 10% of the buttons in the remote control.\nSpeaker E: So all the buttons we have to put are, have to have a use, a real use and not a medium.\nSpeaker A: Okay.\nSpeaker A: So fewer buttons may be good.\nSpeaker E: But many buttons and, and you use able buttons.\nSpeaker D: Okay.\nSpeaker D: But what kind of remote controls do you think that?\nSpeaker D: Sorry.\nSpeaker D: What kind of task was it?\nSpeaker D: It was a TV.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Most, most is TV.\nSpeaker A: Yeah.\nSpeaker A: But in fact, we, it seems that we are going to make a TV remote control according to new requirements I receive from the management board.\nSpeaker A: I will present them in the following.\nSpeaker A: Ah, good.\nSpeaker E: Okay.\nSpeaker E: You can go.\nSpeaker E: So there are the frustrations expressed by users.\nSpeaker E: So they said they lost, uh, often the remote control in the room.\nSpeaker E: So they want to have a way to, to find it.\nSpeaker A: To find it.\nSpeaker E: Yeah.\nSpeaker E: Um, and, um, a lot of the time they, it takes too much time to learn how to use a new remote control.\nSpeaker E: So they want something very, very simple and easy to use.\nSpeaker E: And the remote controls are bad for.\nSpeaker E: What is the other side?\nSpeaker E: Um, other side.\nSpeaker D: Yeah.\nSpeaker D: Yeah, you're, you're a wrist can become painful.\nSpeaker D: You can have talent in this.\nSpeaker A: Ah, yeah.\nSpeaker A: Yeah.\nSpeaker D: I didn't knew that.\nSpeaker D: If you also fit up on the computer in a strange position.\nSpeaker A: Okay.\nSpeaker A: So we have to make it, uh, more ergonomic.\nSpeaker D: Yeah.\nSpeaker D: But, uh, yeah.\nSpeaker A: Have to say, it's your job.\nSpeaker A: Oh.\nSpeaker A: Uh, sorry.\nSpeaker A: But the message from Microsoft.\nNone: Okay.\nSpeaker E: Before that, I have some, some, some, some thing.\nSpeaker E: Yeah.\nSpeaker E: Uh, to say before.\nSpeaker E: Yeah.\nSpeaker E: Um, we know that, uh, that as I use, uh, a lot, uh, remote control.\nSpeaker E: Um, to, to change channel.\nSpeaker E: Yeah.\nSpeaker E: Um, and, um, to, to change the volume selection of the.\nSpeaker E: Okay.\nSpeaker E: And, uh, and not, uh, a lot for setting, for setting the, the channels and, uh, things, things like that.\nSpeaker E: Okay.\nSpeaker E: Um, it's better to put, uh, uh, something very easy to that and, uh, yeah.\nSpeaker A: This function should be very accessible.\nSpeaker A: Accessible, yes.\nSpeaker A: Okay.\nSpeaker A: This is the main function.\nSpeaker A: Okay.\nSpeaker E: So then we ask some questions to them and, um, we ask this question.\nSpeaker E: If they prefer an LCD screen or on the remote function remote control and if they, um, pay more for a speech recognition in remote control and you can go, we have here the results of the first question of the questions.\nSpeaker E: So you know that, um, for the younger, it's very important to have this and speech recognition.\nSpeaker E: And the others is not so important, but, uh, we know that, uh, uh, people between 15 and 25 are people who watch a lot TV and who, who can use a lot this, uh, okay.\nSpeaker E: So maybe we, we can have a speech recognition.\nSpeaker A: Yeah.\nSpeaker A: Maybe this is important.\nSpeaker A: Yeah.\nSpeaker A: Maybe those, uh, like those teenager customer can, uh, advise their parents to buy this equipment.\nSpeaker A: And so we can, we have to take care of that point of view, I think, or so.\nSpeaker E: Mm.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: And after is the conclusion.\nSpeaker E: So as we say before, I think, uh, uh, uh, remote control lightning in the dark.\nSpeaker E: It's, it's a good thing.\nSpeaker E: Okay.\nSpeaker E: So many, many buttons like we, we said before, easy to use, uh, a way to find it easily in the room and, uh, resistance to, to short camp to, to.\nSpeaker C: Okay.\nNone: And I, okay.\nSpeaker E: These are.\nSpeaker E: I don't know if you see something else, important.\nSpeaker B: I'm just thinking of something.\nSpeaker B: Yeah.\nSpeaker B: Um, we want to have a, no, I don't know if this is a good idea.\nSpeaker B: We want to have a general remote control for everything.\nSpeaker A: No, no, no, no.\nSpeaker A: We, it seems that we know want to have a TV remote control.\nSpeaker A: Okay.\nSpeaker A: From the management port I receive an email because it would be costly.\nSpeaker A: Uh, and, and also it would take more time to develop to have a general generic remote\nSpeaker D: control. I, it's not true.\nSpeaker D: I think that it would be too long to develop.\nSpeaker D: I think that should be the same.\nSpeaker A: Oh yeah.\nSpeaker A: Because I received that email from management board and this seems to tell that, that if we want to be on the market as early as possible, we should focus on TV more.\nSpeaker A: It seems that the market is more important.\nSpeaker A: So maybe it's a good decision.\nSpeaker D: I don't know what's your opinion.\nSpeaker D: I'm not the, I'm gonna, I should know a bit more about how fast we can design.\nSpeaker D: I don't think.\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Finished tonight.\nSpeaker B: Okay.\nSpeaker B: But basically, yeah, maybe I can continue with my, my opinion.\nSpeaker B: My presentation would be okay.\nSpeaker B: But I think we have some technical problems.\nSpeaker B: So I'm just going to describe briefly what we do in the remote control.\nSpeaker A: Maybe you can go to the whiteboard if you have some drawings to do.\nSpeaker B: Do I have, oh yeah.\nSpeaker B: I have enough cables.\nSpeaker D: Okay.\nSpeaker D: I feel like a dog is.\nSpeaker B: Okay.\nSpeaker B: So I'm just going to describe in fact for, for a remote control.\nSpeaker B: This is quite easy.\nSpeaker B: We just have, sorry.\nNone: Okay.\nSpeaker B: Are you okay?\nSpeaker B: Like that.\nSpeaker B: I'm just going to describe.\nSpeaker B: Basically we have a battery, a power supply here.\nSpeaker B: After that we just have a user interface.\nSpeaker B: Let's say that something like that, which could be LCD, let's say, or an array of push button, something like that.\nSpeaker B: Push button or LCD.\nSpeaker B: After that we, we feed that into an electronic chip.\nSpeaker B: So I say new, see, and I feed that to LED, which is infrared, which is an infrared component.\nSpeaker B: And so what we, for myself, for us, this is quite easy.\nSpeaker A: You see the central unit?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: It's just a chip which does all the numerical computation, numerical computation according to you display.\nSpeaker B: And so for us, this is quite easy.\nSpeaker B: We just need to define what we want to do when the user interface wants something.\nSpeaker B: And after that we just do the coding to send that to the, not the, to the television.\nSpeaker B: Okay.\nSpeaker B: So for us, this is quite easy.\nSpeaker A: Okay.\nSpeaker A: So this is quite easy.\nSpeaker A: Yeah.\nSpeaker B: And we just have to define the processing power that we need, especially if we want to do some speech recognition in that case, that we are going to use more.\nSpeaker B: This will take more time to develop.\nSpeaker B: Yeah, of course.\nSpeaker B: Of course.\nSpeaker B: But for a standard one, this is really easy.\nSpeaker B: It's a question of one month and so on.\nSpeaker A: Do you have a, you speak about voice control?\nSpeaker A: No, no, no.\nSpeaker A: I say standard button one.\nSpeaker B: Yeah.\nSpeaker B: So the standard remote control takes maybe one month to do that.\nSpeaker D: So the only time problem is the voice recognition.\nSpeaker D: Yeah, definitely.\nSpeaker A: So do you have any idea of how long it would take to have voice recognition?\nSpeaker B: No, I would say.\nSpeaker B: Yeah.\nSpeaker B: It's about eight months to have the first results.\nSpeaker A: Okay.\nSpeaker A: So it's a bit long.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nNone: I can.\nSpeaker A: One month for the standard one with button.\nSpeaker A: Yeah.\nSpeaker A: Even if we have a LCD display.\nSpeaker B: Yeah, even.\nSpeaker B: I mean, this is really standard devices now.\nSpeaker B: And eight for speech recognition.\nSpeaker B: Okay, yeah.\nSpeaker A: Okay.\nSpeaker A: So we can take this into account.\nSpeaker A: So we think it would be good to go for like speech recognition.\nSpeaker D: But we don't have time to market.\nSpeaker B: And also how much?\nSpeaker B: I think.\nSpeaker B: I think we should.\nSpeaker B: During the kickoff meeting, you said that we shouldn't, we shouldn't go up to.\nSpeaker B: 12.5 euros.\nSpeaker B: Yeah.\nSpeaker B: So how many units should we sell to ever?\nSpeaker A: Well, each unit is sell 25 euros.\nSpeaker B: Yeah, but how many?\nNone: Yeah.\nSpeaker D: How much do you get?\nSpeaker D: How much do you buy one million units?\nSpeaker D: How much do you get?\nSpeaker D: Yeah.\nSpeaker D: How much do you cost for one hundred thousand?\nSpeaker B: Usually, I get this is less than that two dollars per chip.\nSpeaker D: Okay.\nSpeaker D: Because we are.\nSpeaker D: Because we are for a powerful one that has a good enough for a speech recognition.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker A: We can.\nSpeaker A: So.\nSpeaker A: It seems that we want to sell like four million units from the first meeting.\nSpeaker A: Okay.\nSpeaker D: Form a little.\nSpeaker D: Okay.\nSpeaker A: Maybe we can look at the new requirement I received from the management board and discussed all functions we want to have.\nSpeaker B: I just had a question.\nSpeaker B: Do you want to continue with your present?\nSpeaker A: Yeah.\nSpeaker A: I will continue.\nSpeaker A: Well, ask your question if you want.\nSpeaker B: You say that I don't remember my heart, but 30% of the people say that it's quite difficult to use the remote control.\nSpeaker E: Yeah.\nSpeaker B: Do they say that this is difficult, but for the same reason or do they have other reason?\nNone: Oh.\nSpeaker B: To, to, to keep in mind maybe to access to that money you should do something like that.\nSpeaker E: Yeah.\nSpeaker E: I think they say that it's difficult to learn how to use it.\nSpeaker E: But when you know how to use it, it's okay.\nSpeaker B: And it's not intuitive.\nSpeaker B: Yeah.\nSpeaker B: But yeah, maybe and what about if we design a remote control which can be configured as you want.\nSpeaker B: You say that I want, I have six buttons.\nSpeaker D: A lot of people are, if you have the LCD screen, you can do completely the way you want because the buttons also would look the way you want them.\nSpeaker D: But also it would be hard to configure.\nSpeaker D: I mean, imagine.\nSpeaker D: So it's really something for the expert user.\nSpeaker D: So I mean, there are markets and markets.\nSpeaker D: I think the young people that are Christine here said that have a, yeah, it is.\nSpeaker D: So for our young people, it would be cool.\nSpeaker D: They are able to use it.\nSpeaker D: Maybe their parents will not, but they will configure it.\nSpeaker D: I guess I don't know if there's a study.\nSpeaker A: Maybe it would be more complex to configure it to be simple than creating a simple project.\nSpeaker D: Yeah.\nSpeaker D: I think that if we make something that simple and needs to use that means that it will be exactly the same as everything else.\nSpeaker D: Right.\nSpeaker D: Yeah.\nSpeaker D: Otherwise, if it's different, then of course it will have some of it has to learn to use it first.\nSpeaker A: Yeah.\nSpeaker A: But also we, we see that that most people find it, find remote controls to complex because they have too many buttons and they mainly use only channels and volume buttons.\nSpeaker A: So we may just make a very easy to use remote control with mainly those buttons.\nSpeaker A: Yeah.\nSpeaker A: And maybe also some lightning stuff to, because most people find also hard to, to find the remote controls, lose it, et cetera.\nSpeaker A: These are, these two points are the main frustration.\nSpeaker A: So maybe if we design something very simple and easy to find when lost, it will add a serious competitive advantage without making something too complex and too long to develop.\nSpeaker A: So, but let, let us see first the new requirement.\nSpeaker A: So we don't have to, so this, this is a, is a, is a, this is in the same direction as we were speaking.\nSpeaker A: So we don't have to make a very complex remote controls to access teletext and stuff like that.\nSpeaker D: Just one button.\nSpeaker A: Yeah, but then you have to, you have to define the buttons to surf amongst pages.\nSpeaker D: Just write the, write the numbers.\nSpeaker A: Yeah.\nSpeaker D: So, wait.\nSpeaker D: We're at with the channel keys, right?\nSpeaker A: Yeah.\nSpeaker A: So anyway, we don't have to include this feature because it's, it's not used anymore by users.\nSpeaker A: They prefer to.\nSpeaker D: Yeah.\nSpeaker D: I'm sure that I don't like, but I don't see, just one button.\nSpeaker A: Yeah.\nSpeaker A: So, I don't know.\nSpeaker A: If one button is still one more button, if, if, if we want to make it very simple, we have to reduce the number of buttons compared to the, to our competitors.\nSpeaker A: Well, anyway, I have this point we can discuss also.\nSpeaker A: So as I told before, it would be better if it's only for the TV because we want to be quick on the market.\nSpeaker A: And then also we have to make very clear that this, this remote control is, is part of, of our products and show our co-operators logo and, and colors and the, and the design as well so that they identify it as one of our products.\nSpeaker A: So this is the, the key point.\nSpeaker A: So before finishing, we can define what would be the characteristic of the, of the control, of the remote control and which button do we need, which function do we want, etc.\nSpeaker A: Capital.\nSpeaker A: So, do you, so, so from, from the marketing expert, I think a key aspect is the easy to use aspect is should be very simple and most button are never used.\nSpeaker A: Only 10% of the button are, are used often.\nSpeaker A: So I think we have to do something very simple and I think we all agree on that point.\nSpeaker D: No, well, if it is going to be just a TV remote control, it's going to be very simple.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So, yeah, the key point here is simple.\nSpeaker A: Maybe so few buttons, channel volume control and what, what function do you see?\nSpeaker D: Well, if you want to get as simple as possible and just let the very remote control, there is no other fraction of the country.\nSpeaker A: Yeah.\nSpeaker A: Maybe switch TV on and off.\nSpeaker D: No, you want to keep it always on so that advertising can, rather than you can come back to us.\nSpeaker A: Yeah.\nSpeaker A: Something.\nSpeaker A: So, a volume, maybe a mute button and an off button and that's all.\nSpeaker B: I know that some, you say that many people are doing plenty of, a lot of zapping.\nSpeaker B: Mm-hmm.\nSpeaker B: I know, I discovered that when I did a quick look.\nSpeaker B: They do not, so they do something which is quite nice now.\nSpeaker B: You have a button, you press it and this is the previous channel which is come back.\nSpeaker A: Yeah, this is cool.\nSpeaker A: Maybe we can include that also, previous, previous channel button.\nSpeaker A: So we have like channel button, the previous channel button, the volume button, press a mute button and just the traditional on off button.\nSpeaker D: And of course, the channel changing buttons.\nSpeaker D: Yeah.\nSpeaker C: I took a lot of that.\nSpeaker D: How should we implement them?\nSpeaker D: Because they could be numerical only, they could be also incremental.\nSpeaker A: Yeah.\nSpeaker A: So, incremental definitely because zapping you, you switch them.\nSpeaker B: Let's say that we can do something like that.\nSpeaker B: This is incremental.\nSpeaker B: But when you press it for a long time, you go five by five.\nSpeaker B: You go five?\nSpeaker B: Two faster?\nSpeaker B: Two faster.\nNone: Mm-hmm.\nSpeaker D: It's an interesting idea that if you press it for a long time, it does something else in general.\nSpeaker D: So if you have your 10 buttons for the numerical buttons and you have, instead of having just one memory, you have a few percent for a long time.\nSpeaker D: No, it doesn't work.\nSpeaker D: Maybe we should have also a digi, but have a complete keyboard and like console commands.\nSpeaker D: That's it.\nSpeaker D: Yeah.\nSpeaker A: Okay, maybe we have also to have the jits only incremental.\nSpeaker D: I don't know because if you have to, if you want to turn between three times for example, then you cannot work with just number and then incremental.\nSpeaker A: Yeah.\nSpeaker A: Because you have your previous channel button, if you have incremental only, it doesn't work, because the previous channel is either minus one or plus one.\nSpeaker A: So I think we need also digits.\nSpeaker A: Okay.\nSpeaker A: Maybe we can make very obvious the channel and volume button and smaller button done there with the digits.\nSpeaker D: Yeah, one user usually you would like to press the same button on the time.\nSpeaker B: Yeah.\nSpeaker B: If you can do something like that, we can design the remote control to have access.\nSpeaker B: You know, some remote control have protection.\nSpeaker B: And so you, you, yeah.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Just all just thinking there is a, I mean, you know, there's, there's some mask with a wheel like this.\nSpeaker D: It's just having the up down buttons.\nSpeaker D: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Kind of just stick.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Maybe we can have a wheel for incremental.\nSpeaker A: Okay.\nSpeaker A: So that I have the digit on the lower side that can be closed.\nSpeaker A: So I say protected.\nSpeaker A: And yeah, I think this is the lower side.\nSpeaker D: I think you have to turn it.\nSpeaker B: And do we, do we have a ball?\nSpeaker B: Yeah.\nSpeaker B: Not a wheel, but a ball.\nSpeaker A: And you say, I would say the wheel is better.\nSpeaker A: Because it is the expert of, yeah.\nSpeaker D: Because it's not like a volume, it's smooth, yeah, it's only 10 to one by one.\nSpeaker D: Yeah.\nSpeaker D: So you have to, the user should like to, should fill the, the, the script.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker A: Yeah.\nSpeaker A: That's a good idea.\nSpeaker A: And while also we have to decide the, so it should be lightning in the dark, I think because most people lost their remote control.\nSpeaker A: It's quite easy.\nSpeaker B: And we do that with backlight on the, if it's a blue, a blue lady and we serve that.\nSpeaker A: And do we put an LCD display?\nSpeaker A: Because it was important for young customers, if you remember.\nSpeaker D: I think it's all important if you have multi-function.\nSpeaker D: Yeah.\nSpeaker D: If you have multi-function, then there's no point in having the, actually, just increase the cost.\nSpeaker D: Okay.\nSpeaker D: The user doesn't have an advantage.\nSpeaker A: So no LCD.\nSpeaker A: And so no speech as well because it would delay too much the development.\nSpeaker D: Okay.\nSpeaker D: The development goes as well.\nSpeaker D: Yeah.\nSpeaker D: But it would be cool.\nSpeaker D: Because the user could say a CNN for example.\nSpeaker D: A CNN.\nSpeaker D: Yeah.\nSpeaker A: That would be cool.\nSpeaker A: But eight months is really long.\nSpeaker A: Maybe we can just, 10 years.\nSpeaker A: So I will, so we will move to next meeting.\nSpeaker A: So in after the launch break, you are the individual action you are required to do, but you will be recalled those actions by email I think.\nSpeaker A: But you can take notes if you want.\nSpeaker A: But while the instruction will be sent.\nSpeaker A: So thank you for your suggestion.\nSpeaker A: And I will make a summary of that meeting that I will put in the shared folder.\nSpeaker A: You can't see.\nSpeaker A: And then you will be able to see what has been said on this meeting and what has been decided.\nSpeaker A: Maybe for next meeting, send me your presentation before by email so that we can see them all together.\nSpeaker D: That would be easier.\nSpeaker D: What is the folder that people use?\nSpeaker D: It works.\nSpeaker D: No.\nSpeaker A: No, it did not work.\nSpeaker D: She sent it to me by email.\nSpeaker A: So maybe this is better to send it by email.\nSpeaker A: Okay.\nSpeaker A: What is your email?\nSpeaker A: So yeah.\nSpeaker A: I'm, it's in the first email.\nSpeaker A: So I'm participant one.\nSpeaker A: I'm participant one.\nSpeaker A: Where is that?\nSpeaker A: It's here.\nSpeaker A: participant one at enemy.\nSpeaker A: Okay.\nSpeaker A: So see you after lunch break.\nSpeaker D: Can't think of.\nSpeaker D: Okay.\nNone: So next time we should try to fight.\nNone: You don't fight anyMusic.\nNone: you just fight a detENT FX\n"}, {"meeting_id": "TS3009a", "summary": "The meeting was mainly about the exchange of the preliminary ideas about a new TV remote control project. They first got used to the writeboard and drew some animals with it.Project manager wanted to sell this remote control for twenty five Euro and expected profit will be around fifty million Euro. Afterwards the team raised different opinions about features of the new remote controls and related questions about market potential. Among all the features, having a good look and size were unanimously considered to be the most essential.", "dialogue": "None: I'm going to hang it open up with me.\nNone: I don't know.\nSpeaker D: I'm throwing away my toothpick.\nSpeaker D: Hi there.\nSpeaker A: Ew...\nSpeaker A: Eh...\nNone: right?\nSpeaker B: Do you have something called Asus Forum?\nSpeaker D: Nice.\nNone:agu Yeah, what the...\nSpeaker B: I've made a presentation, but I'll open it on the smart board, so we can all see it.\nSpeaker B: So it's in the project documents because that's what we can find here.\nSpeaker B: So, very nice.\nSpeaker B: Well this is called the Kick-Off Meeting, so I'm the project manager, so I have to fill it in.\nSpeaker B: So, we'll do the opening.\nSpeaker B: And then we'll meet each other, what we already do, so that's not very much trouble.\nSpeaker B: I'll show you the tools we have here, so we can all use them.\nSpeaker B: Then we'll look at the project plan from real reaction.\nSpeaker B: We'll discuss about our first adhes about the project, and then we'll close the meeting and then we can individually do our things and then we'll get back here.\nSpeaker B: So this is the opening.\nSpeaker B: We'll design a new division remote control.\nSpeaker B: You've heard that already I think, so we want it to be original.\nSpeaker B: So a nice new design.\nSpeaker B: Trendy, it's also for young people.\nSpeaker B: We have to just make it modern and friendly, so size does matter.\nSpeaker B: And well, it has to have the right buttons on the right place, those kind of things.\nSpeaker B: There are, there happen to be three stages, and going into a functional conceptual and detailed design.\nSpeaker B: So every time we do some individual work, get meeting, talk about it, and then go into the next phase.\nSpeaker B: That's just it.\nSpeaker B: We have these two smart words.\nSpeaker B: As I just showed, there's a project management folder, a project document folder on the desktop.\nSpeaker B: It just works exactly the same as a printer.\nSpeaker B: You just click on the folder and you open everything, you put in it with your laptop.\nSpeaker B: So you can make words, excel, everything.\nSpeaker B: The rest also works the same, so when you open a note pad, you just get your piece.\nSpeaker B: You can draw.\nSpeaker B: This is a drawing board.\nSpeaker B: You have these different functions on the board.\nSpeaker B: You can see them there.\nSpeaker B: So you have a nice pen.\nSpeaker B: And it works just like a bull pen.\nSpeaker B: This is just a bull going to.\nSpeaker B: I want to, of course it doesn't work anymore.\nSpeaker C: Maybe you should try to write on the big white.\nSpeaker B: Yes, it works.\nSpeaker B: Eraser, yeah.\nSpeaker B: It's wonderful.\nSpeaker B: Fantastic.\nSpeaker B: We can, well, you can save a file.\nSpeaker B: So if we draw, we have to save everything, don't throw anything away.\nSpeaker B: Just we can start a new one.\nSpeaker B: And we just go on and don't throw anything away, just let them all stand here.\nSpeaker B: We can delete, but we don't do that.\nSpeaker B: You can here select a pen, you can draw anything you want.\nSpeaker B: It's a bit childish, you have to write.\nSpeaker B: It's not as fast as you know it, but it does work sometimes.\nSpeaker A: Well, it's just like a normal paint.\nSpeaker B: So let's go.\nSpeaker B: Alright, yep.\nSpeaker B: Well, we are designers, so we have to have a more, a smart board.\nSpeaker B: So that's fantastic.\nSpeaker B: Well, this speaks for itself.\nSpeaker B: We're going to try it.\nSpeaker B: So we all are going to draw a nice animal on this board, not Mahari.\nSpeaker D: Alright, your favorite animal.\nSpeaker A: Yes, your favorite.\nSpeaker A: So I'm going to have to draw a kangaroo, but I'm going, I'm not going to.\nSpeaker B: I'm just going to draw a nice beast.\nSpeaker B: I don't know.\nSpeaker A: But I'm going to be like, oh.\nSpeaker A: That's not.\nSpeaker C: I hope this was part of the assignment and not your assignment.\nSpeaker C: I hope this was part of the assignment and not your personal enjoyment.\nSpeaker A: I just said it's not my idea, but I am the recognition.\nSpeaker A: Officially, this is my idea.\nSpeaker A: I understand.\nSpeaker D: We're not losing time though.\nSpeaker D: We're losing time.\nSpeaker B: The first meeting is just a bit loose.\nSpeaker B: It's not.\nSpeaker B: A bit meeting in China.\nSpeaker B: Well, nice.\nSpeaker B: Yes.\nSpeaker B: Yes.\nSpeaker B: I hope our industrial design does this better because this is just a count on it.\nSpeaker A: No.\nSpeaker A: A few.\nSpeaker A: Yes, yes, yes.\nSpeaker A: Hippo.\nSpeaker A: Well, I should be making it, but now I think it's a mouse for a rat.\nSpeaker C: I don't think so.\nSpeaker C: Oh, I know it.\nSpeaker C: Well, what is it?\nSpeaker A: It's a hedgehog.\nSpeaker A: Yeah.\nSpeaker A: You're a difficult English word.\nSpeaker A: I didn't knew it myself.\nSpeaker C: Well, I'm amazed about your drawing skills.\nSpeaker B: How characteristics sum it up.\nSpeaker A: Well, it's very painful.\nSpeaker A: Those guys.\nSpeaker A: So we can just...\nSpeaker A: We're going back.\nSpeaker A: And now our industrial design can draw its...\nSpeaker A: I am the industrial design.\nSpeaker C: No, it's a favorite animal.\nSpeaker C: Thank you.\nSpeaker C: Chief, I'm the industrial design.\nSpeaker B: Oh, but marketing design.\nSpeaker C: I think...\nSpeaker C: It resembles the animal drawn by Dick Bruner.\nSpeaker D: It's Nanji.\nSpeaker A: Nanji, what kind of animal is that?\nSpeaker A: I think.\nSpeaker C: Can I say it?\nSpeaker A: Yeah, sure.\nSpeaker A: It's...\nSpeaker A: It's a hard rib rat.\nSpeaker D: It's a rabbit.\nSpeaker D: Well, looks very nice, right?\nSpeaker A: Yeah.\nSpeaker A: It looks amazing.\nSpeaker B: No, no, no.\nSpeaker B: What are you going to do?\nSpeaker E: The one who erase it.\nSpeaker E: No, no, no.\nSpeaker B: Save it and start a new...\nSpeaker B: Save it?\nSpeaker B: These are very important...\nSpeaker C: These are very important documents, of course.\nSpeaker C: Yeah, well, we have to save everything.\nSpeaker B: So, now the next one...\nSpeaker B: You go, man.\nSpeaker B: And then save it and start a blank document.\nNone: Yeah.\nSpeaker B: There's also different colors and different...\nSpeaker B: Well, penwifes.\nSpeaker B: The line, the thickness.\nSpeaker B: So, well, you shoot a triad, but...\nSpeaker B: I should have made mine a white rabbit.\nSpeaker B: Well, you could have, but...\nSpeaker C: He deliberately draws with an animal we don't know the English word for.\nSpeaker C: What the...\nSpeaker C: It looks like...\nSpeaker C: Just a duck.\nSpeaker C: Looks like the beast from Sesame Street.\nSpeaker D: Pino.\nSpeaker D: Yeah.\nSpeaker D: Big bird.\nSpeaker D: It's the...\nNone: It's the...\nNone: Is the plane?\nNone: Right.\nSpeaker A: That's a bird.\nSpeaker A: But what kind of bird?\nSpeaker D: It doesn't draw...\nSpeaker B: Do we have to...\nSpeaker B: That easy...\nSpeaker B: You have to push harder.\nSpeaker B: Yeah.\nSpeaker B: Just a bit childish, bit shorter.\nSpeaker C: But we have to name the specific species of the bird.\nSpeaker C: No, I don't.\nSpeaker E: It's just a bird.\nSpeaker E: Well, wonderful.\nSpeaker B: Well, what a safe document.\nSpeaker B: And then...\nSpeaker B: And then...\nSpeaker B: And then a new blank document for Tys.\nSpeaker B: Tys will choose a new color and a new pen wife.\nSpeaker B: So why do I have to do the difficult tasks?\nSpeaker B: No.\nSpeaker B: Well, first.\nSpeaker B: And then you go to four mods, I think.\nSpeaker B: Current color.\nSpeaker B: Current color.\nSpeaker B: And I knew a...\nSpeaker B: I like...\nSpeaker B: They don't have pink.\nSpeaker B: Oh.\nSpeaker C: I think this is...\nSpeaker B: And a new pen wife.\nSpeaker B: Also format.\nSpeaker B: It's not like in paint.\nSpeaker B: Line wife.\nSpeaker B: And you can choose line with with with with each other.\nSpeaker C: 50.\nSpeaker C: And then I can draw.\nNone: Yeah.\nSpeaker A: So just the way you do it.\nSpeaker B: It's quite easy.\nSpeaker A: It's supposed to get...\nSpeaker A: I have a pussy.\nSpeaker C: The line width is too thick.\nSpeaker B: Well, then you change it.\nSpeaker B: Any raised things?\nSpeaker C: It's big.\nSpeaker C: It smiles nicely.\nSpeaker A: Super big.\nSpeaker C: Now I have to change the line.\nSpeaker C: One.\nSpeaker C: These are whiskers, you know.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: Well, I think it's obvious right now.\nSpeaker B: Yeah.\nSpeaker B: Alright.\nSpeaker B: It's cat.\nSpeaker D: No, it looks great.\nNone: Meow.\nSpeaker C: Meow.\nSpeaker C: If this is enough, you're...\nSpeaker A: Well, well, just save it.\nSpeaker C: I'll save it.\nSpeaker B: Alright.\nSpeaker B: Save.\nSpeaker B: And start a new blank document.\nSpeaker C: Yeah.\nSpeaker C: Blank.\nSpeaker B: Yep.\nSpeaker B: So that's what we're going to use when we need it.\nSpeaker C: Well, I feel comfortable now.\nSpeaker A: Thanks for the service.\nSpeaker A: I feel totally eddy now.\nSpeaker A: Yeah.\nSpeaker A: That's it.\nSpeaker A: We're one big, happy family now.\nSpeaker A: Yeah.\nSpeaker A: Well, then the serious stuff.\nSpeaker B: We want to set it at 25 euros internationally.\nSpeaker B: So we don't know what exactly it is in dollars, but...\nSpeaker B: Change of euros.\nSpeaker B: Our profit aim is worldwide.\nSpeaker B: 50 million euros.\nSpeaker B: So I didn't exactly calculate how much we have to sell.\nSpeaker B: We want to keep it at our costs at 12 and a half euros.\nSpeaker B: So keep that in mind when we talk about our materials and stuff.\nSpeaker B: Marketing research.\nSpeaker B: Now, then we can sit down and discuss what do we think about our current remote controls.\nSpeaker B: First, this is about design, about aim, the markets, etc.\nSpeaker B: But we can sit down because presentation can wait.\nSpeaker B: We can take notes.\nSpeaker B: And well, who has some remarks about the current remote controls?\nSpeaker C: Well, I didn't have to prepare anything about...\nSpeaker C: It's not my text to talk about experience with current remote controls.\nSpeaker B: Well, just we're at four.\nSpeaker D: I think we just have one.\nSpeaker D: It's important to look at the remote controls of our competitors.\nSpeaker D: Yeah.\nSpeaker D: Get the good points, try to merge them into one universal remote control.\nSpeaker D: Yeah.\nSpeaker D: On our corporate site, I saw a new DVD player we're going to produce.\nSpeaker D: Maybe it's important to make it compatible with the DVD player.\nSpeaker D: There would be a nice idea, yes.\nSpeaker D: So you can use your television and your DVD player with the same remote control?\nNone: Yep.\nSpeaker D: Furthermore, it's important to make it acceptable for the whole world, for different cultures, maybe.\nSpeaker D: Because we want to set off, well, 50 million.\nSpeaker B: Yes, 50 million is our aim to profit.\nSpeaker D: A lot of people have to be able to use it.\nSpeaker C: But the buttons have to have international recognizable buttons.\nSpeaker C: Yeah, that's right.\nSpeaker C: Numbers and that every culture and people in every country can recognize.\nSpeaker B: I'll make notes and then maybe...\nSpeaker B: Oh, I put it in the project folder when I'm done just now.\nSpeaker E: Well, we should not add too many buttons.\nSpeaker E: Yeah, yeah, yeah, remote.\nSpeaker E: That's right.\nSpeaker B: You don't use the half of them.\nSpeaker B: That's the same culture, international.\nSpeaker D: Maybe you can make one button to switch between DVD player and TV.\nSpeaker D: And make the other buttons multi-functional.\nSpeaker C: Yeah, so it doesn't become too complicated with too much buttons.\nSpeaker B: Yeah, right.\nSpeaker B: Yep.\nSpeaker B: And maybe we even have more than just a DVD player.\nSpeaker B: Don't we have other...\nSpeaker E: Yeah, we should make a compatible stereo.\nSpeaker B: We also just released a TFT...\nSpeaker B: Yeah, so I saw.\nSpeaker D: Kind of standard television.\nSpeaker D: Yeah.\nSpeaker D: So it also works on that.\nSpeaker E: Yeah.\nSpeaker E: And I think the people who will buy our remote already have some experience with remote.\nSpeaker E: So we can keep...\nSpeaker E: Motion.\nSpeaker C: People do.\nSpeaker B: Well, yeah.\nSpeaker C: It doesn't have to be...\nSpeaker C: It doesn't have to be.\nSpeaker E: It's a new remote and you don't buy a remote if you don't have anything to control with it.\nSpeaker B: Except if we deliver it together with our DVD.\nSpeaker B: Yeah, right.\nSpeaker D: We need to keep it consistent with other...\nSpeaker B: Yeah.\nSpeaker B: Because we look at competitors and if we pick up the good things about that and give it a nice design.\nSpeaker B: Yeah.\nSpeaker D: But it's...\nSpeaker D: That's different.\nSpeaker C: We use it.\nSpeaker E: At the same time.\nSpeaker E: Yes.\nSpeaker E: Yeah.\nSpeaker E: Yeah, we could use another form or shape or...\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: The shape will have to be recognized.\nSpeaker C: I thought about the...\nSpeaker C: Like most remote controls are a long box shaped thing.\nSpeaker C: You can make it triangle shaped.\nSpeaker C: That's not very recognizable.\nSpeaker E: Or more oval or something.\nSpeaker E: Oval.\nSpeaker A: And...\nSpeaker A: We can use it as a gamepad.\nSpeaker A: Oh, it's new.\nSpeaker A: Nothing to add.\nSpeaker A: One hand has the beer.\nSpeaker A: The other hand.\nSpeaker E: Now the young people want something different.\nSpeaker E: Well, we already...\nSpeaker D: One of our aims is that it has to be original.\nSpeaker B: And...\nSpeaker B: It has to be...\nSpeaker B: So...\nSpeaker C: Yeah.\nSpeaker C: But you still have to know it's remote and not another...\nSpeaker C: Well, there has been done a lot of research about remote.\nSpeaker B: We have to...\nSpeaker B: We can imagine because it's a long time on the market.\nSpeaker B: Yeah.\nSpeaker B: So the form will have been tested out.\nSpeaker B: So the long box shaped thing must be...\nSpeaker C: Yeah, usual.\nSpeaker C: Yes.\nSpeaker C: Or else they would have been...\nSpeaker A: For me personally, I have a lot of remotes at home.\nSpeaker B: But those ones that have a round ending and...\nSpeaker B: Well, just in a square middle.\nSpeaker B: I don't like to use them.\nSpeaker B: I have to...\nSpeaker B: It has to fit my hand.\nSpeaker B: Yeah.\nSpeaker B: When it falls over it and I just have...\nSpeaker B: And then the better...\nSpeaker B: Yeah.\nSpeaker B: And I use most...\nSpeaker C: Yeah, a long box shaped.\nSpeaker C: Yeah.\nSpeaker B: You have to...\nSpeaker B: It fits your hand.\nSpeaker B: Then you just push the button that you use most.\nSpeaker B: I don't agree with the long box shaped.\nSpeaker D: Why not?\nSpeaker D: That's the way we custom made for the hand.\nSpeaker D: Yeah, that's a few.\nSpeaker D: No, those new DVDs.\nSpeaker B: But it's just for the hand.\nSpeaker C: If you want to have those.\nSpeaker C: You can make it another shape.\nSpeaker C: But then you have...\nSpeaker E: If you look at new Philips day-to-day spades with...\nSpeaker E: New displays.\nSpeaker E: Yes, it is.\nSpeaker E: Yes, it is.\nSpeaker E: They are...\nSpeaker E: The new remotes aren't box shaped.\nSpeaker E: What else?\nSpeaker D: That's all fashion.\nSpeaker D: Yeah.\nSpeaker E: So what do you suggest then?\nSpeaker E: Well, most of them are...\nSpeaker E: Some with thicker at the end.\nSpeaker E: Yeah, right.\nSpeaker E: And get thinner towards the other end.\nSpeaker E: Well...\nSpeaker C: But it's still the long box.\nSpeaker C: But then with some round forms to fit your hand.\nSpeaker B: It has...\nSpeaker B: It still...\nSpeaker B: That's shaped that it fits your hand.\nSpeaker B: Yeah, yeah, alright.\nSpeaker C: But it's still...\nSpeaker C: It's still the sort of box.\nSpeaker C: It has round forms, but in the end it's still the box.\nSpeaker C: So that's what I mean.\nSpeaker D: Yeah, but it has to fit your hand.\nSpeaker C: Yeah, I understand.\nSpeaker C: But it's two books.\nSpeaker C: Now I don't mean an entire box like...\nSpeaker C: Convenient square, but also with round edges, of course.\nSpeaker C: But at the end it's still this long...\nSpeaker C: Yeah, it should be...\nSpeaker C: Box shaped with...\nSpeaker C: Convenient round shapes.\nSpeaker C: Right.\nSpeaker D: To fit like this.\nSpeaker D: Yeah.\nSpeaker D: And then...\nSpeaker D: Yeah, but I thought about something like that.\nSpeaker D: Between different systems like the D player.\nSpeaker D: And so you can...\nSpeaker D: A big recognizable button on top or something.\nSpeaker D: Yeah, right.\nSpeaker D: And I do think we have to keep this kind of for there...\nSpeaker D: Yeah.\nSpeaker D: With the numbers.\nSpeaker B: We have five minutes left.\nSpeaker B: So...\nSpeaker C: The buttons should also be not too small, not too big, of course.\nSpeaker C: Yes.\nSpeaker C: Not too close together.\nSpeaker D: But it should be possible to make it apparent that there are two functions for every button.\nSpeaker D: Yeah.\nSpeaker D: There has to be some space between the buttons.\nSpeaker C: Yeah, of course.\nSpeaker C: To have icons to explain the different functions.\nSpeaker C: Right.\nSpeaker D: Yeah.\nSpeaker D: And maybe you should use colors.\nSpeaker E: Colors, yeah.\nSpeaker E: On the screen maybe we can just like on cell phones.\nSpeaker E: Those...\nSpeaker E: You know, flanges.\nSpeaker E: Yeah.\nSpeaker E: You can put on them.\nSpeaker E: That's kind of true.\nSpeaker E: So you can customize your own remote with different colors or...\nSpeaker E: Yeah, right.\nSpeaker E:...with paint jobs.\nSpeaker E: I don't know.\nSpeaker C: Sounds nice.\nSpeaker C: Yes.\nSpeaker C: I think we have to talk also about the materials for...\nSpeaker B: Well, it's just about...\nSpeaker B: I have here.\nSpeaker B: So I think individually we'll have to come up with ideas for the next meeting about these materials and markets, etc.\nSpeaker B: Already thought about some things.\nSpeaker D: It's important to notate all the decisions that they make.\nSpeaker D: Yes.\nSpeaker D: Sorry.\nSpeaker B: Well, 30 minutes we have.\nSpeaker B: Mm-hmm.\nSpeaker B: This is about what we're going to do.\nSpeaker B: You'll get specific instructions when you're back in your room.\nSpeaker B: So it's logical, I think.\nSpeaker B: No problem.\nSpeaker B: Oh.\nSpeaker B: And that's all.\nSpeaker B: So we'll just get a notice that the meeting is over.\nSpeaker B: Yeah.\nSpeaker B: Now we can still talk about the materials.\nSpeaker C: We have some minutes left.\nSpeaker C: I wrote down the case should be plastic, of course, hard plastic.\nSpeaker C: Yeah.\nSpeaker C: The button should be rubber.\nSpeaker D: I think it shouldn't be too ugly.\nSpeaker B: No, no.\nSpeaker B: I have had remotes that they had the function of the buttons was about a layer over the buttons.\nSpeaker B: And when I had used it much, it was gone.\nSpeaker A: Yeah.\nSpeaker B: So it has to be made in the buttons, I think.\nSpeaker B: It has to not be used.\nSpeaker C: Mm-hmm.\nSpeaker C: And of course there are several electrical cables in it to connect things.\nSpeaker B: Is there a universal way of transmitting from the remote to the televisions?\nSpeaker B: It's all about...\nSpeaker B: I think so.\nSpeaker B: It's not that it's in a standard way.\nSpeaker C: Yeah.\nSpeaker C: It would go in for red beams and in for red beams.\nSpeaker B: Yeah.\nSpeaker B: But you can have, of course, different between DVDs and televisions.\nSpeaker C: It's that we make a universal remote.\nSpeaker C: So it has to work with all kinds of brands.\nSpeaker C: And...\nSpeaker E: With actual V's are mostly made in China and that sort of kind of thing.\nSpeaker A: Probably, yes.\nSpeaker A: China rules.\nSpeaker C: And, well, yeah, I've wrote something down about how it works.\nSpeaker C: The user presses a button and with an infrared beam.\nSpeaker C: But the signal's still a fishing set accordingly, but that's pretty obvious.\nSpeaker D: Are we going with the frontest idea?\nSpeaker B: Well, I think we can look into that in the next video.\nSpeaker B: We should make it universal.\nSpeaker D: Yeah.\nSpeaker D: And you can always use a front on it, you know?\nSpeaker D: You can use it just plain.\nSpeaker D: Yes.\nSpeaker D: But you can...\nSpeaker B: Well, just...\nSpeaker B: You can't...\nSpeaker B: A normal front with it, but you can change them when you buy it.\nSpeaker B: Right.\nSpeaker B: And we shoot, dispatch those kind of front fronts around the world.\nSpeaker B: So...\nSpeaker E: We make profit with them and it's a way to make them train you.\nSpeaker B: Well, those 50 million don't sell itself.\nSpeaker B: So we have to make some extra effort like from...\nSpeaker C: Yeah, but the standard front will be just gray or something.\nSpeaker C: Yeah, it's a symbol color, not very fleshy color.\nSpeaker B: Well, it has to fit the television and DVD sets we are going to sell.\nSpeaker B: So if they are black and silver...\nSpeaker B: Pink will make them black and silver.\nSpeaker B: So...\nSpeaker E: Standard.\nSpeaker D: But people often don't like bright colors or something.\nSpeaker D: We have to make it gray or blue.\nSpeaker D: Well, young people, yeah, but then you can use front.\nSpeaker E: Yeah, yeah.\nSpeaker C: All right.\nSpeaker C: Yeah.\nSpeaker C: I must not forget my pen that time.\nSpeaker B: Oh, yeah.\nSpeaker B: If you have a television room for little children and you make...\nSpeaker B: If you buy a pink front, they'll love it.\nSpeaker B: Yeah, that's true.\nSpeaker B: I think that's marketing research.\nSpeaker B: That's the best for them.\nSpeaker B: You can ask...\nSpeaker B: Yeah, yeah.\nSpeaker D: I will first go.\nSpeaker D: Yes.\nSpeaker B: Well, I usually in the face you can...\nSpeaker B: Maybe I don't know what your specific instructions will be, but probably about the position of the buttons and those kind of things, and what buttons will have to be on the remote.\nSpeaker B: And you will look into the technical design and form, I think.\nSpeaker C: Yeah, also the look and feel of the remote source on my desk.\nSpeaker C: Yeah.\nSpeaker E: Right.\nSpeaker E: What's the URL of the website?\nSpeaker C: Yeah, I was wondering that too.\nSpeaker C: You went to the company website.\nSpeaker C: If you start up, it starts to finish.\nSpeaker B: Finish meeting now!\nSpeaker B: Oh, all right.\nSpeaker B: Well, we're going to back...\nSpeaker B: I can do our rooms.\nSpeaker B: Well, that's great.\nSpeaker B: The next meeting is in 30 minutes.\nSpeaker B: But I think it will be a little bit worn.\nSpeaker B: Yeah, right.\nSpeaker E: To get over here.\nSpeaker B: I'll have to restore my desktop.\nSpeaker A: It's totally broken.\nSpeaker A: It's the half of the normal size.\nSpeaker A: Right, she'll have it all done.\nSpeaker A: Oh.\nSpeaker A: No!\nSpeaker A: Oh!\nSpeaker A: Oh!\nSpeaker A: Oh!\nSpeaker A: Oh!\nSpeaker A: Oh!\nSpeaker A: It was a nice meeting.\nSpeaker A: No, it's a nice meeting.\nNone: Yeah, yeah, yeah.\nSpeaker C: All right, I'll see you in 30 minutes now.\nSpeaker A: Yeah, I'll see you.\nSpeaker A: Yeah.\n"}, {"meeting_id": "TS3005b", "summary": "Project Manager began functional design meeting with reviewed last meeting and looked at new project requirements. The group did a presentation and made discussion about conceptual remote control design. They decided to use the LCD screen in the design as it was popular for the market and hard to damage. Because of the high price of LCD, the group put up a backup design with buttons instead of LCD as a solution. After group discussion, remote control would be designed with buttons shape and functions such as channels selection and speech recognition.", "dialogue": "None: Hello.\nNone: Good morning.\nNone: Again.\nNone: Again.\nNone: Again.\nSpeaker A: Again.\nSpeaker C: Good morning.\nNone: Again.\nSpeaker D: One question.\nSpeaker D: Two and a number.\nNone: A four.\nNone: Okay.\nSpeaker A: All set.\nSpeaker A: Good.\nSpeaker A: Okay.\nSpeaker A: Let's see what we can find here.\nSpeaker A: Okay.\nSpeaker A: A very warm welcome again to everyone.\nSpeaker A: Here we are already in our functional design meeting.\nSpeaker A: The opening, which we are doing now.\nSpeaker A: In the special note, I'm project manager, but on the meetings I'm also the secretary, which means I will make minutes as I did as the previous meeting.\nSpeaker A: And I'll also put these as fast as possible in the project folder so you can see them and review what we have discussed.\nSpeaker A: If I'm right, there are three presentations, I guess each one of you has prepared one.\nSpeaker A: Good.\nSpeaker A: And we will also take a look at new project requirements if you haven't heard about them yet.\nSpeaker A: And then of course we have to take a decision on the remote control functions and we have some more time, 40 minutes, but I think we will need it.\nSpeaker A: Well, I don't know who wants to go first with this presentation.\nSpeaker A: Go first.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker E: Should I go first with the user?\nSpeaker E: Let's see.\nSpeaker E: Everybody already has it.\nSpeaker B: You can adjust it.\nSpeaker D: Okay.\nSpeaker A: And one question, your name, Denny, is it with AEA?\nSpeaker A: AEA.\nSpeaker A: Okay.\nSpeaker A: Thank you.\nSpeaker D: Okay.\nSpeaker D: I wanted to explain the working design of a remote control.\nSpeaker D: It's possible if you want to design one of those.\nSpeaker D: Well, it basically works as I wrote down in this little summary.\nSpeaker D: When you press a button, that's when you do, for example, when you want to turn up the volume, a little connection is made.\nSpeaker D: The rubber button just presses on a little pin plate, which makes a connection that you can use.\nSpeaker D: A connection that gives the chips, which is mounted beneath those, that plastic of a rubber button, sends that connection has been made and knows what button you press, for example, the volume up or volume down button.\nSpeaker D: The chip makes Morse code like signal, which then is signaled to several transistors.\nSpeaker D: Which sends the signal to a little lat.\nSpeaker D: You know what a lat is?\nSpeaker D: Okay.\nSpeaker D: And that makes the infrared light signal, which is sent to the television set, which has a sensor in it to send the signal of the infrared.\nSpeaker D: And that's basically how it works.\nSpeaker D: The findings that I found searching up some detail information about remote controls are that they are very easy to produce.\nSpeaker D: This piece is possible to make them a mass production because it is as easy as printing a page, just a fiberglass plate.\nSpeaker D: It's covered with some coatings and chips.\nSpeaker D: And the technology is already available.\nSpeaker D: We don't have to find out how remote controls have to work or how to make some chips that are possible to transmit those signals.\nSpeaker D: I made a little animation about how remote control works.\nSpeaker E: Innovation.\nSpeaker B: There is something turning.\nSpeaker A: Yeah, it's a little bug.\nSpeaker D: It's in the smartboard.\nSpeaker D: Well, the subcomponent I suppose that you understand what the subcomponent is.\nSpeaker D: In this example, it's the button.\nSpeaker D: When it is pressed down, the switch is switched on.\nSpeaker D: So with the wire is sent to the chip in cooperation with the battery, of course, because to make a signal possible, you have to have some sort of electronic infrared light.\nSpeaker D: Yes.\nSpeaker D: After it's being composed by the chip, the signal is transported to the infrared bulb.\nSpeaker D: From there, it signals a Morse code-like signal to the bulb in the television cell.\nSpeaker D: I wrote down some personal preferences about the remote control.\nSpeaker D: Of course, it is very handy if the remote control is handheld, so you don't have to wind it up or something.\nSpeaker D: It's very light to use it.\nSpeaker D: I personally prefer that it would become available in various colors and easy to use buttons, but I suppose that one of the other team members thought of that too.\nSpeaker D: It is possible for several designs and easy to use buttons.\nSpeaker D: Perhaps soft touch, touch screen buttons because the rubber buttons are always slightly damaged.\nSpeaker D: So the numbers on the buttons are not possible to read anymore.\nSpeaker D: And as I said before, we can make several designs.\nSpeaker D: Okay.\nSpeaker D: That's my contribution to this meeting.\nSpeaker A: I'll go next.\nSpeaker A: Please.\nSpeaker A: So.\nSpeaker B: Spoke it.\nSpeaker B: Well, my name is Emke Ljung and I looked at the technical functions design of the remote.\nSpeaker B: I did this by looking at examples of other remote controls, how they look and information from the web that I found.\nSpeaker B: What I found was that the actual use of remote control is to send messages to the video set, how you describe them.\nSpeaker B: And it can be all sorts of messages, turn it on, turn it off, change the channel, just fill in that kind of thing, play video text, but also play CD if you use your CD player.\nSpeaker B: There are some examples.\nSpeaker B: You can see they are very different.\nSpeaker B: The one has got all the functions that you could possibly need in a lot of buttons, etc.\nSpeaker B: And the other is more user friendly, little, it's big buttons and not all the stuff you can do with it, but the essential stuff is there.\nSpeaker B: I guess you could better, you should look at the user centered approach because the user centered approach because the customers have to use them and if they don't think it's usable, they won't buy it.\nSpeaker B: A lot of buttons they may think from idle meets as much as that.\nSpeaker B: Personal preferences is a simple remote with the basic functions that you can need that you could use, but keep in mind the new functions of TV, what we discussed earlier, split screen, is that a function that you should have because all the TVs will have them or because only a few ones isn't really necessary.\nSpeaker B: And then I would make so that you could use it on more than one appliance.\nSpeaker B: If you have one that does the video, it could also work with the stereo because play is play and stop, stop and let's read it.\nSpeaker B: You could reuse the buttons so you don't have to have a lot of buttons for anything.\nSpeaker B: And it should be user friendly, clear buttons and not too much.\nSpeaker B: And that's just my presentation.\nSpeaker E: Okay, so, we're going to discuss the functional requirements of the remote.\nSpeaker E: That means that functions user want to have on the remote control or just the users actually.\nSpeaker E: The method I prefer is we're going to look which section of the users we're going to focus on more.\nSpeaker E: Are the younger people going to buy remote control or the elderly people?\nSpeaker E: And that section we're going to focus and adjust the remote more to that section than the whole user section.\nSpeaker E: Some data. Younger people from 16 to 45 years are more interested in features like LCD screens, speech recognition, etc.\nSpeaker E: And we possess about two-thirds of the market from in that range of H.\nSpeaker E: The elderly people from 45 years to 65 years are not that much interested in features and we possess less than two-thirds.\nSpeaker E: That's two-fifth of the market share in that area.\nSpeaker E: Okay, findings. 50% of the users lose their remote often.\nSpeaker E: So we don't have to make it very small like a mobile phone or something but somewhat bigger than small so you don't lose it that much anymore.\nSpeaker E: 75% of the users also find it ugly and 75% of users have a lot.\nSpeaker E: So the buttons should be that small or should be that complex because you have to search for buttons which one you're going to use.\nSpeaker E: Important issues about the remote. I think it would be better with the personal reference but okay.\nSpeaker E: Remote control have to have a low power usage because 75% of the users only step one time an hour.\nSpeaker E: So the power usage is also one time an hour so with a high power usage we would use a lot of batteries.\nSpeaker E: The following buttons and the channel buttons are the two most important buttons on the remote control so those have to be find very easily and have to be somewhat like bigger etc.\nSpeaker E: It has also we have to find easily when the label is gone.\nSpeaker E: My colleague also knows that that label should be scratched off or who be off the slate.\nSpeaker E: So if that's the problem you also have to find it easily on the remote.\nSpeaker E: Buttons like what our colleague said have to be minimalized or should be covered or in LCD screen.\nSpeaker E: The LCD screen is easy because you have the LCD screen.\nSpeaker E: You have the first options put one option and then you have all the buttons of that option so the other options will be gone and you don't see the buttons.\nSpeaker E: So LCD screen should be easy but an LCD screen, the problem with the LCD screen is that elderly people from 45 to 65 years don't use the LCD screen.\nSpeaker E: So we have to keep that in mind that if you're going to implement LCD screen don't have to make it that hard to learn or to use.\nSpeaker D: LCD screen as in touch screen?\nSpeaker E: Yeah touch screen.\nSpeaker E: The last but not least, younger people are more critical about the features because they use the remote control often more often and are more technical than all the people.\nSpeaker E: And all the people spend more money and easily on the remote control.\nSpeaker E: So we have to keep in mind to focus not a lot, not that much on the younger people but also somewhat on elderly people.\nSpeaker E: And on my personal references I don't have any more time to comment that but like I said LCD screen is easily to use because you can implement a lot of buttons in one remote with not that much buttons.\nSpeaker E: And it should be easy to use especially the following buttons, the channel buttons and the number buttons to step through the channels.\nSpeaker E: And that is it.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Thank you.\nSpeaker A: Well thank you all.\nSpeaker A: I don't know.\nSpeaker A: Did everyone receive an email with the new project requirements?\nSpeaker A: No.\nSpeaker A: No.\nSpeaker B: I did not.\nSpeaker B: Perhaps there is.\nSpeaker A: And then I think it's a good thing that I made a separate slide of them so you can all read them.\nSpeaker A: Oh, well not in this presentation.\nSpeaker A: It should be in there.\nSpeaker A: Well I can tell you them from my laptop.\nSpeaker A: The data text has become outdated since the popularity of the internet.\nSpeaker A: So that's the first thing I think we should pay less attention to the data text.\nSpeaker A: Remote control should only be used for the television otherwise the project becomes more complex which endangers the time to market and of course we'll make it more costly I think.\nSpeaker A: Our current customers are within the age group of 40 plus.\nSpeaker A: New products should reach a new market with customers that are younger than 40 and you talked about it before.\nSpeaker A: And a last point but also very important.\nSpeaker A: Our corporate image should stay recognizable in our products which means that our corporate color and slogan must be implemented in the new design.\nSpeaker A: So we have to keep that in mind.\nSpeaker A: Well according to our agenda it's time to take a decision on the remote control functions.\nSpeaker A: So who has any idea what should be on it and what shouldn't?\nSpeaker B: You said it should only be with one appliance or with one the only the TV.\nSpeaker B: Yeah, only with video also.\nSpeaker B: Only for television.\nSpeaker A: Makes it a lot easier.\nSpeaker E: So yeah then you can requirements, no functions.\nSpeaker B: Then it should have on off.\nSpeaker E: Yeah, stand by basics, stand by a volume channel, one till two zero numbers on it.\nSpeaker E: The data text doesn't have to be.\nSpeaker E: Other functions.\nSpeaker B: Yeah, I can say from one number to two.\nSpeaker B: Yeah, yeah.\nSpeaker E: I think it's easy to implement the button with a special, which especially do that because some TVs if you press this one and then the two.\nSpeaker E: It makes it be between five seconds you make twelve.\nSpeaker E: That's relaxed.\nSpeaker D: And there are some models that don't accommodate that function.\nSpeaker D: So the Philips television makes plus one and that indeed to press one and then two to make an easy reach, to reach them twelve.\nSpeaker D: But all the television makes use of those buttons where you first press that button and then press two digits to.\nSpeaker B: So you tap that one.\nSpeaker E: Main targets, HR were 45 plus or?\nSpeaker A: Well, new products should reach a market with customers that are younger than 40.\nSpeaker E: And now we have every customer.\nSpeaker E: Because younger people have 16 to 25 age or 81% interest in LCD screen from 26 to 35 to 66% and 36 to 45 to 55%.\nSpeaker E: So I think because most of the print plate will be broken for about two years, you have to press very hard to go to the next channel with the LCD screen.\nSpeaker E: It's easy because you only have to wipe the screen for a fingerprint.\nSpeaker D: And then you can use that.\nSpeaker D: That collides with our mission to make it very cheap.\nSpeaker D: Because LCD screens are very expensive.\nSpeaker D: Touch screen.\nSpeaker D: Probably even more.\nSpeaker D: True.\nSpeaker D: Well, it's possible to make an LCD screen.\nSpeaker D: How was the information?\nSpeaker E: Yeah, it only says that this percentage like LCD screen, because it says that younger age between 16 and 45 highly interested features more critical.\nSpeaker D: Perhaps we should focus on the LCD screen.\nSpeaker E: Yeah, because our target is 16 to 45.\nSpeaker D: Yeah, but will we not exceed our reduction cost?\nSpeaker E: Yeah, you don't know how much cost LCD screen is it possible to find out?\nSpeaker E: I don't have any costs here.\nSpeaker E: I only have percentages.\nSpeaker B: What if you do an LCD, do we have any buttons?\nSpeaker D: An LCD screen is just like a drone here.\nSpeaker D: It displays several buttons.\nSpeaker D: For example, if you want the minimal use buttons such as channel and volume, it displays four buttons on the screen.\nSpeaker D: And it's possible to press them down just like a touchscreen.\nSpeaker B: So you can adjust which buttons you want on them.\nSpeaker E: Yeah, it's possible to adjust the audio settings, press audio on the touchscreen and you get the buttons for audio settings.\nSpeaker E: So the other buttons are gone.\nSpeaker E: Yeah.\nSpeaker A: So we're going for an LCD screen?\nSpeaker E: Yeah.\nSpeaker A: I think it's the most easy thing.\nSpeaker A: And I'm hoping that when we produce a lot, it won't be too expensive.\nSpeaker B: We had 1250, I guess?\nSpeaker A: Yeah, 1250?\nSpeaker B: Yeah, I don't know how expensive.\nSpeaker B: Any guesses?\nSpeaker D: Well, I suppose if the young people are interested in LCD screens, we surely make them.\nSpeaker D: And if that is our market share to our goal to deliver those remotely.\nSpeaker B: You also said that we should not only focus on the younger people, but also the older.\nSpeaker B: Yeah, but...\nSpeaker B: Will they use it if it's only as an LCD screen?\nSpeaker E: It's 46 to 45, 33% and 56 to 65 to 12%.\nSpeaker E: But our project requirements are the new products will be reached for new markets with customers that are younger than 40.\nSpeaker B: But you don't want to eliminate the other ones?\nSpeaker E: No, they're not now.\nSpeaker B: But if they also buy them, that's all right.\nSpeaker E: Yeah, but market share for 40 years and younger is higher than 65 in younger.\nSpeaker C: All right.\nSpeaker A: Okay, so LCD it is?\nSpeaker A: Yeah.\nSpeaker A: Okay, let's try to...\nSpeaker E: Yeah, if it gets too much to expensive, then we should be sticking to rubber buttons.\nSpeaker D: Yeah, I think that they will send you some information about the cost of LCD.\nSpeaker E: No costs at all.\nSpeaker D: So if you receive an email about that, can you post it in our project map?\nSpeaker E: I think that should... I think we all get the costs of everything.\nSpeaker D: Because you are the marketing expert.\nSpeaker E: I'll post it.\nSpeaker B: I figured out the backup plan that we would use.\nSpeaker B: Sure, sure.\nSpeaker B: Yeah.\nSpeaker E: Yeah.\nSpeaker C: Okay, but for now, LCD.\nSpeaker E: Then you have 75% of use finding it ugly.\nSpeaker B: LCD?\nSpeaker B: Yeah.\nSpeaker E: And 80% of the users would spend more money with it when a remote would look fancy.\nSpeaker B: Oh, this is a problem.\nSpeaker E: It'll fancy with the LCD screen.\nSpeaker B: Yeah, but they don't like it.\nSpeaker B: They think it's ugly when it has an LCD.\nSpeaker E: Yeah, just the play remotes, not specific LCD remotes.\nSpeaker B: All right, that's all the new setup.\nSpeaker A: Yeah, maybe you can make something fancy out of an LCD remote because it's new as far as I know.\nSpeaker A: Yeah, of course.\nSpeaker E: And then you have the other thing that's 75% sep a lot, but that's not a question with the LCD screen.\nSpeaker E: The only thing you have to do is wipe the screen off of one of these in time to get all your fingerprints off of it.\nSpeaker C: No.\nSpeaker C: Okay.\nSpeaker A: Okay, what else does a remote need?\nSpeaker A: A mute button.\nSpeaker A: Mute button.\nSpeaker A: Mm-hmm.\nSpeaker E: I think.\nSpeaker E: The most important thing on the remote control are channel selection, volume selection, and power usage.\nSpeaker E: Yeah.\nSpeaker E: And the Taylor text, but that is not of the question.\nSpeaker E: Other things are...\nSpeaker B: I think you put a button of Taylor text for people who want to use it.\nSpeaker B: Yeah, it would be.\nSpeaker B: I mean, I've got a big remote.\nSpeaker B: Yeah.\nSpeaker A: I'd have to fill it.\nSpeaker A: Yeah, and we could make a separate menu on the LCD.\nSpeaker A: Yeah.\nSpeaker E: And also less important things are screen settings, audio settings, and channel settings.\nSpeaker B: Yeah, less important, because I think they should be there.\nSpeaker D: Yeah, it should be there, but not...\nSpeaker D: But in a sub-press menu.\nSpeaker D: Yeah, sub-menu.\nSpeaker D: I think it's also important to make possible to...\nSpeaker D: Anycon in English?\nSpeaker D: And to not use batteries and use...\nSpeaker D: But...\nSpeaker D:...acry batteries to...\nSpeaker C: I would be very proud.\nSpeaker D: Yeah, sure.\nSpeaker D: Yeah, so you can mount the...\nSpeaker D:...the remote control to charge it.\nSpeaker B: We should think of it 1250.\nSpeaker E: Yeah, but we don't...\nSpeaker D: I don't know why it says...\nSpeaker D: Yeah, so...\nSpeaker D: Okay, because when you get an LCD screen, you run it on batteries.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker B: The batteries will be...\nSpeaker B:...the very...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker D:...the...\nSpeaker B:...the...\nSpeaker B:...the...\nSpeaker B:...the very...\nSpeaker B:... phr capital for extra that you have...\nSpeaker B:...first to turn the remote on, and then you can...\nSpeaker E: Yeah, I think it's not...\nSpeaker E:...that easy, because I don't think people like to...\nSpeaker E:...that you have to do it first, and then use it.\nSpeaker E: I think it's better than...\nSpeaker E: The ticket shuts down, the remote shuts down...\nSpeaker E:...the...\nSpeaker D:......code standard, when you know users...\nSpeaker B:... hijack!\nSpeaker B: Automatically...\nSpeaker B: Okay, I have to rearrange this.\nSpeaker D:... a few minutes or so here, not a few minutes.\nSpeaker A: Yeah, and maybe a low battery indicator\nSpeaker E: Before our when it's get again gets empty they have plenty of time to recharge it I've put it in a charger\nSpeaker A: We are going for a recharge\nSpeaker E: Yeah Because when you're watching TV you're zapping you have to put it in a rechargeer\nSpeaker D: When you're done with\nSpeaker E: Yeah, okay, but don't be watching you sure that the batteries go of our six hours five\nSpeaker B: Yeah\nSpeaker D: But you also forget to buy batteries Yeah, you can you can't use it so I or have to be sure that the batteries lost a couple of days with the church\nSpeaker B: So I think the battery should should work a lot longer than a couple of yeah, but but you have a LCD screen\nSpeaker D: My power uses so It should be a standard move to put your remote control in the charger when you're done watching television. That's also a Great advantage because you can't lose it anymore Because you are obliged to put in the charger and not to leave any couch between some cushions\nSpeaker E: Okay\nSpeaker B: You're near your couch Because otherwise you have to yeah, just a small device\nSpeaker E: Like definitely charge or something and just just the cable or even\nSpeaker D: Charging when you come out it on yeah something like that\nSpeaker E: Okay, well easy to use also My chair speaker speech recognition I\nSpeaker D: For the otherly you could make it possible to enlarge the screen to make it possible to not Display a button at 10 points. I think that this will be standard I have a bunch of options you have to specify because we can look at perhaps 40 buttons at the screen but in the otherly I only look at two buttons\nSpeaker A: Something about speech recognition. Yeah, it says\nSpeaker E: Also 91% 60 to 25 26 to 35 years 76% and 36 to 45 35%\nSpeaker D: Let's leave out all the remote controls and just put a microphone That's an idea It has to be remote control\nSpeaker B: No, they want to talk into the remote control\nSpeaker E: I think the LCD screen should be sufficient\nSpeaker A: Yeah, that's a lot. That's the option should be why not? Why not?\nSpeaker E: I think it's better to have LLC screen because in the area of 36 to 45 we have about 30% of the market share in Our hands and 55 of those people want LCD screen at 35 on speech recognition So I think it's better to keep it with LCD screen\nSpeaker A: Would it be useful to implement both?\nSpeaker E: Yeah, with the costs of allowed\nSpeaker B: I don't know that can be done with the cost of 12\nSpeaker E: If it should be done if it could be done\nSpeaker B: Yeah, I would like to implement that you say volume up and then\nSpeaker D: Certain systems already exist I think\nSpeaker B: Then you also have to have different languages if we go international It's used to do French and English\nSpeaker E: But that should be also coming with some software also with LCD screen Because I think it's a different written volume is different within than So I hear you're something\nSpeaker B: You can use icons for the The speaker and the That's better than language\nSpeaker D: So you want to hear it's international\nSpeaker E: Okay, what else?\nSpeaker A: So no speech recognition?\nSpeaker B: Well, if it could be done with\nSpeaker D: It should be done, it could be done\nSpeaker A: And then with different languages\nSpeaker B: Yeah, that should be\nSpeaker D: That's not so difficult I already used several Voice-operated systems and they are all possible to No, that's difficult\nSpeaker B: Well, you should suggest everything\nSpeaker E: Every language of dialects Yeah, I think it's very difficult\nSpeaker B: You have to speak so that it can understand\nSpeaker E: I think it can be implemented but maybe\nSpeaker B: You can use that as an option Yes, an option, yes\nSpeaker E: Money left or something 50 years, yes\nSpeaker D: Let's do speech Speech recognition\nSpeaker A: Okay, so we only do it when we have enough money left Okay, well, I've written down an honor off button Volume selection channel selection digits from 1 to 0 From 0 to 9 A digit button to switch between 1 and 2 digits Mute button, a separate menu for a teletext A battery indicator We're going to use a docking station And probably LCD And if there's enough money, speech recognition And the possibility to enlarge buttons or to have large buttons\nSpeaker B: With teletext, it wasn't really important No, but Also now I've colored I don't know if we should implement that Color?\nSpeaker B: Yeah, when you press the red button You go to page 102 and we press the I don't know if we should implement that Because it says that teletext is not really important Short cuts Yeah, short cuts\nSpeaker E: I'll take which we could also implement the audio settings, screen settings And channel settings but at sub menus\nSpeaker C: Yeah\nSpeaker E: The mainly, if you turn the remote control on You have to see 1 to 0 channel volume And if you want to use the teletext screen or audio then you can press it\nSpeaker D: It should be available but not directly available Yeah, okay\nSpeaker A: Okay, so not too much the teletext support But in a separate menu\nSpeaker B: So actually it is there But it's just not ready there\nSpeaker D: It's not ready there So it doesn't confuse the user We have to search It has to be easy to use I'll search If you want to use the teletext You can press the teletext button And then the options become available\nSpeaker E: The side of it\nSpeaker A: Okay, but no more buttons or functions Though what else can you do with that?\nSpeaker A: And we forgetting something I've got to cut you examples here\nSpeaker B: But I don't think there's anything we should\nSpeaker E: Play pause doesn't need to be\nSpeaker B: We don't have a video on this\nSpeaker E: So if you have a voice issue or presentation You could check the other remote controls Technical functions\nSpeaker B: Hey, you could hear Oh, the other\nSpeaker E: Which was real\nSpeaker B: And the technical functions They're a bit small You could be able to shoot stretch them Because\nSpeaker E: Bing! Yeah, yeah, yeah, yeah Technical functions Okay\nSpeaker B: I guess we've got a more\nSpeaker E: Uh, I think I got you half only You have a mute play Yeah, let's ask you for the video Free slow Yeah, the zoom buttons\nSpeaker B: Yeah, white screen\nSpeaker E: High screen Different things you have\nSpeaker B: I'll let you put them on now Yeah, different too Yeah With that should also be a sub menu\nSpeaker E: Yeah, it should be available But that is a screen setting or something\nSpeaker B: Yeah So we should also implement a screen setting\nSpeaker E: Yeah, screen settings, audio settings The other settings you have\nSpeaker B: Yeah\nSpeaker E: General settings\nSpeaker B: Yeah So you can probably Those four and of course the main Yeah, so first you see the main And the other ones you can go to\nSpeaker E: Like depth screens or something or I don't know\nSpeaker B: I hope we can do this Yeah, actually it depends\nSpeaker D: On what kind of television you got Because if you don't got a white screen Television, you don't need to be I don't need to be Screen settings\nSpeaker E: They don't use it\nSpeaker D: And if the television does not support Such operations\nSpeaker B: So you don't have to use that tab\nSpeaker E: Yeah Yeah So you leave it Hello Yeah Yeah Or it could be possible to have a Standard version of the remote Yeah Expanded version\nSpeaker B: And do we want to win different colors or The buttons, you say?\nSpeaker B: Colors\nSpeaker D: Colors I think the main color of the Remote Control is the color of the LCD screen Because we don't want to love itself And if I\nSpeaker E: Have a device of how many percent? 80% I think it's ugly We'll spend more money if it looks fancy\nSpeaker D: Okay, so use a lot of people\nSpeaker B: Perhaps you can make a adjustable front Like with the Un- I just want phones You can Okay You can Okay So you're a system But I think that's\nSpeaker D: Well Make it a little bit different colors\nNone: Yeah Red, white, blue, black And a sea Gray She's your first Simpsons versions And she press a button\nSpeaker E: It turns green Yeah Leave\nSpeaker C: Well\nSpeaker A: Five minutes Single for our Final five minutes So I have the things I just read Then we have separate menus for Data Text Screen Settings Audio Settings And well\nSpeaker E: Channel Settings\nSpeaker B: Alright So you can program that too Yeah\nSpeaker A: Okay So you can\nNone: Yeah Perhaps you should Throw them in one pile So options And then use them Yeah Otherwise you have all those data Perhaps they'll text Not\nSpeaker E: Or like If we menu button\nSpeaker A: Also a separate menu\nSpeaker E: Or also a menu button First menu Then you have Main All the settings\nSpeaker A: Okay, but we can work it out Yeah So later I guess So we're having a general menu With the most used functions Data Text Screen Settings Audio Settings Channel Settings And maybe there are options for the remote itself Like large icons or small icons What else?\nSpeaker E: I think Because we don't have a lot of buttons on the one screen I think the buttons Yeah, but Like we only have the general button The volume button Those buttons Yeah\nSpeaker C: There's also an ESD Right Yeah Right\nSpeaker B: So we don't have any normal buttons No, no normal buttons\nSpeaker E: Yeah Maybe only the on and off button\nSpeaker A: We don't need the special options menu for the remote itself No Okay\nSpeaker B: Oh well, you should be able to set which TV you have If you have a TV\nSpeaker D: Of course you need the settings button Or a settings option for the remote control\nSpeaker B: Yeah But is it an idea to use what you said Normal on and off button for the TV?\nSpeaker B: No, no, no\nSpeaker E: Because we discussed that you could charge it Otherwise it It jumps to standby mode\nSpeaker B: Yeah, but not for the remote but for the TV\nSpeaker E: Yeah, the TV of course I think that's the best thing is that To implement that one in the menu with the volume\nSpeaker B: But they're not as normal but the ESD\nSpeaker A: Well, maybe there should be a separate button Apart from the ESD Because you can't turn it on when the ESD is off So how do you turn the thing on that?\nSpeaker A: So we are Why you just tap it?\nSpeaker A: Yeah, you just tap it That's proofing Okay, then it's turn it off And then the television is on also or just remote?\nSpeaker A: No, just remote\nSpeaker E: The television doesn't have to be on That's where you can press on\nSpeaker D: Yeah, it should be a standby mode\nSpeaker E: Yes, standby then Press on the remote Press on and then TV should be Affailable\nSpeaker B: Or not I don't know whether it's handling to have a normal on button Just for the TV so you can turn it on And then you can choose channel Otherwise you I don't know whether or not That's\nSpeaker D: A normal button on the remote control\nSpeaker B: Yeah, or normal To turn it on Or you should put it in the ESD screen\nSpeaker D: Yeah, because when you touch the ESD screen When it is in standby mode It should pop on Why would it be handy to have an on button?\nSpeaker B: I guess if you use the ESD screen You first have search where it's the on button And then you then turn And then the TV goes on But if you have a normal on button on the remote Then you do it on And then you search the channel Which you want\nSpeaker E: Yeah, but I think the remote control If you press tap the screen It always should jump to the screen Which has the volume button Channel button And of course also the on and off button\nSpeaker D: I think it looks a lot more fancy Yeah, I think so too\nSpeaker A: I don't have any buttons on No, actually we're going to create a button less No buttons at all Okay, well that might be a unique selling point Remote If you can afford it Well, I guess Let's put on the discussion to our next meeting Because we're running out of time For now we're having a lunch break And then there will be Half an hour for the next share of individual work I will write minutes If I can create them out of this And put them in the project document folder And here are the individual actions for the other roles And of course specific instructions will be sent to you again By your personal coach Luckily as we are Okay, well thank you very much You just want to have a nice lunch Food\nSpeaker B: Should we put this back in our room? I think so\nSpeaker D: Yeah, I think so I think it's a small piece\nNone: Yeah Yeah But look, there you go Yep, yep, yep Yeah That's right Oh, sorry s So...\n"}, {"meeting_id": "ES2016d", "summary": "Firstly, Project Manager reviewed the overall process that the team designed the remote control that they had viewed the fashion trends and incorporated the key features. After that, User Interface introduced the prototype of remote control based on the previous discussion of its function. The prototype was yellow like a banana with a simplest quick on-off button and could be used both-handed. Secondly, Marketing designed an evaluation test according to criteria found of the market trend as well as the customers' complaints and the team gave one to seven points to the shape, colour, material, controls and functions of the product. Thirdly, the team found that the current product exceeded the budget, so the team decided to completely discard the single button and LCD. Lastly, the team gave positive feedback on the project and the process.", "dialogue": "None: o This is our last meeting.\nSpeaker C: I'll go ahead and go through the minutes from the previous meeting.\nSpeaker C: And then we'll have the prototype presentation.\nSpeaker C: Then we will do an evaluation.\nSpeaker C: We'll see what we need to have under the criteria for the evaluation.\nSpeaker C: Then we'll go through the finance and see if we fall within the budget.\nSpeaker C: Then we'll do the evaluation and then we can finish up after that with changes that we'll need to make.\nSpeaker C: Or hopefully everything will fall right in line.\nSpeaker C: See minutes from the last meeting.\nSpeaker C: We looked at the trends.\nSpeaker C: We had the fashion trends that people want to fancy look and feel.\nSpeaker C: It was twice as important as anything else.\nSpeaker C: They liked fruit and vegetables and the new styles.\nSpeaker C: And a spongy feel.\nSpeaker C: So we were talking about trying to incorporate those into our prototype.\nSpeaker C: They wanted limited buttons and simplicity.\nSpeaker C: Then we looked at the method for coming up with our own remote.\nSpeaker C: Looking at other devices.\nSpeaker C: The iPod.\nSpeaker C: We really like to look at that.\nSpeaker C: We also had the kids remote for a simple idea.\nSpeaker C: A two part remote, which was what we were originally looking at.\nSpeaker C: And then there was talk of speech recognition becoming more predominant and easier to use.\nSpeaker C: But I think we still decided not to go with that.\nSpeaker C: Then we looked at the components.\nSpeaker C: The materials for the case, the different energy sources, the different types of chips.\nSpeaker C: And made a decision on what we were going to use to make our remote.\nSpeaker C: And basically.\nSpeaker C: How what we were making for the prototype.\nSpeaker C: So I'm going to leave it at that and let you guys.\nSpeaker C: Take over.\nSpeaker C: The prototype.\nSpeaker C: Yeah, do you need this?\nSpeaker E: No.\nSpeaker E: There he is.\nSpeaker E: I'll have a blog about it.\nSpeaker E: Yeah, basically we went with the color yellow.\nSpeaker E: Working on the principle of a fruit, which was mentioned, basically designed around a banana.\nSpeaker E: But it would be held in such a fashion.\nSpeaker E: Whereas obviously it wouldn't be that floppy because this would be hard plastic.\nSpeaker E: These would be like the rubber, the rubber grips.\nSpeaker E: So that's how helpful the hell we grip.\nSpeaker E: The ergonomics would be done with this scroll wheel.\nSpeaker E: And this here represents the screen where you'd go through.\nSpeaker E: And the simplest function would be almost identical to an iPod.\nSpeaker E: And that one way through channels, that way with other way through channels.\nSpeaker E: Volume up and down.\nSpeaker E: And then to access the more complicated functions.\nSpeaker E: So you press that and go through the menus.\nSpeaker E: It's that simple.\nSpeaker E: That's just represents the infrared beam.\nSpeaker E: That's a simple or an off switch.\nSpeaker E: I don't know we could use the voice.\nSpeaker E: That blue bit should be yellow.\nSpeaker E: That would be where the batteries would be.\nSpeaker E: And that's about it.\nSpeaker E: It's a simple.\nSpeaker E: We can make it.\nSpeaker E: Is anything else?\nSpeaker D: That's what we have.\nSpeaker D: That's plastic.\nSpeaker D: Plastic covered with rubber.\nSpeaker D: We might add some more underneath here.\nSpeaker D: Maybe give it a form.\nSpeaker D: I mean we're supposed to hold it like that.\nSpeaker D: But she grabbed it.\nSpeaker D: Yeah, it's just made.\nSpeaker D: It's definitely just rough.\nSpeaker D: Left hand in order.\nSpeaker D: And they just both.\nSpeaker D: Exactly.\nSpeaker D: These both might as well.\nSpeaker D: The actual thing might be smaller.\nSpeaker D: Think about the bottom.\nSpeaker D: Yeah.\nSpeaker D: Either put either one on either side.\nSpeaker D: Not too at all.\nSpeaker D: Just a quick on-off button.\nSpeaker D: That's pretty important.\nSpeaker D: So you don't have to fit over that.\nSpeaker D: Okay.\nSpeaker D: That's not.\nSpeaker D: I'd say a bit smaller would probably be nice.\nSpeaker D: Yeah.\nSpeaker D: Play with that over there.\nSpeaker E: Nice flimsy because it's made out of heavy.\nSpeaker E: Right.\nSpeaker E: Do you like to?\nSpeaker B: Well done.\nSpeaker C: Kind of a banana.\nSpeaker E: No, whether or not it would fall into the costume and everything I suppose.\nSpeaker E: Would they scroll and they all see the.\nSpeaker C: Well, luckily we are going to find out.\nSpeaker C: Or not luckily.\nSpeaker C: Do you have a marketing.\nSpeaker B: I do presentation for us.\nSpeaker B: Okay.\nSpeaker B: So first I'll just discuss some of the criteria.\nSpeaker B: That I found just based on the past.\nSpeaker B: Turn the reports that I was looking at earlier.\nSpeaker B: And then we'll do a group evaluation of the prototype.\nSpeaker B: And then we will calculate the average score to see how we did.\nSpeaker B: So the criteria we're going to be looking at.\nSpeaker B: Are the complaints.\nSpeaker B: That we heard from the users who were interviewed earlier.\nSpeaker B: So.\nSpeaker B: We're going to be doing it based on a seven point scale.\nSpeaker B: And one is going to mean true that we did actually achieve that.\nSpeaker B: With seven being false we did not achieve that.\nSpeaker B: Okay.\nSpeaker B: So for the first one we need to decide did we solve the problem of the users who complained about an ugly remote.\nSpeaker C: I think it's definitely different than anything else out there.\nSpeaker C: So if they think that what is out there is ugly then yes.\nSpeaker C: I would say most definitely it's bright.\nSpeaker C: But still has your traditional black.\nNone: It's curved.\nSpeaker C: There's no sharp angles to it.\nSpeaker D: Not angular.\nSpeaker D: It comes to the ergonomics, the form and stuff.\nSpeaker D: Yes, that's definitely more beautiful than your average.\nSpeaker D: However the color.\nSpeaker D: We don't have to say in that.\nSpeaker D: We have to say that we have to say that we have to solve the problem completely.\nSpeaker D: Because the color is ugly.\nSpeaker D: There's nothing you can say about that.\nSpeaker D: I much prefer something like brushed chrome.\nSpeaker D: Yeah.\nSpeaker D: That form.\nSpeaker D: Yeah.\nSpeaker C: Something more modern to go.\nSpeaker C: Right.\nSpeaker C: A modern color to go with the modern form.\nSpeaker C: But it's different.\nSpeaker D: You don't want your three feet huge LCD display in your living room that's hanging from the wall to be controlled with something like that.\nSpeaker B: Okay.\nSpeaker B: So do you think since we this wasn't assigned criteria?\nSpeaker B: Do you think maybe we should put it somewhere in the middle of them?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: What do you think?\nSpeaker C: Three, four.\nSpeaker C: Four.\nSpeaker C: Four.\nSpeaker C: Okay.\nSpeaker C: Very non-committal four.\nSpeaker B: Okay.\nSpeaker B: The second one.\nSpeaker B: Did we make it simple for new users?\nSpeaker D: It's very intuitive.\nSpeaker D: I think.\nSpeaker E: Yeah.\nSpeaker E: That was the main name.\nSpeaker E: The main names that we had.\nSpeaker D: One.\nSpeaker D: One?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Do the controls now match the operating behavior of the users?\nSpeaker E: Yeah.\nSpeaker E: We brought it down to basically four controls where they did the most common channel and volume.\nSpeaker E: And then the other one was just a matter of just going further.\nSpeaker C: Just scrolling through and selecting.\nSpeaker A: So one?\nSpeaker C: Yeah.\nSpeaker C: I think that's a one.\nNone: Okay.\nSpeaker A: Okay.\nSpeaker B: The fourth one.\nSpeaker B: How about the problem of a remote being easily lost?\nSpeaker B: One of the number one.\nSpeaker D: Something that big and that yellow you just don't know.\nSpeaker D: What do you want to know?\nSpeaker E: I'm going to lose.\nSpeaker E: But if we were to, if we were that speech recognition that we could maybe just use that solely for the finding thing.\nSpeaker E: That was what we mentioned.\nSpeaker C: So if we incorporate speech recognition into it, then just to use to find it.\nSpeaker E: If it was lost, but like I said, you lose something so yellow so easily.\nSpeaker E: And it's not going to fall like a rectangle.\nSpeaker E: Slick down behind things.\nSpeaker E: That's going to be a difficult shape.\nSpeaker C: Well, it is quite bright.\nSpeaker E: Yeah.\nSpeaker E: Maybe in the middle again, three or four or something.\nSpeaker E: Okay.\nSpeaker E: I mean, you know, losing things is one of those things that people can lose.\nSpeaker E: Yeah.\nSpeaker E: I mean, a million ways you can pick it up and walk away with it and then you lost it.\nSpeaker C: But if we do go with the with the speech recognition, then it, then our scale goes up quite a bit.\nSpeaker C: Oh, yeah.\nSpeaker C: Yeah.\nSpeaker C: Probably probably two.\nSpeaker C: You know, if we eliminate the fact that, you know, it's impossible to guarantee that it's not going to be lost.\nSpeaker C: Let's say two.\nSpeaker C: With speech recognition, which of course may be changed depending on budget.\nSpeaker D: Yeah.\nSpeaker D: So we could add an extra feature actually, which makes this thing raise hell when you remove it too far from television.\nSpeaker D: Yeah.\nSpeaker D: We could add that, but that's nothing we have thought of.\nSpeaker C: Which, which may be cheaper than speech recognition.\nSpeaker C: It would just stay.\nSpeaker C: Yeah, true.\nSpeaker E: But I mean, just those whistling clappens, cravings, you have that shape.\nSpeaker E: So I can't be that.\nSpeaker E: And the space on this anti anti-thas Sort of proximity.\nSpeaker D: Or suitcases and stuff where you have one piece that's attached to your luggage, another piece that starts beeping that can't cost much.\nSpeaker D: Yeah.\nSpeaker D: So that can also easily be integrated because these things are small enough to hide.\nSpeaker D: So you have one piece, you have to glue somewhere behind your TV.\nSpeaker C: Stick it behind your TV and the other.\nSpeaker C: I'm afraid that you don't accidentally lose that piece.\nSpeaker C: Right.\nSpeaker C: That'd be tough.\nSpeaker D: Well, also your remote would alarm you if somebody stole your TV to your television.\nSpeaker D: Ran off of that without taking the beautiful remote control.\nSpeaker D: Yeah.\nSpeaker B: So, are we adding one of these two features?\nSpeaker B: Let's add one of those features and say yes.\nSpeaker B: Okay.\nSpeaker B: So we're back to a one.\nSpeaker B: Or two.\nSpeaker B: Two.\nNone: Two.\nSpeaker B: Okay.\nSpeaker B: Are we technologically innovative?\nSpeaker E: I'd say so.\nSpeaker E: I don't get my name up.\nSpeaker E: It's all just souls with screen.\nSpeaker E: It's all just stolen.\nSpeaker D: Yeah.\nSpeaker C: But there's not a lot of yellow.\nSpeaker C: There's not a lot of yellow.\nSpeaker C: Of course I wasn't really forced to take that color.\nSpeaker B: Two.\nSpeaker B: Okay.\nSpeaker C: I don't know the way you are that innovative.\nSpeaker C: It's too true.\nSpeaker B: But how many robots do you see like this?\nSpeaker C: If we added the screaming factor.\nSpeaker C: Then we go up.\nSpeaker C: I would say we're probably up four.\nSpeaker B: Really?\nSpeaker B: Okay.\nSpeaker B: It's going to hurt us.\nSpeaker B: Okay.\nSpeaker B: Spongy material.\nSpeaker B: Yeah, we do have that.\nSpeaker D: We have some splashes.\nSpeaker D: It's native.\nSpeaker E: Okay.\nSpeaker D: It's not a one though.\nSpeaker D: One would be the whole thing.\nSpeaker C: Yeah, because it's only got to fold and stuff.\nSpeaker C: Like these parts of the grips and perhaps the back side, the bottom underneath on the back.\nSpeaker D: So lots of four.\nSpeaker C: Probably four at most.\nSpeaker B: Lastly, did we put the fashion in electronics?\nSpeaker B: Yes.\nSpeaker B: Yes.\nSpeaker C: Your fashion is but is Carmen Miranda?\nSpeaker C: You betcha.\nSpeaker D: Well, the recent fashion is rather displayed in the LCD.\nSpeaker D: It's true.\nSpeaker D: And you operated the form on the color, but it definitely is.\nSpeaker E: What we were told.\nSpeaker E: Definitely.\nSpeaker E: Okay.\nSpeaker B: All right.\nSpeaker B: Now I just got to calculate.\nSpeaker B: 17.\nSpeaker B: 17.\nSpeaker B: 17.\nSpeaker B: 17.\nSpeaker E: 17.\nSpeaker B: Jesus and Amas, You bet whistlebl Right? How does that work?\nSpeaker D: I'm impressed I can't do that without a calculator.\nSpeaker D: No, I'm fine.\nSpeaker B: It's been a while.\nSpeaker C: And what is the acceptable criteria?\nSpeaker C: Is there like a scale that we have to hit?\nSpeaker B: Oh no, I just told me to pick my own criteria and have you guys evaluate it.\nSpeaker B: Basically.\nSpeaker C: All right then.\nSpeaker B: So that's that.\nSpeaker C: Okay, well, let's see.\nSpeaker C: Now we get to do the budget numbers.\nSpeaker C: You didn't know that you were going to have a budget.\nSpeaker C: But we do.\nSpeaker E: Okay.\nSpeaker E: It's been going a long time.\nSpeaker E: It's 2.4 to 8.\nSpeaker E: I said it was too strong.\nSpeaker E: Oh my gosh.\nSpeaker E: 2.4 to 8.\nSpeaker E: Yeah, we're going to have that.\nSpeaker C: So I have here 50% be pretty.\nSpeaker C: Yeah.\nSpeaker C: We want a 50% profit on this.\nSpeaker C: I can't really see that very well.\nSpeaker C: Just about finished quickly.\nSpeaker C: 12 and a half euros is what's supposed to cost us.\nSpeaker C: Okay, so.\nSpeaker C: It's too much.\nSpeaker C: Well, let's see.\nSpeaker C: The.\nSpeaker C: This.\nNone: Okay.\nNone: All right.\nSpeaker C: All right, so.\nSpeaker C: I don't know if you guys can read that enough.\nSpeaker C: I can't sit on my glasses on.\nSpeaker C: But so we've got the energy source.\nSpeaker C: There's a four, five, six categories.\nSpeaker C: We have energy source electronics case.\nSpeaker C: Then we have case material supplements interface type.\nSpeaker C: And then button supplements.\nSpeaker C: Okay, so.\nSpeaker C: First of all, energy source.\nSpeaker C: We pick battery.\nSpeaker C: And how many batteries do we think this will probably take?\nSpeaker C: Probably.\nSpeaker C: So either two or four.\nSpeaker C: Two.\nNone: Two.\nNone: Like it.\nSpeaker D: Four is going to be too heavy.\nSpeaker D: So that's not our problem.\nSpeaker D: People can change it every month.\nNone: They won't.\nSpeaker C: Oh, I thought I thought I bought it.\nSpeaker C: This is consumerism.\nSpeaker C: All right.\nSpeaker C: So for the electronics, our choices are simple chip on print, regular chip on print, advanced chip on print, sample sensor, sample speaker.\nSpeaker D: Advanced chip, that's the advanced chip on print.\nSpeaker C: Okay.\nSpeaker C: We have one of those.\nNone: Okay.\nSpeaker C: Then the case is a.\nSpeaker C: Probably that's double curved.\nSpeaker C: Yes.\nSpeaker C: Case materials are.\nSpeaker C: Plastic.\nSpeaker C: Plastic.\nSpeaker C: I guess it's two since one for the top one for the bottom.\nSpeaker C: Is that right or is it just one?\nSpeaker C: Oh, maybe it's one because of the.\nSpeaker D: It's just one single mode.\nSpeaker D: Yeah.\nSpeaker C: I guess it doesn't matter because the price on that one is zero.\nSpeaker C: Right.\nSpeaker C: Special color.\nSpeaker D: That's not a special color.\nSpeaker C: Especially ugly color.\nSpeaker C: Interface type.\nSpeaker C: We have push button scroll wheel interface.\nSpeaker C: Integrated scroll wheel push button LCD display.\nSpeaker C: So we actually have the LCD display.\nSpeaker C: And then is it the integrated or.\nSpeaker E: Absolutely.\nSpeaker E: Yes.\nSpeaker E: Okay.\nSpeaker C: Button supplement special color.\nSpeaker C: Special form.\nSpeaker C: Special material.\nSpeaker D: We could of course make the buttons would.\nSpeaker D: Same how can they.\nSpeaker B: Really love or titanium.\nSpeaker B: The cost is all.\nSpeaker C: But we only have one button.\nSpeaker C: So really we shouldn't be charged.\nSpeaker C: We shouldn't be charged anything.\nSpeaker C: The button supplements.\nSpeaker C: Yeah.\nSpeaker E: Okay.\nSpeaker C: We're going to leave that one blank because we run on a LCD.\nSpeaker C: Scroll.\nSpeaker C: So our total is 15.5.\nSpeaker C: Which I believe.\nSpeaker C: Is.\nSpeaker C: Three euros over.\nSpeaker D: So we go for the hand.\nSpeaker C: So the only thing better than a banana shape remote is one of the.\nSpeaker C: One that you shake.\nSpeaker E: What if we completely took out the one single button we've got on.\nSpeaker E: And just had a scroll wheel interface.\nSpeaker E: And the LCD's way.\nSpeaker E: I suppose the LCD's way is the one that's putting it off a bit.\nSpeaker C: Yeah.\nSpeaker C: Well, because we have to have both, right?\nSpeaker D: That's a basic.\nSpeaker D: It also depends on the software on the television you can have.\nSpeaker E: The information that's been translated on the screen.\nSpeaker C: You could maybe take out the LCD's.\nSpeaker D: If it comes up on the computer itself, on the TV itself.\nSpeaker D: And you're not needed.\nSpeaker D: There you go.\nSpeaker C: There we go.\nSpeaker C: There we go.\nSpeaker C: Twelve point five.\nSpeaker C: Okay.\nSpeaker C: So we just remove our screen here.\nSpeaker E: Big a dial.\nSpeaker E: Easy to use.\nSpeaker E: Easy to use on.\nSpeaker C: Besides, look at what the LCD does to our lovely remote.\nSpeaker C: Back to the design room, boys.\nSpeaker D: Take away a heck of water.\nSpeaker B: What's the blue part?\nSpeaker E: That was just, we ran out of yellow.\nSpeaker D: There you go.\nSpeaker E: There you go.\nSpeaker B: It was simpler.\nSpeaker E: Looks more like a banana.\nSpeaker E: Feels a few little bit out there.\nSpeaker C: Okay.\nSpeaker C: Cost under twelve point five zero.\nSpeaker C: No.\nSpeaker C: We redesigned it.\nSpeaker C: Now it's yes.\nSpeaker C: Next slide.\nSpeaker C: Project evaluation.\nSpeaker C: Project process satisfaction with, for example.\nSpeaker C: Creativity leadership team work means new ideas found.\nSpeaker C: So I guess that.\nSpeaker C: I think that perhaps the project evaluation is just supposed to be complete by me.\nSpeaker C: But I'd like to hear your thoughts.\nSpeaker B: What did you think of our project process?\nSpeaker C: We don't quite well.\nSpeaker E: Team work.\nSpeaker D: Right from the start of the day.\nSpeaker C: Yeah.\nSpeaker C: We started off a little week.\nSpeaker C: Our leadership was quite weak in the beginning.\nSpeaker C: But as the day went along, we had more idea of what we were doing.\nSpeaker C: Room for creativity.\nSpeaker C: I think it was interesting as you guys brought up more information and studies that we were right on with a lot of those things.\nSpeaker C: You guys work together.\nSpeaker C: Well team.\nSpeaker C: And the means.\nSpeaker C: Which was the white board depends.\nSpeaker C: I had some problem with the pen.\nSpeaker B: I think.\nSpeaker B: Minus your power point.\nSpeaker B: Minus your power point.\nSpeaker D: My father.\nSpeaker D: No, I know.\nSpeaker C: People I worked for.\nSpeaker C: Have a list of employees that you would like fired.\nSpeaker B: Yes.\nSpeaker C: Yes.\nSpeaker C: For the remote.\nSpeaker C: Maybe no.\nSpeaker C: For technology.\nSpeaker C: Closing.\nSpeaker C: Cost or within the budget.\nSpeaker C: Project is evaluated.\nSpeaker C: Complete final questionnaire and meeting someone.\nSpeaker C: That's it.\nSpeaker C: I still have to do my minutes.\nSpeaker C: So there will probably be another.\nSpeaker C: Question air coming up.\nSpeaker C: And then we'll have to check with the main boss whether we can.\nSpeaker C: What goes on after that.\nSpeaker C: That's the end of our meeting.\n"}, {"meeting_id": "TS3004d", "summary": "This meeting was about the prototype presentation of the remote control. Firstly, the group had a prototype discussion on the remote control's scroll wheel, standard place, the round shape, the multiple speech buttons, the menu buttons, and the text button with hard plastic material. Next, the group evaluated remote's look, ease of use, and technological innovation based on different criteria. Lastly, the group discussed that the production price would be 12.3 Euros, and the target price would be 20 Euros. Finally, they posed a quick challenge on using the magic pen and the board to draw design features.", "dialogue": "Speaker D: you should please want to sit down and show you it's not safe yet okay\nSpeaker C: so\nNone: okay Oh beautiful drawing\nSpeaker D: so this is our agenda you you're going to show your prototype presentation after me oh yeah so these are the oh okay yeah so these are the last notes we are made if anything doesn't look right to say to me then I don't have to put it in the report\nSpeaker C: you know we're doing the speech recognition because didn't have enough\nSpeaker D: time to design the inside as well okay but still possible financially so if you want to okay yeah what none then we're gonna put it in okay just yeah just so\nSpeaker B: we have to design the inside and but yeah or are we making a slide open like underneath or full open slide open is quite usual for remote control yes like underneath you can slide it open you can maybe that's stronger as well yeah\nSpeaker C: yeah that's it's okay it's better when you have a lot of room inside so you can make it very easy to use because you can write a lot of comments size okay so this is okay I'll go to you know we're gonna use an advanced chips then okay\nSpeaker D: so that's how much that is but okay for this yeah I think so I don't know now you have a different chip for speech recognition okay so I really calculated that and still in the budget so you can show your prototype if you want this\nSpeaker B: yeah together yeah let's do it together okay yeah we just made a word file with the basic elements the look and feel model well the form the case as drawn there simply a square with round corners so that's basically it the material should be hard plastic color changeable and also transparent and color and transparent or just transparent I don't know then the elements we have the functions are just basic like I've pointed a mirror mute function on off function text functions this switch channel okay cool and this is the the numpad and the logos over here in the mic okay and the scroll will know you operate that with your pointing finger so you hold it like this in your right hand and so how many things do you\nSpeaker D: need for for the microphone speech recognition yes speech recognition\nSpeaker C: there's only one button to say specification of that but I can imagine that you have to input your voice or something yeah maybe maybe you have to configure it yes you need options to finger it and often that you don't need\nSpeaker D: them anymore so you can put it on the back as well if you know you can all\nSpeaker B: are on the slides function I don't know okay well we haven't had time to design\nSpeaker C: that the slides we also don't know how many buttons are required or what kind of buttons but you have a lot of room if you can slide it open yeah I can imagine you need at least four buttons or something so the position yeah you write to you wrote this name the main zap buttons are more central this was the most important thing so best replace best reach place on off buttons text buttons mute buttons are together and that's a place they usually are easy to find the on off button is bit bigger so it stands out that way you don't have to make it right because it will show up score wheels on the left side it's basically the best standard place for school yeah but it's not impossible to\nSpeaker B: use it if you're left-handed so yeah because you can use your thumb then\nSpeaker A: just just one thing now you need to have more one tooth you've got one two three four five six seven eight nine yeah but you missed the no the zero yeah two\nSpeaker C: is try that's the other that then if I'm just so you get that okay it's rather important yeah okay this we just missed that but well just I'll get back to later the foreign well you have taken that from the iPod other proper technical device so should be popular with the buttons creating if you that are all round shapes not around these corners okay so that now you get a bit round feeling no use hard plastic since that allows us to use 2D buttons non-robor buttons color changeable more and the backlight thing the thing that lights up we have decided in the the channel buttons there's a little color around it okay and also in the numpad there was also color of light behind it and so\nSpeaker D: we still can do you still can choose what color you want yeah how do you want to implement that just to the maybe on the second level as well yeah these are just\nSpeaker B: basic functions so all the non-basic okay just roll the second level because\nSpeaker A: we need that okay there's one function I use daily and it's not on the basic functions it's to switch to your scarred PlayStation or DVD player okay maybe we use this button yeah maybe or you can lay it beneath and the other yeah as\nSpeaker D: well you should make video device like maybe just just draw second level one and say all options that are still left or something a second level like new\nSpeaker B: no just just here and there okay how does the second level come out it slides\nSpeaker C: I think yeah from from the you can do it as it claps open but I think that's not solid no that's right yeah that's right so it doesn't even have to slide all the way up so what do we need the speeds functions buttons yeah just menu menu button with maybe arrows so you can roll in the navigate the menu scarred well I think we can even put we have one for the zero and one for the one more digit so you can also you have one of the other one so this is the video which I know PlayStation etc that's used pretty often yeah like you have\nSpeaker B: PlayStation you used every day basic external or something you want to save that files well drawing so here are multiple speech buttons I don't know how many\nSpeaker C: doesn't really matter what else what else menu buttons what yeah like a normal like a normal like this yeah within the middle of a menu button yes okay well you don't have any anything on how many buttons speech you cry so you can't really design that okay maybe one button to switch the\nSpeaker A: color of your let's yeah yeah and you can hold it you can hold it and then the\nSpeaker C: color switch yeah multiple switch first again the color switch again maybe okay\nSpeaker D: oh we just make it three buttons all the colors on it red okay enough place and\nSpeaker A: then we can do that we can put as here color buttons and then we choose green blue and red yeah okay okay that's so did we miss anything yeah maybe some some\nSpeaker C: text some text that's more will that it is volume okay yeah but there's one\nSpeaker A: there's one text I hear yeah yeah okay there's more that's the one that you use if you search for a page like 700 and it's counting from 100 to 200 you will switch to your television and back yeah do you do you think of that yeah but\nSpeaker B: then you can switch back to normal tell it tell text you just switch it off and then just put it on a sec well we thought of text button and if you press it again you get the state the state the three stages yeah but but if you are in a second stage the third stage is switch tell text off so you can switch back\nSpeaker A: from second to turn off just don't just remember where it was it doesn't clear the page if you turn tell text on you set 700 and you turn it off then the next time you turn it on it still stays okay okay yeah but it's maybe it's not\nSpeaker D: the way that's a functionality for the television yeah I think as well but\nSpeaker B: that's maybe one yeah no the remote can send like the code for 700 page 700 to\nSpeaker C: the television yeah if you switch it on control in the yeah but just search\nSpeaker D: every time again that's what what happening if you do it like that yeah\nSpeaker C: that's true television it's good so yeah do we need to fix that or no that's what\nSpeaker A: those new TVs do collect all the pages yeah those memories I'm not not every\nSpeaker B: every television so yeah this logo for volume yeah\nSpeaker A: so that's it this is pretty much it yeah yeah I thought about one thing debatons from material already now just like your telephone yeah just hard plus okay because if you use it a couple of years some sometimes the numbers on the on the buttons are a slide away and maybe we can describe the numbers below or above or shall we just turn it on the buttons I think just if you do it above\nSpeaker C: above or below it takes that's too much place\nSpeaker B: yeah I think you have a problem more often with rubber buttons rubber\nSpeaker D: buttons yeah okay fine yeah okay yeah I don't know what this means but I think we just evaluated this one yeah I made some criteria okay you made some good\nSpeaker A: okay evaluate our model okay I did it don't think if it's right did I shall show it you're some usability criteria okay I know all criteria as we just argue about\nSpeaker B: in the bottom look at feel evaluation is yeah\nSpeaker A: information presentation it's not here it doesn't matter it only had two pages or something well I looked in the reports from the marketing strategy or of the new needs and the market the Italians how they think about it and the research about the the the users and that kind of stuff and I made some criteria and we have to test the criteria from one to zero we should we can give it a number and then we can give ourselves an average for our okay model and this these are all I found or I wrote down and we have to discuss about if we give it the one or a seven I think\nSpeaker D: if you have a kind of eyeball idea why beautiful yes we are the difference\nSpeaker A: between a beautiful and fancy look and feel is the the outside beautiful like the iPod or something and fences more like the the flashing lights and the\nSpeaker C: colors and that kind of stuff okay yeah I think we do if it's really you can if you can get the iPad look then it's beautiful yes what kind of what kind of\nSpeaker A: basic colors the basic colors are black or green or yellow\nSpeaker C: yeah how do we make maybe company colors just black black and yellow yellow light no not really but just a bit of yellow black white maybe like white\nSpeaker A: also like and what cars should the buttons be because different colors\nSpeaker D: this is just the same as the the cover but also the light can you change those to with switch no make them no just make them black or gray or something great\nSpeaker A: just dark gray okay okay so what number do we give a beautiful beautiful is really subjective because it has to do lots with the colors well we have changeable phones changeable phone so F4 F1 for everyone is something you just\nSpeaker D: give it a one it's okay it's perfect I think it's just what you want right now\nSpeaker C: it's hard to decide for us but yeah so it's a lot better than current remote\nSpeaker A: controls okay okay the fancy look and feel that's about our flashing lights and the background lights from the buttons yeah okay and we can change the color\nSpeaker B: so that's really fancy I think yeah one more thing are we changing or are they there are back lights on the slight panel too or no backlight slight panel yeah no\nSpeaker D: no it's only on the number behind the numbers and the switch channel is I mean\nSpeaker A: this here yeah but it's not necessary it's pretty cool if you slide it open it lights up that's that's really fancy but I don't know if it's reachable of course\nSpeaker C: it's reachable then we do it just some light to light it all up you can see what's really there not just not really on the buttons or something just green\nSpeaker B: light or from blue light but yeah yeah just backlight not not the buttons and and the normal backlights also not the buttons but behind the buttons so the\nSpeaker A: buttons are just great yeah okay fine so I think it's very fancy yeah\nSpeaker C: yeah I've done a lot of detail in light so yeah and you can also choose in light\nSpeaker D: so yeah this is one it's okay I think it's one okay next this is a difficult\nSpeaker A: one we don't know we don't know it about the I mean show tested but I it's very\nSpeaker D: easy to use but the second layer is not easy to use that's no but you don't\nSpeaker B: have to use that and you don't have to pay attention to that second layer that's\nSpeaker C: that's the main thing that's so good about I think it's easy to use learnable\nSpeaker A: learnable it's not not as fast as a usual remote control well I think it goes I think I think the scroll wheel is very handy but the first time you get this thing in your hands is not to use the scroll wheel I think you must seek for it and up or down but the rest of it is very easy because so so I think it's\nSpeaker B: very clear so few information that you can easily decide what button is for\nSpeaker C: what function second part like speech that's that will be hard to learn so it\nSpeaker D: is learnable and in first place it's very easy to use and I think it's scroll wheels you use it as well if you have ever used a different kind of\nSpeaker A: we've got to some two or three new things and maybe maybe learnable is in compare of old fashioned remote controls so we have speech the scroll wheel and the the slide you must slide it that's not normal at the normal remote controls I think learnable is less than easy to use because it's used comes after\nSpeaker C: I think it's very cool so just give it to maybe three then learnable no but\nSpeaker D: definitely better much better than than normal then one you showed is just all\nSpeaker A: buttons and you don't know okay new features technological innovative the speech function and the speech function new will the scroll wheel and the slide I think the slide is not new no I only saw it in a telephone not in an\nSpeaker D: oh I already have a VCR and it's from 1988 and the other slide and also why the\nSpeaker A: buttons come out as well okay okay and the lightning that's cool yeah for lighting is the scroll wheel I think speech speech is new different colors so yeah different fronts for remote control I think that's new too yeah so we have a pretty new there are no games only that's not a one it's a two again it's not\nSpeaker C: LCD and we also have the home station we are forgetting about that now but yeah\nSpeaker B: oh yeah that's why I'm so rechargeable now we didn't draw that to one but that's\nSpeaker D: more like yeah just draw it afterwards yeah that's just a normal simple thing if you can you save it on the same in the same map as the other ones in the the project map just say save us say that's not it's not that's not it well it's\nSpeaker B: already in the folder like number seven oh yeah okay yeah smart words you have a\nSpeaker D: this one this one is not yet in there I think it is\nSpeaker C: yeah I think it is\nSpeaker D: stealth is all the reason and your features so we give it a two or also again and one no I think you if you have games on it then then you get you're one we are not extraordinary new or something targeted audience we are the targeted\nSpeaker A: audience do we like it we we search for a young group yeah but young enough 40 so we are exactly the targeted group yeah but did we reach with our style the\nSpeaker D: targeted audience that's my question you get the fancy things for younger people yeah aesthetic things for older people we've got a one for fancy look and feel\nSpeaker B: and that's what attracts the yeah young audience so yeah that's the only point\nSpeaker A: is that we don't have that's already that's not that's this question yeah\nSpeaker B: but it does basically not not handy and I don't think I don't see yeah this so\nSpeaker A: targeted that we didn't follow the latest trends no yeah you could make a\nSpeaker E: I think you could make a front a front that's like a like a night or something\nSpeaker C: or my computer well food and vegetables you get different you get different\nSpeaker A: front so we had like a fruit banana kind of front but Spongy will never be no so we give herself a free or something okay so that's 11 that's what's our first 11 divided by six yeah this 1.83 up perfect score we're not\nSpeaker C: too hard on ourselves this is a power indicator okay you can see how far it's\nSpeaker A: charged up and then you need to a button to call it to let it be to go oh yeah\nSpeaker B: that's still yeah but we have to make a speaker then too if you want to make it be maybe I just skip that one it's no one that in but we can we can do it underneath the logo that's usable if you do it's really useful yeah the speaker\nSpeaker C: is very small as well yeah okay it's so I just got a financial\nSpeaker B: you say safety yeah I did save it\nSpeaker D: no okay if I look at this one here the production cost of it if I forgot anything say to me just a battery there are some things that they didn't mention because recharge is not on the list but okay so I think we are pretty much in the right direction because it's 12.3 euros okay but it's a regular chip and and a sample speaker okay so that's both no that's perfect 12.3 yeah and\nSpeaker B: single curve it but but is it does it include a home station or no that's not on the list but can we make that for 20 cents yeah we can probably I just the\nSpeaker D: button supplements I was wondering if this special color maybe was different fronts but standard front one but yeah special form yeah but maybe we have to yeah it's okay special color you can skip this one because it's all quite normal yeah you get different ones at all so you can put a recharge here\nSpeaker A: this is expensive the sample speaker yeah it's for it's four euros oh damn is\nSpeaker C: that included yeah that's kind of weird that we then we need to use we get this\nSpeaker D: information now afterwards because yeah yeah okay so this is pretty much hit this one damn solo sales so I was just I just want to yeah we just made it so we can do the project if you evaluation now for everything together okay we can do some discussion about this was there room for creativity sure yes okay\nSpeaker C: but also room for creativity yeah yeah I think so I think everyone share the\nSpeaker B: early so yeah if we got an eye mark for innovativeness or innovativity yeah we probably have things about it\nSpeaker A: we could make a lot of different remote controls yeah it's creativity\nSpeaker D: okay so the leadership was the real leadership of course the worst okay\nSpeaker B: what do you have to say about that yeah just normal discussion I think not yeah yes one leader or something one leader to check the time yeah and make notes yeah\nSpeaker D: I know so more like a secretary yeah okay next one team work the third meeting I think that one was pretty hard you were not oh we're not yeah agree with\nSpeaker B: every now we had so much information that we get through email and just I\nSpeaker C: think we got the wrong information at the wrong time I think that was the biggest problem yeah like the prices if you knew that before we could have yeah that's\nSpeaker A: really really really quicker yeah because the prices could be 20 euro or\nSpeaker C: something and if you had a good news then we would discourage yeah yeah finance so\nSpeaker B: we're basically basically just lucky to get the price right but the team\nSpeaker A: where it was okay yeah everybody could speak their opinion I think everyone\nSpeaker C: listened to each other yeah like marketing said then we had to include\nSpeaker D: them in the design yeah okay yeah what I have to say about means the smart board is okay did you pen is horrible if you use it but if you want to download it to\nSpeaker C: your computer it doesn't work just doesn't work no digital pen smart board would be very nice to work with if it's worked really real yeah it would be\nSpeaker A: faster drowning so more hard to make I think yeah it should be more accurate\nSpeaker C: precise and I think it would be great if you could edit it from just with a mouse when it's hitting yeah not just pointing out the same for the\nSpeaker D: presentation for the presentation you can do it from here that's my cheese and standing there yeah and see yeah I'm just point with a mouse no use to draw\nSpeaker C: on the board itself it's just slows down yeah just old-fashioned kind of black\nSpeaker B: board style but yeah they might as well do it in normal computer style even hard as you draw like this and it's far too slow this way yeah okay you could\nSpeaker D: draw on it but not as main function no I guess I think yeah okay so we made it in time\nSpeaker A: we did it the budget yeah new ideas found new ideas what's that four four four I\nSpeaker D: don't know what I mean just think if you to gather or to work together or new ideas for remote control probably no for the project format though favorites\nSpeaker A: for your text no for the prototype new ideas yeah but still you couldn't make\nSpeaker B: a fancy you couldn't make a prototype out of this because we don't have any sizes and move but it's for the next team yeah what you can possibly do that in such a short time I think yeah just this is the ID face yeah just brainstorming basically details so are we finished yes I think just just write a final report\nSpeaker D: okay quite early yeah no we have only four minutes left it's okay well do we\nSpeaker C: have to do them we were done at four o'clock it's now quarter past three so\nSpeaker A: I should take some pictures okay let's play my sweeper\nSpeaker C: one two three four five six seven cameras\nSpeaker B: so did we'll see it yeah now we can look at this this is we're probably not supposed to look at this yeah from the previous group they went for\nSpeaker C: for you know but also different devices yeah then then else deep would be any here are the basic functions and here selecting the devices yeah I do agree English is not so hard by the way no I'm breaking a girls record here well later got a problem Paul you do yes you have to make a choice no enjoy choice whoa that's pretty quick just take one yeah to the side it's the low one what's this a bomber no open one is the bomb this the bomb yes for in a row\nSpeaker A: no that's too much work come on said previous work yeah this one I challenge\nSpeaker C: you\nSpeaker A: that's so stupid not that doesn't work no you got to use the magic pen\nSpeaker C: hmm what if I put one yeah that's stupid we'll see okay I don't agree hmm so you had two choices that's gonna be drawn or not hmm too bad I'll put it here yeah then we're going to put it there damn no one wins difficult choice either here or there this is a very interesting\nSpeaker B: design\nSpeaker A: I mean it's just the same as normal they done and it has a LCD yeah I think then do\nSpeaker C: it correctly it's too good okay wait I'm going to draw something you must okay I'm going to guess what you're drawing okay blank no not a new one now I did just don't say I know how yeah you have to use the pen you have to save everything you know that thing not everything yeah everything pen select pen and once you know what we do in our spare time okay it's a house a plant no it's only you can know it only I can know it it's it's very hard to draw okay that makes it easier I think I know no wrong I think I know what you're trying to do but it's wrong already it's very you missed the right five\nSpeaker A: a little bit maybe\nSpeaker C: you where you're wrong now the entrance warning finish meeting now all right okay\nSpeaker A: the entrance is more to the left warning warning you're correct okay that I think\nSpeaker C: this part yeah I think I made another mistake oh yeah there are a lot of mistakes because the walls are thick like this no not that one but there's a big hole here\nSpeaker A: as well oh here that's the kind of big mistake they're walking behind the walls\nSpeaker C: warning finish meeting now guys I think we have to finish the meeting okay\nSpeaker A: this is a hard one\nSpeaker B: what does it say fill in the questionnaire what now come on okay okay no more chit chat over there yeah\nSpeaker C: I'm gonna be so lonely clean it up later this is that's my native face look up\nSpeaker A: this edited smiley\nNone: oh you\n"}, {"meeting_id": "TS3009c", "summary": "This meeting is about some final decisions on the design of the remote control. It started with presentations on the conceptual design from Marketing, Industrial Designer, and User Interface, and had a discussion on the decisions about the control. The Project Manager put a file in his folder to place the kind of decisions they should take.", "dialogue": "None: That's a goal.\nNone: That's a room for the report.\nSpeaker B: Hello guys.\nSpeaker B: The pin of the board is calibrated again.\nNone: So it should work now.\nNone: That's it.\nSpeaker B: Yeah.\nSpeaker B: Great.\nNone: I've started working on the end report.\nSpeaker B: Okay.\nSpeaker B: Door is closed.\nSpeaker B: Well, let's begin.\nSpeaker B: We have as much time as the last meeting.\nSpeaker B: We'll have to hurry up.\nSpeaker B: We'll start with the presentation again.\nSpeaker B: You.\nSpeaker B: Show.\nSpeaker B: This one I think.\nSpeaker B: Yeah.\nSpeaker B: I'll show you the notes.\nSpeaker B: We'll get your presentation again on the conceptual design.\nSpeaker B: Then we'll have to decide about the remote control concepts.\nSpeaker B: I've put a file in the project management folder which says exactly what kind of decisions we should take.\nSpeaker B: So this time we know what to decide about.\nSpeaker B: And then we'll close again.\nSpeaker B: Well, these are some examples but we'll talk about them later.\nSpeaker B: First look at your presentations.\nSpeaker B: Router will start again.\nSpeaker D: All right.\nNone: Trend watch.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: Right.\nSpeaker D: So, we'll have to take a look at the training.\nSpeaker D: It's very important that the control is fancy looking and good feeling.\nSpeaker D: This because of our last model was very functional.\nSpeaker D: But people didn't like that.\nSpeaker D: So our new model must be very good looking.\nSpeaker D: That's something you have to take a look at.\nSpeaker D: All right.\nSpeaker D: And also the menus and things like that.\nSpeaker D: They have to feel great.\nSpeaker D: There's minus two times here because this is the most important point.\nSpeaker D: This is two times less important.\nSpeaker D: Same for this one.\nSpeaker D: Technological innovations that's regarded very highly to such as an LCD screen and speech acknowledgement as we talked about earlier.\nSpeaker D: So we have to have something like that.\nSpeaker B: LCD and our front.\nSpeaker D: Right.\nSpeaker D: The last point is easy to use.\nSpeaker D: Well, I think that speaks for itself.\nSpeaker D: I don't know who's going to look at that.\nSpeaker D: Easy to use?\nSpeaker B: Well, easy to use is a bit, I should...\nSpeaker B: Contradicting with the first functional is not an issue and easy to use.\nSpeaker B: Well, we have to choose one of them.\nSpeaker D: I think we have to go for the first one.\nSpeaker B: It's the most important one.\nSpeaker D: So we have to take that one.\nSpeaker D: Okay.\nSpeaker D: So it isn't very important that it works easy.\nSpeaker C: It has to look great.\nSpeaker C: It can be easy to use.\nSpeaker B: Well, we'll look at...\nSpeaker A: Yeah, yeah, yeah, you can make a very complicated remote anyway.\nSpeaker A: So easy to use.\nSpeaker D: But the most important thing is that it looks great and people say from, wow, that's a great concept.\nSpeaker D: All right.\nSpeaker D: All right.\nSpeaker D: These are the new colors of this year.\nSpeaker D: So it must be very bright, very colorful.\nSpeaker D: People like this.\nSpeaker D: So we have to think in this direction.\nSpeaker D: So set your mind to it.\nSpeaker D: Well, findings, fashion update.\nSpeaker D: Fruit and vegetables are cool.\nSpeaker D: I'm told.\nSpeaker D: The group we are targeting is very pleased with fruit and vegetables.\nSpeaker D: So we might consider in front of...\nSpeaker D: In that sort of way.\nSpeaker D: Furthermore, material, that's your part, should be very strong.\nSpeaker D: I was thinking of something like, well, iron plate over it, maybe in a color or something that looks really flashy, but it is also strong.\nSpeaker D: All right.\nSpeaker D: And that's also for the younger public.\nSpeaker B: The handy thing about our phones is that we can follow these trends every year.\nSpeaker B: This year it's fruit, but next year it's something totally different.\nSpeaker D: Yeah, that's great.\nSpeaker D: But I think we can all make the fronts of titanium or something really thin.\nSpeaker D: So it looks very heavy, but you can still use it very easily.\nSpeaker E: Yep.\nSpeaker E: All right.\nSpeaker D: Well, the don'ts.\nSpeaker D: All the people like dark colors and simple shapes.\nSpeaker D: Well, we don't want all the people.\nSpeaker D: We want young people.\nSpeaker D: So we're going to turn that around.\nSpeaker D: We're going to have real cool shapes and lots of colors.\nSpeaker D: Right?\nSpeaker D: We don't want wood.\nSpeaker D: Yeah, among the old people.\nSpeaker D: So that's it for me.\nSpeaker B: All right.\nSpeaker B: Nice.\nSpeaker B: Well, Tyse, show us.\nSpeaker A: Right, I'm going to tell you something about components design.\nSpeaker A: Again, I've put up the specification properties.\nSpeaker A: So the different components of the device and the materials.\nSpeaker A: I've heard several things, so I'll have to change that on the way.\nSpeaker A: But the case, I suggested in a previous meeting, hard plastic.\nSpeaker A: But as you indicated, it should be strong.\nSpeaker A: It should feel strong.\nSpeaker A: So maybe plastic is not sufficient.\nSpeaker A: We should move through something.\nSpeaker B: Maybe it is, but it doesn't look strong.\nSpeaker A: Well, yeah, hard plastic is, of course, pretty tough, but it doesn't have a really, really tough look.\nSpeaker B: We still have to look at our price, of course.\nSpeaker B: Also, we want to announce the D, window, etc.\nSpeaker A: But we'll return to that.\nSpeaker A: The buttons, of course, rubber, I think, everyone agrees.\nSpeaker A: Electrical cables, copper is all pretty basic stuff.\nSpeaker A: The chip made of silicon, I guess.\nSpeaker A: I think that's the best way to do it.\nSpeaker A: Infrared lettuce, just a simple bulb.\nSpeaker A: Then I've had a few findings, made a few findings, a target audience product style.\nSpeaker A: It's generally the case that senior and wealthy people, above 45 years old, like, as you said, particularly the traditional materials, such as wood, and materials such as that.\nSpeaker A: They're also like straightforward shapes and luxurious style.\nSpeaker A: Of course, that's not our thing, so this is things we must not do.\nSpeaker A: And then we have young and dynamic people, which is, of course, our group, the people we aim at.\nSpeaker A: Under 45 years old, they like soft materials with primary color.\nSpeaker A: So, often, there is, of course, again, a bit of contradiction with our material choice of what you said, that should be hard and strong looking.\nSpeaker A: But they like soft materials, so we might have to consider that.\nSpeaker A: Also, they like curved round shapes, so not to do formal like the other people want.\nSpeaker A: And if also finding, but not very applicable here, that sports and gaming devices, such as disc men's for joining, those kinds of devices, game devices, should define the characteristics of the device.\nSpeaker A: But since we don't have really a sports or gaming device, we don't really have to consider that.\nSpeaker B: Sports, they're...\nSpeaker B: I actually won on your LCD window.\nSpeaker B: That's nice.\nSpeaker B: I also have results.\nSpeaker D: We keep coming back to the front.\nSpeaker A: Several examples of...\nSpeaker A: I'm sure you can get a clear picture of what I mean.\nSpeaker A: These are the basic older people's stuff.\nSpeaker A: It's not very interesting, very classical looking, but that's not what we want.\nSpeaker A: We have these kind of things.\nSpeaker A: I don't know what exactly they are.\nSpeaker A: You know you recognize shapes, it's very primary colors, bright colors, round shapes.\nSpeaker A: You also see this device, it's very round.\nSpeaker A: Fruity.\nSpeaker A: Fruity, of course.\nSpeaker A: It's terrible.\nSpeaker A: It's true.\nSpeaker A: And round shapes, brand new colors.\nSpeaker A: You can see it over here.\nSpeaker A: And of course, this famous device.\nSpeaker A: I guess you know something, some devices like this.\nSpeaker A: So to give you an idea of...\nSpeaker A: Strong look, this.\nSpeaker A: This has a strong look.\nSpeaker A: Although it's plastic, it's great to look to give the iron look.\nSpeaker D: And it's round.\nSpeaker D: But then you're losing your fruity colors.\nSpeaker B: Yeah.\nSpeaker B: You can make that middle ring, you can make it on the color.\nSpeaker B: That's true.\nSpeaker A: We can really make a round remote control.\nSpeaker A: I don't think that's very practical.\nSpeaker A: But it's important to think about the color.\nSpeaker A: Because if you make it gray or silvery looking, it does make it a lot more...\nSpeaker A: It does make it looking a lot more strong.\nSpeaker A: Because if you look at this, it doesn't look very strong.\nSpeaker A: But this is plastic.\nSpeaker D: But it does look strong.\nSpeaker D: The results are the feel of the material is expected to be strong.\nSpeaker A: The feel.\nSpeaker D: All right.\nSpeaker C: Well, if you...\nSpeaker C: It doesn't have to be strong.\nSpeaker C: Well, the feel.\nSpeaker B: You maybe you shoot some colored titanium or something.\nSpeaker B: So it looks fruity, but it feels strong.\nSpeaker D: Right.\nSpeaker D: I agree.\nSpeaker A: Then I have some more findings about the energy energy source of the device.\nSpeaker A: I suggest the basic battery.\nSpeaker A: I got some other energy sources, of course.\nSpeaker A: But solar energy is not very practical inside the house.\nSpeaker A: Because you don't have a lot of sun.\nSpeaker A: And kinetic and dynamos are not very practical, I think, for a simple remote.\nSpeaker A: It's a bit...\nSpeaker A: Oh, that's not...\nSpeaker A: No titanium.\nSpeaker A: That's a bit much.\nSpeaker A: And I also suggest as a shape, a double curved case, the disadvantages of that, that you can use no titanium.\nSpeaker A: That's the information I received.\nSpeaker A: If you use the curved case, a curved case, a double curved...\nSpeaker A: What do you mean double curved?\nSpeaker A: Now this, to give the more modern look.\nSpeaker A: Now the shape, yeah, a curved case.\nSpeaker A: Yeah, I think this is sort of a triangle shape, the bottom or something.\nSpeaker A: If more modern look, not plain, long box style, but...\nSpeaker A: Double curved.\nSpeaker A: I don't know.\nSpeaker A: I'll draw it, but maybe later.\nSpeaker A: And it makes...\nSpeaker A: It gives it the more European-like shape.\nSpeaker A: Anyway, for the buttons, simple push buttons, no other eye, no difficult scroll things or things like that, because it makes more complex and expensive.\nSpeaker A: And as we agreed, we don't use a speaker or a sensor or a speech.\nSpeaker A: Right.\nSpeaker A: And it controls the device, because it makes it also more complex and expensive.\nSpeaker A: But we do use an LCD screen, so we do have to consider, we have to use a more advanced chip, which is more complex and expensive, but it's worth the trouble, I think.\nSpeaker C: The box can be made of in a soft material, because people like that.\nSpeaker B: It's soft material.\nSpeaker B: And rubber is a soft material, I guess.\nSpeaker A: Yeah, right.\nSpeaker A: Soft enough.\nSpeaker A: So that's basically what I'm going to talk about.\nSpeaker B: Okay, we'll take that, and then Mike.\nNone: Okay.\nSpeaker B: Yeah.\nSpeaker C: Well, now the method, we will include the buttons as we discussed earlier.\nSpeaker C: And in also days, the screen will be implemented.\nSpeaker C: We must decide where this meeting is.\nSpeaker C: There are new developments in speech recognition systems, and they are already being used on coffee machines.\nSpeaker C: And, well, they're cheap, so we could use them now.\nSpeaker C: That's interesting.\nSpeaker C: It's not really speech recognition.\nSpeaker C: It's more like you can talk to the chip, record a message, and record an answer.\nSpeaker C: And then once you talk to the remote then, he will answer with the pre-recorded message that you will have.\nSpeaker C: So, we can say, hi, Mike, and you have recorded hi, Mike, back, and you will get there.\nSpeaker B: Okay.\nSpeaker B: But you can also say that when you say something, it does something.\nSpeaker B: It doesn't, though.\nSpeaker B: It does not do anything.\nSpeaker D: It makes it just an extra function.\nSpeaker B: Understand, but there's no functionality for remote.\nSpeaker C: No, but that's the gadget they were.\nSpeaker A: But it's young people.\nSpeaker A: It's nice for young people.\nSpeaker C: They like it.\nSpeaker C: You should really include that one.\nSpeaker C: If it's cheap, yeah.\nSpeaker C: Well, as I said earlier, I think the LCD screens will be positioned at the lower end of the remote.\nSpeaker C: The buttons for screen width and channel settings and that kind of stuff.\nSpeaker C: We can also do that kind of functions shown in the LCD screen.\nSpeaker C: Instead of extra buttons, I think young people and, yeah, well, every user would like that.\nSpeaker C: The buttons should be positioned the same way as they are on conventional remote, I think, for the learnability and, well, to keep it recognizable.\nSpeaker C: For its recognition, it can be implemented.\nSpeaker C: I do an example, but it did not work quite the way I...\nSpeaker A: Can you draw it now?\nSpeaker C: I can draw it again, and I know what I did wrong.\nSpeaker C: I didn't take the notebook box.\nSpeaker D: How do you give input to the menu on the LCD screen?\nSpeaker C: With the up and down and...\nSpeaker D: All right, so you have a menu button and then you can go.\nSpeaker B: I will draw what I have on the screen.\nSpeaker B: And OK button.\nSpeaker C: Yeah, right.\nSpeaker C: I should make the button.\nSpeaker C: If it works.\nSpeaker B: Just...\nSpeaker B: There is already a blank, yes.\nSpeaker C: Yeah.\nSpeaker C: So, why?\nNone: You have to do it.\nNone: I don't know.\nSpeaker D: You have to push hard.\nSpeaker A: I suggest a banana shape because of the fruity fashion.\nSpeaker A: No.\nSpeaker B: Yellow and...\nSpeaker A: That's out.\nSpeaker A: Yeah, all right.\nSpeaker A: Just a hunch.\nSpeaker C: Well, these little buttons are a bit difficult to draw.\nSpeaker C: Those are the channel buttons.\nSpeaker C: Yeah, just the numbers.\nSpeaker A: And then below is the LCD screen.\nSpeaker C: These will be bigger in the real design.\nNone: All right.\nSpeaker C: This must be the OK button used to interact with the LCD screen.\nSpeaker C: And with this you can go through the menus.\nSpeaker C: Your video button should be a part button because you want to use it fast within one click.\nSpeaker C: And you want to use the speaker for the...\nSpeaker C: This button can also be the menu button.\nSpeaker C: How do you get out of the menu again?\nSpeaker B: Maybe I...\nSpeaker A: By pressing the menu button again.\nSpeaker A: By pressing the menu button again, you go out.\nSpeaker A: It's usually the kind of...\nSpeaker B: Yes, because when you push menu, you get in.\nSpeaker B: And then you have to push OK when you get to a choice.\nSpeaker A: But you can press menu again to get out.\nSpeaker B: Well, that's also the OK button.\nSpeaker C: No, you should have the extra menu button.\nSpeaker B: Or you can put in the LCD window and option, get out.\nSpeaker B: Exit.\nSpeaker C: Now, once you have an extra menu button, you don't need that extra option.\nSpeaker D: Well, we need a recording button for the speech part.\nSpeaker C: If we decide to implement it.\nSpeaker A: Why would you put it then?\nSpeaker A: Where is the recording, the microphone?\nSpeaker C: Well, that could be anywhere. That's very small. It could be down here.\nSpeaker C: Well, I'm not here. I suggest here.\nSpeaker C: But that's just a little speaker at the back.\nSpeaker C: Well, the speaker in the microphone, I think, are the same little whole thing.\nSpeaker E: I understand.\nSpeaker A: But we could do what is perhaps more expensive.\nSpeaker A: Well, speaker on the back or something.\nSpeaker C: There are already being implemented in coffee machines, so they won't be very expensive.\nSpeaker C: All right.\nSpeaker C: This is my suggestion.\nSpeaker B: Well, OK.\nSpeaker B: Then let's have a look at the decisions we are going to have to make.\nSpeaker C: I think, as you can see, the LCD screen does look better at the lower end.\nSpeaker C: I agree.\nSpeaker D: That's fine.\nSpeaker D: That's fine.\nSpeaker A: Well, yeah.\nSpeaker C: Oh, that's one of my examples.\nSpeaker C: I can live with it.\nSpeaker C: Yes.\nSpeaker C: But I did not like it very much.\nSpeaker C: Well, these are quite obvious, very ugly remotes.\nSpeaker A: They don't look fruity enough.\nSpeaker A: No, they're not really.\nSpeaker A: They're all black.\nSpeaker C: Yeah.\nSpeaker C: Well, not all.\nSpeaker B: Hey, that one that I like.\nSpeaker D: It's a special one, but it doesn't look strong.\nSpeaker A: But it doesn't...\nSpeaker A: With the colors, it's been the way we're going to.\nSpeaker C: Yeah, right.\nSpeaker C: This is terrible.\nSpeaker E: It's too much.\nSpeaker A: It's not too childish, of course.\nSpeaker C: These are the LCD screens.\nSpeaker C: I think we should, if it's possible, one with colors.\nSpeaker C: But I don't know.\nSpeaker C: This is expensive.\nSpeaker E: All right.\nSpeaker B: All right.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: OK.\nSpeaker C: Even more.\nSpeaker B: No.\nSpeaker D: But they're going for a strange form.\nSpeaker D: No, it's not too strange.\nSpeaker A: Not too strange.\nSpeaker A: Not too strange.\nSpeaker A: It's still nice.\nSpeaker D: The underside, you can make it more round.\nSpeaker D: Yes.\nSpeaker B: Where the LCD is.\nSpeaker B: A kind of bridge.\nSpeaker D: I don't know if you know the no-key at the telephone.\nSpeaker D: With the round thing at the bottom.\nSpeaker D: Something like that.\nSpeaker D: No?\nSpeaker B: Well, I have at home a remote.\nSpeaker B: If a bridge is just a half round, a half circle, and then it falls exactly over the hand.\nSpeaker B: And that's very nice.\nSpeaker B: That feels comfortable.\nSpeaker D: Yeah.\nSpeaker D: But people like something new.\nSpeaker D: Exotic.\nSpeaker A: Yeah.\nSpeaker A: We have friends.\nSpeaker A: I will design it.\nSpeaker A: We design it later.\nSpeaker A: So we'll get to that later.\nSpeaker B: All right.\nSpeaker B: Where did I put it?\nSpeaker B: There's a specific shape.\nSpeaker B: Conceptual face.\nSpeaker B: I think this is it.\nSpeaker B: I got this from our friends.\nSpeaker B: So.\nSpeaker B: Our sources.\nSpeaker B: Yeah.\nSpeaker B: Conceptual design.\nSpeaker B: These are a few examples which we have to decide about.\nSpeaker B: All the materials from the case.\nSpeaker B: The electric cable.\nSpeaker B: That's all your side of the story.\nSpeaker B: Your bag.\nSpeaker B: Yeah.\nSpeaker B: Now, from the user interface, your package.\nSpeaker B: Where in the world it's more like the buttons where they have to come.\nSpeaker B: And a bit of.\nSpeaker B: Yeah.\nSpeaker B: Well, a bit of design.\nSpeaker B: Yeah.\nSpeaker C: This is what we've just done.\nSpeaker C: But we should decide about these now.\nSpeaker B: All right.\nSpeaker B: Yeah.\nSpeaker B: Material.\nSpeaker B: So, most important.\nSpeaker B: So, as you said, Fruity is in netwell, polarity cells good.\nSpeaker B: Wow.\nSpeaker B: These facts.\nSpeaker A: So we have to put it in one document.\nSpeaker A: Yes.\nSpeaker B: So, if we go through.\nSpeaker A: Copy-based this story into a word documented and put the answers after the subjects.\nSpeaker A: Yeah.\nSpeaker A: Not everything.\nSpeaker C: We have to decide what we think.\nSpeaker A: Yeah.\nSpeaker C: But all these examples are of a coffee machine.\nSpeaker C: Yeah.\nSpeaker A: Well, why should I override them?\nSpeaker A: Yeah.\nSpeaker E: Yeah.\nSpeaker A: Well, in case that's me, I suggest, well, what do I suggest actually?\nSpeaker B: What kind of properties should it have?\nSpeaker B: Well, we just listened.\nSpeaker B: I think we did.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Why don't we use titanium or hard, yeah.\nSpeaker C: Some kind of metal for the.\nSpeaker C: Do you know the new remote?\nSpeaker C: Except the front.\nSpeaker C: Just like.\nSpeaker C: Yeah, I understand.\nSpeaker A: Yeah.\nSpeaker A: So, we have tight the front.\nSpeaker A: Yeah, the front is the most important.\nSpeaker A: Yeah, but the non-removable elements of the remote, so not the front, could be titanium to get that strong look.\nSpeaker A: And then the front is made of plastic.\nSpeaker A: And you can put that on and off and switch it.\nSpeaker D: Yeah.\nSpeaker D: But the feel of plastic isn't strong.\nSpeaker D: No, but you have titanium, of course.\nSpeaker A: So you have the best of both worlds.\nSpeaker C: Okay.\nSpeaker C: So you have to do a remote in your hand like this, so you feel titanium.\nSpeaker A: And of course, yeah, you have the plastic front end, but you also have the titanium.\nSpeaker D: Fronts are cheaper than.\nSpeaker A: Yeah, of course, but yeah, you have to make.\nSpeaker C: And I don't know if you can make steel just any way you want it to.\nSpeaker D: Yeah, but it's expensive.\nSpeaker A: I guess so, titanium, I think.\nSpeaker A: Bendable.\nSpeaker A: Huh?\nSpeaker C: Bendable.\nSpeaker C: Well, any color.\nSpeaker A: I'm going to my source.\nSpeaker A: It's totally possible to make an entire, titanium is available and we can make an entire remote out of it.\nSpeaker C: Yeah, you paint it in the color you want.\nSpeaker C: Yeah, the plastic is the color you want.\nSpeaker C: Yeah, right.\nSpeaker D: So we're going for titanium back and a.\nSpeaker A: I think that's a great front.\nSpeaker B: Okay.\nSpeaker B: Well, I'm going to put it in here because so we can.\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker B: Solid fuel.\nSpeaker B: And trim the new cell material.\nSpeaker B: And hard plastic.\nSpeaker A: Yeah, for the front and titanium for the back.\nSpeaker A: For the non-removable part.\nSpeaker D: But then you have the problem.\nSpeaker D: When you have titanium back, you can switch it.\nSpeaker D: When you want to add a color on the front, it doesn't match.\nSpeaker C: Well, but.\nSpeaker C: Titanium is neutral.\nSpeaker A: Titanium isn't very, it doesn't, it's not a curse.\nSpeaker A: No, titanium is very, no, no, no, but titanium is a very basic color.\nSpeaker A: And it doesn't really matter if you have a purple front on it or in front.\nSpeaker A: I don't know if you disagree, but I think it doesn't matter.\nSpeaker C: Yeah.\nSpeaker C: Funky, too many colors.\nSpeaker A: And even if it doesn't match, people like the customers.\nSpeaker A: Okay.\nSpeaker B: Well, the electric cable is just normal.\nNone: The beans come from the ones in the third thing.\nSpeaker B: Excuse me.\nSpeaker B: The electric cable is just a coffee.\nSpeaker C: We don't use them in electric cable.\nSpeaker C: Inside, but this is for a coffee machine.\nSpeaker A: Yeah, well, inside the remote control.\nSpeaker C: A couple of, it's not worth spending.\nSpeaker C: Oh, well.\nSpeaker D: So external.\nSpeaker A: Oh, external?\nSpeaker A: Yeah.\nSpeaker A: We do, we do.\nSpeaker A: Coffee grime doesn't it?\nSpeaker A: Yeah.\nSpeaker A: Never mind the coffee grime.\nSpeaker B: Well, all the inside work of our remote is standard work.\nSpeaker B: So.\nSpeaker B: Yeah.\nSpeaker B: Right?\nSpeaker B: The chip is normal in the silicon.\nSpeaker B: The buttons are normal.\nSpeaker B: That's all right.\nSpeaker B: Okay.\nSpeaker B: So that's just easy.\nSpeaker B: We decided about that.\nSpeaker B: Just by looking at our competitors and our earlier remote.\nSpeaker B: The conceptual specification of the user interface.\nSpeaker B: Well, we have our beautiful drawing.\nSpeaker C: I've got a better one here.\nSpeaker C: All right.\nSpeaker B: Well, you can put that into the shared folder and then I'll put it in our end report.\nSpeaker C: I will work this out.\nSpeaker C: Yes.\nSpeaker B: You can.\nSpeaker B: You can put some which button is what?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: The trend watching.\nSpeaker B: Proteries shows.\nSpeaker B: Good.\nSpeaker B: These days.\nSpeaker B: And what do we think that fruit and bright colors are?\nSpeaker A: Yeah.\nSpeaker D: I think we can launch a couple of packages.\nSpeaker D: You can buy a different kind of machine, but it's the same thing with another front.\nSpeaker B: Yes.\nSpeaker B: So it's an implement.\nSpeaker B: You can also implement the fronts from movies that are very hot.\nSpeaker B: Yeah, right.\nSpeaker C: That's for later on.\nSpeaker C: The fronts you can do anything with.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But if you launch five different packages like I put mini.\nSpeaker C: For the initiative launch.\nSpeaker A: Yeah.\nSpeaker A: For the launch.\nSpeaker A: It's good marketing.\nSpeaker A: And then a couple of basic colors.\nSpeaker A: Not very.\nSpeaker A: Yeah.\nSpeaker A: Not too.\nSpeaker D: You can always take another pick.\nSpeaker B: Yeah.\nSpeaker B: We should not give them the most lovely fronts.\nSpeaker B: When they were later on.\nSpeaker B: Right.\nSpeaker B: Come on.\nSpeaker B: We're basic.\nSpeaker B: We still have to make those 50 million.\nSpeaker B: Very boring.\nSpeaker A: The most boring.\nSpeaker B: You can give them a.\nSpeaker C: It's possible.\nSpeaker C: We can experiment with it.\nSpeaker C: Yeah.\nSpeaker A: They want more.\nSpeaker A: Yeah.\nSpeaker A: You can have some basic colors.\nSpeaker A: And then we come with special patterns on them.\nSpeaker A: Red and green.\nSpeaker C: You can give them.\nSpeaker C: Yeah.\nSpeaker B: All right.\nSpeaker B: Well, the buttons, et cetera, we get from Mike.\nSpeaker B: This fruit and bright colors.\nSpeaker B: Yeah.\nSpeaker B: Well, I think we'll have to in the next half an hour.\nSpeaker B: We'll have to specify the different types we want to launch when we.\nSpeaker B: Well, introduce our remote.\nSpeaker A: We still have to make the real basic design.\nSpeaker D: Because we have the sketchbook.\nSpeaker D: We must remember that fancy look and feel is the most important thing.\nSpeaker D: Yes.\nSpeaker D: Also, it won't sell.\nSpeaker D: All right.\nSpeaker B: Well, I'll have to pick four.\nSpeaker B: I get another warning from five minutes.\nSpeaker B: I'm going to get.\nSpeaker B: Where is my mouse?\nSpeaker B: Where is my mouse?\nSpeaker B: I lost my mouse.\nSpeaker D: Oh, yeah.\nSpeaker B: This is it.\nSpeaker D: I'm really cool.\nSpeaker B: Well, this we have.\nSpeaker B: Basic stuff.\nSpeaker B: Interface we have.\nSpeaker B: Supplements LCD.\nSpeaker B: Maybe a cheap voice recording.\nSpeaker B: Well, you should be good.\nSpeaker B: Yes.\nSpeaker B: All right.\nSpeaker A: Individual actions.\nSpeaker B: We have the industry designer.\nSpeaker B: Use your interface, Mike.\nSpeaker B: You're going to work together on a prototype drawing on the smartboard.\nSpeaker C: We can do that.\nSpeaker A: Together?\nSpeaker A: Not together.\nSpeaker C: That's what I got to hear.\nSpeaker B: We'll get it to hear the thing.\nSpeaker B: I think there's a smartboard.\nSpeaker B: You can take the smartboard.\nSpeaker B: This is the smartboard.\nSpeaker B: So, take it to our rooms.\nSpeaker B: You're going to make a prototype.\nSpeaker B: That includes specifying the buttons.\nSpeaker B: It's a search engine.\nSpeaker B: You'll get it on your laptop.\nSpeaker B: You will go and do something else.\nSpeaker B: That's fun.\nSpeaker B: Project information.\nSpeaker B: Are you going to do?\nSpeaker B: I don't know what product evaluation exactly means, but you'll get the specifications.\nSpeaker D: We don't have product yet.\nSpeaker B: That's why I got it.\nSpeaker B: You're fired.\nSpeaker B: How long do we still have?\nSpeaker A: Well, can we talk about something else?\nSpeaker B: Yes.\nSpeaker A: No, I don't know anything.\nSpeaker C: We'll use round balance or square ones.\nSpeaker C: Round, I think.\nSpeaker C: To make it as round as possible.\nSpeaker C: These buttons are more...\nSpeaker D: So, you can see you up and down.\nSpeaker D: But now I see the...\nSpeaker C: This program.\nSpeaker C: We can't, up and down this program and left and right.\nSpeaker C: Well, that depends.\nSpeaker C: Well, we...\nSpeaker A: We'll just give them and we'll design it and let it give comments on it.\nSpeaker B: It doesn't work anymore.\nSpeaker B: We can save them.\nSpeaker D: You can click the corner.\nSpeaker B: It's a bit...\nSpeaker B: It's really pretty.\nSpeaker B: Well, we still have more than five minutes.\nSpeaker B: So, what are we going to do?\nSpeaker B: I ask for a minute.\nSpeaker B: Don't have a clue.\nSpeaker B: Let's start the design.\nSpeaker B: Oh, we can decide how we implement the feeling from our company into the remote.\nSpeaker A: Well, yeah, I think a logo, our company logo and the slogan should be...\nSpeaker A: Also, just start with...\nSpeaker A: Why not?\nSpeaker A: If there's enough space, then you can put the wheel.\nSpeaker C: I would say only the logo.\nSpeaker C: Yeah, me too.\nSpeaker C: Too much text and...\nSpeaker A: Well, our slogan is not very long.\nSpeaker A: It's just as simple.\nSpeaker C: What is the fashion in electronics?\nSpeaker B: We put the fashion in electronics.\nSpeaker D: We keep adjusting to the fashion with our fronts.\nSpeaker A: But you can put it on the back of the titanium part.\nSpeaker A: The logo and the...\nSpeaker C: Yeah, the logo should be on the top, I think.\nSpeaker B: On the top.\nSpeaker A: Yeah.\nSpeaker B: In the right top corner.\nSpeaker B: Right corner, yes.\nSpeaker B: Right corner.\nSpeaker A: In the back, you can put...\nSpeaker B: Maybe here in the middle.\nSpeaker B: At the bottom, you can put the logo with the text.\nSpeaker E: Yeah.\nSpeaker D: The logo shouldn't be exchangeable when you...\nSpeaker D: Yeah, it should be hard on the board.\nSpeaker C: You can...\nSpeaker C: The logo can be on every front.\nSpeaker D: Yeah, but you can scratch it off or something.\nSpeaker D: It's better if you have...\nSpeaker C: Yeah, but then you must...\nSpeaker C: Carved into the material.\nSpeaker C: Push it in or something.\nSpeaker B: Now, you can carve it into the titanium at the back.\nSpeaker A: I like the idea of the slogan on the...\nSpeaker A: On the thing, but...\nSpeaker A: Am I the only one?\nSpeaker C: Well, management.\nSpeaker C: I like it.\nSpeaker C: Not on the front side, I think.\nSpeaker B: The front side?\nSpeaker C: No, no, no, no.\nSpeaker C: No, no, no.\nSpeaker C: Push it in.\nSpeaker A: I think it's a nice idea to make it more recognizable.\nSpeaker A: Yeah.\nSpeaker A: Next to the logo, you have to slogan.\nSpeaker A: I agree.\nSpeaker A: To make more of the...\nSpeaker A: But not too big.\nSpeaker A: Not too big.\nSpeaker A: No, very small.\nSpeaker A: Not here.\nSpeaker A: On the entire back, but just very small.\nSpeaker A: But readable and out, of course.\nSpeaker A: Okay.\nSpeaker A: Yes, slogan.\nSpeaker A: But we'll take that...\nSpeaker C: With us and take the design.\nSpeaker C: Over here.\nSpeaker C: The logo.\nSpeaker A: Or maybe here in the middle.\nSpeaker A: What we'll decide like.\nSpeaker B: With logo and logo.\nSpeaker B: Also on the front, but not X change.\nSpeaker B: So, oh my God.\nSpeaker B: When changing fronts.\nNone: All right.\nSpeaker B: Do you see a bit of the...\nSpeaker A: Five minutes left.\nSpeaker A: Of the titanium on the front.\nSpeaker A: Maybe if we make this lower part, titanium, the front is the upper part.\nSpeaker A: And the bit with LCD screen.\nSpeaker A: So, a bit of titanium between?\nSpeaker B: No, no, between.\nSpeaker A: That's pretty cool.\nSpeaker A: But then you have two parts of four.\nSpeaker A: Two parts.\nSpeaker A: Two parts to make this titanium too.\nSpeaker A: Yeah.\nSpeaker D: The entire...\nSpeaker D: You got problems with LCD.\nSpeaker D: Like dust in it.\nSpeaker D: Things like that.\nSpeaker D: When you exchange all the fronts.\nSpeaker B: I already have all kinds of...\nSpeaker B: And then not filled between the...\nSpeaker B: Mobile.\nSpeaker C: And then the lower part is titanium.\nSpeaker C: This is titanium.\nSpeaker C: I think that's nice.\nSpeaker D: Some kind of way.\nSpeaker B: Yeah.\nSpeaker B: Round forms.\nSpeaker A: Yeah.\nSpeaker A: I like that a bit of a titanium also.\nSpeaker A: Yeah.\nSpeaker A: A bit like your mobile phone.\nSpeaker A: Maybe you can show it as the...\nSpeaker A:... those things.\nSpeaker A: A bit like this.\nSpeaker D: Yeah.\nSpeaker D: You can make it go...\nSpeaker D:...run this corner too, where the logo is.\nSpeaker D: Because there's a lot of time.\nSpeaker A: I...\nSpeaker A: That's a nice touch here.\nSpeaker A: A little corner of titanium.\nSpeaker C: Oh yeah.\nSpeaker B: That's all the sounds pretty.\nSpeaker B: So this...\nSpeaker A: I think I'm going to buy it.\nSpeaker A: This is the...\nSpeaker A: We want it.\nSpeaker B: And it's only 25 euros.\nSpeaker B: Come on.\nSpeaker A: That's a bit too much.\nSpeaker A: No, it's nothing.\nSpeaker A: No, no, no.\nSpeaker A: I think it looks pretty nice.\nSpeaker A: Of course.\nSpeaker A: Because it's my design.\nSpeaker A: Our design.\nSpeaker A: Take all the credit.\nSpeaker B: Well, if you are going through it together, you'll get your specifications on your laptop and then...\nSpeaker B: Yeah.\nSpeaker B: We'll stay here, I guess.\nSpeaker D: But there's a problem.\nSpeaker D: We can take a blank one.\nSpeaker D: What can we?\nSpeaker B: Well, I think we can...\nSpeaker B:... I just proved it.\nSpeaker C: Well, we can erase an animal, I guess.\nSpeaker C: Yeah.\nSpeaker C: That's right.\nSpeaker A: But don't erase my cat.\nSpeaker C: Yeah.\nSpeaker C: All fourth one is empty.\nSpeaker C: Oh!\nSpeaker A: I want to preserve it.\nNone: Thank you.\nSpeaker D: Yeah, this one is empty.\nSpeaker D: Yeah, it's empty one.\nSpeaker B: Oh, we have a pre-empty one.\nSpeaker B: What are you doing, Chief?\nSpeaker B: So, you can draw.\nSpeaker C: Yeah, I think we have to wait.\nSpeaker E: No, I'll get the message.\nSpeaker A: I'll get the message.\nSpeaker A: But I don't...\nSpeaker A: Do we have to stay here?\nSpeaker A: I think now.\nSpeaker A: I think you can...\nSpeaker B: Well, maybe you can keep your laptop here.\nSpeaker B: Maybe.\nSpeaker B: Or get your mouse.\nSpeaker B: Because it's...\nSpeaker B: The hard power from a bone.\nSpeaker B: Hard to work with these plates.\nSpeaker D: We'll have to talk here.\nSpeaker D: It's not relaxing.\nSpeaker B: I always have a mouse next to my laptop.\nSpeaker B: I hate these.\nSpeaker B: Yeah, touchpad.\nSpeaker A: Yeah, touchpad.\nSpeaker A: We can do a touchpad on our remote.\nSpeaker A: Yeah, right.\nSpeaker A: Just kidding.\nSpeaker A: Oh, you have to have a bit of humor in the...\nSpeaker A:... in the discussions too.\nSpeaker A: Look, I see it becomes too...\nSpeaker A: And in your remote control.\nSpeaker B: So we put a touchpad on it and say, Ha ha!\nSpeaker B: Alright.\nSpeaker B: This is no friction.\nSpeaker B: We know you hate this, but...\nSpeaker A: But the youth doesn't care about functionalities.\nSpeaker B: No.\nSpeaker B: But they do want some gadgets.\nSpeaker B: It's all about cool things.\nSpeaker B: What are we doing?\nSpeaker B: And it will sell.\nSpeaker D: We will be rich.\nSpeaker D: Bless you.\nSpeaker B: Well, we will...\nSpeaker B: Our bosses will be rich.\nSpeaker C: We've done too much in the previous meetings.\nSpeaker C: Too much?\nSpeaker A: Yes.\nSpeaker C: We've got nothing to do now.\nSpeaker A: Well, that's not bad, isn't it?\nSpeaker A: No.\nSpeaker A: I think that's good.\nSpeaker A: We all had our talk and...\nSpeaker A: We agree, I guess, on several things.\nSpeaker A: Most things.\nSpeaker C: It's the best remote in...\nSpeaker B: So...\nSpeaker B: Oh, Heiswier...\nSpeaker B: Oh, Heiswier, he is totally over again.\nSpeaker E: Well...\nSpeaker D: No, man.\nSpeaker D: You just have to push harder.\nSpeaker A: Yeah, you have to push harder.\nSpeaker B: But when I start here...\nSpeaker B: It's here.\nSpeaker B: So...\nSpeaker C: You get really close to the screen with your hand.\nSpeaker C: And I don't think it's...\nSpeaker D: Recalibrated.\nSpeaker D: Yes.\nSpeaker A: Where's the good old chalkboard with...\nSpeaker D: Yeah.\nSpeaker A: Greenboard with...\nSpeaker A: How do you call it chalk?\nNone: Yeah.\nSpeaker C: You can better draw a design on this.\nSpeaker A: Yeah, maybe.\nSpeaker A: Yeah, but we can't.\nSpeaker A: Maybe if that thing is recalibrated, we can draw a thing quick.\nSpeaker A: Before it...\nSpeaker B: Yeah.\nSpeaker A: He goes kill again.\nSpeaker B: Well, finished meeting now.\nSpeaker B: Finished meeting now.\nSpeaker B: Alright, now we know what to do, shall we?\nSpeaker B: We'll stay here.\nSpeaker B: Yeah.\nSpeaker C: Oh, we'll get here.\nSpeaker C: Maybe you know.\nSpeaker A: No.\nSpeaker A: Get away.\nSpeaker A: Can we get a meal here?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Um...\nSpeaker B: I'm not sure.\nSpeaker B: Maybe, yeah.\nSpeaker B: Well, you'll have to go ahead.\nSpeaker B: Oh, I'll just have a few seconds.\nSpeaker B: Yeah.\nSpeaker B: That's a good idea.\nSpeaker B: Alright.\nSpeaker D: Have fun, lads.\nSpeaker D: Well, happy holiday.\nSpeaker A: Okay.\nNone: Don't do too little.\nNone: Oh, man.\nNone: Make us a nice design.\nNone: Going to relax.\nNone: I think we have to design here.\nNone: Or maybe this room is only for discussion.\nNone: But how can we cooperate every...\nNone: Well, it's a little off again.\nNone: And also, do we have to...\nNone: Either industry designer and...\nNone: We have to design now together the...\nNone: Yeah.\nNone: Yeah.\nNone: And you would leave later?\nNone: Yeah.\nNone: Shh.\nNone: Talking me.\nNone: Yeah.\nNone: They stay here and work here.\nNone: You're just staying here.\nNone: Alright.\nNone: Exactly.\nNone: Okay.\nNone: Get out of here.\nNone: Good luck.\nNone: Alright.\nNone: You, uh...\nNone: We'll draw.\nNone: I think.\nNone: You draw the main design.\nNone: I'm the better artist, I guess.\nNone: That's true.\nNone: Oh.\nNone: Oh.\nNone: Give me one more piece.\nNone: Yeah.\nNone: Of course.\nNone: I really like this.\nNone: Yeah.\nNone: So, these green are at the bottom.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: I think we have to put somewhere also, uh...\nNone: There, there, away so we can decide things for ourselves.\nNone: Uh, Mike and...\nNone: Tays and coaches.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: I think we have to put somewhere also, uh...\nNone: There, there, away so we can decide things for ourselves.\nNone: Uh, Mike and...\nNone: Tays and coaches.\nNone: And company.\nNone: Get into your room.\nNone: Maybe that's not fair fair fair fair fair.\nNone: Give me one more piece.\nNone: Come on.\nNone: Come on.\nNone: Come on.\nNone: Mm-hmm.\nNone: Should work now.\nNone: Mm-hmm.\nNone: Alright.\nNone: Alright.\nNone: Thank you.\nNone: Thanks.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\n"}, {"meeting_id": "IS1000d", "summary": "Firstly, Industrial Designer introduced both the cheap one and the other, the expensive prototype of remote control based on the previous discussion of its function. Both the devices had the special shape, like the surf-board. The first prototype was a pretty simple design with LCD display and a conventional layout of buttons. The second prototype was advanced in its speech recognition function and looked like a fashion mobile phone. Secondly, Marketing designed an evaluation test, focusing on its look and feel, innovation and ease of use and the team were asked to give one to seven points to each feature of the product to compare the two prototypes. Thirdly, the team calculated the cost of both the prototypes and found that the second one had exceeded the budget to a great extent. As a result they discussed and voted a lot to discard the advanced functions and eliminate the number of the push buttons, which made the budget under control.", "dialogue": "None: I guess you can do it.\nSpeaker B: Did you get my email with the stats?\nNone: I'm going to change off.\nNone: I'm going to change off soon.\nNone: You're all right.\nNone: It's okay.\nSpeaker A: I'm going to change off soon.\nNone: You're all right.\nNone: It's okay.\nNone: I guess I have to change the pen otherwise.\nNone: Can you come through to it?\nNone: Don't know.\nSpeaker A: Maybe the pen is supposed to go over the seats.\nSpeaker A: Maybe the seat is four rather than the person.\nSpeaker B: Do you think?\nSpeaker B: Yeah.\nNone: I'm going to do this.\nSpeaker B: Yep.\nSpeaker B: I'm going to do this.\nSpeaker B: I'm going to do this.\nSpeaker B: Yeah.\nSpeaker B: I'm going to do this.\nSpeaker F: Okay.\nSpeaker F: So, Marshall is late again.\nSpeaker F: Probably an important man.\nSpeaker F: So, well, it is important for him to be here.\nSpeaker F: Yeah.\nSpeaker F: So, what I do?\nSpeaker F: You did work together.\nSpeaker A: Yeah.\nSpeaker A: So, I would be able to summarize our meeting.\nSpeaker A: Yes.\nSpeaker A: I think it would be very important if the as main designer will be important that he is here.\nSpeaker F: Yes.\nSpeaker A: Basically, we designed two items.\nSpeaker F: Yes, but we...\nSpeaker A: Can we have a phone or can someone...\nSpeaker F: Yes, we should phone him.\nSpeaker F: It's really well designed.\nSpeaker F: When he's not here, we will just have to continue.\nSpeaker F: So, just for a record, I will take notes again.\nSpeaker E: Okay.\nSpeaker F: And...\nSpeaker F: Well...\nSpeaker F: First thing, I was...\nSpeaker F: I got an email from my superior again that we really should stay within the budget.\nSpeaker F: The budget of the 12 euro and 50 cents.\nSpeaker F: He said to me, well, when you stay in, it's good.\nSpeaker F: When you don't stay in, you have to redesign.\nSpeaker F: There is no negotiation possible in this matter.\nSpeaker F: So, we have to consider that.\nSpeaker F: Good.\nSpeaker F: So, maybe Anna, you can have your first thing.\nSpeaker B: Oh, we had an evaluation until we had a design.\nNone: Hello.\nSpeaker F: Okay.\nSpeaker F: Nice to meet you.\nSpeaker F: Nice to meet you.\nSpeaker F: Sure.\nSpeaker F: Great, great.\nSpeaker F: Well, maybe then you can start now with your designs.\nSpeaker A: Yeah.\nSpeaker A: So, I will start by the basic one that fits into eight euros, actually.\nSpeaker A: Right, seven, eight euros.\nSpeaker A: And, well, first, for both, they have a special shape.\nSpeaker A: Maybe the designer can explain better than me.\nSpeaker A: But it's like a surfboard.\nSpeaker A: And you are supposed to surf to browse, to surf TV, maybe the web.\nSpeaker A: And it's kind of interesting shape because unconsciously, people want to surf.\nSpeaker A: And also, it's not too far from a mobile.\nSpeaker A: So, people are used to that kind of shape.\nSpeaker A: Don't take care too much about the color because we will...\nSpeaker A: No, we are supposed to give some offers.\nSpeaker A: So, here would be basically the infrared eye.\nSpeaker A: Hi.\nSpeaker A: Let the LED...\nSpeaker A: The on-off button.\nSpeaker A: Here would be the volume on the left.\nSpeaker A: Okay.\nSpeaker A: So, easy to turn on and off.\nSpeaker A: And so, this is a very cheap version.\nSpeaker A: So, there are...\nSpeaker A: Maybe you can carry on with you.\nSpeaker C: Also, you have browsing the channels, actually.\nSpeaker C: So, you can go up and down the channels.\nSpeaker C: If you have a video or something, it can forward.\nSpeaker A: How can you change from VCR to TV, by the way?\nSpeaker C: Oh, no, no, no.\nSpeaker C: This is a model with just a TV one.\nSpeaker C: No, no, sorry.\nSpeaker C: This is a standard TV one.\nSpeaker C: We are not talking about that.\nSpeaker C: So, and then we have usually...\nSpeaker C: There are 12 keys, but we know that we have only 10 digits.\nSpeaker C: And then we have an extra two for having...\nSpeaker C: Or giving an option for having more than one channel.\nSpeaker C: And the other one is for the teletext or something you want to go out through from that.\nSpeaker B: Okay, so, it's a very basic...\nSpeaker C: It's a very basic minimal thing which is also available in the market.\nSpeaker C: Actually, that's what it is that they...\nSpeaker C: And it costs us to build this about 8 euros.\nSpeaker F: Except for the special shape, the surfing board.\nSpeaker F: It has a quite conventional layout of buttons.\nSpeaker C: So, this one model and...\nSpeaker C: Can I say?\nSpeaker C: Yeah, sure.\nSpeaker B: Yeah, I like the volume control.\nSpeaker C: That's good.\nSpeaker C: This is a magic one, but right now we don't want to talk about that.\nSpeaker C: It is a very futuristic.\nSpeaker C: It's like a brain, machine interface and all this stuff.\nSpeaker C: We are thinking about it in the future.\nSpeaker C: It can come.\nSpeaker C: So, it doesn't actually have buttons.\nSpeaker C: So, that...\nSpeaker C: Then what we look...\nSpeaker C: Do you want to say?\nSpeaker F: Yeah.\nSpeaker F: This is a model, yeah.\nNone: I just...\nSpeaker F: Look how it feels.\nSpeaker F: Oh, yeah.\nSpeaker F: It really feels like...\nSpeaker F: Like a mobile phone.\nSpeaker F: Yeah.\nSpeaker F: Just sign.\nSpeaker F: Yeah, see.\nSpeaker F: I really want to talk to it.\nSpeaker F: But...\nSpeaker F: Yeah, you want to talk to that?\nSpeaker F: So, but you continue with your gr...\nSpeaker C: So, well, then the...\nSpeaker C: This is a more...\nSpeaker C: A little...\nSpeaker C: A smooth and it gives you a lot of functionality in this way.\nSpeaker C: So, all we have...\nSpeaker C: You see that there are only 6 keys, but don't worry.\nSpeaker C: They are doing the job of tool keys actually here.\nSpeaker C: And so, they have more space, actually.\nSpeaker C: And it's easy to...\nSpeaker C: Use this.\nSpeaker C: And...\nSpeaker C: You have a...\nSpeaker C: So, this is a standard URL infra-red I.\nSpeaker C: And then you have a power button.\nSpeaker C: It should volume what you have.\nSpeaker C: And then, other than that, you have a channel up and down and the flow of pause or...\nSpeaker C: Slow-those.\nSpeaker C: Yeah.\nSpeaker C: And then, you have LCD display here.\nSpeaker C: And this is the functional thing which can change, like...\nSpeaker C: It's a toggle switch, which could change the function.\nSpeaker C: Say...\nSpeaker C: It's a DVD player to tell if it's easy.\nSpeaker C: Yeah, I can change it to...\nSpeaker C: Instead of having many switches.\nSpeaker F: And then you get feedback, fire, LCD...\nSpeaker C: LCD can display, what is that on that?\nSpeaker C: And...\nSpeaker C: Well, you can have an integrated microphone over here.\nSpeaker C: This is the...\nSpeaker C:... or an button here.\nSpeaker C: So...\nSpeaker C: Which can...\nSpeaker C: Basically, you want to do a speech recognition and the channel or whatever information can be displayed here directly on the...\nSpeaker C:...on your display.\nSpeaker C: And here is a...\nSpeaker C:... small LED, which is like blinking one, which tells you like...\nSpeaker C: Are you running out of the battery and which is...\nSpeaker C:... can be useful for the locating as I was talking earlier.\nSpeaker C: Okay.\nSpeaker C: And...\nSpeaker C: Well, then we have a cover, basically.\nSpeaker C: Basically, you don't need much other time this.\nSpeaker C: When you need, you can use it.\nSpeaker C: And this gives additional fascinating that tomorrow you want, you can add a tactile thing to this cover, you know.\nSpeaker C: For example, design and...\nSpeaker F: But...\nSpeaker F: But...\nSpeaker F:... India...\nSpeaker F:... when this is closed...\nSpeaker F: Yeah.\nSpeaker F:... will also cover up the LCD screen.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker F: Yeah, it's basically to do that.\nSpeaker F: But the LCD screen, I mean, is a very...\nSpeaker F:...\nSpeaker F:... well, an eye-attracting feature with...\nSpeaker F: Actually, when you're watching the TV...\nSpeaker C: Actually...\nSpeaker C: When you're watching anything or listening to them, you hardly care about what is getting displayed here, you know.\nSpeaker C: You want to...\nSpeaker C:... and this gives a protection to the LCD, actually, giving a cover to that, actually.\nSpeaker C: It gives a protection because when it falls down or something...\nSpeaker C:... it is...\nSpeaker C: It's more robust.\nSpeaker C: It's more robust that way.\nSpeaker C: Yeah, okay.\nSpeaker C: Yeah.\nSpeaker C: And you have very good chances.\nSpeaker A: It's low weight.\nSpeaker A: You have to see.\nSpeaker A: The components we put inside it is very low weight.\nSpeaker A: Yeah.\nSpeaker A: So the cost is actually a bit more.\nSpeaker A: It's...\nSpeaker A: It's 16 frames.\nSpeaker C: 16 euros.\nSpeaker A: 16 euros, okay.\nSpeaker B: So that's where outside the budget then.\nSpeaker A: Then it's out of budget.\nSpeaker A: The main point we talk about that with our manufacturer.\nSpeaker A: And they say basically that the ASR system would be...\nSpeaker A:... something like three frames per item.\nSpeaker A: Three euros.\nSpeaker A: So you also...\nSpeaker A: That's on top of the 16 or is it part of that?\nSpeaker B: No, no, no, part of that.\nSpeaker B: So it takes it down to 13 euros without the state check ignition?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker C: Yeah.\nSpeaker C: Well, if you can have...\nSpeaker C: If you have new more ideas, we can add new more...\nSpeaker C:... some more keys if you want to, you know.\nSpeaker C: I think...\nSpeaker A: Yeah, we should stick with a number of keys because if we add too much...\nSpeaker A: Yeah, it should not be clapped, turning up everything.\nSpeaker C: What's this one on the side?\nSpeaker C: That's for the... it's kind of a LED for indicating your battery.\nSpeaker C: Okay.\nSpeaker C: And as well as it's like a blinking one, you can keep it the same.\nSpeaker B: I like the shape of them.\nSpeaker B: I do like the size and the shape.\nSpeaker F: Before talking about the money and what's possible and what is not possible, maybe you can give us your...\nSpeaker F: Okay.\nSpeaker B: Maybe we run the evaluation on both of the products, both of these two.\nSpeaker F: Yes, the evaluation and also the evaluation criteria.\nSpeaker F: So what is important to look at?\nSpeaker B: Best guess what we've talked about already.\nSpeaker B: From the marketing point of view, we just want to make sure that we've taken into account...\nSpeaker B: Well, just...\nSpeaker F: Yeah.\nSpeaker F:... do it quickly if we all already...\nSpeaker B: So it's just a short list of criteria on the things that we identified as being important to selling the product.\nSpeaker B: We just go through these and rate them as a group, and then at the end we'll make an evaluation based on that.\nSpeaker B: So I just average this score of those items.\nSpeaker B: So these are the things we're identified as being important.\nSpeaker B: The three things were look and feel, innovation and ease of use.\nSpeaker B: They're three important components.\nSpeaker B: Appealing to the correct demographic.\nSpeaker B: So using those things in the right way to appeal to our demographic.\nSpeaker B: And then following the company motto, following the fashion trends and putting the end of the product as well.\nSpeaker B: So we'll go through and put through those on the two products now and discuss them further and then evaluate them.\nSpeaker F: No, why not discuss it now?\nSpeaker F: Okay.\nSpeaker B: So I've just got through on the whiteboard, I guess.\nSpeaker B: I'm just going to come in.\nSpeaker B: So the first one was really very far below budget.\nSpeaker B: Would you want to take the price down of the end product according to that or just have a high profit on it?\nSpeaker B: It was running a make up to eight euros and we have a...\nSpeaker F: Well, my personal view is when this one is eight euros, we must think how can we improve it.\nSpeaker F: And then, I mean, you must just see it as we can still spend this four and a half euro.\nSpeaker F: And because the selling price is already quite fixed on 25 euros.\nSpeaker F: So we just have to offer as much as value for the customer he can have for 25 euro.\nSpeaker B: Okay, so looking for innovation.\nSpeaker B: So bad.\nSpeaker C: And now it's easy to use.\nSpeaker A: Easy to use.\nSpeaker A: Ojon's target.\nSpeaker A: And trends.\nSpeaker B: Are we following the idea of using the removable covers on these?\nSpeaker B: Is that part of my thumb?\nSpeaker F: We can still discuss that.\nSpeaker F: And together with evaluating this, we might come with new ideas in adding things or removing options because they are too expensive.\nSpeaker F: I've received a framework which we can do this.\nSpeaker F: This is the latest price of our production unit for several components.\nSpeaker F: So we can see whether the price is within the 12 euro worth 50 cents.\nSpeaker F: So maybe we can start with this calling this one.\nSpeaker B: Yep.\nSpeaker B: So that's the thing.\nSpeaker F: This is the first design.\nSpeaker B: Another one's green.\nSpeaker B: Yes.\nSpeaker B: Okay, so look and feel.\nSpeaker B: Well, one is a brick and a pen again.\nSpeaker B: There is another pen.\nSpeaker B: Yeah, get that one.\nSpeaker B: One's bad and seven's the best.\nSpeaker B: So one's true and seven's false.\nSpeaker B: One's the best.\nSpeaker B: Okay.\nSpeaker B: So I'll just gather one to seven.\nSpeaker F: Okay.\nSpeaker F: So look and feel.\nSpeaker F: Well, you already feel it pretty much, I think.\nSpeaker F: In my opinion, if you really feel it's very good.\nSpeaker F: It's very good in your hand.\nSpeaker F: So I would consider two or maybe even one for feels.\nSpeaker F: But that's just half.\nSpeaker F: We should also consider the look.\nSpeaker F: And then it looks quite conventional.\nSpeaker F: Don't you agree?\nSpeaker B: Okay.\nSpeaker B: On the scale, it's between functional and fancy.\nSpeaker B: Basically, we're looking at.\nSpeaker F: So say five.\nSpeaker F: It's my opinion, but I don't know what to do.\nSpeaker C: Well, I will give it to my wife.\nSpeaker C: We have any way the way we have designed this like the surf.\nSpeaker C: As we say, you know, it doesn't look as fine, but still, I will give four in that game.\nSpeaker C: Yeah, four maybe.\nSpeaker F: Oh, yeah.\nSpeaker F: Then we should have one for.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nNone: Yeah.\nNone: That's not.\nSpeaker F: Okay.\nSpeaker F: Can you maybe fix the other?\nSpeaker A: Yeah.\nSpeaker A: You should print like this, not like this.\nSpeaker A: When you.\nSpeaker C: No, it's the.\nSpeaker C: No, can you get the batteries?\nSpeaker C: No, no, that's falling down.\nSpeaker B: That's not the ink.\nSpeaker B: That one.\nSpeaker B: Yeah.\nSpeaker B: That's not the ink.\nSpeaker B: That's the, that, that one.\nSpeaker A: Yeah.\nSpeaker C: That's not.\nSpeaker C: No, no, it's not that.\nSpeaker C: It's here to close the battery.\nSpeaker A: Okay.\nNone: Yeah.\nSpeaker C: No, it should be.\nSpeaker C: No, I think it's lost a battery.\nSpeaker B: Right.\nSpeaker B: It's right.\nSpeaker B: No, it's.\nSpeaker B: It's still right, but it wouldn't pick it up with the sensors.\nSpeaker C: I think that another battery there.\nSpeaker C: Oh, yeah.\nSpeaker C: You got a second.\nSpeaker B: Troy.\nSpeaker F: Oh, we got a tell.\nSpeaker F: Yes, it has a.\nSpeaker F: Perfect.\nSpeaker B: Is that working?\nSpeaker B: Did it come out?\nSpeaker F: Good.\nSpeaker B: Good.\nSpeaker B: Because we'll be able to see it still even if it wasn't working.\nSpeaker B: It's just normal whiteboard marker, but it wouldn't be picked up in the actual whiteboard.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: So.\nSpeaker F: Then then.\nSpeaker B: The other one.\nNone: Wow.\nNone: I think it's slightly better.\nSpeaker B: It's hard to tell from us the first scene, but.\nSpeaker F: When we want to include, I'm doubting about this component.\nSpeaker F: It breaks in your game.\nSpeaker C: No, actually, this is not going to protrude, actually.\nSpeaker C: Okay.\nSpeaker A: It's not a button.\nSpeaker A: It's a lead.\nSpeaker C: It's a lead, actually, which will be coming in a call.\nSpeaker A: Actually, yeah, it should be embedded.\nSpeaker A: It should be embedded there.\nSpeaker A: Yeah.\nSpeaker A: So it won't be really, you know, protruding or something.\nSpeaker A: You can push it again.\nSpeaker A: You can push it.\nSpeaker B: Yeah.\nSpeaker B: The other thing is, it's the left hand one protruding because if people are left handed, the one needs other hands.\nSpeaker B: Yeah.\nSpeaker B: It works so well.\nSpeaker C: No, it's not protruding, actually.\nSpeaker C: It'll go embedded into that.\nSpeaker B: Well, I'd say two or three for that one, personally.\nSpeaker B: Which was three than two.\nSpeaker F: I think the look is better, but the feel is worse.\nSpeaker F: So I would also say this is four.\nSpeaker F: Okay.\nSpeaker F: But what do you think?\nSpeaker C: Oh, it's fine.\nSpeaker C: I think.\nSpeaker C: No, just that the feel is that you, right now you don't see the feel because, I don't follow them, but if you press it quite inside, now like this.\nSpeaker C: Now it's embedded one.\nSpeaker E: Yeah.\nSpeaker C: This is how the embedded one is.\nSpeaker F: Yeah, it's basically the same shape.\nSpeaker C: Yeah.\nSpeaker C: It's basically the same thing, actually.\nSpeaker C: You will be.\nSpeaker C: Except that in this...\nSpeaker C: The very CD makes it better.\nSpeaker A: Yeah.\nSpeaker C: And you may have a slight thing for to follow.\nSpeaker B: Yeah, it's definitely more fancy than that one.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker B: I would say two or three.\nSpeaker B: Okay.\nSpeaker B: So consensus, two or three?\nSpeaker F: Two.\nSpeaker F: Two.\nNone: Yeah, two is good, yes.\nSpeaker B: So two.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Innovation.\nSpeaker B: First one.\nSpeaker A: There is no innovation.\nSpeaker A: The first one compared to what exists in the market.\nSpeaker C: No, but except for the design of the surf.\nSpeaker B: Yeah.\nSpeaker B: What feature should we actually include?\nSpeaker B: Including a location kind of thing?\nSpeaker B: I'm trying to find it.\nSpeaker B: I don't know.\nSpeaker B: I think it's one of the feel.\nSpeaker B: But there's no actual innovation in that at all.\nSpeaker B: It's just a straight out remote control.\nSpeaker F: Yeah.\nSpeaker F: The only innovation is to shape.\nSpeaker F: Yeah.\nSpeaker F: So there's no...\nSpeaker B: That's a looking field thing though.\nSpeaker B: It's not a technological innovation.\nSpeaker B: Yeah.\nSpeaker B: So I'd be up for seven for innovation.\nNone: Yeah.\nSpeaker A: And the second one is really state of the art.\nSpeaker A: Yeah.\nSpeaker A: In terms of innovation.\nSpeaker A: And with many more functionalities, can open and impose the...\nSpeaker A: Yeah, it is.\nSpeaker A: Yeah.\nSpeaker F: LCD screens.\nSpeaker F: Yeah.\nSpeaker F: That's...\nSpeaker F: Well, quite.\nSpeaker A: Any of the squirting buttons and menu and programmable device behind this.\nSpeaker A: You could put it at one or two.\nSpeaker A: Okay.\nSpeaker D: That's not right.\nSpeaker B: Yeah.\nSpeaker B: So what are the innovations?\nSpeaker B: This got the LCD screen.\nSpeaker A: Yeah.\nSpeaker A: Automatics feature recognition.\nSpeaker A: Is that in this one though?\nSpeaker B: Is this...\nSpeaker B: Because this is the...\nSpeaker B: No.\nSpeaker B: There were two different options we discussed then.\nSpeaker B: We discussed the one that was in budget and the one that was out of budget.\nSpeaker F: We discussed it as you all designed it and then we all tried it.\nSpeaker B: Okay.\nSpeaker B: So the cost for these were...\nSpeaker B: Get in the budget.\nSpeaker B: What was the cost for the first one?\nSpeaker B: Eight euros?\nNone: Eight.\nSpeaker A: Actually, we have to check again.\nSpeaker A: Yeah, I do.\nSpeaker B: And this one was 16 euros.\nSpeaker C: 16 euros.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker B: Innovation for this one is two.\nSpeaker B: One.\nSpeaker C: It's not two.\nSpeaker C: I would say two.\nSpeaker A: Two.\nSpeaker A: And to be one, what would we do in it?\nSpeaker A: Actually, I don't see...\nSpeaker A: Why?\nSpeaker A: Okay.\nSpeaker A: One would be without a button.\nSpeaker A: Yeah.\nSpeaker B: Well, this feature recognition is a very good innovation.\nSpeaker B: I believe it's a...\nSpeaker B: So maybe...\nSpeaker F: It's a...\nSpeaker F: With the speech recognition.\nSpeaker A: It's using speech recognition.\nSpeaker F: Okay.\nSpeaker F: Give that.\nSpeaker F: Give it a one.\nSpeaker B: It works.\nSpeaker F: Yeah.\nSpeaker F: Then it's...\nSpeaker F: I see one.\nSpeaker B: Okay.\nSpeaker B: Ease abuse.\nSpeaker A: So the first one is re-standouts, so everybody...\nSpeaker A: Is used to it, yeah.\nSpeaker A: Can use it.\nSpeaker C: They are used to it.\nSpeaker B: Yeah.\nSpeaker B: So that's maybe a two for ease abuse.\nSpeaker C: Yeah.\nSpeaker A: Here, the main...\nSpeaker A: The other one is quite easy.\nSpeaker A: Though...\nSpeaker C: Though it has more functionality, I think it shouldn't be for the user to learn it, actually.\nSpeaker B: So maybe a three or four.\nSpeaker C: Yeah.\nSpeaker C: Actually, in fact, I think it will be...\nSpeaker A: Yeah.\nSpeaker A: Maybe sometimes people get scared with the number of buttons.\nSpeaker A: Yeah, but...\nSpeaker A: There's...\nSpeaker A: Like I would say three...\nSpeaker A: Four.\nSpeaker A: Okay.\nSpeaker C: Well, we have reduced the key, actually.\nSpeaker C: You see.\nSpeaker C: Three is fine.\nSpeaker C: Yeah.\nSpeaker A: Because it's...\nSpeaker A: It's not like a big one with 100 buttons.\nSpeaker C: Actually, the user has to put some effort to do use that, actually.\nSpeaker C: It's not so easy.\nSpeaker C: Like this one, the normal.\nSpeaker F: Yeah, but then when he used to it, it's quite easy.\nSpeaker F: It's quite easy.\nSpeaker C: Yeah.\nSpeaker C: Initially there, there was a lot of effort.\nSpeaker C: Okay.\nSpeaker F: Yeah.\nSpeaker B: Good.\nSpeaker B: So three, how will it go to the target demographic?\nSpeaker B: So we're still thinking 20 to 40 year olds?\nSpeaker A: 20 to 40, yes.\nSpeaker A: This one would be for grandmothers.\nSpeaker C: No, this would...\nSpeaker C: I would give this model to the whole people.\nSpeaker C: Yeah, one mother's, yeah.\nSpeaker B: So it completely changed our demographic there.\nSpeaker B: It's not part of the funky young thing.\nSpeaker B: Except for the surfing shape.\nSpeaker B: I mean, that's something with...\nSpeaker F: As an appeal on this group, I think...\nSpeaker B: If it was the very bottom price range, it was like between this and another one, which did the same thing.\nSpeaker B: Then I can imagine it being a part of the demographic, yeah.\nSpeaker B: But it's still...\nSpeaker F: After this, we can consider, for instance, making this more attractive to the demographic.\nSpeaker B: So I've got room, but some budget there to add a few things to it.\nSpeaker F: But as it is now, I would say...\nSpeaker F: Six?\nSpeaker F: Yeah.\nSpeaker E: Oh, yeah.\nSpeaker C: Do you agree?\nSpeaker C: I like the movie.\nSpeaker C: Yeah, because it's so important.\nSpeaker A: It was written that it was so important, the look and taking care of.\nSpeaker A: It targets the right range of people.\nSpeaker A: Yeah.\nSpeaker C: But if you sell it in the market, it's going to be cheap, actually.\nSpeaker C: Yeah.\nSpeaker B: But it's going to be cheap, whatever, though.\nSpeaker B: It was set with it.\nSpeaker C: And people can still decide to use a cheaper one.\nSpeaker A: Yeah.\nSpeaker C: Instead of a...\nSpeaker A: But for us, we have a concern that we need to sell at 25 euros.\nSpeaker B: Yeah.\nSpeaker B: There's nothing that would make me spend an extra few euros in that one rather than another one.\nSpeaker F: Just paying 25 euros.\nSpeaker F: I mean, it's not going to be cheaper.\nSpeaker C: Okay, so in that case, well, it's fine.\nSpeaker C: Okay.\nSpeaker B: And the demographic of the second one?\nSpeaker C: The demographic of the...\nSpeaker B: It's got the toys in it.\nSpeaker B: It's got the LCD screen.\nSpeaker F: I think it's better because of the LCD screen, which really can appeal only on the...\nSpeaker F: Yeah.\nSpeaker F:... on the yep sign on the...\nSpeaker A: Yeah.\nSpeaker A: And if you want to target...\nSpeaker A: Yeah, if we wish to sell four millions of this, I think for these agents, we need absolutely the LCD screen.\nSpeaker A: Mm-hmm.\nSpeaker F: So, you also need screen tracks.\nSpeaker F: Yeah.\nSpeaker B: Especially if we can have an LCD screen and a low range product and that's good.\nSpeaker B: Yeah.\nSpeaker F: Maybe that's something to consider.\nSpeaker F: Okay.\nSpeaker F: So, what...\nSpeaker B: I'd probably go with three again for that one.\nSpeaker C: Yeah, I think it has more market.\nSpeaker C: Even one and two.\nSpeaker C: Yeah.\nSpeaker B: You know, I've saved you two.\nSpeaker B: Yeah.\nSpeaker C: Two.\nSpeaker C: Two, yeah.\nSpeaker C: Tomorrow, this will be more appealing because you can add a lot of investigation on that.\nSpeaker C: Mm-hmm.\nSpeaker C: Because then you have it...\nSpeaker C: Yeah.\nSpeaker C: You have a lot of things which you can include for the people.\nSpeaker B: Yeah.\nSpeaker B: And especially, I guess, this is the speech recognition as well.\nSpeaker B: And that makes it more appealing, it's more of a new fun toy.\nSpeaker C: We have to practically test it.\nSpeaker C: The field test will tell you how good...\nSpeaker C: Yeah.\nSpeaker F: Okay, you find a point.\nSpeaker F: And...\nSpeaker B:... that's fine.\nSpeaker B: So, the trends were the fruit and vegetables and there's bungee feel.\nSpeaker E: Mm-hmm.\nSpeaker B: So, as it is, not really doing either of them.\nSpeaker A: I think spongy...\nSpeaker A:... that means that it goes in the water.\nSpeaker B: Well, spongy could be part of the field of the buttons as well.\nSpeaker B: I know some...\nSpeaker B:... who tried the mobile phones that have got the kind of spongy buttons?\nSpeaker B: Not exactly spongy, but...\nSpeaker B: I'm thinking one of the knockiest, it's got like...\nSpeaker B:... you can't...\nSpeaker B:... have got individual buttons, it's got just a...\nSpeaker B:... bit on it and so it can...\nSpeaker B:... that feels kind of spongy.\nSpeaker A: This one includes this feature, right?\nSpeaker A: Spongy buttons.\nSpeaker A: Yeah, it's sort of...\nSpeaker C: Yeah, it's sort of...\nSpeaker C:... that way it can be.\nSpeaker C: Okay.\nSpeaker A: And maybe in the corner we can make it...\nSpeaker A:... a foot and a veggie.\nSpeaker B: But that's if you're using the covers.\nSpeaker F: And then we can...\nSpeaker F: Oh, is this...\nSpeaker F: Yes, we can consider...\nSpeaker F: Is it possible, do you think, to make a cover for such a phone?\nSpeaker B: Well, they make up for mobile, so it can't be that much more.\nSpeaker C: I do want to cover that, actually.\nSpeaker F: Just with a flexible plastic...\nSpeaker B: So you've got the option of having different colours or different textures?\nSpeaker B: This is possible.\nSpeaker C: Uh...\nSpeaker C:... it's...\nSpeaker C:... any shows?\nSpeaker C: Yeah, I think it should be possible, like what you do with the mobile.\nSpeaker F: Or just two things which can be put on each other.\nSpeaker F: Yeah, if I do like...\nSpeaker A:... not careful.\nSpeaker F: Yeah.\nSpeaker F: Yeah, I do like it.\nSpeaker F: Alright.\nSpeaker F: So...\nSpeaker F:... maybe we can...\nSpeaker F:... but we have to decide that we can put the...\nSpeaker F:... fancy look of vegetables, for instance, to do these covers.\nSpeaker F: And now try to invest in the features.\nSpeaker F: I think the...\nSpeaker B:... if we do the covers, that's really going with the company's philosophy of having the fashion in electronics.\nSpeaker B: It lets you follow the latter's fashion.\nSpeaker B: Even next year when Frid and Besh will not be out, you'll still be able to cover that and then it'll still be in fashion.\nSpeaker A: Yeah.\nSpeaker A: Because sometimes, look at this computer.\nSpeaker A: This laptop is all black.\nSpeaker A: And it's quite conventional.\nSpeaker A: And sometimes people don't like too much flashy colours, like...\nSpeaker A:... there's some presented here.\nSpeaker A: So maybe we could do...\nSpeaker A:... like in the range of the set of...\nSpeaker A:... what we propose, a black one.\nSpeaker A: Yeah.\nSpeaker A: Very standard one.\nSpeaker A: Yeah.\nSpeaker A: And then the requirements of such people that want to really stand out things.\nSpeaker B: And then you have the option of having different colours.\nSpeaker B: Yeah.\nSpeaker F: So that would make the trends equal, so we...\nSpeaker F:... we really don't have...\nSpeaker B: Yeah.\nSpeaker B: That's the feature that could be included in either of them.\nSpeaker B: Although it is more with the spandy buttons on the other ones.\nSpeaker B: Yeah.\nSpeaker A: So a point better for the...\nSpeaker A: Okay.\nSpeaker A: So...\nSpeaker B: Two and three, or one or two?\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker D: Say one, two.\nSpeaker F: One, two.\nSpeaker A: Okay.\nSpeaker F: So I like to see...\nSpeaker F:... so...\nSpeaker F:... these are one of the spandy buttons.\nSpeaker F: Yeah.\nSpeaker A: So there's...\nSpeaker A:... one of the spandy.\nSpeaker A: Hmm.\nSpeaker F: I see.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: So the average of that is...\nSpeaker B:... three, six, nine, about a five.\nSpeaker B: So...\nSpeaker B:... just added.\nSpeaker F: One by six.\nSpeaker B: One by six.\nSpeaker B: One by six.\nSpeaker B: One by five.\nSpeaker B: One by six.\nSpeaker C: One by eight, yeah.\nSpeaker B: This one, eleven, thirteen, nineteen, twenty-one.\nSpeaker B: Oh, five is four point two.\nSpeaker B: One point two.\nSpeaker B: Very good.\nSpeaker A: Okay.\nSpeaker B: But we still have a very different price to those two.\nSpeaker B: So they're not really comparable yet anyway.\nSpeaker F: We must try to get them closer.\nSpeaker F: Yeah.\nSpeaker F: Both of them here.\nSpeaker F: Or we just have to choose and adapt.\nSpeaker F: Because when we choose for this one, we have to...\nSpeaker F:... we have to make it more attractive.\nSpeaker F: And when we choose for this one, we have to make it more cheap.\nSpeaker B: Well, it's easy to make that one cheaper by just taking this bit recognition out.\nSpeaker B: That'll basically take us down to the budget.\nSpeaker F: But I...\nSpeaker F: Now, did you work with the same prices as I have here?\nSpeaker A: So I give, you know, I just give a call with the money fixture and I expand them and they turn me.\nSpeaker A: This could be possible for six euros.\nSpeaker A: Unfortunately, we didn't see this shit.\nSpeaker A: Yeah.\nSpeaker A: Matthew.\nSpeaker A: So maybe we have to recap with this one.\nSpeaker C: Have you had to reconsider everything with this?\nSpeaker C: Yeah.\nSpeaker F: Well, yes.\nSpeaker F: Well, you reconsider.\nSpeaker F: So let's try to model this phone in this heat.\nSpeaker F: What kind of energy source?\nSpeaker F: We didn't speak about that.\nSpeaker F: It's a normal battery or...\nSpeaker C: Yeah, it might... it'll need more than a conventional one.\nSpeaker C: It won't be a... just...\nSpeaker C: Maybe you might use a AA battery actually.\nSpeaker C: What do you say?\nSpeaker A: For this one, it's a normal battery.\nSpeaker F: Yes, just one battery.\nSpeaker F: Okay.\nSpeaker F: Electronics.\nSpeaker F: Now, given speech recognition, I think you should go for your fans chip.\nSpeaker A: Yeah.\nSpeaker A: Some person.\nSpeaker A: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Simple sensor, yes.\nSpeaker F: Simple.\nSpeaker F: Yeah, this one.\nSpeaker F: Yes, this one.\nSpeaker F: Okay.\nSpeaker F: Case.\nSpeaker F: Um...\nSpeaker F: So...\nSpeaker F: Cloud...\nSpeaker C: Double Cloud, yeah.\nSpeaker C: I see.\nSpeaker C: Double car.\nSpeaker B: Yeah, it's going to be more than just the basic case, definitely.\nSpeaker A: So which one are we talking to?\nSpeaker C: Are you talking about these or that?\nSpeaker C: Yes, we are talking about...\nSpeaker F: But they have the same shape, but...\nSpeaker B: Yeah.\nSpeaker B: So it's going to be not basic cases.\nSpeaker F: So this would be double carves?\nSpeaker E: Mm-hmm.\nSpeaker F: Yeah.\nNone: Okay.\nSpeaker F: Uh, Plashtik would be a...\nSpeaker F: Yeah, the basic one.\nSpeaker F: Yeah.\nSpeaker A: Zero.\nSpeaker A: Zero French.\nSpeaker C: Special carves.\nSpeaker C: Uh...\nSpeaker F: Special car, now we leave to the carves.\nSpeaker B: Uh...\nSpeaker B: So now we're going...\nSpeaker B: Bush Button or LCD...\nSpeaker B: LCD display.\nSpeaker A: Yes.\nSpeaker A: I see the display.\nSpeaker A: Yeah.\nSpeaker A: It's okay.\nSpeaker B: Just say, LCD display.\nSpeaker B: Is that price per unit or for the whole...\nSpeaker F: Yeah, now this is per unit.\nSpeaker F: So with names...\nSpeaker C: There's a number of components.\nSpeaker C: Yeah, we made me the scroll.\nSpeaker C: We'll write for that.\nSpeaker A: No, but for this one, it's true.\nSpeaker C: No, for that one also.\nSpeaker A: The author?\nSpeaker A: So, yeah, that's the scroll.\nSpeaker F: One, two, three, four, five, six, seven, eight, nine.\nSpeaker F: Yes, well, probably.\nSpeaker F: So this comes to 18.\nSpeaker B: And that's without any special button supplements.\nSpeaker C: Yeah, one scroll really.\nSpeaker C: So...\nSpeaker B: So we'd have special colour, special form, special material, and all of them.\nSpeaker B: I don't understand about them.\nSpeaker F: So, I think...\nSpeaker F: But do you agree that this...\nSpeaker F: Wait a minute.\nSpeaker A: It's not double curved.\nSpeaker A: It's single curved, right?\nSpeaker A: Because there is no...\nSpeaker A: But I thought it would be curved on too.\nSpeaker B: So it's curved on the size and curved on the top of the bottom.\nSpeaker B: Yes, I understand what you mean here.\nSpeaker E: I understand what you mean here.\nSpeaker A: So, there is only...\nSpeaker A: There is nothing like...\nSpeaker A: You know, in the other stuff...\nSpeaker A: You're talking about concave curves.\nSpeaker A: Yeah, concave.\nSpeaker A: So I think we can put...\nSpeaker A: You think a single curved?\nSpeaker A: A single curved, like 16.\nSpeaker A: That makes 17.\nSpeaker A: And whether the...\nSpeaker A: Buttons, we have tried buttons, are you sure?\nSpeaker A: Yes.\nSpeaker A: We have more, we've got the scroll wheel on the side.\nSpeaker B: So I had a bad extension.\nSpeaker A: Bad extension.\nSpeaker A: Bad extension.\nSpeaker A: We have...\nSpeaker F: We haven't talked about a lot, but that's not...\nSpeaker F: A lot is very inexpensive, I believe, but it's not in the list.\nSpeaker B: We've got a scroll wheel for the bottom, don't we?\nSpeaker B: Or is it some other thing that's not on there?\nSpeaker F: Is this a scroll wheel or is this a sort of button which can be pressed on two sides?\nSpeaker F: So far.\nSpeaker B: Okay, we've only got five minutes left, guys, so we need to wrap it up pretty fast.\nSpeaker A: Yes, you can scroll to it.\nSpeaker B: So this is more than the...\nSpeaker B: And the cost you gave for 16 years?\nSpeaker A: Okay, so based on that, yeah.\nSpeaker A: Where is that?\nSpeaker A: Yes, I'll get some of this picture.\nSpeaker C: That is the sample sensor and sample speaker.\nSpeaker F: So...\nSpeaker A: Let's do that.\nSpeaker C: We just need that, actually.\nSpeaker C: Where... we need one.\nSpeaker F: We could go for the... for the... for a simple chip, but then we can't have the speech recognition, yes?\nSpeaker A: No, we can't, yeah.\nSpeaker A: So when we...\nSpeaker F: But this would be cutting the speech recognition.\nSpeaker A: Yeah, but if you have the near-the-rcd, you can choose... select between... you know, universal between audio, TV and VCR.\nSpeaker A: And this needs a... needs an advanced chip.\nSpeaker A: Transchip.\nSpeaker A: Right?\nSpeaker A: Hi.\nSpeaker A: Oh, you're on a chip?\nSpeaker C: I think it's going to be...\nSpeaker A: I think it's going to be...\nSpeaker A: I think it's regular.\nSpeaker A: Yeah, it's going to be...\nSpeaker A: You can do that.\nSpeaker F: With the regular chip, yeah.\nSpeaker F: Yeah, it's a regular chip.\nSpeaker F: Yeah, it's a regular chip.\nSpeaker A: Yeah, it's a regular chip.\nSpeaker A: So... and what about the number of buttons?\nSpeaker F: Just... that... maybe...\nSpeaker F: Well, we can just say...\nSpeaker F: Matthew.\nSpeaker E: Uh...\nSpeaker A: What is this?\nSpeaker A: When...\nSpeaker A: When you look at this... this... item...\nSpeaker C: Yeah?\nSpeaker F: Okay, look, we just... when we just want to... to cut the number of buttons, we have to make 7 to fit in 12... 1250.\nSpeaker F: So is it possible?\nSpeaker B: I... but the 7 basic buttons... or 7 buttons that any adds on...\nSpeaker B: Without special colors or form or material.\nSpeaker C: Well, we... then we have to ask the user to press itself.\nSpeaker A: You're gonna make a phone to your boss saying 2550 is really...\nSpeaker A: Pffft...\nSpeaker A: Ha ha ha ha ha ha...\nSpeaker F: No, no, no, ha ha ha...\nSpeaker B: Ah ha...\nSpeaker B: So the LCD place is three francs.\nSpeaker B: Sorry, three euros.\nSpeaker B: Five-sef.\nSpeaker A: And we don't want to change that right.\nSpeaker A: And we really want to add the LCD otherwise we wouldn't get the market.\nSpeaker A: Otherwise...\nSpeaker F: You are very normal...\nSpeaker F:...thing like this.\nSpeaker F: Yeah.\nSpeaker B: So... 12 euro 50.\nSpeaker B: We got two off of the battery.\nSpeaker B: We can't do anything about that.\nSpeaker B: So, 1050.\nSpeaker B: What else do you have to write the display?\nSpeaker B: That's 750.\nSpeaker B: So, a 750 to use for the case and for the buttons.\nSpeaker E: And the chips.\nSpeaker B: So, the chips.\nSpeaker B: Up there already.\nSpeaker B: So, we're gonna have to scale this down to get within budget.\nSpeaker B: There's no doubt about that.\nSpeaker A: So, what each of us think about the...\nSpeaker A:...because it's a major point in the LCD.\nSpeaker A: Do you think it's important?\nSpeaker C: Oh, we could even replace them by buttons.\nSpeaker A: Because sometimes when you watch the TV in fact you have a big display and maybe you don't need one more on your hands.\nSpeaker A: I don't know.\nSpeaker A: Actually, it depends.\nSpeaker C: It depends what kind of functionality you want to add into it.\nSpeaker C: For example, if you add two more functionality, then better you add two more buttons there.\nSpeaker C: Instead of LCD display.\nSpeaker C: Which is that is going to bring the cost by two euro at least.\nSpeaker B: I think unless we can really drive these prices down, we need to get rid of the LCD display.\nSpeaker B: Okay.\nSpeaker C: So, we can get rid of it and then add...\nSpeaker C:...but we want that.\nSpeaker A: On the market point of view, yeah.\nSpeaker A: What do you think LCD is a major feature?\nSpeaker A: For the price, it's going to be what we can afford.\nSpeaker B: And it's looking like we can't afford the LCD display.\nSpeaker B: There's no way we can get it in there.\nSpeaker F: I think we have to come to a decision now.\nSpeaker F: I think what we just do is fold.\nSpeaker F: About the LCD display.\nSpeaker A: You assume you want a democratic voice?\nSpeaker A: Yes.\nSpeaker F: One man, one fold.\nSpeaker F: So, who thinks the LCD display should be in it?\nSpeaker B: I'd like to be in, but I can't see it happening.\nSpeaker B: I think, but...\nSpeaker B: You have power, a veto anyway, as project manager.\nSpeaker C: Yeah, but the only thing is that what is the multiple functionalities you want to include with that.\nSpeaker B: Well, we have to make a decision now.\nSpeaker F: Yes.\nSpeaker F: So, having an LCD display is just a very limited amount of buttons.\nSpeaker F: Is that acceptable?\nSpeaker F: Can your functions be implemented in...\nSpeaker A: Yes.\nSpeaker F: You agree?\nSpeaker F: So, having seven buttons instead of twelve.\nSpeaker F: Yeah.\nSpeaker F: So, that would be cutting, say about these buttons.\nSpeaker A: Because one...\nSpeaker A: Show me that.\nSpeaker A: Actually, we could, in fact, remove these three buttons and have three possibilities for each of the three here.\nSpeaker A: Here one at the middle and...\nSpeaker B: I think we're really losing as of years.\nSpeaker B: Okay.\nSpeaker C: That would create another problem.\nSpeaker C: But the people to use it is not going to be easy.\nSpeaker C: Okay.\nSpeaker F: So, I think we should go to LCD screen.\nSpeaker F: That's my opinion.\nSpeaker F: No, it's okay.\nSpeaker C: You cut the LCD scene and introduce two more buttons.\nSpeaker B: Okay.\nSpeaker B: So, LCD is out.\nSpeaker B: The speed rig out now, we've...\nSpeaker A: The speed rig condition is ours.\nSpeaker F: Yeah.\nSpeaker F: Because of the budget.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: Now we can just...\nSpeaker B: So, I'll be basically back to the original one now.\nSpeaker B: Back to the first version.\nSpeaker B: Which turns out to be on budget exactly pretty much with these new costings.\nSpeaker B: So, just look at that one and look at that one now.\nSpeaker F: Yeah.\nSpeaker F: I think we just go for this one.\nSpeaker F: That now 12 euro is the price.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker F: So, we're glad that it's 50 cents.\nSpeaker F: Yeah.\nSpeaker A: Cheap reduction.\nSpeaker A: So...\nSpeaker A: We wouldn't need the really expert designers.\nSpeaker A: Yeah.\nSpeaker A: Because they're home.\nSpeaker A: Yeah.\nSpeaker B: So, we can go back and talk to the suppliers.\nSpeaker B: We see if we can drive the prices down to add a few more things in.\nSpeaker B: But that's all we can do with the restrictions we have at the moment.\nSpeaker F: Okay.\nSpeaker F: Good.\nSpeaker F: Then we have for the change.\nSpeaker E: Thank you.\nSpeaker E: Okay.\nSpeaker E: That was it.\nSpeaker E: That's it.\nNone: Thanks.\nNone: Cool.\nNone: Okay.\nNone: Okay.\nSpeaker D: Okay.\nNone: Okay.\nNone: Finally, the design.\nNone: I need a left sheet.\n"}, {"meeting_id": "TS3012d", "summary": "The discussion in this meeting focused on detailed design. First of all, the prototype was presented, which was pretty much like what the group had agreed before, including the yellow colour, cover and buttons made by rubber, etc. However, what they had exceeded their expected budget. Therefore, they had discussions on which features were okay to be taken out. They finally agreed on removing speech recognition, and the special colour. Then they did an evaluation of their product, and overall they did well. To sum up this teamwork experience, the Project Manager emphasized the importance of the communications. At last, they had a discussion on the product name, and agreed on \"triple R\".", "dialogue": "Speaker B: Okay, things are expensive.\nNone: Form blah blah, bite.\nSpeaker E: In other words.\nSpeaker C: Got it, we don't have any changes, do we?\nSpeaker C: So no.\nSpeaker A: Okay.\nNone: There we go.\nSpeaker B: Okay, here we are again.\nSpeaker B: Detail design.\nSpeaker B: Oh, come on.\nSpeaker B: Well, I forgot to insert the minutes, but it's about the same thing we discussed before.\nSpeaker B: It opened anyway, I think.\nSpeaker B: Other design.\nSpeaker B: Anyway, we took rubber as the material last time.\nSpeaker B: We also know that you just been busy with it.\nSpeaker B: Took the advanced shift, to implement the advanced features.\nSpeaker B: Well, we discussed the design.\nSpeaker B: Those sharp corners, we rounded it off, like you see on the other screen, which is fine.\nSpeaker B: Barely, we agreed that the color should be yellow and black.\nSpeaker B: Yellow in the back, because it's more trendy than black anyway.\nSpeaker B: So, I'm okay, yeah.\nSpeaker B: We agreed that we would implement both LCD and speed recognition, but I'll get to that in a moment.\nSpeaker B: Because some changes in the financials have letters and implications anyway.\nSpeaker B: So, like I said, we had no insight into finances, no prices, but we have them now.\nSpeaker B: And it's bad.\nSpeaker B: Anyway, we are, uh, prototype presentation.\nSpeaker B: Well, first, you guys built the prototype, so you could present that.\nSpeaker A: But, um, let's see, what we had to do.\nSpeaker B: Now, you know, just go ahead and present the, we'll, we'll scrap it later because.\nSpeaker D: I think it's more or less the same.\nSpeaker C: It's basically what we agreed upon.\nSpeaker C: Not just a little bit.\nSpeaker A: Not much.\nSpeaker A: I didn't change that much.\nSpeaker B: No, no, no, no.\nSpeaker B: I didn't expect any of it.\nSpeaker B: You just colored it.\nSpeaker C: Final design.\nSpeaker C: Basically, into what we discussed, covered buttons, what we made of rubber, yellow color, black components, as you can see right over here.\nSpeaker C: I'll let the menu.\nSpeaker C: I'll let the menu.\nSpeaker C: I'll show you a different type of color for the menu, a bit darker yellow, so that it really shows in this keypad.\nSpeaker C: If you put them all black, it's not really that good of contrast.\nSpeaker B: So, and I suppose the yellow is not printed on the, on the rubber.\nSpeaker B: It's part of the rubber, I suppose.\nSpeaker B: Probably.\nSpeaker B: Yeah.\nSpeaker C: I think that's more durable anyway than printed onto.\nSpeaker C: I guess it's more easier to just paint it on the rubber than to the integration store again.\nSpeaker C: Okay.\nSpeaker C: So, we have, it's a bit round shaped.\nSpeaker C: That's what we had.\nSpeaker C: We chose the buttons to be teletext, okay button, favorite channel, and the mute.\nSpeaker C: So, that's basically what we chose there.\nSpeaker C: Okay.\nSpeaker D: This is just a description of what we see there.\nSpeaker B: That's pretty much it.\nSpeaker B: Okay.\nSpeaker B: That's my time to ruin everything.\nSpeaker B: Well, not ruin everything, but.\nSpeaker B: Oh, sorry.\nSpeaker B: No.\nSpeaker B: Finances.\nSpeaker B: That's what we have here.\nSpeaker B: What you drew.\nSpeaker B: We have battery power.\nSpeaker B: We have advanced chips and the sensor, sample sensor for speak recognition anyway.\nSpeaker B: So, which you see the, which is one of the most expensive parts.\nSpeaker B: So, well, we have seen one curve in design, rubber design, and we had a special color, suppose yellow is a special color.\nSpeaker B: So, it's just half a year for, you got push buttons and an LCD display.\nSpeaker B: Gives us a total of 17 euros in production cost, which is higher than the 12 and a half that we are permitted to use.\nSpeaker B: So, easy.\nSpeaker B: What do we scrape?\nSpeaker B: I think the best solution that I came up with is just to take out the speech recognition.\nSpeaker B: I'd say that too.\nSpeaker B: Because the LCD has more support on customer side.\nSpeaker B: There are 91% of the people or something like that, but 90% who favor an LCD display and only 60% that favor speech recognition.\nSpeaker C: I think it's harder to have a extra function with the speed tempo, which you can't do with a normal remote control, which people already do.\nSpeaker B: So, I took that out.\nSpeaker B: So, I was still stuck with 13, so I had to take out the special color, I suppose.\nSpeaker B: And, yeah, I didn't see anything else I could take out.\nSpeaker B: Yeah, I should take out the push buttons, but we need those.\nSpeaker B: So, generally what I came up with.\nSpeaker B: In order to have production cost of 12 and a half euros, scraped speech recognition and the separate covers can account for the, if people want it.\nSpeaker B: Well, just, then we'll just do it in black.\nSpeaker B: We'll just, delivered in black, it has all the function that it's supposed to have.\nSpeaker B: And if you want the custom design, then you can buy the separate covers.\nSpeaker B: You make it, or whatever you want.\nSpeaker C: I tend to disagree with you on that, because the trend issue was a big issue when we started designing this.\nSpeaker C: So, can't we just basically extend it to 13?\nSpeaker B: Okay, let's just see what we know we have to be under 12 and a half.\nSpeaker B: It's not, there's no go, we go over 12 and a half.\nSpeaker D: Okay, but there's another problem.\nSpeaker D: If you take another cover, for instance, black, then we also need another button frame, because black and black doesn't work, obviously.\nSpeaker B: I think you, that's what you were assigned to do really to see how both those work together.\nSpeaker B: Yeah.\nSpeaker B: So, I think, yeah, I think it's a good way to, to help people, to make, to keep product 22, just keep, just make new covers for it, like we agreed before.\nSpeaker B: I agree.\nSpeaker B: And everything that's left is the basic function that we want our product to have, because the expensive parts are in the advanced chip, but we need that for the LCD display.\nSpeaker B: Then again, we have the LCD display, which is also expensive, but let's go together.\nSpeaker B: And yeah, we could take out the curve.\nSpeaker B: Oh, say, let's, we could take out the curve, indeed.\nSpeaker B: We could take out the curve. Is that an option?\nSpeaker B: Yes.\nNone: For you?\nSpeaker C: I think the other is more important than really the curve, because if you just end up with an entirely black,\nSpeaker B: I think it's, it'd be just ruin it, but the fact that I took that decision or took this example, actually not really the decision, but the example is because we do offer the, the possibility of adding your own custom cover so you can change them into any color you want.\nSpeaker B: So, it's just you deliver basic remote control with the possibility to change it into whatever you want.\nSpeaker D: Give me the note also, change the material. We take plastic for the basic cover.\nSpeaker B: You can take plastic, but I, that's something that's stuck into my mind is that something, something that really can't forward from the marketing research is that people like the, the squishy feeling of the spongy feeling of the, we can put that in, it really makes it, also makes it different from the existing remote control, and it's all plastic. So, which in, in term, rubber would increase durability because it doesn't break.\nSpeaker D: But what do you then suggest we lose? Because we have to lose two things and, I guess.\nSpeaker B: I, like I said, I lost the speech recognition and I lost a special color, which would make this.\nSpeaker B: Okay, and that's enough. Black and gray. Yeah, that's, that's enough.\nSpeaker B: So, black and gray is okay. I guess those are the basic colors.\nSpeaker C: Which we can fabricate. I think those people want to pay for it.\nSpeaker E: So, they want, why, why do we have to keep us on the 12th?\nSpeaker B: To ensure the profit. That's, that's the order. We're just, we're the project team and we got our orders from the, from, from the cost of our company, which say we don't want to spend more than 50 for this.\nSpeaker B: But that's not, that's not our decision to take. We have a budget of 1250 per product.\nSpeaker C: We need to stick to that.\nSpeaker B: Stick to that. I don't think it's really bad either. I mean, if we, we have the, the backup of, or the backup design thing.\nSpeaker B: I hope the people would like it. To have, I think they would do. I think they do like, we, we agreed upon that the, that the, the, the, the cover thing was a nice idea because you could have all sorts of designs.\nSpeaker B: Well, at the same time just manufacturing one product. One basic product, which you can turn into any, any taste you want.\nSpeaker B: I think it's the best solution.\nSpeaker B: To make those custom covers for the design aspect and keep the functionality between, of within the boundaries of your, of your budget.\nSpeaker D: That's what should make clear to our customer that we had to do this to stay under the cost. And that, that they know that this is an option and that we had to drop the option to stay under the cost that they know that.\nSpeaker B: Well, I don't think, yeah.\nSpeaker B: Perhaps it doesn't mean anything to the customer. Like, like, we don't care that you have to.\nSpeaker D: No, but perhaps I think, okay, the cover is such a nice idea. Let's, that, that then they allow us to make some more cost.\nSpeaker B: We can, we can at least tell them we didn't get that.\nSpeaker B: You don't know that.\nSpeaker B: I think it's, it should either be a pack, maybe it should, this should be sold in, in this, in stores with, with a standard cover or something.\nSpeaker D: No, I'm not talking about that customer, but the one that has given us the order to design this.\nSpeaker D: We could at least make it like this, but like you said, and then tell them, okay, we had to drop this and death just that you know.\nSpeaker D: It isn't still an option, but not for this price.\nSpeaker B: Yeah, it's true.\nSpeaker B: So actually, it's not that much of an increase, but yeah, we kind of contact them.\nSpeaker B: It's just the order that we got. So that's what we got to go with.\nSpeaker B: So it's either once final, just to get it, just to get it through final, it's either turned into plastic, drop the squishy feel, make it, make it more breakable, or turn it yellow.\nSpeaker B: So it's, so we have to decide on.\nSpeaker D: Let's say lose the curve.\nSpeaker B: I said lose the curve.\nSpeaker B: Oh, that's true. We could lose the curve.\nSpeaker B: Yeah, I forgot that. Sorry.\nSpeaker B: The curve.\nSpeaker C: So which curve is that? That's basically just this one.\nSpeaker B: Just the banana curve.\nSpeaker E: Yeah, so we could still have the comfort.\nSpeaker B: No, no, that would be a curve inside the thing, I guess.\nSpeaker B: Now, would you then would just be a straight remote, just like, like that?\nSpeaker B: Which would, yeah, turn it to something far more ordinary.\nSpeaker B: We could make it yellow then, but-\nSpeaker C: You second that.\nSpeaker B: You second that, you second, do we lose the curve?\nSpeaker C: No, that it would turn out to be a pretty straightforward remote control. Okay, yeah. It's not really-\nSpeaker B: So I think it would be a good idea to keep the curve, to separate from the rest of the remote control world, so to speak. So we keep the curve.\nSpeaker B: So the only solution is either to use the- lose the yellow or lose the rubber.\nSpeaker B: And I'm in favor of keeping the rubber because it has more advantages than the color yellow has.\nSpeaker C: I would say, I would agree with you on the color because that's an extra option.\nSpeaker C: An extra service we can deliver for a little bit more money.\nSpeaker B: So, yeah, I guess people are willing to pay for that, so I think we can take that option just with the idea in the back of our head that you can customize your remote control.\nSpeaker B: So I think that would still make it a nice product.\nSpeaker B: Okay, we're fine on that, so it's too bad. We can make the whole super thing, but- Anyways, we're here.\nSpeaker B: Yeah.\nSpeaker B: Basically what we discussed.\nSpeaker B: Just now.\nSpeaker B: I sort of expected that everything would turn out this way, but because everything cannot be for free.\nSpeaker B: I think it was too bad we didn't have the financial info the last time because that was really essential, because we spent entire stage designing a product of which we had no- idea what it would cost.\nSpeaker B: So, just something- I think it's really nothing stuck with reality actually.\nSpeaker D: Me too. I felt that blind throughout the project.\nSpeaker D: Because in the beginning I had no list of- Yeah, I think it would have been.\nSpeaker B: And then I had no list of available finances.\nSpeaker B: At least the last meeting I would have expected had to have that.\nSpeaker B: So, let's see if it sells.\nSpeaker B: I mean, I suppose this sells because it's very extended, but- I hope it sells.\nSpeaker B: I suppose it sells because it's good.\nSpeaker B: I mean, it's got everything for the reasonable price because we didn't know what it's going to cost anyway.\nSpeaker E: Okay.\nSpeaker E: Let's evaluate the product of us, our design.\nSpeaker E: I have some methods, requirements and scale-off.\nSpeaker E: I will present some statements and we will decide together what, if it's true or false.\nSpeaker E: Okay.\nSpeaker E: Then we see if the requirements of the user are fulfilled or not.\nSpeaker E: And I will make a new blank sheet.\nSpeaker E: So, the buttons, the look and feel, I thought it was okay, but that forms the settings screen, audio and channel.\nSpeaker E: They're stuck on the menu.\nSpeaker B: We are not- For the menu.\nSpeaker B: I think those are totally met because we really took them for the- They have the feel they want, they have the simplicity they want.\nSpeaker B: I think it's very well met.\nSpeaker B: Either two or one maybe.\nSpeaker B: I think we took that everything they wanted into consideration.\nSpeaker B: So, it could either be a two or one.\nSpeaker E: So, one and a half.\nSpeaker E: Which is not an option.\nSpeaker E: Oh yeah, let's create our own option.\nSpeaker E: Look and feel it everywhere, it's true.\nSpeaker E: So, and the next one.\nSpeaker E: Yeah.\nSpeaker E: When it's lost, you can find it.\nSpeaker B: Even for deaf people.\nSpeaker B: I don't think it's perfect, but we did everything possible to get it back.\nSpeaker B: Because it's stuck in your couch.\nSpeaker B: You can't see the light.\nSpeaker B: Maybe you can't hear it.\nSpeaker B: We tried.\nSpeaker B: So, I think that deserves a one.\nSpeaker A: Definitely.\nSpeaker E: That's okay then.\nSpeaker E: And the next one.\nSpeaker E: How is that?\nSpeaker E: We don't have a manual.\nSpeaker E: Yeah.\nSpeaker E: I think that's part of it.\nSpeaker D: It could be a little bit more difficult than a normal module.\nSpeaker D: But then again, it's for young people.\nSpeaker E: Yeah, and LCD tells a lot about-\nSpeaker C: And it's pretty straightforward. Navigation, not keys to navigate through the LCD manual.\nSpeaker E: I think it won't be a big problem.\nSpeaker E: So, it's a one or- One.\nSpeaker C: No.\nSpeaker A: I think what we did-\nSpeaker B: For the other phones, there was no issue on making manual actually. No, it really discusses.\nSpeaker B: I don't think it takes no.\nSpeaker B: It really doesn't take time to learn.\nSpeaker B: I think we took it so easy.\nSpeaker B: We have so little buttons, everything speaks for itself really.\nSpeaker B: So, I think it's two or one.\nSpeaker B: I guess maybe it's a two because the LCD is a little new.\nSpeaker B: There are some options hidden under the menu button.\nSpeaker B: So, I might make this a two instead of a one, I guess.\nSpeaker E: And LCD, you have to see it.\nSpeaker A: So, just make that a two.\nSpeaker E: Oh.\nSpeaker E: It's a little bit learning.\nSpeaker E: Okay.\nSpeaker E: Yeah.\nSpeaker E: It's a little bit same, but it tells you- Can you just see it in a good way?\nSpeaker B: I think so.\nSpeaker B: I think it's perfect.\nSpeaker B: The where it is, what it can do.\nSpeaker B: If it's useful.\nSpeaker B: I think so.\nSpeaker E: Yeah.\nSpeaker E: What are we displaying on the LCD screen?\nSpeaker E: Just only the channels and-\nSpeaker D: In the menu, sir. Things like brightness and those kind of things we've put in the menu.\nSpeaker D: Because we have no buttons for those.\nSpeaker C: There are basically menu options indeed.\nSpeaker E: Oh, in the LCD screen.\nSpeaker B: I mean, the LCD screen, the small screen.\nSpeaker B: And for the channel selection.\nSpeaker B: I thought it was- I thought the people wanted previews on there.\nSpeaker B: I'm sure that even possible.\nSpeaker B: Because it requires quite a bit of bandwidth.\nSpeaker B: I don't think it's possible really, but they didn't really define what should be used for.\nSpeaker B: Maybe a TV guide or something.\nSpeaker B: I think, for example, like TV guides, I think that's- Yeah.\nSpeaker B: That you can transmit through it and everything.\nSpeaker B: Just for extra information on your programs.\nSpeaker E: It must be clear.\nSpeaker B: It also thinks like- It must be for menus.\nSpeaker B: Or how about preferences of your- A lot of configuring your remote control for your favorite channel.\nSpeaker B: For people.\nSpeaker B: How do you configure that?\nSpeaker B: So that could be done by the LCD display.\nSpeaker B: I think it's good.\nSpeaker B: Maybe it's down to one because we're not using it perfectly.\nSpeaker B: We didn't give it- I don't think- if we're discussing now, we gave it enough thought though.\nSpeaker B: I think we should just lower it.\nSpeaker B: Maybe it's a three though.\nSpeaker B: You could have used it more effectively.\nSpeaker B: Probably.\nSpeaker E: Yeah, indeed.\nSpeaker E: So everybody's agree with the free on it.\nSpeaker B: We are using it, but it's not poorly used, but it's not efficiently used I think.\nSpeaker B: We could have even lost the selection button and done everything via LCD selection.\nSpeaker B: Yeah.\nSpeaker B: Now it's just extra.\nSpeaker B: To illustrate it.\nSpeaker B: Yeah.\nSpeaker B: Extra features.\nSpeaker B: Okay.\nSpeaker B: Free.\nSpeaker D: Nah, it's not really only an extra menu.\nSpeaker D: Nothing.\nSpeaker D: Seven.\nSpeaker E: Seven.\nSpeaker E: That's a- Can you talk to remote control?\nSpeaker B: Or we can say- Well, we can't talk anymore.\nSpeaker B: Yeah.\nSpeaker B: So we scraped that.\nSpeaker C: Or we could say neutral because we scratched the-\nSpeaker B: It's just to be a prick, but if you can talk to your remote control, it doesn't do anything. You can talk to it.\nSpeaker E: Not with the speech recognition.\nSpeaker E: Yeah.\nSpeaker E: All the trends and all the colors anymore.\nSpeaker E: So-\nSpeaker B: Well, we did take everything in consideration, of course. Yeah.\nSpeaker B: I think- I think it's okay.\nSpeaker E: But the colors, we don't have special colors.\nSpeaker B: No, we don't have the colors.\nSpeaker B: But we took into consideration the fact that it's customizable to the fashion.\nSpeaker B: Yeah, we don't have it.\nSpeaker C: So-\nSpeaker B: We don't have it. We do have it.\nSpeaker B: It's just sold as a package.\nSpeaker B: It's not part of the basic product.\nSpeaker D: Changing covers is also trends.\nSpeaker B: That's what I call trending.\nSpeaker B: I mean, the shape is trendy.\nSpeaker B: The functions are trendy.\nSpeaker B: It's just the colors that are not supporting the basic model.\nSpeaker B: Yeah.\nSpeaker B: Because yeah, it's just not affordable.\nSpeaker C: Maybe we should go with it too then.\nSpeaker C: Because it's not perfect.\nSpeaker C: Because we can't do it initially.\nSpeaker B: It's possible that you have to pay extra for the possibility of having a different color.\nSpeaker E: Oh, well.\nSpeaker E: Hmm.\nSpeaker E: The two, right?\nSpeaker E: The last one.\nSpeaker E: That's all score.\nSpeaker E: Overall.\nSpeaker E: It's a three.\nSpeaker E: 16.\nSpeaker E: 16.\nSpeaker E: Two points.\nSpeaker A: Two points.\nSpeaker E: Seven or something.\nSpeaker E: I don't know.\nSpeaker C: Sixteen divided by six.\nSpeaker C: Two, two, third.\nSpeaker E: So we can say it's the product is.\nSpeaker E: It's okay.\nSpeaker B: It's okay.\nSpeaker B: But mostly it's influenced by the fact that we didn't have enough resources to implement speech recognition.\nSpeaker B: Because yeah, that gives you a seven which ruins your average.\nSpeaker B: Without that it would be like it would be under two.\nSpeaker B: So I think even with this it's reasonable.\nSpeaker C: Yeah, if we make it into a four as in neutral because we didn't implement it so we can't say that we.\nSpeaker C: Yeah.\nSpeaker C: And it's really not well implemented.\nSpeaker C: We come out on average of two one eight.\nSpeaker B: I think it's two.\nSpeaker B: Okay.\nSpeaker C: Which is pretty good.\nSpeaker C: Yeah, two three.\nSpeaker C: At least on the positive side.\nSpeaker C: Yeah, definitely have done better if we've had more resources.\nSpeaker B: Yeah, I think it's probably.\nSpeaker B: I do admit that we did miss a little or didn't talk enough about the possibility of the LCD display.\nSpeaker B: We could have used it more efficiently.\nSpeaker B: We just didn't think of it that way.\nSpeaker E: Yeah, we've scored a few.\nSpeaker B: So like I said, changing channels, everything hidden in your LCD display.\nSpeaker B: So you just need the navigation buttons to do everything.\nSpeaker D: I think for this price, this is really a reasonable product.\nSpeaker B: I think we did very well.\nSpeaker B: Even if you look at this, we did quite well.\nSpeaker E: With an LCD screen.\nSpeaker B: Just looking for improvements what you could have improved.\nSpeaker D: If the people really want speech recognition then they must be prepared to pay more.\nSpeaker D: Because it's a combination.\nSpeaker B: I should get kids and just stick them in TV and say change the channel.\nSpeaker D: But for this price, you cannot ask that.\nSpeaker B: It's not affordable.\nSpeaker B: You should lose the LCD screen probably.\nSpeaker B: I think the LCD screen is more worth that speech recognition.\nSpeaker B: It's also more attractive.\nSpeaker B: Definitely.\nSpeaker B: Okay.\nSpeaker E: How's that?\nSpeaker E: Yeah.\nSpeaker B: So that's the final product without the speakers.\nSpeaker B: Let's see.\nSpeaker B: What was the left of that?\nSpeaker B: Another one.\nSpeaker F: Yeah.\nSpeaker B: We evaluate the products.\nSpeaker B: We're not.\nSpeaker B: General projects.\nSpeaker B: For example, I thought we were pretty creative.\nSpeaker B: We took the whole new approach of making exchangeable cover for example.\nSpeaker B: I thought it was pretty creative.\nSpeaker B: It was never ever listed somewhere.\nSpeaker E: Favorite channel?\nSpeaker B: Anyways, leadership is up to you.\nSpeaker B: Perhaps I screwed up.\nSpeaker B: For this speech recognition.\nSpeaker B: That's not from the inside.\nSpeaker B: I think we did pretty well as teamwork.\nSpeaker B: It was hard to work with one another if you cannot communicate in the meantime.\nSpeaker B: When I got the input for the financial results, initially I wanted to contact USA.\nSpeaker B: Look, you're doing the wrong thing.\nSpeaker B: You're wasting your time now because we're implementing stuff that we cannot afford.\nSpeaker B: It would be better if there was more communication between the director.\nSpeaker B: That's what we usually do.\nSpeaker B: Either call or email someone.\nSpeaker D: That was too bad.\nSpeaker D: It was the same thing that I had in the beginning.\nSpeaker B: It was using materials that I didn't have.\nSpeaker B: It didn't have or didn't know what they cost or what.\nSpeaker B: There was just too little information about what things actually cost and if you could use them.\nSpeaker B: That was a little unclear.\nSpeaker B: I think a smartboard was pretty cool.\nSpeaker B: It was easier to share them.\nSpeaker C: For actual design, I'd say the response time should be a little bit higher.\nSpeaker B: The response time is very bad.\nSpeaker B: It's good to visualize everything, but I think the response time could be a lot better.\nSpeaker C: The digital pen was definitely better to draw my ideas.\nSpeaker C: I know.\nSpeaker D: There's also one problem with this I noticed.\nSpeaker D: You have to finish a page.\nSpeaker D: No, you don't have to.\nSpeaker A: No, you don't.\nSpeaker B: I didn't check the finish button.\nSpeaker B: I just did you just copy it or whatever.\nSpeaker B: Only if you check the notes or press done, then it exports the word automatically.\nSpeaker B: That's not necessary to check either one of those two.\nSpeaker B: Okay, but I made a few.\nSpeaker B: You can just preview your page in the program.\nSpeaker B: Okay, but I made three pages.\nSpeaker D: They were not finished.\nSpeaker D: When the third one was finished, I wanted to download it.\nSpeaker D: Then it was not possible anymore.\nSpeaker D: You have to close all the pages before going further.\nSpeaker B: Okay, before starting a new page.\nSpeaker D: So you cannot work on one page at the same time?\nSpeaker D: That's not possible.\nSpeaker A: You have to finish it completely, then download it, then start a new one.\nSpeaker D: That's not very handy.\nSpeaker D: If you know that, then it's not possible.\nSpeaker A: Yeah, I understand it more.\nSpeaker B: Okay.\nSpeaker B: Any new ideas?\nSpeaker B: Yeah, more communication between those things.\nSpeaker B: I noticed that communication is very important because if you get new information, it's essential for the other teammates to know that as soon as possible.\nSpeaker B: Because you would avoid doing extra work because you were doing extra work now working on the speech recognition limitation, both on the technical and on the design side.\nSpeaker B: So I think that could have been better, but I think it's more of a setting here that you can communicate than somewhere else.\nSpeaker C: Yeah, it could possibly be a more real-time information base.\nSpeaker C: So we can all see.\nSpeaker B: Yeah, I think so.\nSpeaker B: Which information is available for you?\nSpeaker B: I'm not sure you got spammed as well.\nSpeaker B: I get to spend like every two minutes.\nSpeaker B: There was another email about master classes or something, which were totally useless actually.\nSpeaker B: I thought I should probably look into them.\nSpeaker B: They were all useless.\nSpeaker C: I personally did not have that.\nSpeaker C: That's probably your distribution.\nSpeaker C: What I also did not really assume you had that as well is that we finished up the design and then we checked the website.\nSpeaker C: And then there was just extra information.\nSpeaker E: After five minutes.\nSpeaker C: There was a little delay in that.\nSpeaker B: I didn't have any more information.\nSpeaker B: It's just always the same here.\nSpeaker B: So that's kind of an email change, but not for me.\nSpeaker B: I had no extra information to go.\nSpeaker B: I couldn't do any research myself.\nSpeaker B: I could have done a little extra work probably.\nSpeaker B: But I was busy enough anyway.\nSpeaker B: Any ideas found?\nSpeaker A: Or is that probably?\nSpeaker C: How much time do we have for this anyway?\nSpeaker A: I have no clue.\nSpeaker B: If the project is evaluated and it was within budget, we should celebrate.\nSpeaker C: Bring out the beer.\nSpeaker C: Champagne.\nSpeaker A: One more about it.\nSpeaker B: I'm not sure what we should still do.\nSpeaker B: I don't see, but all your tasks are finished, right?\nSpeaker D: I have no more emails.\nSpeaker D: It's being very silent now.\nSpeaker B: I still have the total report to finish up.\nSpeaker B: I think we took very little time now.\nSpeaker B: We're in agreement.\nSpeaker B: The design is okay.\nSpeaker B: One thing we missed though.\nSpeaker D: We have a product name.\nSpeaker D: It's better than a serial number.\nSpeaker D: Sony TR-something.\nSpeaker B: Think of a catchy name.\nSpeaker B: Until the beef.\nSpeaker E: Fruit name?\nSpeaker E: Banana remote.\nSpeaker B: It's the front.\nSpeaker B: It's not yellow.\nSpeaker C: I was going for the RC Deluxe, but it's not a catchy name or anything.\nSpeaker B: It's more?\nSpeaker D: No, it's not something with numbers.\nSpeaker D: Numbers are so meaningless to the panel.\nSpeaker C: Something with our company name.\nSpeaker C: Anything with that?\nSpeaker C: A real website which will help us out.\nSpeaker D: Real reaction.\nSpeaker B: Real reaction.\nSpeaker B: The future RC.\nSpeaker B: Step into the future of remote controlling your TV.\nSpeaker C: Is that a name or a campaign?\nSpeaker C: That's a catchy slogan.\nSpeaker C: Control your remote control.\nSpeaker B: I go for future RC.\nSpeaker D: The real reactor I don't find that bad at all?\nSpeaker B: Real reactor.\nSpeaker B: It makes me think of different products in a remote control.\nSpeaker B: Real reaction.\nSpeaker C: Stepping.\nSpeaker D: I'm looking for things in the name.\nSpeaker E: The triple R.\nSpeaker D: Real reaction.\nSpeaker D: Triple R.\nSpeaker E: Do you mean it like this?\nSpeaker E: Real reaction.\nSpeaker E: Remote control.\nSpeaker B: It should be longer because it's not a product name.\nSpeaker B: Triple R.\nSpeaker B: Triple R.\nSpeaker E: The triple R.\nSpeaker E: The triple R.\nNone: The triple R.\nSpeaker B: The triple R.\nSpeaker C: I think I like it like this more.\nSpeaker B: Triple R or triple RC.\nSpeaker C: I like it.\nSpeaker B: I like it.\nSpeaker B: I like it like this.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker D: The triple R.\nSpeaker D: The triple R.\nSpeaker D: Triple R.\nSpeaker C: The triple R.\nSpeaker C: The triple R.\nSpeaker E: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker B: The triple R.\nSpeaker A: The triple R.\nSpeaker D: The triple R.\nSpeaker F: The triple R.\nSpeaker E: The triple R.\nSpeaker E: The triple R.\nNone: The triple R.\nNone: The triple R.\nNone: The double R.\nNone: The triple R.\nSpeaker B: Could you guys draw me a picture of the file design to put on the cover of the report?\nSpeaker D: I'll try to do it all over again.\nSpeaker C: I don't really know whether we can save it as a picture of it.\nSpeaker D: Is it okay if I try?\nSpeaker E: I'll put it by need information.\nSpeaker D: It's okay.\nSpeaker D: It has been saving something.\nSpeaker D: Yes, I can export as you push.\nSpeaker D: Is it okay if I try not to put this, whatever I want.\nSpeaker D: My document is the wrong one.\nSpeaker E: Is it not that we're placing?\nSpeaker E: No.\nSpeaker E: Maybe it's not on the network of the rest.\nSpeaker C: I wouldn't take that one now.\nSpeaker D: That's pity.\nSpeaker D: That means that we have to draw it again.\nSpeaker D: Are you going to do that?\nNone: Yes.\nSpeaker D: Yes.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: Yes, that's correct.\nSpeaker E: Okay.\nSpeaker D: No.\nSpeaker D: Oh, it's exposed.\nSpeaker E: Oh, yeah.\nSpeaker D: Okay.\nSpeaker E: Can I see scores?\nSpeaker E: Oh, of course.\nSpeaker E: Sorry.\nSpeaker E: One, one, two, three.\nSpeaker E: Okay.\nSpeaker E: Overall two.\nSpeaker E: Yes.\nSpeaker C: I see you later.\nSpeaker E: Yeah.\nNone: Okay.\n"}, {"meeting_id": "Bro008", "summary": "The project was finally coming to a conclusion and the team was getting ready to share their results. They achieved an error rate reduction for the task, but the error rate in itself was still significant. The team also had to think about how they could make their model efficient for it to be deployed as they continue with future research. They discussed some new directions and were reminded that while the initial phase was over, much work lied ahead of them.", "dialogue": "Speaker D: Okay, so I got these results from CIFON.\nSpeaker D: So I think that we might hear later today about other results.\nSpeaker D: And there are some other very good results compared to other results from other places.\nSpeaker D: I'm sorry.\nSpeaker D: I got this from you.\nSpeaker D: And then I sent a note to CINNAL about because he has been running some other systems other than the XEOG.\nSpeaker D: I want to see what that is.\nSpeaker D: But, you know, so we'll see what is comparatively later.\nSpeaker D: But it looks like most of the time even though it's true that the overall number for damage is we didn't improve it.\nSpeaker D: If you look individually, what I really say is that there's looks like out of the six cases between the different kinds of matching conditions.\nSpeaker D: Out of the six cases, there's basically a couple of states about the same.\nSpeaker D: Three work gets better and one work gets worse.\nSpeaker E: Actually, for the damage, there's still some kind of mystery because when we use the straight features, we are not able to get this nice number with the XEOG I1.\nSpeaker E: We don't have this 93.78.\nSpeaker A: 89.44.\nSpeaker E: So, there's probably something wrong with the feature that we get from Boji and Sunili is working on.\nSpeaker D: Oh, let me have a little time on that actually.\nSpeaker D: We have a little bit of time on that actually.\nSpeaker D: Yeah, there it sounds.\nSpeaker D: When do you folks leave?\nSpeaker E: Sunday.\nSpeaker D: Sunday.\nSpeaker D: Saturday, midnight or something.\nSpeaker D: That would be good.\nSpeaker D: That would be good.\nSpeaker D: And, you know, whenever anybody figures it out, they should also, for sure, email Heenik because Heenik will be over there telling people we did some of these.\nSpeaker D: So, we'll hold off on that a little bit. Even with these results as they are, it's really not that bad.\nSpeaker D: It looks like the overall result, as they are now, even without any bugs being fixed, is that on the other tasks, we had this average of 49% or so improvement.\nSpeaker D: And here we have somewhat better than that in the Danish and so on.\nSpeaker D: Worst than that in German.\nSpeaker D: But, I mean, it sounds like one way or another of the methods that we're doing can reduce the error rate of mail capture down by 4th of them to a half of them.\nSpeaker D: So, I'm not depending on that.\nSpeaker D: Yeah, the exact case.\nSpeaker D: So, that's good.\nSpeaker D: I mean, I think that one of the things that Heenik was talking about was understanding what was in the other really, the proposals and trying to see if what should ultimately be proposed to some combination of things.\nSpeaker D: Because there's things that they are doing there that we certainly are not doing.\nNone: And there's things that we're doing that they're not doing.\nSpeaker C: Like the things.\nSpeaker D: How much better was the best system than ours?\nSpeaker D: Well, we don't know yet.\nSpeaker D: I mean, first place there's still this thing to work out.\nSpeaker D: And second place, second thing is that the only results that we have so far before were really development and results.\nSpeaker D: And this community that's of interest, it's not like everything is being pinned on the evaluation set. But for the development set, our best result was a little bit short of 50%.\nSpeaker D: And the best result of any system was about 54.\nSpeaker D: Where these numbers are the relative reduction in word error rate.\nSpeaker D: Oh, okay.\nSpeaker D: The other systems were somewhat lower than that.\nSpeaker D: There was actually, it was much less of a huge range than there was in a RR1.\nSpeaker D: There were systems that basically didn't approve things.\nSpeaker D: And here the worst system still reduced their rate by 33%.\nSpeaker D: So, you know, sort of everybody is doing things roughly 30 years and half a year is being created and bearing on different tests and so forth.\nSpeaker D: So, I think it's probably a good time to look at what's really going on.\nSpeaker D: And see if there's a way to combine the best ideas while at the same time not blowing up the amount of resources used.\nNone: That's critical.\nSpeaker C: Do we know anything about who's wasn't that had the lowest on the Devset?\nSpeaker D: There were two systems that were put forth by a combination of French telecom and Alcatel.\nSpeaker D: And they differ in some respects.\nSpeaker D: But one was called the French telecom, Alcatel system.\nSpeaker D: It was called the Alcatel French telecom system.\nSpeaker D: It was the biggest difference.\nSpeaker D: And they both did very well.\nSpeaker D: So, my impression is they also did very well on the evaluation set.\nSpeaker D: But we haven't seen any minor results from that.\nSpeaker C: And they used the main thing that they used with spectral subtraction?\nSpeaker D: There was a couple pieces to it.\nSpeaker D: There was a spectral subtraction.\nSpeaker D: And there was some modification of the capture of parameters.\nSpeaker E: Yeah, actually something that's close to kept strumming subtraction.\nSpeaker E: But the way the mean is adapted, it's signal dependent.\nSpeaker E: So basically the mean is adapted during speech and not during silence.\nSpeaker E: But it's very close to kept strumming subtraction.\nSpeaker D: And we've done exactly that sort of thing to look in speech only to try to measure these things.\nSpeaker D: So it looks like they did some reasonable things.\nSpeaker D: And we did unreasonable things.\nSpeaker D: Because we like to try strange things.\nSpeaker D: And our things work too.\nSpeaker D: It's possible that some combination of these different things that we're done will be the best thing to do.\nSpeaker D: But the only caveat to that is that everybody is being real conscious of how much memory, how much CPU they're using.\nSpeaker D: Because these standards are supposed to go on cell phones with moderate resources, both your specs.\nSpeaker C: Did anybody do anything with the models as an experiment?\nSpeaker D: They didn't report it.\nSpeaker D: I think everybody was focused elsewhere.\nSpeaker D: Now one of the things that's nice about the CPU did is we do have a filter in which leads to a production of the bandwidth and the modulation spectra, which allows us to down sample.\nSpeaker D: So as we know that we have a reduced transmission rate for the bands.\nSpeaker D: Now it was reported the first time out. It said the same amount because for convenience sake, in a particular way, this is being tested.\nSpeaker D: They were repeating the packets.\nSpeaker D: They had 2400 bits per second, but they were literally creating 4800 bits per second.\nSpeaker C: So you could have had a repeat count in there or something.\nSpeaker D: This was just a funny thing to fit into the software that was testing the errors, channel errors, and so on.\nSpeaker D: So in reality, if you put this system into the field, it would be 2400 bits per second.\nSpeaker D: So that's a nice feature of what we did.\nSpeaker D: But we still have to see how it all comes out. And then there's the process, which is a lot together.\nSpeaker C: When is the development set? I mean the test set results do. Like the day before you leave or something?\nSpeaker D: Probably the day after they leave, but we'll have to stop it the day before.\nSpeaker D: I think the meeting is on 13th or something.\nSpeaker D: Yeah, it's Tuesday.\nSpeaker D: And the results are due like the day before you leave or something.\nSpeaker D: Yeah, probably one.\nSpeaker D: I think they are. Yeah, so since we have a bit farther to travel.\nSpeaker D: I'll have to get that a little quicker.\nSpeaker D: I mean, it's just tracing down these bugs. I mean, just exactly this sort of thing.\nSpeaker D: Why these features seem to be behaving differently in California than in Oregon.\nSpeaker D: I guess something to do with electricity shortage. We can have enough electrons here.\nSpeaker D: But I think the main reason for having, you know, it takes to run the two test sets in just in computer time.\nSpeaker D: It's just a day or so.\nSpeaker D: And so I think the whole reason for having as long as we have, which is like we can have this because of bugs like that.\nSpeaker D: So we're going to end up with these same kind of sheets that have the percentages.\nSpeaker E: Yeah, so there are two more calamity machines. I guess it's the same sheets.\nSpeaker E: Yeah, it's the same sheets.\nSpeaker D: Yeah. So I'll just regard these numbers. That's good.\nSpeaker E: So you can try to push for trying to combine different things.\nSpeaker D: Well, let's.\nSpeaker D: Yeah, I mean, I think the question is, is there is there some advantage?\nSpeaker D: I mean, you could just take the best system and say that's the standard. But the thing is that if different systems are getting at good things, begin with in the constraint of the resources.\nSpeaker D: If there's something simple that you could do.\nSpeaker D: For instance, I think very reasonable to have a standard for the terminal side.\nSpeaker D: And then for the server side, say here's a number of things that could be done.\nSpeaker D: So everything we did could probably just be added on to what Alcatel did.\nSpeaker D: We're pretty well with them too.\nSpeaker D: So that's one aspect of it.\nSpeaker D: And then on the terminal side, I don't know how much memory and CPU it takes.\nSpeaker D: But it seems like the filtering.\nSpeaker D: I mean, the VAD stuff they both had, right?\nSpeaker D: So, and they both had some kind of online normalization.\nSpeaker D: So it seems like the main difference there is the filtering.\nSpeaker D: And the filtering, I think, if you can, it shouldn't take a lot of memory to do that.\nSpeaker D: And I also wouldn't think the CPU would be much either.\nSpeaker D: So if you can add those in, then you can cut the data right now.\nSpeaker D: So it seems like the right thing to do is to, on the terminal side, take what they did if it does seem to generalize well to German Danish.\nSpeaker D: Take what they did, add in a filter and add in some stuff on the server side.\nSpeaker D: And that's probably a reasonable standard.\nSpeaker E: They are working on this already because some filter will be that you are trying already to put some kind of...\nSpeaker D: Yeah, so that's the thing.\nSpeaker D: That would be ideal, if they could actually show that in fact the combination of some sort would work even better than any other system chat.\nSpeaker D: And then it would be something to discuss on the meeting.\nSpeaker D: But not clear what will go on.\nSpeaker D: I mean, on the one hand, sometimes people are just anxious to get a standard out there and you can always have another standard after that.\nSpeaker D: But this process is going on for a while already.\nSpeaker D: Might just want to pick something and say, okay, this is it.\nSpeaker D: And then that's a standard.\nSpeaker D: Standards are always optional, it's just that if you disobey them, then you risk not being able to sell your product.\nSpeaker D: And people often work on new standards, well-known standards, and so on.\nSpeaker D: So it's not final, even if they declare a standard.\nSpeaker D: The other hand, they might just say they just don't know enough yet to declare a standard.\nSpeaker D: You will become experts on this, more firm or than me, but this particular standard is brought since once you go to this meeting.\nSpeaker D: So I'd be interested hearing your thoughts now.\nSpeaker D: I mean, you're almost done.\nSpeaker D: I mean, you're done in the sense that maybe you'll get some new features from snail and we'll rerun it.\nSpeaker D: But other than that, you're basically done.\nSpeaker D: So you're just hearing your thoughts about where you think we should go from this.\nSpeaker D: I mean, you're trying a lot of things in a hurry.\nSpeaker D: And if we can back off from this now and sort of take our time with something that is doing things quickly, be quite so much the constraint, what you think would be the best thing to do.\nSpeaker E: Well, first, to really have a look at the speech from this database is because we tried several things, but we didn't really look.\nSpeaker E: What's happening?\nSpeaker E: Where is the noise?\nSpeaker D: It's a novel idea.\nNone: Look at the data.\nSpeaker D: More generally, I guess, what is causing the degradation.\nSpeaker D: Yeah, yeah.\nSpeaker E: Actually, there is one thing that generally we think that most of the errors are within phoneme classes.\nSpeaker E: So I think it could be interesting to see if it, I don't think it's still true when we add noise.\nSpeaker E: So I guess the confusion, the confusion, matrices are very different when we have noise.\nSpeaker E: When it's clean speech.\nSpeaker E: And probably there is much more between classes errors or noisy speech.\nSpeaker E: So, perhaps we could have a large gain just by looking at improving the recognition, not phoneme, but phoneme classes simply.\nSpeaker E: Which is a simpler problem perhaps, but which is perhaps important for noisy speech.\nSpeaker D: So the other thing that strikes me just looking at these numbers is just taking the best cases.\nSpeaker D: Some of these, of course, even with all of our wonderful processes, still our horrible kinds of numbers.\nSpeaker D: Just take the best case, the well matched, the German case after, or well matched, Danish, after we, kind of numbers we're getting are about 8% error per digit.\nSpeaker D: Yeah, this is obviously not usable.\nSpeaker D: I mean, you have 10 digits.\nSpeaker D: Oh, no, not very bad.\nSpeaker D: Now then you get it right.\nSpeaker D: So, I mean the other thing is that, and also part of what's nice about this is that this is almost realistic database. I mean, it's still not people who are really trying to accomplish something.\nSpeaker D: But within the artificial setup, it isn't noise artificially simulated.\nSpeaker D: It's real noise condition.\nSpeaker D: And the training, I guess, is always done on close talking?\nSpeaker E: No, actually, actually the well matched condition is still quite difficult.\nSpeaker E: They have all these data from the close mic and from the decent mic, from different driving conditions, open window, close window, and they take all of this, and they take 70% for training and 30% for testing.\nSpeaker E: So training is done on different conditions and different microphones and testing also is done on different microphones and conditions.\nSpeaker E: So probably if we only take the closed microphones, I guess the research should be much, much better than this.\nSpeaker D: I see.\nSpeaker D: Okay, that's better.\nSpeaker E: So there is this, the mismatch is the same kind of thing, but the driving conditions, I mean the speed and the kind of road is different for training and testing.\nSpeaker E: And the last condition is closed microphone for training and this done for testing.\nSpeaker D: Okay, so the highly mismatched case is in some sense a good model for what we've been typically talking about when we talk about added noise.\nSpeaker D: And so it does correspond to a realistic situation in the sense that people might really be trying to call out telephone numbers or something like that in their cars.\nSpeaker D: They're trying to connect to something.\nSpeaker E: Actually, yeah, it's very close to clean speech training because the closed microphone and noisy speech testing.\nSpeaker D: And the well-matched condition is what you might imagine that you might be able to approach if you know that this is the application, you're going to record a bunch of people in cars and so forth, do these training.\nSpeaker D: And then when you sell it to somebody, it will be a different person with a different car and so on.\nSpeaker D: So this is somewhat optimistic to view on it.\nSpeaker D: So the real thing is probably somewhere in between the two. But even the optimistic one is working.\nSpeaker D: Yeah, right.\nSpeaker D: That's sort of the dominant thing is even say under development set stuff that we saw the numbers that Alcatel was getting, which was the best single numbers.\nSpeaker D: It just wasn't good enough for real system.\nSpeaker D: So still a lot of stuff to do.\nSpeaker D: And I don't know.\nSpeaker D: So looking at the data, what's the characteristic of the thing?\nSpeaker D: What are your thoughts about what else you're thinking about?\nSpeaker A: A lot of things. Because we're trying a lot of things and we're not working.\nSpeaker A: We remove this. Maybe we try again with the articulatory feature.\nSpeaker A: I don't know exactly because we tried with some one experiment and some work and forgot it.\nSpeaker A: I don't know if it's a trip because maybe to better some step of the general diagram.\nSpeaker A: I don't know if it's a trip to think what we can improve.\nSpeaker D: Yeah, because a lot of times it's true. There were a lot of times when we tried something and it didn't work right away.\nSpeaker D: Even though we had an intuition that there should be something there, so then we would just stop it.\nSpeaker D: One of the things, I don't remember the details on, but I remember at some point when we were working with a second stream in which I had a fast filtering and a cap stream.\nSpeaker D: In some case you got, well, it was an MSG-like thing, but it was an MSG.\nSpeaker D: I think in some case you got some little improvement, but it was sort of a small improvement and it was a big added complication, so you dropped it.\nSpeaker D: But that was just sort of one try, right? Just took one filter through it there, right?\nSpeaker D: And it seems to me that if that isn't an important idea, it might be that one could work at it for a while as you're saying.\nSpeaker D: And you had the multi-band thing, so there's an issue with that.\nSpeaker D: Barry is going to be continuing working on multi-band things as well.\nSpeaker D: We were just talking about some work that we were interested in, kind of inspired by the stuff where Larry saw the learning and articulatory feature, I think in the case of his paper, with Sonderance based on multi-band information, where you have a combination of gradient learning and, yeah.\nSpeaker D: So I think that this is a neat data set, and then, as we mentioned before, we also have the new digit set coming up from recordings in this room.\nSpeaker D: So there's a lot of things to work with.\nSpeaker D: Yeah, what I like about it in a way is that the results are still so terrible.\nSpeaker D: I mean, they're much better than they were. You know, we're talking about 30 to 60% error rate reduction, and it's really great stuff to do that relatively short time.\nSpeaker D: But even after that, it's still poor that we could use it.\nSpeaker D: I think that's great, and also because, again, it's not something... sometimes we've got terrible results by taking some data and are officially involving it with some new response or something to take it very one point of binding on downstairs into the basement.\nSpeaker D: It's a hallway that is very reverberant, and we made some recordings there, and then we made a simulation of the room acoustics there and applied it to other things.\nSpeaker D: But it was all pretty artificial, and how often would you really try to have your most crucial conversations in this very reverberant hallway?\nSpeaker D: So this was nice about the Aurora data and the data here is that it's sort of a realistic room situation, acoustics situation, with terms of noise and reflections and so on.\nSpeaker D: It's something that's still relatively realistic, it's still very hard to do it well.\nSpeaker E: Yeah, so... well... actually, that's why we... well, it's a different kind of data, we're not used to work with this kind of data.\nSpeaker E: That's why we should have a little more closer look at what's going on.\nSpeaker E: So this would be the first thing, and then of course try to... well, kind of debug what was wrong when we do a lot of tests on the MLG, particularly on the multiband.\nSpeaker E: Yeah.\nSpeaker D: Yeah. Yeah, I think there's lots of... it's a good thing to do with this.\nSpeaker D: So... so let's... I guess... you can see as well.\nSpeaker C: What do you think?\nSpeaker C: About... anything?\nSpeaker C: About other experiences.\nSpeaker C: Now I'm interested in looking at the experiments where you use data from multiple languages to train to neural net.\nSpeaker C: I don't know how far... or if you guys even had a chance to try that, but that would be something to be interesting to me.\nSpeaker E: Yeah, but again it's the kind of thing that you were thinking that it would work, but it didn't work.\nSpeaker E: And... sorry.\nSpeaker E: Not a bug, but something wrong in that we...\nSpeaker E: Right.\nSpeaker E: And... something wrong, perhaps, in the... just in the fact that the labels are... what work best is the end-leabled data?\nSpeaker E: So, yeah. I don't know if we can get some end-leabled data from other languages.\nSpeaker E: Yeah.\nSpeaker E: It's not so easy to find.\nSpeaker E: Right.\nSpeaker E: But that would be something interesting to me.\nSpeaker D: Yeah.\nSpeaker D: Also, there was just a whole notion of having multiple nets that were trained on different data.\nSpeaker D: So one form of different data is from different languages, but the other...\nSpeaker D: Well, in fact, in those experiments there wasn't so much combining multiple nets. It was a single net that had different.\nSpeaker D: Yeah.\nSpeaker D: So first thing is, would it be better if they were multiple nets?\nSpeaker D: So the second thing is never mind the different languages, just having different acoustic conditions, rather than training them all up and one,\nNone: would it be helpful to have different ones? So that was a question that was kind of raised by Mike Chares, these were seen in that case in terms of through preparation.\nSpeaker D: Sometimes it might be better to do that.\nSpeaker D: But I think we know.\nSpeaker D: So, all right. So next week we won't meet because you'll be here.\nNone: And when are you two getting back?\nSpeaker A: I'm...\nSpeaker A: Sunday because it's less expensive than practice.\nSpeaker A: I'm not a ticker.\nSpeaker E: Right. I'll be back Tuesday.\nSpeaker C: Where is the meeting?\nSpeaker D: Amsterdam, I think.\nSpeaker D: So we'll skip next week and we'll meet two weeks from now.\nSpeaker D: And I guess the main topic will be telling us what happened.\nNone: Yeah.\nSpeaker A: Yeah.\nSpeaker D: So, yeah. Well, we don't have anything else we should turn off the machine and say we don't ask you for.\nSpeaker C: Should we do digits first?\nSpeaker D: Oh yes, digits. Yeah, good point.\nSpeaker C: Okay. Transcript 37913810904 02007 11704 240921 3613509 495 60607 85680 97 0509 0505 0701009 1 2 3 0 7 9 8 8 0 5 7 6 3 9 7 7 1 8 8 9\nSpeaker D: Transcript 3751-3770 8 557 306 9 6 6 0 0 0 1 2 0 4 2 5 5 6 6 6 7 8 2 8 9 0 0 1 1 2 3 4 3 8 6 4 4 8 3 5 6 7 0 4 3 0 0 7\nSpeaker B: Transcript 3671-36904 5004 27203 8530510 9718 476 0 9 4 5 103 2 4 115 5 4 9 9 2 6 6 3 7 7 9 8 8 0 0 2 8 5 9 0 7 0 0 0 8 1 2 2 8 6 8 3 5 3 7 10 9 4 5\nSpeaker E: Transcript 6 9, yeah, sorry. Transcript 3691-3714-5 6 9 4 5 015-032-1720-2624-304-405-4 617-hoho 758-628-3 8 8 4 9 8 0 0 0 19 8 1 0 0 3 2 8 0 9 4 2 5 4 5\nSpeaker A: Transcript number 3771-3790 9 1 0 8 0 4 6 9 0 0 0 6 4 7 6 5 9 9 0 8 5 7 6 7 2 8 5 6 8 3 6 7 4 1 3 9 8 9 0 0 1 1 I'm sorry 1 1 1 3 1 4 4 7 5 7 7 3 6 9 6 5 7 8\n"}, {"meeting_id": "Bdb001", "summary": "Meeting participants wanted to agree upon a standard database to link up different components of the transcripts. The current idea was to use an XML script, but it quickly seemed that other options, like a pfile or ATLAS, are more suitable. The reason being that they would make it easier to deal with different linguistic units, like frames and utterances. Eventually, the team was skeptical of using something that would be hard to learn, like ATLAS. Nonetheless, they wanted to explore their options. The meeting finished with some discussion about handling annotations.", "dialogue": "Speaker E: Yeah, we had a long discussion about how easy we want to make it for people to bleep things out.\nSpeaker E: So Morgan wants to make it hard.\nSpeaker F: It doesn't.\nSpeaker F: I didn't even check yesterday when I started it.\nSpeaker E: I don't know if it doesn't like low-three.\nSpeaker F: You know, I discovered something yesterday on these wireless ones.\nSpeaker F: You can tell if it's picking up breath noise and stuff.\nSpeaker E: Yeah, it has a little indicator on it on the AAF.\nSpeaker F: So if you breathe under, breathe and then you see AAF go off, then you know.\nSpeaker F: Pick up your mouth nice.\nSpeaker D: Oh, that's good.\nSpeaker D: Because we have a lot of tests.\nSpeaker D: In fact, if you listen to just the channels and people not talk, it's like...\nSpeaker D: What did you see, Hannibal, recently?\nSpeaker D: It's a very disconquering.\nSpeaker D: Okay.\nSpeaker D: So I was going to try to get out of here in half an hour.\nSpeaker D: Because I really appreciate people coming.\nSpeaker D: And the main thing that I was going to ask people to help with today is to give input on what kinds of database format we should use in starting to link up things like word transcripts and annotations of word transcripts.\nSpeaker D: So anything that transcribers or discourse coders would ever put in the signal with time marks for words and phone boundaries and all the stuff we get out of the forced alignments in the recognizer.\nSpeaker D: So we have this, I think a starting point is clearly the channelized output of Dave Galbert's program, which Don brought a copy of.\nSpeaker D: Yeah, I'm familiar with that.\nSpeaker E: I mean, I sort of already have developed an XML format for this sort of stuff.\nSpeaker E: And so the only question is it the sort of thing that you want to use or not.\nSpeaker E: Have you looked at that?\nSpeaker E: I mean, I had a web page up.\nSpeaker D: So I actually mostly need to be able to link up or.\nSpeaker D: I mean, it's a question both of what the represent.\nSpeaker D: I guess I am going to be standing up and drawing up.\nSpeaker E: Okay, yeah, so you should definitely.\nSpeaker E: So it definitely had that as a concept.\nSpeaker E: So that it has a single timeline.\nSpeaker E: And then you can have lots of different sections, each of which have IDs attached to it.\nSpeaker E: And then you can refer from other sections to those IDs if you want to.\nSpeaker E: So that you start with a timeline tag.\nSpeaker E: Timeline.\nSpeaker E: And then you have a bunch of times.\nSpeaker E: I don't remember exactly what my notation was.\nSpeaker E: Oh, I remember it.\nSpeaker B: Right.\nSpeaker E: Yeah, t equals one point three two.\nSpeaker E: And then I also had optional things like accuracy.\nSpeaker E: And then ID equals t one.\nSpeaker E: And then I also wanted to be able to not specify specifically what the time was and just have a stamp.\nSpeaker E: So these are arbitrary assigned by a program, not by users.\nSpeaker E: And you have a whole bunch of those.\nSpeaker E: And then somewhere further down you might have something like an utterance tag, which has start equals t 17 and equals t 18.\nSpeaker E: So what that's saying is we know it starts at this particular time.\nSpeaker E: We don't know when it ends.\nSpeaker E: Okay.\nSpeaker E: Right. But it ends at this t 18, which may be somewhere else.\nSpeaker E: We say there's another utterance.\nSpeaker E: We don't know what that time actually is, but we know that it's the same time as this end time.\nSpeaker E: Yeah, whatever you want.\nSpeaker B: So you're specifically defining a lot of us?\nSpeaker E: Yes, exactly.\nSpeaker E: And then these also have IDs.\nSpeaker E: Right.\nSpeaker E: So you could have some sort of other other tag later in the file.\nSpeaker E: That would be something like, no, I don't know.\nSpeaker E: Noise type equals dorslam.\nSpeaker E: And then you could either say time equals a particular time mark, or you could do other sorts of references.\nSpeaker E: So, or you might have a prosody.\nSpeaker E: Prosody, right?\nSpeaker E: D, t.\nSpeaker D: It's an O instead of an I, but the D is good.\nSpeaker D: You like the D?\nSpeaker E: It's a good D.\nSpeaker E: You know, so you could have some sort of type here.\nSpeaker E: And then you could have the utterance that it's referring to could be U 17 or something like that.\nSpeaker D: Okay, so I mean, that seems great for all of the encoding of things with time and...\nSpeaker D: I guess my question is more, what do you do with, say, a forced alignment?\nSpeaker D: We've got all these phone labels.\nSpeaker D: And what do you do if you just conceptually, if you get transcriptions where the words are staying, but the time boundaries are changing because you've got a new recognition output.\nSpeaker D: Sort of what's the sequence of going from the waveforms that stay the same, the transcripts that may or may not change.\nSpeaker D: And then the utterance, which are the time boundaries that may or may not change.\nSpeaker B: That's actually very nicely handled here because you could, you could, all you'd have to change is the time stamps in the timeline without...\nSpeaker B:...to propagate all of them.\nSpeaker E: That's why you do that extra level of indirection, so that you can just change the timeline.\nSpeaker B: Except the timeline is going to be huge.\nSpeaker B: Yes.\nSpeaker D: Yeah, especially if you have a phone level.\nSpeaker D: We have a phone level back.\nSpeaker E: Yeah, I don't think I would do this for phone level.\nSpeaker E: I think for phone level, you want to use some sort of binary representation because it'll be too dense otherwise.\nSpeaker D: Okay, so if you were doing that and you had this sort of companion thing that gets called up for phone level, what would that look like?\nSpeaker E: I would use just an existing way of doing it.\nSpeaker B: It's just a matter of it being bigger.\nSpeaker B: But if you have, you know, barring memory limitations, I mean this is still...\nSpeaker E: It's parsing limitations.\nSpeaker E: I don't want to have this text file that you have to read in the whole thing to do something very simple for.\nSpeaker B: Oh no, you would use it only for purposes where you actually want the phone level information.\nSpeaker D: So you could have some file that configures how much information you want in your XML or something.\nSpeaker E: I mean, I imagine you'd have multiple versions of this depending on the information that you want.\nSpeaker E: What I'm wondering is whether I think for word level, this would be okay.\nSpeaker E: For word level, it's all right.\nSpeaker E: For lower than word level, you're talking about so much data that I don't know.\nSpeaker E: I don't know if that...\nSpeaker D: I mean, we actually have so one thing that Donna is doing...\nSpeaker D: Gladeth is a big key.\nSpeaker D: We're running for every frame you get a pitch value, not only one pitch value, but different kinds of pitch values.\nSpeaker E: Yeah, I mean, for something like that, I would use p-file or any frame level stuff I would use p-file.\nSpeaker E: Meaning?\nSpeaker E: Well, or something like it.\nSpeaker E: It's...it's...it's...it's...it's...it's...it's...it's a format for frame level representation of features.\nSpeaker D: Okay. That you could call...that you would tie into this representation with like an ID.\nSpeaker E: Or there's a particular way in XML to refer to external resources.\nNone: Okay.\nSpeaker E: So you would say refer to this external file.\nSpeaker E: So that external file wouldn't be in...\nSpeaker F: But what's the advantage of doing that versus just putting it into this format?\nSpeaker E: More compact, which I think is better.\nSpeaker E: I mean, if you did it at this...\nSpeaker D: You don't want to do it with that.\nSpeaker E: Anything at frame level, you had better encode binary or it's going to be really painful.\nSpeaker B: Or you just...I mean, I like text formats. You can always...\nSpeaker B: Jesus, then.\nSpeaker B: And you know, decompress them on the fly if space is really...\nSpeaker F: I was thinking the advantage is that we can share that one frame.\nSpeaker E: You're talking about gigabyte-sized files.\nSpeaker E: You're going to actually run out of space in your file system for one file.\nSpeaker E: Right? Because you have two gigabyte limit on most OSs.\nSpeaker B: I would say, okay, so frame level is probably not a good idea.\nSpeaker B: But for phone level stuff, it's perfectly...like phones are syllables...\nSpeaker B: Well, they're every five frames though.\nSpeaker D: So something like that.\nSpeaker B: But most of the frames are actually not speech.\nSpeaker B: So, you know, people don't...\nSpeaker B: Yeah, but we actually...\nSpeaker B: Look at the average number of phones in an English word is...\nSpeaker B: I don't know, five maybe.\nSpeaker B: So look at it.\nSpeaker B: Number of words times five.\nSpeaker D: So you mean, pause phones take up a lot of the long pause phones?\nSpeaker D: Yeah, exactly.\nSpeaker D: Okay, that's true, but you do have to keep them in there.\nSpeaker E: So I think it's debatable whether you want to do phone level in the same thing.\nSpeaker E: Okay.\nSpeaker E: But I think anything at frame level, even P file is true for both.\nSpeaker E: So I would use something tighter than P files.\nSpeaker E: Are you familiar with it?\nSpeaker D: I haven't seen this particular format.\nSpeaker B: I've used them.\nSpeaker B: I don't know what their structure is.\nSpeaker F: But I mean, P file for each frame is storing a vector of capstroller, PLP values, right?\nSpeaker F: It's whatever you want, right?\nSpeaker E: Actually, so that...what's nice about the P file, it...built into it is the concept of frames, utterances, sentences, that sort of thing, that structure.\nSpeaker E: And then also attached to it is an arbitrary vector of values.\nSpeaker E: Oh.\nSpeaker E: And it can take different types.\nSpeaker E: So they don't all have to be floats.\nSpeaker E: You know, you can have integers and you can have doubles and all that sort of stuff.\nSpeaker E: That sounds about what I...\nSpeaker E: Right, and it has a header, it has a header format that describes it to some extent.\nSpeaker E: So the only problem with it is it's actually storing the utterance numbers and the frame numbers in the file, even though they're always sequential.\nSpeaker E: And so it does waste a lot of space.\nSpeaker E: But it's still a lot tighter than ASCII.\nSpeaker E: And we have a lot of tools already to deal with it.\nSpeaker D: Do you...okay.\nSpeaker D: Is there some documentation on the...\nSpeaker D: Yeah, there's a ton of it.\nSpeaker E: Man pages and source code and...\nSpeaker E: I mean, that sounds good.\nSpeaker D: I was just looking for something...\nSpeaker D: Not a database person, but something sort of standard enough that...\nSpeaker D: You know, if we start using this, we can give it out.\nSpeaker D: Other people can work on that.\nSpeaker E: Yeah, it's not standard.\nSpeaker E: I mean, it's something that we developed at ICSI.\nSpeaker E: But...\nSpeaker D: But it's been used here.\nSpeaker E: But it's been used here and...\nSpeaker E: And, you know, we have a well-configured system that you can distribute for free.\nSpeaker F: And...\nSpeaker F: I mean, it must be the equivalent of whatever you guys use to store your computed features in, right?\nSpeaker B: Yeah, we have... actually, we use a generalization of the sphere format.\nSpeaker B: But...\nSpeaker B: Yeah, so there's something like that.\nSpeaker B: But it's...\nSpeaker E: I think the word is HTK-DU for features.\nSpeaker E: Or does it even have a concept of features?\nSpeaker B: I mean, the topic has their own feature format.\nSpeaker B: It's called like SSD or something like that.\nSpeaker E: Yeah, I'm just wondering, would it be worthwhile to use that instead?\nSpeaker D: Yeah, this is exactly a good decision.\nSpeaker F: And people don't typically share this kind of stuff, right?\nSpeaker F: I mean, they generate their own.\nSpeaker D: Yeah.\nSpeaker D: Actually, I just...\nSpeaker D: You know, we've done this stuff on prasadix.\nSpeaker D: And three or four places I've asked for those prasadic files.\nSpeaker D: And we just have an ASCII.\nSpeaker D: All right.\nSpeaker D: But a frame by frame, which is fine.\nSpeaker D: But it gets unwieldy to go in and query these files with really huge files.\nSpeaker D: Right.\nSpeaker D: And we could do it.\nSpeaker D: I was just thinking if there's something...\nSpeaker D: And again, if you have a...\nSpeaker D: We're out of frame values, right?\nSpeaker E: If you have a two-hour long meeting, that's gonna...\nSpeaker E: They're quite large.\nSpeaker D: Yeah, I mean, they're very much enormous.\nSpeaker D: These are 10-minute switchboard conversations.\nSpeaker D: Right.\nSpeaker D: So it's doable.\nSpeaker D: It's just that you can only store a feature vector at frame by frame.\nSpeaker D: And it doesn't have any kind of...\nSpeaker F: Is the sharing part of this a pretty important consideration?\nSpeaker F: Or is that just sort of a nice thing to have?\nSpeaker D: I don't know enough about what we're gonna do with the data.\nSpeaker D: But I thought it would be good to get something that other people can use or adopt for their own kinds of encoding.\nSpeaker D: And we have to make some decision about what to do.\nSpeaker D: And especially for the prosody work, what it ends up being is you get features from the signal.\nSpeaker D: And of course, those change every time your alignments change.\nSpeaker D: So you rerun a recognize you want to recompute your features and then keep the database up to date.\nSpeaker D: You change a word or you change an utterance boundary segment, which is gonna happen a lot.\nSpeaker D: And so I wanted something where all of this can be done in an elegant way.\nSpeaker D: And that if somebody wants to try something or compute something else, that it can be done flexibly, doesn't have to be pretty.\nSpeaker D: It just has to be easy to use.\nSpeaker E: Yeah, the other thing we should look at Atlas, the NIST thing, and see if they have anything at that level.\nSpeaker E: I mean, I'm not sure what to do about this with Atlas because they chose a different route.\nSpeaker E: I chose something that there are sort of two choices.\nSpeaker E: Your file format can know about, know that you're talking about language and speech, which is what I chose in time, or your file format can just be a graph representation.\nSpeaker E: And then the application has to impose the structure on top.\nSpeaker E: So what it looked like Atlas chose is they chose the other way, which was their file format is just nodes and links.\nSpeaker E: And you have to interpret what they mean yourself.\nSpeaker D: And why did you not choose that?\nSpeaker E: Because I knew that we were doing speech and I thought it was better if you're looking at a raw file to be for the tags to say it's an utterance, as opposed to the tag to say it's a link.\nSpeaker D: But other than that, are they compatible?\nSpeaker D: I mean, you could sort of...\nSpeaker E: Yeah, they're reasonably compatible.\nSpeaker E: You could probably translate between them.\nSpeaker E: Yeah, that's...\nSpeaker E: So...\nSpeaker E: So, well, the other thing is, if we choose to use Atlas, which maybe we should just do, we should just throw this out before we invest a lot of time in it.\nSpeaker D: So this is what the meeting is about, just sort of, how to...\nSpeaker D: Because we need to come up with the database like this just to do our work.\nSpeaker D: And I actually don't care as long as it's something useful to other people what we choose.\nSpeaker D: So maybe it's...\nSpeaker D: You know, you have any idea of how to choose because I don't...\nSpeaker D: Yeah.\nSpeaker E: I mean, I chose this for a couple of reasons. One of them is that it's easy to parse.\nSpeaker E: You don't need a full XML parser.\nSpeaker E: It's very easy to just write a purl script to parse it.\nSpeaker B: As long as each tag is on one line...\nSpeaker E: Exactly, exactly, which I always do.\nSpeaker D: And you can have as much information in the tag as you want, right?\nSpeaker E: Well, I have it structured, right?\nSpeaker E: So each tag has only particular items that it can take.\nSpeaker D: But you can add to those structures if you...\nSpeaker E: If you have more information. So what this would say is that instead of doing this, you would say something like a link...\nSpeaker E: Start equals some node ID...\nSpeaker E: And equals some other node ID...\nSpeaker E: And then type would be utterance.\nSpeaker E: So it's very similar.\nSpeaker D: So why would it be a waste to do it this way if it's similar enough that we can always translate...\nSpeaker F: It probably wouldn't be a waste. It would mean that at some point if we wanted to switch, we'd just have to...\nSpeaker F: Write a translator, if that's...\nSpeaker E: Since they're developing a big...\nSpeaker E: I don't think that's a big deal.\nSpeaker E: They're developing a big infrastructure.\nSpeaker E: And so it seems to me that if we want to use that, we might as well go directly to what they're doing rather than...\nSpeaker B: Yeah, so that's the question.\nSpeaker F: I mean, how stable is there...\nSpeaker F: Are they ready to go?\nSpeaker E: I looked at it...\nSpeaker E: I looked at it a couple of years ago, probably a year ago, when we first started talking about this.\nSpeaker E: And at that time, at least, it was still not very complete.\nSpeaker E: And so specifically, they didn't have any external format representation at that time.\nSpeaker E: They just had sort of conceptual node annotated transcription graph, which I really liked.\nSpeaker E: And that's exactly what this stuff is based on.\nSpeaker E: Since then, they've developed their own external file format, which is this sort of thing.\nSpeaker E: And apparently, they've also developed a lot of tools, but I haven't looked at them.\nSpeaker D: Maybe I should...\nSpeaker D: Would the tools run on something like this?\nSpeaker D: And translate them anyway?\nSpeaker E: I mean, I guess that's the question.\nSpeaker E: What would worry me is that maybe we might miss a little detail.\nSpeaker E: That would make it very difficult to translate from one to the other.\nSpeaker B: They're conceptually close.\nSpeaker B: And they already have or will have tools that everybody else will be using.\nSpeaker B: I mean, we might as well.\nSpeaker E: So I'll take a closer look at it.\nSpeaker D: So that would really be the question is just what you feel is in the long run the best thing.\nSpeaker D: Because once we start sort of doing this, we don't actually have enough time to probably have to rehash it out again.\nSpeaker E: The other way that I sort of established this was as easy translation to and from the transcriber format.\nSpeaker E: Right.\nSpeaker D: But...\nSpeaker D: I mean, I like this.\nSpeaker D: This is sort of intuitively easy to actually read as easy as it could be.\nSpeaker D: But I suppose that as long as they have a type here that specifies...\nSpeaker E: It's almost the same.\nSpeaker E: The point is with this though is that you can't really add any supplementary information.\nSpeaker E: So if you suddenly decide that you want...\nSpeaker E: You have to make a different type.\nSpeaker E: Yeah, you'd have to make a different type.\nSpeaker D: Well, if you look at it, I guess in my mind, I don't know enough Jane would know better about the types of annotations.\nSpeaker D: But I imagine that those are things that...\nSpeaker D: Well, you guys mentioned this that could span any...\nSpeaker D: It could be in its own channel.\nSpeaker D: It could span time boundaries of any type.\nSpeaker D: It could be instantaneous.\nSpeaker D: Things like that.\nSpeaker D: And then from the recognition side, we have back traces at the phone level.\nSpeaker D: If you could handle that, it could handle states or whatever.\nSpeaker D: And then at the prosely level, we have frame sort of like actual feature files, like these P files or anything like that.\nSpeaker D: And that's sort of the world of things that I...\nSpeaker D: And then we have the align channels.\nSpeaker D: Very coarse.\nSpeaker B: It seems to me you want to keep the frame level stuff separate.\nSpeaker B: Yeah, I definitely agree.\nSpeaker D: And I wanted to find actually a nicer format or a maybe a more compact format that we used before.\nSpeaker D: Just because you've got 10 channels or whatever and two hours of a meeting.\nSpeaker B: How would you represent multiple speakers in this framework?\nSpeaker B: Were you just represent them as you would have like a speaker tag or something?\nSpeaker E: There's a speaker tag up at the top, which identifies them.\nSpeaker E: And then the way I had it is each turn or each other, and so I don't even remember now, had a speaker ID tag attached to it.\nSpeaker E: And in this format, you would have a different tag, which would be linked to the link.\nSpeaker E: So somewhere else you would have another thing that would be...\nSpeaker E: Let's see, would it be a node or a link?\nSpeaker E: And so this one would have an ID, is link 74 or something like that.\nSpeaker E: And then somewhere up here you would have a link that was referencing L74 and had speaker item.\nSpeaker D: Actually, the channel I think that...\nSpeaker D: Well, channel is what the channel lies out.\nSpeaker E: This isn't quite right. I have to look at it again.\nSpeaker B: But so how in the next format do we express a hierarchical relationship between, say, and utterance and the words within it?\nSpeaker B: So how do you tell that these are the words that belong to that utterance?\nSpeaker E: You would have another structure lower down than this, that would be saying they're all belonging to this ID.\nSpeaker F: So each thing refers to the utterance that it belongs to?\nSpeaker F: Right.\nSpeaker E: So each utterance could refer to a term, and each term could refer to something higher up.\nSpeaker D: And what if you actually have... Right now what you have as utterance, the closest thing that comes out of the channel lies is the stuff between the segment boundaries that the transcribers put in or that T-Lo put in.\nSpeaker D: Which may or may not actually be like... It's usually not the beginning and end of a sentence.\nSpeaker D: Well, that's why I didn't call it sentence.\nSpeaker D: Right. So it's like a segment or something.\nSpeaker D: So I mean, as soon as this is possible that if you have someone annotates the punctuation or whatever when they transcribe, you can say, you know, from the beginning of the sentence to the end of the sentence from the annotations, this is a unit.\nSpeaker D: Even though it never actually, it's only a unit by virtue of the annotations at the word level.\nSpeaker D: Sure. I mean, so you would have yet another tag.\nSpeaker D: That would be a tag somehow.\nSpeaker E: You'd have another tag which says, this is of type sentence.\nSpeaker D: But it's just not overtly in the... Because this is exactly the kind of...\nSpeaker D: I think that should be possible as long as the... But what I don't understand is where in this type of file that would be expressed.\nSpeaker E: Right. You would have another tag somewhere.\nSpeaker E: Well, there are two ways of doing it.\nSpeaker E: Two ways of doing it.\nSpeaker E: And ID is S whatever.\nSpeaker E: And then lower down, you could have an utterance.\nSpeaker E: So the type is utterance equals UT.\nSpeaker E: And you could either say that... No, I don't know.\nSpeaker F: I take this as a thing.\nSpeaker E: Can you say that this is... You would just have a part of this or do you say this is part of this?\nSpeaker E: I think... But you would refer up to the sentence.\nSpeaker D: They're actually overlapping each other.\nSpeaker B: The thing is that something may be a part of one thing for one purpose and another thing of another purpose.\nSpeaker B: You have to have another type then, I guess.\nSpeaker E: Well, I think I had better look at it again because I...\nSpeaker B: So, for instance... There's one more level of indirection that I'm forgetting.\nSpeaker B: So, once you have a word sequence and you have two different segmentations of that same word sequence, I say one segmentation is in terms of, you know, sentences.\nSpeaker B: And another segmentation is in terms of, I don't know, prosodic phrases.\nSpeaker B: And let's say that they don't nest.\nSpeaker B: So, you know, a prosodic phrase may cross two sentences or something.\nSpeaker B: I don't know if that's true or not.\nSpeaker D: Well, it's definitely true with the segment. That's what I exactly meant by the utterances versus the sentence.\nSpeaker B: So, you want to say this word is part of that sentence and this prosodic phrase.\nSpeaker B: But the phrase is not part of the sentence and neither is the sentence part of the phrase.\nSpeaker E: I'm pretty sure that you can do that, but I'm forgetting the exact level of nesting.\nSpeaker B: Two different pointers from the word up, one level up.\nSpeaker E: So, what you would end up having is a tag saying, here's a word and it starts here and it ends here.\nSpeaker E: And then lower down you would say, here's a prosodic boundary and it has these words in it.\nSpeaker E: And lower down you'd have, here's sentence and it has these words in it.\nSpeaker D: So, you would be able to go in and say, you know, give me all the words in the bound, in the prosodic phrase, and give me all the words in it.\nSpeaker E: So, I think that's the word. Let me look at it again.\nSpeaker B: Okay.\nSpeaker B: The other issue that you had was, how do you actually efficiently extract, find and extract information in a structural of this type?\nSpeaker B: So, you gave some examples, like.\nSpeaker D: Well, and I mean, you guys might, I don't know if this is premature because I suppose once you get the representation, you can do this, but the kinds of things I was worried about is...\nSpeaker D: No, that's not clear.\nSpeaker B: Well, okay, so you can do it, but can you do it?\nSpeaker B: I mean, I can't do it, but you know.\nSpeaker B: Well, you're going to do this, you're going to want to do this very quickly, or else you'll spend all your time sort of searching through various...\nSpeaker D: Right, you need to sort of a paradigm for how to do it, but an example would be, find all the cases in which Adam started to talk while Andreas was talking and his pitch was rising.\nSpeaker D: Andreas is a pitch.\nSpeaker E: Right, I mean, that's going to be, is the rising pitch a feature, or is it going to be in the same file?\nSpeaker D: Well, the rising pitch will never be hand annotated, so all the prasodic features are going to be automatically...\nSpeaker E: I mean, that's going to be hard, regardless, right, because you're going to have to write a program that goes through your feature file and looks for rising pitch.\nSpeaker D: So normally what we would do is we would say, what are we going to assign rising pitch to, are we going to assign it to words, are we going to just assign it to sort of... when it's rising, we have a begin and rise representation, but suppose we dump out this file, we say...\nSpeaker D: For every word, we just classify it as rise or fall or neither.\nSpeaker E: Okay, well, in that case, you would add that to this format.\nSpeaker D: We would basically be sort of taking the format and enriching it with things that we want to query in relation to the words that are already in the file, and then querying it.\nSpeaker B: You want sort of a graph that works at the structural...\nSpeaker E: You have that. There's a standard again in XML, specifically for searching XML documents, structured XML documents, where you can specify both the content and the structural position.\nSpeaker B: Yeah, but it's not clear that that's relative to the structure of the XML document.\nSpeaker E: You use it as a tool. You use it as a tool, not an end user. It's not an end user thing. You would use that to build your tool to do that sort of search.\nSpeaker B: Because here you're specifying a lattice, so the underlying data structure and the whole of the other search can have lattice.\nSpeaker E: That's different from searching. As long as the feature... Well, no, the whole point is that the text and the lattice are isomorphic. They represent each other completely.\nSpeaker E: So that...\nSpeaker D: That's true if the features from your acoustics or whatever that are not explicitly in this are at the level of these types.\nSpeaker E: If you can... Yeah, but that's going to be the trouble no matter what format you choose. You're going to have the difficulty of relating the frame level features.\nSpeaker D: I'm going to figure out what's the best format for this representation. It's still going to be not direct.\nSpeaker D: Another example was where in the language, where in the word sequence are people interrupting? I guess that one's actually easier.\nSpeaker F: What about the idea of using a relational database to store the information from the XML?\nSpeaker F: So you would have XML basically would... You could use the XML to put the data in. And then when you get data out, you put it back in XML.\nSpeaker F: So you use XML as sort of the transfer format. But then you store the data in the database, which allows you to do all kinds of good search things.\nSpeaker E: One of the things that Atlas is doing is they're trying to define an API, which is independent of the backstort.\nSpeaker E: So that you could define a single API and the storage could be flat XML files or a database.\nSpeaker E: My opinion on that is for the sort of stuff that we're doing, I suspect it's overkill to do a full relational database that...\nSpeaker E: Just a flat file and search tools I bet will be enough.\nSpeaker E: But that's the advantage of Atlas is that if we actually decide to go that route completely and we program to their API, then if we wanted to add a database later, it would be pretty easy.\nSpeaker D: It seems like the kind of thing you do if people start adding all kinds of files and whistles to the data.\nSpeaker D: And so that might be good for us to use a format where we know we can easily input that to some database of other people.\nSpeaker E: I guess I'm just a little hesitant to try to go whole hog on sort of the whole framework that that NIST is talking about with Atlas and the database and all that sort of stuff.\nSpeaker E: Because it's a big learning curve just to get going. Whereas if we just do a flat file format, sure it may not be as efficient, but everyone can program in Perl and use it.\nSpeaker E: Right, so as opposed to...\nSpeaker B: I'm still not convinced that you can do much at all on the text on the flat file that the text representation.\nSpeaker B: Because the text representation is going to be not reflecting the structure of your words and annotations.\nSpeaker E: So if it's not representing it, then how do you recover it? Of course it's representing it. That's the whole point.\nSpeaker B: You have to basically, yeah, you can use Perl to read it in and construct a internal representation that is essentially a lattice.\nSpeaker E: Okay, well that was a different point, right? So what I was saying is that...\nSpeaker B: But that's what you'll have to do.\nSpeaker E: Perl, if you want to just do Perl. If you wanted to use the structured XML query language, that's a different thing.\nSpeaker E: And it's a set of tools that let you specify given the DDDD of the document what sorts of structural searches you want to do.\nSpeaker E: So you want to say that you're looking for a tag within a tag within a particular tag that has this particular text in it and refers to a particular value.\nSpeaker E: And so the point isn't that an end user who is looking for a query like you specified wouldn't program it in this language.\nSpeaker E: What you would do is someone would build a tool that used that as a library so that you wouldn't have to construct the internal representations you're seeing.\nSpeaker D: So I think the kinds of questions, at least in the next to the end of this year, are there may be a lot of different ones, but they'll have a similar nature.\nSpeaker D: They'll be looking at either a word level, prosadik, a value, like a continuous value, like this slope of something.\nSpeaker D: But we'll do something where we, some kind of data reduction where the prosadik features are either at the word level or at the segment level or something like that.\nSpeaker D: They're not going to be at the phone level and they're not going to be at the frame level when we get done with sort of giving them simpler shapes and things.\nSpeaker D: And so the main thing is just being able, well, I guess the two goals, one that Chuck mentioned is starting out with something that we don't have to start over, that we don't have to throw away if other people want to extend it for other kinds of questions.\nSpeaker D: And being able to at least get enough information out on where we condition the location of features on information that's in the kind of file that you put up there.\nSpeaker E: And that would do it. I mean, I think that there are quick and dirty solutions and then there are long term big infrastructure solutions.\nSpeaker E: And so we want to try to pick something that lets us do a little bit of both.\nSpeaker D: And especially that the representation doesn't have to be thrown away, even if your tools change.\nSpeaker E: And so it seems to me that I have to look at it again to see whether it can really do what we want.\nSpeaker E: But if we use the Atlas external file representation, it seems like it's rich enough that you could do quick tools just as I said in Pearl.\nSpeaker E: And then later on, if we choose to go up the learning curve, we can use the whole Atlas infrastructure.\nSpeaker E: And that sounds good to me.\nSpeaker D: Which has all that built in.\nSpeaker D: So if you look at that and let us know what you think, I mean, I think we're sort of getting things because I want to get the property work done, but I don't want to waste time getting the, yeah?\nSpeaker E: Well, I wouldn't wait for the formats because anything you pick will be able to translate to another form.\nSpeaker B: Well, maybe you should actually look at it yourself to get a sense of what is your dealing with because Adam might have one opinion.\nSpeaker B: I definitely do.\nSpeaker B: Yeah, definitely.\nSpeaker B: I think the more eyes look at this, the better.\nSpeaker D: Especially if there's, you know, if someone can help with at least the setup of the right.\nSpeaker D: Okay.\nSpeaker A: The right representation.\nSpeaker D: Then, you know, I hope it won't, we don't actually need the whole full blown.\nSpeaker D: Thank you.\nSpeaker D: Thank you.\nSpeaker D: So maybe if you guys can look at it and see what I think we're, I think we're about to end.\nSpeaker D: Yeah, wrapping up.\nSpeaker D: But yeah, sorry, it's a short meeting, but why don't I, is there anything else like that helps me?\nSpeaker E: Well, I think the other thing we might want to look at is alternatives to P file.\nSpeaker E: I mean, the reason I like P files, I'm already familiar with it, we have expertise here.\nSpeaker E: And so if we pick something else, there's learning curve problem.\nSpeaker E: But I mean, it is just something we developed at XE.\nSpeaker B: And so, is there an API?\nSpeaker E: Yeah, there's an API for it.\nSpeaker B: And a bunch of libraries, P file utilities.\nSpeaker B: And so basically the file system,\nSpeaker E: Well, that's going to be a problem, no matter what. You have the two gigabyte limit on the file system size.\nSpeaker E: And we definitely hit that with broadcast news.\nSpeaker B: Maybe you could extend the API to a support, like, splitting up, you know, conceptually one file into smaller files on disk so that you can essentially.\nSpeaker E: Yeah, you know, most of the tools can handle that so that we didn't do it at the API level.\nSpeaker E: We did it at the tool level that most many of them can, you can specify several P files.\nSpeaker E: And they'll just be done sequentially.\nSpeaker D: So I guess, yeah, if you and Don can, if you can show them the P file stuff.\nSpeaker D: Sure.\nSpeaker E: So this may be like that.\nSpeaker E: So it's a file or apropos P file you see.\nSpeaker A: I've looked at it at least briefly.\nSpeaker A: I think what we were doing.\nSpeaker E: I have no idea.\nSpeaker E: I didn't do it.\nSpeaker E: I didn't develop it.\nSpeaker E: You know, I think it was Dave Johnson.\nSpeaker B: So it's all part of the quick net library has all the utilities for.\nSpeaker B: No, P files were around.\nSpeaker B: Maybe for quick net.\nSpeaker B: Oh, we're there.\nSpeaker B: Rapp, right?\nSpeaker F: You work with P files.\nSpeaker F: I don't remember what the P is.\nSpeaker E: But there are not there.\nSpeaker E: The quick net library has a bunch of things in it to handle P files so it works pretty well.\nSpeaker D: And that isn't really, I guess, as important as the main, I don't know what you call it.\nSpeaker D: The main third or third?\nSpeaker F: It only stands for fill.\nSpeaker F: Fill cone.\nSpeaker F: Is it fill file?\nSpeaker D: That's my guess.\nSpeaker D: Well, that's really useful.\nSpeaker D: I mean, this is exactly the kind of thing that wanted to set up.\nSpeaker E: You have been meaning to look at the whole stuff again anyway.\nSpeaker D: I guess it's also sort of a political decision.\nSpeaker D: I mean, if you feel like that's a community that would be good to tie into anyway, then it's something we're doing.\nSpeaker E: And as I said, what I did with this stuff, I based it on theirs.\nSpeaker E: It's just they hadn't actually come up with an external format yet.\nSpeaker E: So now that they have come up with a format, it seems pretty reasonable to use it.\nSpeaker E: But let me look at it again.\nSpeaker D: Okay, great.\nSpeaker D: As I said.\nSpeaker E: Because there's one level, there's one more level of indirection and I'm just blanking on exactly how it works.\nSpeaker E: I got to look at it again.\nSpeaker D: And we can start with, I guess, this input from Dave's, which you had printed out the channelized input.\nSpeaker D: Because he has all of the channels with the channels and the tag and stuff like that.\nSpeaker D: So that would be directly easy and wrapped.\nSpeaker D: And so then it would just be a matter of making sure to handle the annotations that are not at the word level.\nSpeaker D: Where are those annotations coming from?\nSpeaker D: Well, right, Jane would.\nSpeaker C: Are you talking about the overlap on annotations?\nSpeaker D: Yeah, any kind of annotation that isn't already there.\nSpeaker D: Anything in vision.\nSpeaker C: So what I was imagining was, so Dave says, we can have unlimited numbers of green ribbons.\nSpeaker C: And so put a green ribbon on for an overlap code.\nSpeaker C: And since we, I think it's important to remain flexible regarding the time bins for now.\nSpeaker C: And so it's nice to have, however, you know, you want to have it time located in the discourse.\nSpeaker C: So if we tie the overlap code to the first word in the overlap, then you'll have a time marking.\nSpeaker C: It'll be independent of the time bins, however, these evolve shranker, whatever increase, or also you could have different time bins for different purposes.\nSpeaker C: And having it tied to the first word in an overlap segment is unique, you know, anchored, clear.\nSpeaker C: And it would just end up on a separate ribbon.\nSpeaker C: So the overlap code is going to be easy with respect to that.\nSpeaker C: It'll be puzzled.\nSpeaker F: I just, I don't quite understand what these things are.\nSpeaker F: What codes themselves?\nSpeaker F: Overlap codes, I'm not sure what they're.\nSpeaker F: Well, I mean, it's that.\nSpeaker F: Well, it probably doesn't matter.\nSpeaker C: It doesn't mean it doesn't.\nSpeaker C: It's not for the topic of this meeting.\nSpeaker C: No, the idea is just to have a separate green ribbon, you know, and let's say this is the time bin.\nSpeaker C: There's a word here.\nSpeaker C: It's the first word of an overlap segment of any length overlapping with any other word, is segment of any length.\nSpeaker C: And then you can indicate that this here was perhaps a back channel, or you can say that it was usurping of the journey, or you could, you know, any number of categories.\nSpeaker C: But the fact is you have a time tag in a way this independent of the particular time bin that the word ends up in, if it's a large unit or a small unit, or we change the boundaries of the units.\nSpeaker C: Right?\nSpeaker C: It's still unique and fits with the format flexible on all that.\nSpeaker B: It would be nice.\nSpeaker B: This is sort of regarding, it's related but not correctly to the main topic of discussion.\nSpeaker B: But when it comes to annotations, you often find yourself in the situation where you have different annotations of the same word sequence.\nSpeaker B: Okay?\nSpeaker B: And sometimes the word sequence is even differ slightly because they were edited at one place, but not the other.\nSpeaker B: So once this data gets out there, some people might start annotating this for, I don't know, the ILR acts or, you know, topics or what the heck, you know, there's a zillion things that people might annotate this for.\nSpeaker B: And the only thing that is really sort of common among all the various versions of this data is the word sequence, or the proximity.\nSpeaker B: For the time.\nSpeaker B: Or the times, but see if you annotate dialogue acts, you don't necessarily want to, or topics, you don't really want to be dealing with time marks.\nSpeaker B: It's much more efficient for them to just see the word sequence, right?\nSpeaker B: I mean, most people aren't as sophisticated as we are here with, you know, time alignments and stuff.\nSpeaker B: So the point is...\nSpeaker E: Did you mention some names on the...\nSpeaker B: So my point is that you're going to end up with word sequences that are differently annotated.\nSpeaker B: And you want some tool that is able to sort of merge these different annotations back into a single version.\nSpeaker B: Okay.\nSpeaker B: And we had this problem very massively at SRI when we worked a while back on dialogue acts as well as, you know, what was it?\nSpeaker B: Well, all those things were...\nSpeaker B: At times, automatic punctuation and stuff like that.\nSpeaker B: Because we had one set of annotations that were based on one version of the transcripts with a particular segmentation.\nSpeaker B: And then we had another version that was based on a different slightly edited version of the transcripts with a different segmentation.\nSpeaker B: So we had these two different versions which were, you know, you could tell they were from the same source, but they weren't identical.\nSpeaker B: So it was extremely hard to reliably merge these two back together to correlate the information from the different annotations.\nSpeaker E: Yeah. I don't see any way that file formats are going to help us with that.\nSpeaker E: No.\nSpeaker E: It's all a question of semantic.\nSpeaker B: But once you have a file format, I can imagine writing, not personally, but someone writing a tool that is essentially an alignment tool that mediates between various versions.\nSpeaker B: And sort of like, you know, you have this thing in Unix where you have...\nSpeaker B: Diff.\nSpeaker B: W-diff, or...\nSpeaker B: There's the diff that actually tries to reconcile different...\nSpeaker B: And two diffs based on the same original. Something like that. But operating on these lattices that are really what's behind this...\nSpeaker B: Yeah.\nSpeaker E: There's actually a diff library you can use to do things like that.\nSpeaker B: So somewhere in the A4, you would like to have like a merge or some function that merges two...\nSpeaker E: Yeah, I think it's going to be very hard. Any sort of structured, anything when you try to merge is really, really hard because the hard part isn't the file format. The hard part is specifying what you mean by merge.\nSpeaker D: But the one that would work here actually for is that it's more reliable than the utterances is the speaker on and off.\nSpeaker D: So if you have a good...\nSpeaker E: But this is exactly what I mean. Is that the problem...\nSpeaker E: What to tie it to? Yeah, exactly. The problem is saying, what are the semantics? What do you mean by merge?\nSpeaker B: So just to let you know where we clued it by doing both were based on words.\nSpeaker B: So we had two versions of the same words, sprinkles with different tags.\nSpeaker E: And then you did diff.\nSpeaker E: And we did diff.\nSpeaker E: Yeah, that's just what I thought.\nSpeaker E: And that's just how I would have done it.\nSpeaker B: But you know, it had lots of errors and things would end up in the wrong order and so forth.\nSpeaker B: So if you had a more...\nSpeaker B: It was a clued because it was basically reducing everything to...\nSpeaker B: Textual.\nSpeaker B: To textual alignment.\nSpeaker D: But isn't that something where...\nSpeaker D: Whoever...\nSpeaker D: If the people who are making changes say in the transcripts, because this all happened when the transcripts were different.\nSpeaker D: If they tie it to something, like if they tied it to the acoustic segment.\nSpeaker D: If they...\nSpeaker D: You know what I mean?\nSpeaker D: If they tied it to an acoustic segment and we had the time march that would help.\nSpeaker D: But the problem is that exactly as Adam said that you get...\nSpeaker D: You know, you don't have that information or it's lost in the merge somehow.\nSpeaker D: Okay, one question.\nSpeaker C: It seems to me that we will have the official version of the corpus, which will be only one version in terms of the words.\nSpeaker C: We still have an emerging issue, maybe, of coding for done independently of the...\nSpeaker B: And you're going to get that because if the data gets out, people will do all kinds of things to it.\nSpeaker B: And you know, several years from now, you might want to look into the prosody of referring expressions.\nSpeaker B: And someone at the University of Who knows where has annotated the referring expressions.\nSpeaker B: So you want to get that annotation and bring it back in line with your data.\nSpeaker E: But unfortunately, they've also handed it to you.\nSpeaker D: But they've also handed it to you.\nSpeaker D: And so that's exactly what we should somehow when you distribute the data, say that you know, that have some way of knowing how to merge it back in and asking people to write it.\nSpeaker F: What's wrong with doing times?\nSpeaker D: Yeah, time is the...\nSpeaker D: Well...\nSpeaker B: Time is passage.\nSpeaker B: Time...\nSpeaker B: Times are...\nSpeaker B: What if they haven't notated?\nSpeaker E: He's a language-miling person.\nSpeaker E: So imagine, I think his example is a good one.\nSpeaker E: Imagine that this person who developed the corpus of the referring expressions didn't include time.\nSpeaker E: He included references to words.\nSpeaker E: He said that at this word is when it happened.\nSpeaker D: But still they...\nSpeaker D: They figure out the time.\nSpeaker D: Exactly.\nSpeaker E: Sure, but what if they change the words?\nNone: Well, some angry point.\nSpeaker F: You couldn't change the word.\nSpeaker E: Sure, but they can't change the time of the word.\nSpeaker E: The point is that they may have annotated it off a word transcript.\nSpeaker E: That isn't the same as our word transcript.\nSpeaker E: So how do you merge it back?\nSpeaker E: And I understand what you're saying.\nSpeaker E: And I guess the answer is...\nSpeaker E: It's going to be different every time.\nSpeaker E: It's just going to be...\nSpeaker D: You only know the boundary.\nSpeaker E: Exactly what I said before, which is that what do you mean by merge?\nSpeaker E: So in this case where you have the words and you don't have the times, well, what do you mean by merge?\nSpeaker E: If you tell me what you mean, I can write a program to do it.\nSpeaker D: Right, you can merge at the level of the representation that the other person preserved.\nSpeaker D: Right, and that's not all you can do.\nSpeaker D: And that's not all you can do.\nSpeaker D: It's relative ordering and sometimes even that is wrong.\nSpeaker E: So in this one you would have to do a best match between the word sequences.\nSpeaker E: Extract the times from the best match of theirs to yours.\nSpeaker E: And use that.\nSpeaker D: And then it's for that their time marks are somewhere in between, yeah, exactly.\nSpeaker C: But it could be that they just...\nSpeaker C: It could be that they chumped.\nSpeaker C: They lost certain utterances.\nSpeaker C: Right, exactly.\nSpeaker E: So it could get very, very ugly.\nSpeaker D: Definitely.\nSpeaker D: All right, well, I guess I didn't want to keep people too long and Adam wanted people...\nSpeaker D: I'll read the digits if anyone else offers to that be great.\nSpeaker D: And it's not, I guess.\nSpeaker D: More digits the better.\nSpeaker D: Okay, this is.\nSpeaker D: I think a lot is really helpful.\nSpeaker D: I mean Adam and Don will sort of meet and I think that's great.\nSpeaker D: Great.\nSpeaker E: Transcript 2731-2750.\nSpeaker E: 850-51950-61.\nSpeaker E: 0780-1202.\nSpeaker E: 443-4427-556-6600985-765-8849-00191-15614-2748560-34450567.\nSpeaker F: Transcript 1511-1530.\nSpeaker F: 01066-739-310-5881-4538-59434-661134-87011-1243-80729.\nSpeaker F: Scratch that.\nSpeaker F: 90729-01-1394-225-3867-4532-4734-875-2543.\nSpeaker F: 3870-45722-810-951-0980.\nSpeaker C: Transcript 1451-1470-9520-0304-710-103-3307-3240104-6201-751-8512455-9571-0049-0401-103-5401-103-6201-103-6201-751-8512455.\nSpeaker C: 9571-0049-01081790-33613-44580-5251-660-774-378.\nSpeaker D: Transcript 33313350-1377810-2595-384892-5605-9103-08583-1401-2303030-63030-7651-6601-10310-510.\nSpeaker D: 836-406-130-765-837-993-0.\nSpeaker B: Transcripts 36-11-36-30-225-2590-342-456-708-9482-6-03-08801-234.\nSpeaker B: 6-1-757-8453-791-9570-0025-010.\nSpeaker A: Transcript 2971-2990-7798990-902-0101-101-245-2-37445-67091460-7601-126.\nSpeaker A: 460-039-4-01819-958-283-7590-340-7362626-7301.\n"}, {"meeting_id": "Bro021", "summary": "The meeting consisted of participants giving an update on their projects. The team learned that the Eurospeech paper was accepted. PhD C told the team about strange activity in on-line normalization with regards to C-zero and C-one. Grad A shared his work on quals and intention to start cheating experiments. Grad G informed the team that phase normalization was coming along after a bug was fixed, but the results were still not great. Finally, the team discussed Weiner filtering for the Aurora task and the mathematical formulation of the VTS.", "dialogue": "Speaker C: Okay.\nSpeaker F: Somebody else should run this.\nSpeaker F: I'm sick of being the one sort of go through and say, well what do you think about this?\nSpeaker H: Do you want me to run it today?\nSpeaker H: Why should I?\nSpeaker H: Okay.\nSpeaker H: Let's see maybe we should just get a list of items, things that we should talk about.\nSpeaker H: I guess there's the usual updates everybody going around saying, you know, what they're working on, things that happen the last week, but aside from that, is there anything in particular that anybody wants to bring up?\nSpeaker H: No. Okay. So why don't we just go around it?\nSpeaker D: You want to start?\nSpeaker D: All right. We had the first thing maybe that the Euros pitch paper is accepted.\nSpeaker H: This is what's in the paper there?\nSpeaker D: So it's the paper that described basically the system that we're proposed for the one that\nSpeaker H: we submitted the last round.\nSpeaker D: Right. Two under commands seems from the reviewer.\nSpeaker D: Good.\nSpeaker D: Where is it going to be this year?\nSpeaker D: It's Alborg in Denmark.\nSpeaker D: September.\nSpeaker D: Yeah. Well, I've been working mainly on line normalization this week, trying different, slightly different approaches.\nSpeaker D: The first thing is trying to play a little bit again with the time constant.\nSpeaker D: One thing is trying on line normalization with two different means, one mean for the silence and one for the speech.\nSpeaker D: And so I have two recursions which are controlled by the probability of the voice activity detector.\nSpeaker D: This actually doesn't seem to help.\nSpeaker D: So, but well both online normalization approach seems equivalent.\nSpeaker H: Are the means pretty different?\nSpeaker D: Yes. They can be very different.\nSpeaker F: Do you maybe make errors in different places?\nSpeaker D: I didn't look more closely.\nSpeaker D: It might be here.\nSpeaker D: There is one thing that we can observe is that the mean are more different for C0 and C1 than for the other coefficients.\nSpeaker D: Yeah. C1 is...\nSpeaker D: There is strange thing happening with C1 is that when you have different kind of noises, the mean for the silence portion can be different.\nSpeaker D: So, when you look at the trajectory of C1, it's a strange shape.\nSpeaker D: I was expecting that this two mean helps, especially because of the strange C1 shape, which can...\nSpeaker D: You can have a trajectory for the speech and then when you are in the silence, it goes somewhere.\nSpeaker D: But if the noise is different, it goes somewhere else.\nSpeaker D: So, which would mean that if we estimate the mean based on all the signal, even though we have frame dropping, but we don't frame drop everything, this can hurt the estimation of the mean for speech.\nSpeaker D: But I still have to investigate further, I think.\nSpeaker D: The third thing is that instead of having a fixed time constant, I try to have a time constant that's smaller at the beginning of the utterances to adapt more quickly to the...\nSpeaker D: Something that's closer to the right mean.\nSpeaker D: And then this time constant increases and I have a threshold that...\nSpeaker D: If it's higher than a certain threshold, I keep it to this threshold to still adapt the mean.\nSpeaker D: When...\nSpeaker D: If the utterances long enough to continue to adapt after like one second.\nSpeaker D: Well, this doesn't help neither, but this doesn't hurt.\nSpeaker H: Wasn't there some experiment you were going to try where you did something differently for each?\nSpeaker H: I don't know whether it was each male band or each FFT band or something.\nSpeaker H: There's something you were going to...\nSpeaker H: Some parameter you were going to vary depending on the frequency.\nSpeaker H: I don't know if that was...\nSpeaker D: I guess it was...\nSpeaker D: I don't know, maybe it's this idea of having different online normalization for each...\nSpeaker D: Tuning for the different MFCCs.\nSpeaker H: Yeah, I thought Morgan knew brought it up a couple of meetings ago and then it was something about...\nSpeaker H: Then somebody said, yeah, it does seem like C0 is the one that's the major one.\nSpeaker H: I can't remember exactly what it was now.\nSpeaker D: Actually, yeah, it's very important to normalize C0 and much less to normalize the other coefficients.\nSpeaker D: Well, at least with the current online normalization scheme.\nSpeaker D: I think we kind of know that normalizing C1 doesn't help with the current scheme.\nSpeaker D: In my idea, I was thinking that the reason is maybe because of this funny things that happen between speech and silence, which have different means.\nSpeaker D: But maybe it's not so easy to...\nSpeaker F: I really would like to suggest looking a little bit at the kinds of errors.\nSpeaker F: I know you can get lost in that and go forever and not see too much, but sometimes.\nSpeaker F: Just seeing that each of these things didn't make things better may not be enough.\nSpeaker F: Maybe that they're making them better in some ways and worse than others or increasing insurgents and increasing delusions or...\nSpeaker F: helping with noisy case, but hurting in quiet case. If you saw that, then maybe something would occur to you. How to deal with that.\nSpeaker D: All right.\nSpeaker D: Yeah.\nSpeaker D: So that's it, I think, for the online normalization.\nSpeaker D: I've been playing a little bit with some kind of thresholding.\nSpeaker D: As a first experiment, I think what I did is to take to measure the average, no, the maximum energy of each utterance and then put a threshold.\nSpeaker D: Well, this for each band. Then put a threshold. That's 15 dB below. Well, a couple of dB below this maximum.\nSpeaker D: Actually, it was another threshold. It was just adding noise.\nSpeaker D: I was adding a white noise, energy. That's 15 dB below, the maximum energy of the utterance.\nSpeaker D: Yeah. When we look at the MFCC that results from this, there are a lot more smoother.\nSpeaker D: When we compare like a channel zero and channel one utterance, so a clean and the same noisy utterance.\nSpeaker D: Well, there is almost no difference between the capture coefficient of the two.\nSpeaker D: And the result that we have in terms of speed recognition, actually, it's not worse. It's not better neither.\nSpeaker D: But it's kind of surprising that it's not worse because basically you had noise at 15 dB, just 15 dB below, the maximum energy.\nSpeaker H: So why does that smooth things out? I don't understand.\nSpeaker D: Oh, there's less difference. It's whitening.\nSpeaker D: The portion that are more silent, as you add to a white noise that is very high energy, it whitens everything.\nSpeaker D: And the high energy portion of the speech don't get much affected anyway.\nSpeaker D: The other noise, as the noise you add is the same, the shape is also the same.\nSpeaker D: So the trajectory are very, very similar.\nSpeaker F: I mean, again, if you trained in one kind of noise and tested in the same kind of noise, you know, given enough training data, you don't do badly.\nSpeaker F: The reason we have the problems we have is because it's different in training and test.\nSpeaker F: Even if the general kind is the same, the exact instances are different.\nSpeaker F: So when you whiten it, then it's like the only noise to first order the only noise that you have is white noise, and you've added the same thing for training and test.\nSpeaker H: So would that be similar to doing the smoothing then over time?\nSpeaker D: I think it's different.\nSpeaker D: It's something that affects more of the silent portions because, well, anyway, the portion of speech that have high energy are not a lot affected by the noises in the order of database.\nSpeaker D: If you compare the two channels of speech that are during speech portion, the MFCC are not very different.\nSpeaker D: They are very different when energy is lower, like during fricatives or during speech poses.\nSpeaker F: But you're still getting more recognition errors, which means that the differences, even though they look like they're not so big, are hurting your recognition.\nSpeaker D: So it destroys the speech, right?\nSpeaker D: So performance went down?\nSpeaker D: No, it didn't.\nSpeaker D: So in this case, I really expect that maybe the two stream of features are very different.\nSpeaker D: Maybe we could gain something by combining them.\nSpeaker F: Well, the other thing is that you just picked one particular way of doing it. First place is 15 dB down across the utterance.\nSpeaker F: Maybe you'd want to have something that was a little more daft.\nSpeaker F: Secondly, you happened to pick 15 dB and maybe 20 dB better, or 12.\nSpeaker H: What was the threshold part of it?\nSpeaker H: Was the threshold far down?\nSpeaker H: Yeah.\nSpeaker F: Well, he had to figure out how much to add.\nSpeaker F: So he was looking at the peak value.\nSpeaker H: And so what's, I don't understand, how does it go?\nSpeaker H: If the peak value is above some threshold, then you add the noise, or it's below some.\nSpeaker D: I systematically particularly add the noise, but the noise level is just some kind of threshold below the peak.\nSpeaker D: Oh, I see.\nSpeaker D: Yeah, which is not really, no, it actually is just adding a constant to each of the male energy.\nSpeaker D: To each of the male energy.\nSpeaker D: To each of the male energy.\nSpeaker D: So yeah, it's really white noise.\nSpeaker C: Yeah.\nSpeaker F: So then afterwards the log is taken, and that's sort of why the little variation tends to go away.\nSpeaker D: Yeah, so maybe, well, this threshold is still a factor that we have to look at.\nSpeaker D: I don't know, maybe a constant noise addition would be final.\nSpeaker F: Or not constant, but varying over time.\nSpeaker F: In fact, it's another way to go.\nSpeaker D: Yeah.\nSpeaker F: Were you using the normalization in addition to this? I mean, what was the rest of the system?\nSpeaker D: Yeah, it was the same system.\nSpeaker D: It was the same system.\nSpeaker D: Yeah.\nSpeaker D: The third thing is that I play a little bit with finding what was different between...\nSpeaker D: And there were a couple of differences, like the LDF filters were not the same.\nSpeaker D: He had the French-elect complied localization in the system.\nSpeaker D: The nerve of MFCC was different, you used 13 and we used 15.\nSpeaker D: Well, a bunch of differences.\nSpeaker D: And actually, the result that you got were much better on the IDGET, especially.\nSpeaker D: So I kind of investigated to see what was the main factor of this difference.\nSpeaker D: And it seems that the LDF filter is less hurting.\nSpeaker D: So when we put some noise compensation, the LDF filter that's derived from noisy speeches, not more, any more optimal.\nSpeaker D: And it makes a big difference on the IDGETs, trained on clean.\nSpeaker D: If we use the old LDF filter, I mean the LDF filter that was in the proposal, we have like 82.7% recognition rate on noisy speech when the system is trained on clean speech.\nSpeaker D: But when we use the filter that's derived from clean speech, we jumped from 82.7 to 85.1, which is a huge leap.\nSpeaker D: So now the results are more similar.\nSpeaker D: I will not investigate on the other differences, which is like the number of MFCC that we keep and other small things that we can optimize later on.\nSpeaker F: Sure.\nSpeaker F: But on the other hand, if everybody is trying different kinds of noise suppression things and so forth, it might be good to standardize on the piece that we're not changing.\nSpeaker F: So if there's any particular reason to have pick one or the other, which one is closer to what the proposal was that was submitted to Aurora, are they both?\nSpeaker D: I think the new system that I tested, I guess, is closer because it doesn't have less of a front telecom stuff.\nSpeaker B: But the water you tested with at least, you're like, yeah.\nSpeaker F: Yeah, you're trying to add in front telecom, in front of the rest of it, like you said, the number of filters may be different or something.\nSpeaker F: The number of capsule coefficients.\nSpeaker F: Yeah, so I mean, I think we want to standardize there with me.\nSpeaker F: So I should pick something.\nSpeaker D: I think we were going to work with this new system.\nSpeaker B: So the right now, the system that is there in the, what we have in the repository is, it uses 15.\nSpeaker C: Right.\nSpeaker C: Yeah, so.\nSpeaker D: But we will use the LDA filters derived from clean speech.\nSpeaker D: Yeah, yeah.\nSpeaker D: And so actually it's not the LDA filter.\nSpeaker D: It's something that's also short enough in latency.\nSpeaker B: So we have been always using 15 coefficients, not 30.\nSpeaker C: Yeah.\nSpeaker C: Well, that's something.\nSpeaker C: Yeah.\nSpeaker F: I think as long as you guys agree, it doesn't matter.\nSpeaker F: I think we have a maximum of 60 features that we're allowed to sell.\nSpeaker B: Maybe we can at least run some experiments to see whether, once I have this, noise compensation to see whether 13 and 15 really matters or not.\nSpeaker B: Never tested it with the compensation, but without compensation, it was like 15 was slightly better than 13.\nSpeaker B: So that's why we stuck to 13.\nSpeaker D: And there is so this like energy versus zero.\nSpeaker B: Yeah, the larger the devices is zero.\nSpeaker B: But if that's the other thing, I mean, without noise compensation, certainly zero is better than log energy.\nSpeaker B: I mean, because there are more mismatched conditions than the matching conditions for testing.\nSpeaker B: You know, always for the matched condition, you always get a slightly better performance for log energy than C0.\nSpeaker B: But not for, I mean, for matched and the clean condition, both you get log energy.\nSpeaker B: You get a better performance with log energy.\nSpeaker B: Well, maybe once we have this noise compensation, I know we have to try that also, so that we want to go for C0 log energy.\nSpeaker C: You can see that.\nSpeaker H: So do you have more?\nSpeaker H: That's it, I think.\nSpeaker F: You have anything more than this?\nSpeaker F: No, I'm just, you know, being a manager this week.\nSpeaker E: I'll actually get.\nSpeaker E: Still working on my equals preparation stuff.\nSpeaker E: So I'm thinking about starting some cheating experiments to determine the relative effectiveness of some intermediate categories that I want to classify.\nSpeaker E: So for example, I know where voicing occurs and everything.\nSpeaker E: I do a phone recognition experiment, somehow putting in the perfect knowledge that I have about voicing.\nSpeaker E: So in particular, I was thinking in the hybrid framework, just taking those L&A files and setting to 0 those probabilities that these phones are not voicing.\nSpeaker E: So say like I know this particular segment is voicing.\nSpeaker E: I would say go into the corresponding L&A file and solve out the post-series for those phonemes that are not voiced and then see what kinds of improvements I get.\nSpeaker E: And so this would be a useful thing to know in terms of which of these categories are good for speech recognition.\nSpeaker E: So I hope to get those experiments done by a time queues come around in July.\nSpeaker H: So do you just take the probabilities of the other ones and spread them out evenly among them?\nSpeaker E: Yeah, I was thinking, okay, so just set to some really low number of the non-voiced phones and then re-normalize.\nSpeaker H: That would be really interesting to see.\nSpeaker H: So then you're going to feed those into some standard recognized art.\nSpeaker E: Are you going to do a dig answer?\nSpeaker E: Well, I'm going to work with Timit.\nSpeaker H: And then you'll feed those.\nSpeaker H: Sorry, so where are the outputs of the net going to if you're doing follow.\nSpeaker E: Oh, the outputs of the net go into the standard XE hybrid recognized it.\nSpeaker E: So maybe.\nSpeaker H: And you're going to do phone recognition with the net.\nSpeaker E: Right.\nSpeaker E: Another thing would be to extend this to digits or something where I can look at whole words.\nSpeaker E: And I would be able to see not just like phony events, but interphony events.\nSpeaker E: From stop to a vocalic.\nSpeaker E: Something that's transitional.\nSpeaker H: Yeah.\nSpeaker E: Okay.\nSpeaker E: That's it.\nSpeaker H: Okay.\nSpeaker H: So I haven't done a whole lot on anything related to this.\nSpeaker H: It's weak. I've been focusing mainly on meeting recorders done.\nSpeaker H: So, so just that's it on the day.\nSpeaker G: Well, in my last talk last week, I said I tried phase normalization and gotten garbage results.\nSpeaker G: I didn't have long term means of traction approach. Turned out there was a bug in my mat.\nSpeaker G: I tried it again.\nSpeaker G: And the results were better.\nSpeaker G: I got intelligible speech back, but they're still wearing as good as just the fact that they magnitude.\nSpeaker G: The long magnitude means.\nSpeaker G: And also, I'll be talking to.\nSpeaker G: I'm doing some QL about the smart online which model and about coming up with a good model for.\nSpeaker G: Far might use the smart com system.\nSpeaker G: So I'm going to be working on implementing this means of fashion approach from the.\nSpeaker G: Far make system for the smart com system.\nSpeaker G: And one of the experiments we're going to do is.\nSpeaker G: We're going to train the broadcast news net, which is because that's what we've been using so far.\nSpeaker G: And adapted on some other data.\nSpeaker G: And I just want to use data that resembles red speech.\nSpeaker G: Like these picture readings because he feels that the smart com system interaction is not going to be exactly conversational.\nSpeaker G: So actually, I was wondering how long does it take to train that broadcast news net?\nSpeaker F: The big one takes a while.\nSpeaker F: Yeah, it takes two, three weeks.\nSpeaker F: So, but you know, you can get.\nSpeaker F: I don't know if you want to run the big one in the final system because it takes a little while to run it.\nSpeaker F: So you can scale it down by.\nSpeaker F: I'm sorry, it was two, three weeks for training up for the large broadcast news test set training set.\nSpeaker F: I don't know how much you'd be training on.\nSpeaker F: Full.\nSpeaker F: So if you trained on half as much and made the net half as big, then it would be one fourth.\nSpeaker F: A lot of time and it'd be nearly as good.\nSpeaker F: Also, I guess we had, we've had these little discussions, I guess you have an chance to work with it too much.\nSpeaker F: We've had about other ways of taking care of the phase.\nSpeaker F: So, I mean, I guess I was something I could say would be that we've talked a little bit about just doing it all with complex arithmetic.\nSpeaker F: And not doing the polar representation with magnitude and phase, but it looks like there's ways that one could potentially just work with the complex numbers.\nSpeaker F: And in principle, get rid of the effects of the average complex spectrum.\nSpeaker G: And actually regarding the phase normalization.\nSpeaker G: So I did two experiments and one is, so phases get added modulo 2 pi.\nSpeaker G: Because you only know the phase of the complex number of the two, about you modulo 2 pi.\nSpeaker G: And so I thought it first, that what I should do is unwrap the phase, because that will undo that.\nSpeaker G: But I actually got worse results doing that unwrapping using the simple phase unwrapping.\nSpeaker G: I did not unwrapping at all.\nSpeaker F: Yeah. So.\nSpeaker G: And that's all happening.\nSpeaker F: So I'm still hopeful that, I mean, we don't even know if the phase is something, the average phase is something that we do want to remove.\nSpeaker F: I mean, maybe there's some deeper reason why it isn't the right thing to do.\nSpeaker F: At least in principle, it looks like there's a couple potential ways to do one being to just work with the complex numbers.\nSpeaker F: And in rectangular coordinates.\nSpeaker F: And the other is to do a Taylor series.\nSpeaker F: Well, so they work with the complex numbers.\nSpeaker F: And then when you get the spectrum, the average complex spectrum actually divided out as opposed to taking the log and subtracting.\nSpeaker F: So then there might be some numerical issues.\nSpeaker F: I mean, we don't really know that.\nSpeaker F: The other thing we talked a little bit about was Taylor series expansion.\nSpeaker F: And actually I was talking to Dick Carp about a little bit.\nSpeaker F: And since I got thinking about it.\nSpeaker F: The other thing is that you have to do, I think we have to do this on a white board.\nSpeaker F: But I think you have to be a little careful about scaling the numbers that you're taking the complex numbers that you're taking the log off.\nSpeaker F: Because the Taylor expansion for it has square on a cube and so forth.\nSpeaker F: And so if you have a number that is modulus, you know, very different from one, should be right around one.\nSpeaker F: Because it's an expansion of log 1 minus epsilon.\nSpeaker F: One plus epsilon.\nSpeaker F: Well, it's an epsilon squared over 2 and an epsilon cubed over 3 and so forth.\nSpeaker F: So if epsilon is bigger than 1, then it diverges.\nSpeaker F: So you have to do some scaling.\nSpeaker F: But that's not a big deal.\nSpeaker F: Because it's the log of k times the complex number.\nSpeaker F: And you can just the same as log of k plus log of the complex number.\nSpeaker F: So there's converges.\nSpeaker B: Okay.\nSpeaker B: How about you, Sena?\nSpeaker B: So I've been implementing this, we're filtering for this robot task.\nSpeaker B: And I actually thought it was doing fine when I tested it once.\nSpeaker B: It's like using a small section of the code and then I ran the whole recognition experiment with Italian.\nSpeaker B: And I got like, was results then not using it.\nSpeaker B: So I've been trying to find where the problem came from and then it looks like I have some problem in the way.\nSpeaker B: There's some very silly bugs somewhere.\nSpeaker B: I mean, it actually made the whole thing worse.\nSpeaker B: I was looking at the spectrograms that I got.\nSpeaker B: And it's like, it's very horrible.\nSpeaker F: I miss the, I'm sorry, I missed the very first sense.\nSpeaker F: Oh, Ben was on the rest.\nSpeaker F: But what?\nSpeaker B: Oh, yeah, I actually implemented the filter as a module and then I stood it out separately.\nSpeaker B: And it gave like, I just got the signal out and it was okay.\nSpeaker B: So I plugged it in somewhere and then, I mean, it's like I had to remove some part and then plugging it in somewhere.\nSpeaker B: And then in that process, I messed it up somewhere.\nSpeaker B: So, I mean, I thought it was all fine and then I ran it and I got something worse than not using it.\nSpeaker B: So I was like, I'm trying to find what problem came in.\nSpeaker B: It seems to be like somewhere, some silly stuff.\nSpeaker B: And the other thing was, he showed up one day and then I was talking.\nSpeaker B: Yeah, as he's wanted to do.\nSpeaker B: Yeah, so I was actually that I was thinking about doing something about the minute filtering and then Carlos method of stuff.\nSpeaker B: And then he showed up and then I told him and then he gave me a whole bunch of filters what Carlos used for his thesis.\nSpeaker B: That was something which came up and then, so I'm actually thinking of using that also in this, a winner filtering because that is a modified winner filtering approach where instead of using the current frame, he uses adjacent frames also in designing the winner filter.\nSpeaker B: So instead of designing our own new winner filters, I may just use one of those Carlos filters in this implementation and see whether it actually gives me something better than using just the current current frame.\nSpeaker B: Which is in a way, something like the smoothing the winner filter.\nSpeaker B: So I'm like, that's so that is the next thing once this, once I saw this problem out, maybe I'll just go into that also.\nSpeaker B: And the other thing was about the subspace approach.\nSpeaker B: So I like plugged some routines for computing this eigen values and eigen vectors.\nSpeaker B: So just trying to assemble some small block of things which I need to put together for the subspace approach.\nSpeaker B: And I'm in the process of like building up that stuff.\nSpeaker C: And I guess, yeah, because that's it and that's where I am right now.\nSpeaker A: I'm working with VTS.\nSpeaker A: I do several experiments with the Spanish database first, only with VTS and nothing more, not VAD, no LDA, nothing more.\nSpeaker H: What is VTS again?\nSpeaker A: Vectorial, Tyler said to remove the noise.\nSpeaker H: I think I ask you that every single minute.\nSpeaker H: What?\nSpeaker H: I ask you that question every meeting.\nSpeaker F: It's not big of a analysis. It's good to have some cases of the same address.\nSpeaker F: Yeah.\nSpeaker H: What is VTS?\nSpeaker A: VTS.\nSpeaker A: Well, and the question is that, what?\nSpeaker A: Remove some noise, but not too much.\nSpeaker A: And when we put the VAD, the result is better.\nSpeaker A: And we put everything, the result is better, but it's not better than the result that we have with the VTS.\nSpeaker A: You know.\nSpeaker F: I see.\nSpeaker F: So that given that you're using the VAD, also the effect of the VTS is not similar.\nSpeaker F: How much of that do you think is due to just the particular implementation, how much are you adjusting it, or how much do you think is intrinsic?\nSpeaker A: I don't know, because...\nSpeaker D: Are you still using only the 10th-first frame for noise estimation?\nSpeaker A: I do the experiment using only the...\nSpeaker A: Yeah.\nSpeaker A:...doing only one first estimation of the noise.\nSpeaker A: And also I did some experiment doing a line estimation of the noise.\nSpeaker A: Well, it's a little bit better, but not...\nSpeaker D: Maybe you have to standardize this thing also, noise estimation, because all the things that you are testing use different...\nSpeaker D: They all need some...\nSpeaker D: No, it's not a bad idea.\nSpeaker D: They all use a different one.\nSpeaker F: I have an idea.\nSpeaker F: If you're right, I mean, each of these require this.\nSpeaker F: Given that we're going to have, for this test, at least, boundaries, what if initially we start off by using known sections of non-speech for the estimation?\nSpeaker F: Right?\nSpeaker F: Yes.\nSpeaker F: So first place, I mean, even if ultimately we wouldn't be given the boundaries, this would be a good initial experiment to separate out the effects of things.\nSpeaker F: I mean, how much is the poor, you know, relatively unhelpful result that you're getting in this or this or this, is due to some inherent limitation to the method for these tests and how much of it is just due to the fact that you're not accurately finding enough regions that are really noise.\nSpeaker F: So maybe if you tested it using that, you'd have more reliable stretches of non-speech to the estimation from and see if that helps.\nSpeaker A: Another thing is the code book, the initial code book, that maybe, well, it's too clean.\nSpeaker A: Because it's, I don't know, the methods.\nSpeaker A: If you want, I can say something about the method.\nSpeaker A: Yeah, indeed.\nSpeaker A: Because it's a little bit different after the method.\nSpeaker A: But we have, if this is the noise signal, in the log domain, we have something like this.\nSpeaker A: And the idea of this method is to, given, how do you say, I will read because it's other in my English.\nSpeaker A: It's the estimate of the PDF of the noise signal when we have a statistic of the Glinger's speech and a statistic of the noise's speech.\nSpeaker A: And the Glinger's speech, the statistic of the Glinger's speech is from a code book.\nSpeaker A: This is the idea. Well, like this relation is not linear, the methods propose to develop this in the Dr. Taylor series.\nSpeaker A: Approximately.\nSpeaker F: I'm actually just confused about the equations you have up there.\nSpeaker F: So the topic equation is, this is the log domain.\nSpeaker A: Which is the log domain?\nSpeaker A: It's the T, it's equal to log of.\nSpeaker F: But why is what? Why of the spectrum?\nSpeaker A: This is this and this is this.\nSpeaker F: No, no. The top Y is what?\nSpeaker F: Is that the power spectrum?\nSpeaker F: No, is that power spectrum?\nSpeaker F: Yeah, this is the power spectrum.\nSpeaker A: Yeah, it's the power spectrum.\nSpeaker A: So this is the netted, yes.\nSpeaker A: I'll evaluate you.\nSpeaker F: Yeah, okay, so this is the magnitude, the error or something.\nSpeaker F: Yeah.\nSpeaker F: Okay, so you have power spectrum added there.\nSpeaker F: And down here you have, you put the depends on T.\nSpeaker F: But, Billy, this is just, you just mean the log of the one up above.\nSpeaker F: And so that is X times...\nSpeaker A: Yeah, maybe...\nSpeaker A: Well, we can put this...\nSpeaker F: X times one plus...\nSpeaker F: And in minus X.\nSpeaker F: Yeah.\nSpeaker F: And then...\nSpeaker F: And then I see...\nSpeaker F: That's log of X plus log of one plus...\nSpeaker F: Well...\nSpeaker F: Is that right?\nSpeaker F: I'll go and see.\nSpeaker A: Well...\nSpeaker F: I actually don't see how you get that.\nSpeaker A: Well, if we apply the log, we have...\nSpeaker A: It's...\nSpeaker A: Log...\nSpeaker A: It's equal to log of X plus N.\nSpeaker A: Yeah.\nSpeaker A: And well...\nSpeaker A: We can say that...\nSpeaker A: E...\nSpeaker A: Is equal to log of...\nSpeaker A: Exponential of X plus exponential of N.\nSpeaker F: No.\nSpeaker F: No.\nSpeaker C: That doesn't follow.\nSpeaker C: Blow of E every week.\nSpeaker C: Well, this is...\nSpeaker A: This is the time...\nSpeaker A: The time the mind.\nSpeaker A: Well, we have that...\nSpeaker A: We have first that, for example, X...\nSpeaker A: Is equal...\nSpeaker A: Well...\nSpeaker A: This is the frequency domain.\nSpeaker A: And we can put that...\nSpeaker A: The log domain...\nSpeaker A: Log of X...\nSpeaker A: Oh my god, but...\nSpeaker A: Well, in the time the mind, we have an exponential.\nSpeaker A: No?\nSpeaker A: No?\nSpeaker A: Oh, maybe...\nSpeaker F: Yeah, I mean, just never mind what they are.\nSpeaker F: It's just if X and N are variables.\nSpeaker F: Right?\nSpeaker F: The log of X plus N is not the same as the log of E the X plus E the N.\nSpeaker A: Yeah, but this is...\nSpeaker A: I don't...\nSpeaker A: Maybe I can take it offline, but...\nSpeaker A: I can do this incorrectly.\nSpeaker A: The expression that they're bearing in the paper is...\nSpeaker A: Very long.\nSpeaker B: The Taylor series expansion for log 1 plus N by X is...\nSpeaker B: It's X.\nSpeaker B: Yes, the first one.\nSpeaker B: Yeah, I guess.\nSpeaker F: Yeah, because it doesn't just follow as there has to be some...\nSpeaker B: If you take log X into log 1 plus N by X and then expand the log 1 plus N by X into Taylor's...\nSpeaker A: Yeah, this is the...\nSpeaker A: And then...\nSpeaker D: Yeah, but the second...\nSpeaker D: Expression that you put is the first order expansion of...\nSpeaker D: Not something.\nSpeaker A: The first order.\nSpeaker A: No, no, no, it's not the first...\nSpeaker A: We have...\nSpeaker A: We can put that X is equal...\nSpeaker A: I is equal to log of...\nSpeaker F: That doesn't follow.\nSpeaker A: I mean, we can put...\nSpeaker F: That... I mean, the top one does not imply the second one.\nSpeaker F: Because the log of a sum is not the same as...\nSpeaker A: Yeah, yeah, yeah, yeah, but we can...\nSpeaker A: We know that, for some, the log of E plus B is equal to log of E plus log.\nSpeaker F: Right.\nSpeaker F: To B.\nSpeaker A: And we can say...\nSpeaker A: Right, so you can...\nSpeaker A: Yes, it is.\nSpeaker A: And we can put this inside.\nSpeaker A: And then we can...\nSpeaker A: You know?\nSpeaker F: No, but...\nSpeaker A: Yeah?\nSpeaker F: I don't see how you get the second expression from the top one.\nSpeaker F: I mean, just more generally here.\nSpeaker F: If you say log of a plus B, the log of a plus B is not...\nSpeaker F: Or a plus B is not the log of E to the a plus E to the B.\nSpeaker A: No, no, no, no, no, no, no, it's not.\nSpeaker F: Right?\nSpeaker F: And that's what you seem to be saying.\nSpeaker F: No, but this is the same...\nSpeaker F: Right?\nSpeaker F: Because you appear you have the a plus B.\nSpeaker A: No, I say...\nSpeaker A: I have a log of E is equal to log of...\nSpeaker A: In this side is equal to log of X plus N.\nSpeaker A: Right.\nSpeaker A: No?\nSpeaker A: Right.\nSpeaker A: And that's how you go from there to the...\nSpeaker A: And then if I apply exponential to half here...\nSpeaker A: Look, here's what's...\nSpeaker F: I mean C equals a plus B.\nSpeaker D: Of capital Y, right?\nSpeaker D: And then...\nSpeaker D: X...\nSpeaker C: X...\nSpeaker C: This is X inside.\nSpeaker A: Right.\nSpeaker F: We have this, no?\nSpeaker F: Yeah, that one's right.\nSpeaker F: Mm-hmm.\nNone: BEEP BEEP BEEP BEEP\nSpeaker A: BEEP We can put here the set transformation.\nSpeaker F: I see.\nSpeaker F: No?\nSpeaker F: I see.\nSpeaker F: Okay, I understand that.\nSpeaker F: All right, thanks.\nSpeaker A: Yeah, in this case, what we can put here...\nSpeaker A: Y.\nSpeaker F: Okay, so yeah, just by definition, that the individual...\nSpeaker F: So capital X is by definition the same as each of the little X because he's saying that the little X is the log.\nSpeaker A: All right, we can put this.\nSpeaker A: Yeah.\nSpeaker A: And here we can...\nSpeaker F: I think these things are a lot clearer when you can use fonts.\nSpeaker F: Oh, yeah.\nSpeaker F: For fonts.\nSpeaker A: They're saying that which is what...\nSpeaker A: Yeah, yeah, yeah.\nSpeaker A: I understand.\nSpeaker A: Okay.\nSpeaker A: But this is correct.\nSpeaker A: Sure.\nSpeaker A: And now I can do...\nSpeaker A: I can put log...\nSpeaker A: Of EX...\nSpeaker A: Plus log.\nSpeaker F: Yeah, so I understand now.\nSpeaker F: And this is...\nSpeaker G: Yeah.\nNone: Yeah.\nSpeaker D: Right.\nSpeaker G: Let's call that.\nSpeaker A: Right.\nSpeaker A: Okay, thanks.\nSpeaker A: Well, the idea, well, we have fixed this, segue.\nSpeaker F: Okay, so now once you get that one, then you...\nSpeaker F: Then do a first or second order or something, Taylor?\nSpeaker A: Yeah.\nSpeaker A: This is an linear relation that is to develop this in...\nSpeaker A: Backter Taylor's say.\nSpeaker A: Sure.\nSpeaker A: Right.\nSpeaker A: And for that, well, the goal is to estimate a PDF for the noise speech.\nSpeaker A: When we have a statistic for clean speech and for the noise speech.\nSpeaker A: And the way to obtain the PDF for the noise speech is...\nSpeaker A: Well, we know this statistic.\nSpeaker A: And we know the notes is...\nSpeaker A: Well, we can apply first order of the vector Taylor series of the order...\nSpeaker A: The order that we want to increase the complexity of the problem.\nSpeaker A: And then when we have a expression for the...\nSpeaker A: Mean that variance of the noise speech.\nSpeaker A: We apply a technique of minimum mean square estimation to obtain the expected value of the clean speech given the...\nSpeaker A: And this statistic for the noise speech, the statistic for clean speech and the statistic of the noise speech.\nSpeaker A: This only that.\nSpeaker A: But the idea is that...\nSpeaker D: And the model of clean speech is a code book, right?\nSpeaker D: Yeah.\nSpeaker A: We have a code boot with different density Gaussian.\nSpeaker A: We can put the...\nSpeaker A: For the clean speech, the probability of the clean speech is equal to...\nSpeaker C: Well, we got the data.\nSpeaker F: So, how much in the work they reported, how much noise speech did you need to get good enough statistics for to get this mapping?\nSpeaker A: I don't know, sadly.\nSpeaker A: Yeah.\nSpeaker A: I need this...\nSpeaker A: I don't know, sadly.\nSpeaker F: I think what's certainly characteristic of a lot of the data in this test is that you don't have...\nSpeaker F: The training set may not be a great estimator for the noise in the test set.\nSpeaker F: Sometimes it is and sometimes it's on.\nSpeaker A: Yeah, the clean speech, the code book for clean speech, I am using themit.\nSpeaker A: And I have now 64 Gaussian.\nSpeaker F: And what are you using for the noise?\nSpeaker A: I estimate the noises.\nSpeaker A: For the noises, I usually use one Gaussian.\nSpeaker F: And you train it up entirely from non-speech sections in the test?\nSpeaker A: Yes, the first experiment that I do is only to calculate this value.\nSpeaker A: The compensation of the dictionary one time using the noise at the beginning of the sentence, this is the first experiment.\nSpeaker A: And I fix this for all the sentences.\nSpeaker A: Because well, the VDS matters.\nSpeaker A: In fact, the first thing that I do is to obtain an expression for E.\nSpeaker A: Probability is expression of E.\nSpeaker A: That means that the VDS with the VDS we obtain...\nSpeaker A: We obtain the means for each Gaussian and the variance.\nSpeaker A: This is one.\nSpeaker A: This is the compensation of the dictionary.\nSpeaker A: And the other thing that is with this matters is to obtain to calculate this value.\nSpeaker A: Because we can write that the estimation of the Gling speed is equal at a expected value of the Gling speed conditional to the noise seen.\nSpeaker A: The probability of the statistic of the Gling speed and the statistic of the noise.\nSpeaker A: This is the matters that say that we obtain this.\nSpeaker A: And we can put that this is equal to the estimate value of E minus a function that conditional to the naysina.\nSpeaker A: This is the term after the developed this, the term that we obtain.\nSpeaker A: And we can put that this is equal to the naysina minus.\nNone: Okay.\nSpeaker A: I put this name and I can calculate this.\nSpeaker F: What is the first variable in that probability?\nSpeaker F: The Gaussian.\nSpeaker A: This is the...\nSpeaker A: It's not exactly this.\nSpeaker A: It's modified.\nSpeaker A: If we have the Gling speed, we have the probability of a weight for each Gaussian.\nSpeaker A: And now this weight is different.\nSpeaker A: And I need to calculate this.\nSpeaker A: I can't develop an expression that this.\nSpeaker A: I can calculate this value with the statistic of the noise.\nSpeaker A: I can calculate this value with the statistic of the noise speed that I calculated before with the VTS approximation.\nSpeaker A: And what normalizes?\nSpeaker A: And I know everything.\nSpeaker A: When I developed this in Taylor's series, I can't calculate the mean and variance of the...\nSpeaker A: For each of the Gaussian of the dictionary for the noise speed.\nSpeaker A: And this is fixed.\nSpeaker A: If I never do a new estimation of the noise, this mean and variance are fixed.\nSpeaker A: And for each frame of the speed, the only thing that I need to do is to calculate this in order to calculate the estimation of the Gling speed.\nSpeaker A: Giving a noise speed.\nSpeaker F: So I'm not following this perfectly, but are you saying that all of these estimates are done using estimates of the probability density for the noise that are calculated only from the first 10 frames?\nSpeaker F: And never change throughout anything else.\nSpeaker A: This is one of the estimation that I do.\nSpeaker F: Per utterance?\nSpeaker A: Per utterance.\nSpeaker F: So it's done new for each new address.\nSpeaker F: So this changes the whole mapping for every address.\nSpeaker A: Yeah, it's fixed, the dictionary.\nSpeaker A: And the other estimation is when I do the underlying estimation, I change the means and variance of the noise speed.\nSpeaker A: It's time that I detect noise.\nSpeaker A: I do again this, the value, estimate the new mean and variance of the nice speed.\nSpeaker A: And with this new mean and variance, I estimate again.\nSpeaker F: So you estimate it completely forgetting what you had before?\nSpeaker A: Or is there some...\nSpeaker A: No, no, no, no, it's not completely noise.\nSpeaker A: I am doing something like another station of the noise.\nSpeaker F: Now do we know, either from their experience or from yours, that just having two parameters that mean variance is enough?\nSpeaker F: Yeah, I mean, I know you don't have a lot of data to estimate with, but...\nSpeaker A: I estimate mean and variance for each one of the Gaussian of the code, Boo.\nSpeaker F: No, I'm talking about the noise.\nSpeaker A: There's only one Gaussian.\nSpeaker A: I'm using only one.\nSpeaker A: Right.\nSpeaker F: I don't know.\nSpeaker F: And it's only one...\nSpeaker F: Wait a minute, what's the dimensionality of the Gaussian?\nSpeaker A: It's in after the Melvator Bank.\nSpeaker A: So it's 20...\nSpeaker F: So it's actually 40 numbers that you're getting.\nSpeaker F: Yeah, maybe...\nSpeaker A: The original papers say that only one Gaussian for the noise.\nSpeaker F: Well, yeah, but I mean...\nSpeaker F: Yeah, maybe...\nSpeaker F: No paper is a Bible, you know.\nSpeaker F: The question is whether it would be helpful, particularly if you had more...\nSpeaker F: So suppose you did, this is almost cheating, it certainly isn't real time, but if you suppose you used the real boundaries that you were going to refactor given by the VAD and so forth, or I guess we're going to be given even better boundaries than that.\nSpeaker F: And you take all of the non-speech components in an utterance, so you have a fair amount.\nSpeaker F: Do you benefit from having a better model for the noise? That would be another question.\nSpeaker F: Maybe.\nSpeaker F: So first question would be, to what extent are the errors that you still seeing based on the fact that you have poor boundaries for the non-speech?\nSpeaker F: The second question might be given that you have good boundaries, could you do better if you use more parameters to characterize the noise?\nSpeaker F: Also, another question might be, they are using first term only of the vector Taylor series.\nSpeaker F: If you do a second term, does it get too complicated?\nSpeaker F: Yeah, it's quite complicated.\nSpeaker F: Yeah, okay.\nSpeaker F: I want to ask the last question then.\nSpeaker A: For me, it's the first time that I am working with VTS.\nSpeaker F: Yeah, no, it's interesting. We haven't had anybody work with it before, so it's interesting to get your feedback.\nSpeaker A: It's another type of approximation because it's a statistic approximation to remove the noise.\nSpeaker H: Right.\nSpeaker H: Okay, well, I guess we're about done.\nSpeaker H: So some of the digit forms don't have digits.\nSpeaker H: If you ran out and there were some blanks in there, so not everybody will be reading digits.\nSpeaker H: I guess you've got some right Morgan.\nSpeaker H: I have some.\nSpeaker H: I want you to go ahead and start and I think it's just us down here.\nSpeaker C: Okay, switch off the phone.\nSpeaker F: Leave it on.\nSpeaker F: They prefer to have them on just so that they're continuing to get the distant information.\nSpeaker F: Transcript, LDASH 169.\nSpeaker F: 393 095 798 2760 565 867 611 1 8 2 4 2 6 1 9 0 0 2 7 8 3 8 8 1 2 2 6 7 4 1 9 7 9 4 9 8 5 7 8 7 9 0 9 2 375 417 612 3762 6367 2942\nSpeaker E: Transcript LDASH 167 7 5 6 0 5 373 147 4 4 9 2 7 8 1167 38 85 57 8 164 64 8 203 377 5 263 2857 95672 2 9 0 18 32 2623 32 5 2 7 7 9 4 9 6\nSpeaker H: Transcript LDASH 168 7 9 9 9 2 4 8 8 0 6 9 8 7 0 1 8 4 8 0 3 1 7 8 4 1 7 8 2 5 2 3 2 6 5 9 8 3 6 6 5 4 0 4 1 5 0 5 8 8 2 0 357 2 6 0 6 5 9 7 6 2 8 1 9 5 9 8 11 4 2 0 8 6 7 9 8\nSpeaker G: Transcript LDASH 174 4911 8 4 7 9 2 4 1 2 7 8 9 2 5 1 6 6 367 7 3 2 8 9 7 6 2 7 9 3 6 1 0 5 9 3 5 7 5 6 0 4 8 8 8 2 9 8 5 9 6 7 2 5 0 6 9 4 8 2 2 0 19 269 1\nSpeaker F: Okay Okay\n"}, {"meeting_id": "TS3009b", "summary": "This meeting was about the functional design of the remote control. Firstly, the Project Manager mentioned new requirements including one recognizable button in the middle, the handy design, rubber materials and they would not use teletext. Next, Marketing presented on functional requirements including a fancy look, ease of use, and the LCD as well as speech recognition. Then, Industrial Designer introduced technical design and working design on remote interfaces. Lastly, User Interface presented on component design and the group decided to adopt a user-centered approach to reduce the main buttons and put a big circle in the middle with the power button, channel, and volume buttons.", "dialogue": "Speaker C: You know minutes.\nNone: Your screen was also...\nNone: I have great.\nSpeaker A: Oh, I forgot my pen.\nSpeaker F: Oh, I don't need it.\nSpeaker F: My screen is small again.\nNone: I don't know what to do.\nSpeaker C: My screen is big.\nSpeaker C: Who starts?\nSpeaker C: Well, I'll start just with another presentation.\nSpeaker C: So then we can look at the agenda for this meeting.\nSpeaker C: I'll put some new things in the map.\nSpeaker C: This is it.\nSpeaker C: I don't know the shortcut.\nSpeaker C: Well, our functional design meeting, the states were in.\nSpeaker C: Also, all three of you have prepared something about it.\nSpeaker C: We'll just have a look at the notes from the previous meeting.\nSpeaker C: What we thought we had decided.\nSpeaker C: Then we'll look at the three presentations from you.\nSpeaker C: I think you've prepared for all of that.\nSpeaker C: We'll look at the new project requirements.\nSpeaker C: You also have received the new project requirements from our bosses.\nSpeaker C: I've received them with some additional requirements.\nSpeaker C: I think we should show them before your presentations because it's not really smart to include some things we can because of the new requirements.\nSpeaker C: Then we can make some decisions about our remote control functions.\nSpeaker C: We have to decide in this meeting what our function will be.\nSpeaker C: Then we can discuss some more closing.\nSpeaker B: We have 40 minutes for this discussion.\nSpeaker C: Yeah, I think so.\nSpeaker C: Well, Vaka, the closing.\nSpeaker C: We'll not look at that yet.\nSpeaker C: I'll show that to this board.\nSpeaker C: Well, notes for this meeting.\nSpeaker C: I gave a presentation.\nSpeaker C: We familiarized ourselves with the boards.\nSpeaker C: Then we discussed some first ideas.\nSpeaker C: We have to merge the strong points from our competitors and look at their remote controls.\nSpeaker C: We should make it compatible with our new DVD and other releases we have, our technical releases.\nSpeaker C: Not too many buttons.\nSpeaker C: We'll recognize a move button in the middle.\nSpeaker C: Why do you do the most important functions with?\nSpeaker C: They can have two functions because you have a DVD and a television.\nSpeaker C: The design has to fit the hands, the original, but also be familiar.\nSpeaker C: It's one of our ideas.\nSpeaker C: It's just 30 minutes ago, so it's not quite a...\nSpeaker C: Now it's right.\nSpeaker C: But I have to do it.\nSpeaker C: The materials should be hard plastic with rubber from the ties and the labeling of the buttons should be indestructible.\nSpeaker C: It should be very nice.\nSpeaker C: Well, fronts for the remote, just like one more telephone.\nSpeaker C: And the technical aspect and also the labeling of the buttons, the functions should be universal standards.\nSpeaker C: That's just some ideas from the first meeting.\nSpeaker C: It's quite logical.\nSpeaker C: All of it.\nSpeaker C: Now the new project requirements, I'll just show them.\nSpeaker C: I got this mill from our bosses.\nSpeaker C: Well, data text goes out.\nSpeaker C: We will not use data text.\nSpeaker C: Maybe I just ignored a thing.\nSpeaker C: But not data text.\nSpeaker C: It's not my place to disagree against.\nSpeaker C: The second is a bit...\nSpeaker C: Shh, pity.\nSpeaker C: Because we just said we wanted to include DVD and data wanted because of our time we have for this project.\nSpeaker C: So that's a shame because especially for the third requirement, we want to reach people under the 30 years because we don't have those customers a lot at this point.\nSpeaker C: Well, it's a bit pity because those people want to have one remote control for all those technical devices they can...\nSpeaker C: We forget about it.\nSpeaker F: It's just time consuming.\nSpeaker C: Well, and our corporate image should stay recognizable in our products.\nSpeaker C: So we have to use maybe a slogan, maybe a color, and on our remote controls, the design has to be...\nSpeaker C: As we already said, I'm actually familiar.\nSpeaker C: Not only this shape, but also our company.\nSpeaker B: Yeah, we are a real fashion more company.\nSpeaker B: I read it all the...\nSpeaker B: I didn't know what company we were, but we designed it.\nSpeaker B: Especially trendy stuff.\nSpeaker B: So it has to be a modern design. It's important to know when you design it.\nSpeaker C: Yes, I noted our slogan that we have, our company, that's... we put the fashion in electronics.\nSpeaker C: No.\nSpeaker C: So maybe that's a slogan that we can put somewhere on a remote control or something.\nSpeaker C: Alright, then we're going to have three presentations.\nSpeaker C: You want the start?\nSpeaker C: I think I have to start.\nSpeaker C: I didn't see anything about that.\nSpeaker F: I just have to think which follows mine, because I was a bit in a hurry.\nSpeaker F: Okay, well...\nSpeaker F: I think it's this one, but I'm not sure.\nSpeaker F: Gwynne?\nSpeaker C: Hi, if you want to open to PowerPoint.\nSpeaker F: Yes, this one.\nSpeaker F: Yeah.\nSpeaker F: Well, I'm going to tell you something about functional requirements.\nSpeaker F: To start with these points.\nSpeaker F: Next sheet.\nSpeaker F: At first I tell you something about what people dislike about the current controls, because it's a smart thing to exclude those things.\nSpeaker F: Furthermore, it's very important what they do like and what they do use.\nSpeaker F: Then I tell something about the most important issues.\nSpeaker F: So we have to focus on those three things.\nSpeaker F: And in the end I'll show you our target audience or our target product users, customers.\nSpeaker F: Well, the first findings are that people think most controls are very kind of ugly.\nSpeaker F: So that's 75% of the current users.\nSpeaker F: They don't like it, so we might think about fronts in that section.\nSpeaker F: They also say that's about, I thought it was 50%, that more money will be spent on better looking controls.\nSpeaker F: So it's very important that you design a nice looking control.\nSpeaker F: The current user uses his machine just about, well, all the time for a few functions.\nSpeaker F: Almost every user uses the control for just 10% of his capacity.\nSpeaker F: So it's really important to make the buttons for the common tasks kind of big or kind of flashy.\nSpeaker F: Furthermore, it's 75% of the users' Zepsilop.\nSpeaker F: This might be smart to make a big zapping button or something in the middle.\nSpeaker F: So you can reach the system.\nSpeaker F: You can zap away.\nSpeaker F: Yeah, right.\nSpeaker F: Right.\nSpeaker F: A lot of users' users lose their controls in the living room.\nSpeaker F: So it might be sensible to make some kind of a button on your television that your control beeps or something that you can find very easily.\nSpeaker F: I don't know, maybe that's an idea.\nSpeaker F: Because it's a big, I think, 50% of the users' loses its control within the same room.\nSpeaker C: It's actually loose from the television because it can also be used for other televisions.\nSpeaker C: So if you deliver a small click on device that you can put on your television that leaps to your remote control, everyone can use it.\nSpeaker C: Yeah, but if you lose your click on the device.\nSpeaker C: No, you can click it on your television.\nSpeaker F: Yeah, but if somebody else uses it in another room or something.\nSpeaker F: In another room.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: The control is lost in the same room.\nSpeaker F: So, well, it'll be revised.\nSpeaker F: Yeah.\nSpeaker F: Furthermore, the learning time is a problem.\nSpeaker F: 34% thinks it's too difficult to learn.\nSpeaker F: So the learning curve should be very short for the dumps people should be able to use it.\nSpeaker B: I think our user expert should also consider manual for the remote, of course.\nSpeaker F: Yeah, but people don't read manuals.\nSpeaker B: No, they don't read it.\nSpeaker B: They use it to add one, do you think?\nSpeaker A: I don't think.\nSpeaker F: I think you should put more time in the design of pick up and use.\nSpeaker E: That's manual.\nSpeaker E: Yeah, right.\nSpeaker E: Big cost of the user.\nSpeaker E: I know how it's supposed to work.\nSpeaker E: Yeah.\nSpeaker C: Right.\nSpeaker C: You should always be a manual, but it can be very short ahead.\nSpeaker E: And it should be consistent with all the remote control, I think.\nSpeaker C: Yes.\nSpeaker C: Okay.\nSpeaker B: Now, maybe for the, if you don't recognize a button, who do they call?\nSpeaker E: Yeah, right.\nSpeaker E: It should be there, the manual, but not to explain how the remote works.\nSpeaker F: We don't have much time, so it's better to put our attention to the design so you can pick up and use it.\nSpeaker C: We are a design team.\nSpeaker C: We can say to some writer, make a manual.\nSpeaker B: It's not part of the, no.\nSpeaker B: Well, have a look.\nSpeaker F: Next point.\nSpeaker F: Yes?\nSpeaker F: Well, that's about 20% I thought, but the designer should take the, should consider there.\nSpeaker F: Yeah, consider the consequences of using your remote.\nSpeaker F: It should be a good in your hand.\nSpeaker F: Right.\nSpeaker F: This is the most important part.\nSpeaker F: We're, like the requirements that we're going to specify, we're going to target a younger audience.\nSpeaker F: That's about 60% of the market, so it's quite important.\nSpeaker F: Research shows that they like to have a little LCD screen on their, on their zapping device.\nSpeaker F: I thought it was the age between 16 and 20, 99% of the people like that.\nSpeaker F: So it's very important.\nSpeaker F: We should definitely have that in our designs.\nSpeaker C: Well, with 12 euro 50 as production cost we can afford to be in LZB.\nSpeaker F: Yeah, but they think it's really important.\nSpeaker F: So if we want to, if we have a big, if we make lots of, of the stuff, maybe we can buy it very cheap.\nSpeaker F: I don't know.\nSpeaker F: We have to.\nSpeaker B: Well, we'll consider it here.\nSpeaker C: Sure, your chance to look into the costs of those.\nSpeaker C: I don't know.\nSpeaker F: I don't have any information on it.\nSpeaker D: I know.\nSpeaker D: We'll look into that later.\nSpeaker F: Right.\nSpeaker F: And another thing is speech recognition.\nSpeaker F: They also like that, but research is very costly.\nSpeaker E: So that's difficult to realize also.\nSpeaker F: Yeah.\nSpeaker F: But it might be important.\nSpeaker C: We have very demanding clients.\nSpeaker C: It's not yet standard development.\nSpeaker E: We have customers in the world.\nSpeaker F: LCD is more, I think, reachable than speech recognition.\nSpeaker F: Yeah, absolutely.\nSpeaker F: So we might consider LCD screens.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I will consider both and see what we can find.\nSpeaker B: We don't rule them out immediately.\nSpeaker F: Okay.\nSpeaker F: I think that's it.\nSpeaker F: I think it is sensible to take these points in the future.\nSpeaker F: Yes.\nSpeaker F: So you can write.\nSpeaker F: Well, I can still see your presentation.\nSpeaker C: Yeah.\nSpeaker C: Right.\nSpeaker C: Well, next, I don't know who is next.\nSpeaker C: You got a.\nSpeaker C: Can I give a technical talk?\nSpeaker C: Yes.\nSpeaker B: Go ahead.\nSpeaker B: Well, it's my tech to explain what to point out of working design.\nSpeaker B: Yep.\nSpeaker B: We have that here.\nSpeaker B: Okay.\nSpeaker B: How do you enlarge it?\nSpeaker B: So that you have the F5.\nSpeaker D: F5.\nSpeaker D: Yeah.\nSpeaker B: Oh, the one in the same?\nSpeaker B: That's my next button.\nSpeaker B: Well, right.\nSpeaker B: You know who I am and what I do.\nSpeaker B: So we have this.\nSpeaker B: It's a bit unclear because I want to copy paste something.\nSpeaker B: It was originally in black and white, but it became black and purple.\nSpeaker B: But I think you can read it.\nSpeaker B: Yeah.\nSpeaker B: Well, I think it's important for you to realize the basic function of a remote control.\nSpeaker B: Well, you can see maybe you can select it.\nSpeaker F: Oh, yes.\nSpeaker F: In first.\nSpeaker B: And I don't can select.\nSpeaker B: I can select on the.\nSpeaker B: Go to the next page.\nSpeaker B: Oh, not for my.\nSpeaker B: Well, you can read it.\nSpeaker B: It's not.\nSpeaker B: It's a schematic view of how a basic remote control works.\nSpeaker B: You have basically the energy, the power of the remote control and the center, which is the lead, the bulb that sends the infrared beam to the set.\nSpeaker B: And the source is, of course, the user.\nSpeaker B: And the user interface is the buttons, of course.\nSpeaker B: And the user interface sends the different signals of the different buttons to the chip.\nSpeaker B: The chip sends it to the left and the left sends it to the receiver.\nSpeaker B: That's the basic idea.\nSpeaker B: Very basic.\nSpeaker B: Well, I have put it in a couple of basic steps.\nSpeaker B: The remote is basically just waiting for a user to press a key, does nothing until, of course, the key is pressed.\nSpeaker B: The key press is the signal to a chip.\nSpeaker B: The chip sends it to connection and recognize the key.\nSpeaker B: So, well, you understand.\nSpeaker B: The chip produces Morse code, a specific code to indicate that specific button, that is pressed, of course, and it uses transistors in the remote control to amplify and to send that signal again to the lead, which is the bulb, of course.\nSpeaker B: And the lead produces an infrared beam, signals the, well, it's very simple, and signals the, signals to the center on the TV set and the TV set also recognize the signal and performs the assigned task.\nSpeaker C: So, this is also why we have to have a button that says, I'm now busy with a DVD if we've had done that.\nSpeaker B: But we don't.\nSpeaker B: Yeah, exactly.\nSpeaker B: Well, this is the basic function of a remote.\nSpeaker B: I've done a couple of pictures here.\nSpeaker B: It's a very basic one.\nSpeaker B: And the score is if we're going to add an LCD screen to it, it won't look anything like this, but this is very basic, basically the shape of a remote control that has very little buttons.\nSpeaker B: But it's quite, you can easily recognize the buttons.\nSpeaker B: They're far enough apart and anything.\nSpeaker B: It's not very, not very high tech indeed, and not very user friendly.\nSpeaker B: If you look at shape, it's just a simple long box shape.\nSpeaker B: So we have to change a little bit to that.\nSpeaker B: So that comes more user friendly and that problems like Air Z and those kinds of things don't occur.\nSpeaker F: Yeah.\nSpeaker F: I have a table here about the relevance of the buttons.\nSpeaker F: The power button is used very much.\nSpeaker F: Channel selection, volume, and tail attacks.\nSpeaker F: Well, tail attacks is not an option.\nSpeaker F: But I think it's very important to make the power channel and volume buttons near to the thumb so you can't have Air Z consequences.\nSpeaker B: Yeah, because they are the most important buttons and you can immediately make them easy to\nSpeaker F: press.\nSpeaker D: You don't have to look and search for them.\nSpeaker C: If you have the most used buttons all in one place and you keep making the same moves. Well, if you would put it at a different place then you have to move your hands and that's\nSpeaker F: one of the things about Air Z. That's right.\nSpeaker B: You can't have any, every button under the thumb.\nSpeaker C: No, but the most important buttons, maybe you can just put them a bit apart.\nSpeaker C: So you would reject Air Z.\nSpeaker E: Maybe you can make for channel changing two little buttons on the side of the remote so you can just do it like this.\nSpeaker E: Yes.\nSpeaker C: I saw that on mobile telephones they also have those things that are usable.\nSpeaker F: Do people, when they pick up remote, know that they have to do that.\nSpeaker F: It's a new feature.\nSpeaker F: You can make a double feature like a button on the top.\nSpeaker C: Also, if someone picks up the remote, I think it's a good thing.\nSpeaker C: If someone picks up his remote and he touches the side and he's already on the next channel.\nSpeaker A: That's very irritating.\nSpeaker C: That's right.\nSpeaker C: I think.\nSpeaker F: Right.\nSpeaker B: In any case, the basic function should be, indeed, as you say, at the thumb, I think that's a good idea.\nSpeaker B: The less important buttons, like the different channels, the numbers, 1, 3, 4, 5, should be not in reach because they don't use it all the time.\nSpeaker B: Well, it's pretty, basically, as you said, some pictures of the inside workings, but I don't want to get too technical because that's not very useful for your part of the job.\nSpeaker B: So, yeah, you said, this is how it looks from the inside.\nSpeaker B: I still like this.\nSpeaker B: I had to delete this button.\nSpeaker D: I had to make a schematic of the...\nSpeaker C: I had to do a little time, but don't look at this, please.\nSpeaker C: I think it's clear.\nSpeaker C: It's the most important thing.\nSpeaker C: Probably the latest that we started this presentation.\nSpeaker C: I don't know.\nSpeaker F: I think about 20 minutes ago.\nSpeaker A: Yeah.\nSpeaker C: Then we have still the time.\nSpeaker C: But we do have to come to a decision.\nSpeaker C: Yeah, right.\nSpeaker C: So...\nSpeaker E: That's cool, ever.\nSpeaker E: I thought everybody on the website would see the same thing, but that's not the case.\nSpeaker B: Yeah, there are different...\nSpeaker B: We have all different home pages.\nSpeaker E: Oh, yeah.\nSpeaker E: Different links, yeah.\nSpeaker E: For instance, you couldn't see this.\nSpeaker E: Okay.\nSpeaker E: Yeah, well...\nSpeaker E: Yeah.\nSpeaker E: Well, I'm my younger.\nSpeaker E: Usually, in faces, I have the method.\nSpeaker E: Well, I use my own experience with remote to look at the remote on the corporate website, which are these two.\nSpeaker E: Yeah.\nSpeaker E: These are already used.\nSpeaker E: Yes, these are from another manufacturer.\nSpeaker E: All right.\nSpeaker E: This one is engineering-centered, so this one has the most functions and things.\nSpeaker E: This one is user-centered.\nSpeaker E: Mm-hmm.\nSpeaker D: Well, I like users-centered.\nSpeaker D: Yeah.\nSpeaker E: Also the best.\nSpeaker E: We also have...\nSpeaker E: Well, I thought that we would use the option to control the DVD also and tailor text and that kind of stuff.\nSpeaker E: So I thought we would use more or need more buttons than this one, but we have to reject that because it requires...\nSpeaker E: Yes.\nSpeaker E: That's why this is not relevant anymore.\nSpeaker E: I think this is about the maximum number of buttons we'll need.\nSpeaker E: Yeah, right.\nSpeaker E: I kind of like the shape.\nSpeaker E: I think this is what we talked about.\nSpeaker A: Yeah.\nSpeaker B: Yeah, I think you can't really see it from different sides, but I think...\nSpeaker E: No, I'll show this.\nSpeaker E: You can draw it a few...\nSpeaker E: Yeah, I think we should go further with the idea of removable fronts.\nSpeaker E: Mm-hmm.\nSpeaker E: So we can customize...\nSpeaker C: Oh, absolutely.\nSpeaker C: But they all have to have something about the recognition from our company.\nSpeaker C: So we cannot just make someone want...\nSpeaker E: If it's not the whole remote that changes.\nSpeaker C: No, but the side they look at is the front.\nSpeaker C: So if you make a front with just a tiger on it, then our recognition is totally gone.\nSpeaker C: Yeah, that's right.\nSpeaker E: You can put the same symbol on every remote.\nSpeaker D: That's the most...\nSpeaker D: Like everything...\nSpeaker D: We can put it on the backside.\nSpeaker D: Yeah.\nSpeaker E: Well, and...\nSpeaker E: Something like this...\nSpeaker E: Yes.\nSpeaker E: It's recognizable.\nSpeaker E: Okay.\nSpeaker F: We can make a symbol of the company right here and a few with the front on it.\nSpeaker F: There's a hole in the front.\nSpeaker B: Yeah, so don't replace the symbol.\nSpeaker C: Or the lowest part of the remote isn't changed by the front.\nSpeaker C: That's not focused on the front.\nSpeaker E: Okay.\nSpeaker E: Well, so like I said, I thought we'd use more functions.\nSpeaker E: We had to include more functions, but we don't.\nSpeaker E: So...\nSpeaker E: Yeah.\nSpeaker E: I think this is about the maximum number of buttons we need.\nSpeaker E: Maybe some less.\nSpeaker E: Like eject we don't need.\nSpeaker E: Some other buttons we don't need.\nSpeaker B: Mike, can you put that picture for me in the work document file?\nSpeaker E: I will.\nSpeaker E: I think for the remote less is more.\nSpeaker E: Less buttons, the better the design.\nSpeaker E: Yeah.\nSpeaker E: We should go with that concept, I think.\nSpeaker F: All right.\nSpeaker F: I've got another point.\nSpeaker F: Yeah.\nSpeaker F: There are two target audiences.\nSpeaker F: And we've chose for the younger one.\nSpeaker F: Research has shown that it's high-interested in features.\nSpeaker F: They are high-interested in feature.\nSpeaker F: But they are more critical for, yeah, critical.\nSpeaker F: The younger audience.\nSpeaker F: So we must design and control.\nSpeaker F: That really speaks to people.\nSpeaker C: Well, what if we at home have a remote that has the most familiar buttons on the top?\nSpeaker C: And the bottom side of the front has a little clap.\nSpeaker C: You can click away.\nSpeaker C: And then you have much more functions that most people don't use, but some don't do.\nSpeaker B: But I think the most functions underneath that aren't as much.\nSpeaker C: They are, especially if their features are important.\nSpeaker C: They want a lot.\nSpeaker E: But not.\nSpeaker E: What kind of features?\nSpeaker E: I think most of these screens have a false recognition.\nSpeaker E: But most look at the numbers.\nSpeaker F: The newest features are, like I said, our LCD and speech control.\nSpeaker F: Our audience, these people, are very like these features.\nSpeaker F: You see?\nSpeaker F: So we must build in something, or they will go to the concurrent.\nSpeaker F: Our competitors.\nSpeaker F: Our competitors, right.\nSpeaker F: So I do think we have to have some features.\nSpeaker C: Even though they cost a little more.\nSpeaker C: From some calculators, you have those little LCD that you can click on or something.\nSpeaker C: You can click out of the remote.\nSpeaker C: And if that gives you a little bit of set of information.\nSpeaker F: Which program you are watching or something.\nSpeaker C: Those kind of things.\nSpeaker C: Because you also have those program recognition for your PCRs.\nSpeaker C: And if your remote picks that up also, they can display which program you are currently watching.\nSpeaker B: So it just signals the different symbols on the screen.\nSpeaker B: Because if you change to channel 2, you have 2 on the screen and 2 on your...\nSpeaker B: For example.\nSpeaker B: On your LCD screen.\nSpeaker C: It could be such a little that you can click in and out and you have it.\nSpeaker E: Yeah, we should keep that simple too.\nSpeaker B: But should it really be clickable?\nSpeaker B: No, maybe.\nSpeaker B: No, it's really integrated.\nSpeaker B: Just integrate in the right to make it more trendy.\nSpeaker E: You said like this, you can watch something like on some radios in the car.\nSpeaker E: You're walking...\nSpeaker F: Are the answers to something?\nSpeaker B: That's of course a bit more expensive than the basic calculator design.\nSpeaker B: But there's a scrolling text and that kind of thing.\nSpeaker C: I think it's just a script that keeps it rolling.\nSpeaker C: It's five minutes of implementing time.\nSpeaker C: Five minutes, yeah.\nSpeaker F: Yeah, for programming.\nSpeaker F: So I don't think that's the issue.\nSpeaker F: No.\nSpeaker F: Alright.\nSpeaker F: Alright.\nSpeaker C: We go with LCD.\nSpeaker C: Well, I think so, yes.\nSpeaker E: We still need to know how much that will cost.\nSpeaker F: I don't know if I can find that.\nSpeaker E: No, maybe you'll get that information.\nSpeaker C: We can use this board again, I think.\nSpeaker C: We can put some decisions about the controls we want, the issue with it.\nSpeaker C: Where's my presentation?\nSpeaker F: I understand which.\nSpeaker F: We should have a general idea of how it's going to look.\nSpeaker C: Well, we are here now, I think.\nSpeaker C: These I've already given you, so we have to decide on the different remote control functions.\nSpeaker C: So we want to have a small LCD screen.\nSpeaker C: That's special.\nSpeaker B: Shouldn't we start with the most important part?\nSpeaker B: LCD screen.\nSpeaker B: Alright, we should start with the power button.\nSpeaker B: Yeah.\nSpeaker F: How do you zap?\nSpeaker F: You just sit in your chair.\nSpeaker A: Huh?\nSpeaker E: With the LCD screen on top, it gets a bit terrible.\nSpeaker E: It's terrible.\nSpeaker E: Because most remote have some space left at the bottom.\nSpeaker F: Yeah, but that's where your hand form might be.\nSpeaker F: I don't know.\nSpeaker B: And then...\nSpeaker F: We'll try to...\nSpeaker B: Maybe we should centralize the discussion here.\nSpeaker B: I don't know what you're talking about.\nSpeaker F: We're basically with something.\nSpeaker F: Yeah, right.\nSpeaker F: He thinks it's better to put LCD at the bottom.\nSpeaker B: Why do you think it's better at the bottom?\nSpeaker E: Well, because most remotes have some space left at the bottom.\nSpeaker E: That way you can keep the shape recognizable.\nSpeaker E: Sure.\nSpeaker B: But you just can put the whole interface a bit down to the test.\nSpeaker B: Really?\nSpeaker F: I think that the LCD is like small.\nSpeaker F: It's white.\nSpeaker F: It's not high.\nSpeaker F: Well, I think Mike has a point.\nSpeaker C: When I use a remote, I hate the buttons at the bottom.\nSpeaker C: Bottom.\nSpeaker C: Bottom.\nSpeaker C: And I like to use them once.\nSpeaker A: On the top.\nSpeaker C: When I have to...\nSpeaker C: Well, I agree with you.\nSpeaker C: It's also more recognizable.\nSpeaker B: It looks more like a calculator to people.\nSpeaker B: If you have the...\nSpeaker B: Yes, but we don't want that.\nSpeaker B: So, I think that's a good thing.\nSpeaker B: I think that's a good thing.\nSpeaker B: You have the...\nSpeaker C: We don't want that.\nSpeaker C: We don't want them to look like a calculator.\nSpeaker C: It must be a remote.\nSpeaker C: Like our original.\nSpeaker B: Well, you don't have to throw an important aspect, like familiarity completely away.\nSpeaker C: I think it's a bit lighter.\nSpeaker B: I think it's still important to have it at the top.\nSpeaker F: It's more familiar that way.\nSpeaker F: When I draw here, it's a bit off.\nSpeaker C: A little bit.\nSpeaker C: It needs to be calibrated.\nSpeaker C: Let's talk about that later.\nSpeaker B: Maybe you should not append.\nSpeaker B: Maybe that's better.\nSpeaker B: It's special now.\nSpeaker F: We have to make a decision now.\nSpeaker F: We don't have much time.\nSpeaker F: I think we have a few functions.\nSpeaker F: We can put the LCD above it.\nSpeaker F: We still have lots of room at the bottom.\nSpeaker C: I think it should be at the bottom.\nSpeaker C: I'm the designer.\nSpeaker E: Oh, yeah.\nSpeaker F: I just agree.\nSpeaker F: We're too fierce too.\nSpeaker C: What if we first decide the different functions\nSpeaker D: and then look at the design?\nSpeaker B: We were busy with that. We should summon the different aspects of the thing.\nSpeaker B: We have the power button.\nSpeaker F: I think that you two should come to consensus about the LCD.\nSpeaker F: Yes.\nSpeaker F: It's your job.\nSpeaker F: Yeah, of course.\nSpeaker D: Yeah.\nSpeaker D: We have a power button, guys.\nSpeaker B: Are you have to agree?\nSpeaker B: I can say it's like this.\nSpeaker B: I agree.\nSpeaker B: We have a power button,\nSpeaker C: a setting button, LCD window, the number buttons.\nSpeaker B: Yeah. Yeah.\nSpeaker B: Volume.\nNone: Volume control.\nSpeaker B: Yes.\nNone: Let's look at the data.\nSpeaker B: Let's look at your design.\nSpeaker C: The mute button.\nSpeaker C: I love that one.\nSpeaker E: I think we should use something like this.\nSpeaker E: To the general up and general down button.\nSpeaker E: Yes.\nSpeaker E: Yeah.\nSpeaker E: Well, that's the exposure design.\nSpeaker E: The design's role also.\nSpeaker B: On this remote, these controls are for something else.\nSpeaker E: Yes, they are for some video drops on these buttons.\nSpeaker F: You should put that power button channel and volume should have the most importance.\nSpeaker E: Yeah.\nSpeaker E: I think this should be in one big circle in the middle.\nSpeaker F: Yeah, but what he said about RZ was kind of true.\nSpeaker F: When you put them all in the same place, the most used buttons, you're doing the same thing all the time.\nSpeaker F: That's just what RZ is all about.\nSpeaker F: So it might be smarter to put them a little more apart away from each other.\nSpeaker C: So people have to move their hand and they get less complaints of RZ.\nSpeaker B: Yeah, it's also not good to completely stay in one position constantly when one ends.\nSpeaker B: That's what I always do because all my important buttons are in the same place.\nSpeaker B: Good to move from time to time.\nSpeaker E: Yes.\nSpeaker E: People don't like it when their buttons are all over the place.\nSpeaker B: No, no, it would be sent.\nSpeaker B: We have to choose a frequency for button use.\nSpeaker F: Yes.\nSpeaker F: Channels are most used within the hour.\nSpeaker F: You can see the channel buttons should be far apart.\nSpeaker F: I think.\nSpeaker F: Open down.\nSpeaker F: I've been down far apart from each other.\nSpeaker F: You're thinking about RZ.\nSpeaker F: Well, not too much.\nSpeaker A: Look at that.\nSpeaker E: Not too much, but the other two frustrations are far more important.\nSpeaker B: I think you can have the two buttons up and down close together, but you don't have to have volume control.\nSpeaker C: Well, for example, the power button.\nSpeaker E: You can, well, they're used for time to go away.\nSpeaker C: If someone is constantly zapping, it's not going to miss that the power button is not right beside because someone...\nSpeaker C: I don't need to be...\nSpeaker C: So that one can be put away.\nSpeaker F: Yeah, right.\nNone: Yeah.\nSpeaker C: The power button can be...\nSpeaker E: Power button should be left at the top.\nSpeaker C: Yes.\nSpeaker C: Oh, man, five minutes.\nSpeaker C: Yeah.\nSpeaker F: Well, five minutes left.\nSpeaker F: Just make some decisions.\nSpeaker F: Yep.\nSpeaker F: The most important things we have to...\nSpeaker C: How are we going to do it with the game makers?\nSpeaker B: We can use the drawing board now.\nSpeaker B: I think it doesn't work well, but...\nSpeaker B: Well, I think it's pretty nice if we could just draw a simple...\nSpeaker B: I do think we have to take two seconds.\nSpeaker B: You have to take two seconds.\nSpeaker F: You have to take two seconds.\nSpeaker F: You decide that, you decide that.\nSpeaker F: Yes.\nSpeaker F: All right.\nSpeaker C: Well, the LCD...\nSpeaker C: You are industrial.\nSpeaker C: You are user interface.\nSpeaker C: So I think it's going to go to Mike.\nSpeaker C: But you will have to make consensus with...\nSpeaker C: Well...\nSpeaker E: Well, consensus.\nSpeaker E: Well, it's a bit hard because we're going to be individually.\nSpeaker C: That's a bit...\nSpeaker C: We're deciding now.\nSpeaker F: Yes.\nSpeaker F: Top of the bottom.\nSpeaker B: Well, yeah, I still think it's quite important.\nSpeaker B: Yes, it is.\nSpeaker B: We have it at top, so...\nSpeaker D: You say familiarity is important, but...\nSpeaker C: Well, what if we're going to now decide about the functions and the design comes into the next round?\nSpeaker C: OK.\nSpeaker C: But the design round is still to come, eh?\nSpeaker C: All right.\nSpeaker C: All right.\nSpeaker C: All right.\nSpeaker C: These functions...\nSpeaker C: We agree we do have LCD, so...\nSpeaker C: Yes.\nSpeaker C: That's all right.\nSpeaker C: That's...\nSpeaker C: The number function.\nSpeaker C: That's enough.\nSpeaker C: Are we going to do it like on Mike's screen with one button that says I'm going to do two numbers.\nSpeaker C: Two number of digital?\nSpeaker E: No.\nSpeaker C: How do you want it to do then?\nSpeaker C: There's one to three, four, five, six.\nSpeaker C: It has to recognize one as...\nSpeaker C: There can still come more.\nSpeaker E: No, if you...\nSpeaker E: Almost...\nSpeaker E: If you press two numbers shortly after each other.\nSpeaker D: Yes.\nSpeaker D: It recognizes it.\nSpeaker D: So no button for that.\nSpeaker C: That's very easy.\nSpeaker C: No.\nSpeaker C: OK.\nSpeaker C: Anyone, any other functionalities of our...\nSpeaker B: Remote, I think these are the most important things you'll have to picture over there.\nSpeaker B: Yeah.\nNone: Show.\nSpeaker F: Yeah, that's right.\nSpeaker F: You do need the...\nSpeaker F: Milk or...\nSpeaker F: No.\nSpeaker F: Like this one.\nSpeaker F: You doing it?\nSpeaker F: No, we didn't.\nSpeaker C: No, we didn't.\nSpeaker F: No, we didn't.\nSpeaker F: Are we?\nSpeaker C: All right.\nSpeaker C: All right.\nSpeaker C: No, OK.\nSpeaker C: Well, because we can integrate it with any other...\nSpeaker C: Remotes all those buttons on those pictures are irrelevant.\nSpeaker C: Right.\nSpeaker C: So just for a television, is that all we need?\nSpeaker F: Yeah, it's most useful this way.\nSpeaker B: It is heavy-breathing function.\nSpeaker C: The A-text is gone.\nSpeaker C: So all those buttons that are to do with A-text...\nSpeaker C: Oh, screen placing.\nSpeaker C: You have those buttons about...\nSpeaker C: And there's two important buttons we're forgetting.\nSpeaker C: There's a screen.\nSpeaker C: You can make it wider and less wide.\nSpeaker C: And the button that you can go to for your video.\nSpeaker F: Right.\nSpeaker F: I do think we have to put that underneath the clip.\nSpeaker D: Those two, but it's just two.\nSpeaker D: And we make a...\nSpeaker B: Just two under a...\nSpeaker B: That's a bit...\nSpeaker E: We can make a little row of like four buttons down here.\nSpeaker D: Yes.\nSpeaker D: Or...\nSpeaker C: At the top.\nSpeaker C: Your LCD scan is going to go.\nSpeaker B: You can put two or three buttons under another section.\nSpeaker B: That's too complicated.\nSpeaker B: You can just put it somewhere.\nSpeaker B: They aren't used much.\nSpeaker B: Not as much as those other.\nSpeaker B: You can put it somewhere...\nSpeaker B: They can be smaller around.\nSpeaker B: Yeah, a bit smaller.\nSpeaker B: And I think more at the bottom.\nSpeaker B: Yeah.\nSpeaker B: Or at the top.\nSpeaker B: What do you think?\nSpeaker B: Those buttons.\nSpeaker B: About four.\nSpeaker B: I think they should.\nSpeaker C: In an isolated part of the remote.\nSpeaker B: And where?\nSpeaker B: Well, we have it.\nSpeaker B: And we design later where everything goes.\nSpeaker C: Yes.\nSpeaker C: Well, any other...\nSpeaker F: Well, if you...\nSpeaker C: Why don't you video?\nSpeaker C: If you...\nSpeaker C: Go to video.\nSpeaker C: That's always on your remote control.\nSpeaker C: Yeah, but you have the...\nSpeaker C: You have the...\nSpeaker C: The video channel.\nSpeaker F: The 80.\nSpeaker A: That's just zero.\nSpeaker A: Zero.\nSpeaker E: No, not always.\nSpeaker B: Not a...\nSpeaker B: Not a...\nSpeaker B: Zero is a different channel than the video channel.\nSpeaker B: Yeah, but you can...\nSpeaker F: Yeah, but when you step down zero, you get to AFI.\nSpeaker F: Yes, I think that...\nSpeaker C: I don't.\nSpeaker C: I go to 99.\nSpeaker C: Then you press 99.\nSpeaker E: But I think go to video with...\nSpeaker E: That's the button.\nSpeaker B: But it's easy to go...\nSpeaker B: If you are at a channel 55 and you want to go immediately to the video channel, you have to push A to...\nSpeaker B: Now you can go to zero.\nSpeaker B: Zero.\nSpeaker B: More easy to get to...\nSpeaker D: Yeah.\nSpeaker C: But the specific...\nSpeaker C: A video channel.\nSpeaker C: I use it.\nSpeaker F: Yeah, but if we're choosing to incorporate these buttons, you have to have a channel setting.\nSpeaker F: If you have a new TV, you have to set the channels.\nSpeaker F: All these different buttons.\nSpeaker C: Even though I don't have buttons for channels, especially on my remote.\nSpeaker C: Yeah, well, it's different.\nSpeaker C: More great settings.\nSpeaker C: Plus and down.\nSpeaker C: Now you...\nSpeaker E: You have one button from...\nSpeaker E: Set frequency or something.\nSpeaker E: And with plus and minus, you can adjust the...\nSpeaker F: That's the only one we put in there.\nSpeaker C: So we should have one for...\nSpeaker C: For screen...\nSpeaker F: Channel setting.\nSpeaker F: Yeah, to...\nSpeaker B: Yeah, of course, to configure the...\nSpeaker B: Oh, okay button.\nSpeaker C: Oh, you're always in the center, an okay button.\nSpeaker C: Yeah.\nSpeaker C: For your menu.\nSpeaker C: Menu button.\nSpeaker C: You should...\nSpeaker B: Yeah.\nSpeaker B: I think that's important.\nSpeaker B: Channel setting...\nSpeaker B: Blue.\nSpeaker B: And settings to change the brightness.\nSpeaker B: The settings to change the brightness and the contrast.\nSpeaker B: Okay.\nSpeaker B: Channel.\nSpeaker D: Yeah.\nSpeaker C: So...\nSpeaker C: G.\nSpeaker D: Safe.\nNone: G.\nSpeaker B: Yes, say.\nSpeaker B: The menu button is also important.\nSpeaker B: Yes, I have to put it in.\nSpeaker B: Well...\nSpeaker B: I don't see it.\nSpeaker B: Here.\nSpeaker E: Oh, menu.\nSpeaker E: I think things like contrast and brightness should be...\nSpeaker E: In the menu.\nSpeaker E: In the menu, yeah.\nSpeaker B: And you can adjust it with the second buttons or something.\nSpeaker B: Yeah, the...\nSpeaker B: Because the second buttons are used then, if you are in the menu.\nSpeaker C: Guys, we're going to...\nSpeaker C: We're going to go to our rooms and we'll have to decide our things on our own, I think.\nSpeaker C: Great.\nSpeaker C: So, Chief.\nNone: All right.\nSpeaker F: Well, see you.\nSpeaker F: We have lunch time, by the way.\nSpeaker C: Now...\nSpeaker F: I'm hungry.\nSpeaker B: Lunch.\nSpeaker F: Gotten to my belly.\nSpeaker C: We didn't exactly do everything, man.\nSpeaker C: And how many pows is sticking to?\nSpeaker E: See you later, right?\nSpeaker C: So...\nSpeaker C: We don't have any clothes in there.\nSpeaker C: We have clothes in here.\nSpeaker C: Hey?\nNone: Mm-hmm.\nSpeaker C: We have clothes in here.\nSpeaker C: We have a presentation in here.\nNone: We have a presentation in here.\n"}, {"meeting_id": "TS3005a", "summary": "A new remote control project was introduced by Project Manager. Members made self-introductions and began to know each other by guessing animals others drew. Project manager said that the remote control would be sold at 25 Euros and a production cost of 12.5 Euros. Project Manager also hoped to achieve fifty million Euros and international market by new remote control. Then the team discussed the disadvantages of the remote control and designed a new remote control style. Industrial Designer made a desire to design a remote control that can be used in different devices with most used buttons. Project Manager agreed that rather than complicated buttons, friendly to users and minimalist design with necessary buttons would be better.", "dialogue": "None: On the road turn on We're getting by now I can unexplainable This place seems to be different If we Laf Quadretta happened Good morning\nSpeaker D: Good morning.\nNone: Thank you, I have you a chip over there man. Cotto?\nSpeaker C: I have some sushi.\nNone: Yeah, its 1977.\nSpeaker A: Sh Good morning. I'd like to welcome you to our first meeting. I've prepared a little presentation.\nSpeaker A: My name is Marim Plonf and I hope you will introduce yourself in a few minutes as well.\nSpeaker A: I'm the project manager of this project and well I will tell you fur on what actually is the project.\nSpeaker A: This is the agenda for our first meeting. This is the opening. Then we will get, I will hope we will get acquainted with each other.\nSpeaker A: We'll do a little tool training with these two things. We'll take a look at the project plan.\nSpeaker A: There will be time for discussion. Actually we have to discuss because we have to create a product and then we will close this session.\nSpeaker A: First of all, I'd like to introduce you to this room. As you probably have noticed there are little black fields on the table.\nSpeaker A: You have to put your laptop exactly in that field so the little cameras can see your face.\nSpeaker A: There are cameras everywhere around the room especially here for your face.\nSpeaker A: This isn't a pie. It's a set of microphones and there are microphones here also. Please don't be afraid of them. They won't hurt you.\nSpeaker A: I'm the project manager and I'm hoping for a good project. I'd like to hear who you are and what your functions are on this project.\nSpeaker D: Let's start with the ladies. I'm Imkriel and my function is user interface design. I'm a industrial designer and I hope to look forward to a very pleasing end of this project.\nSpeaker C: My name is Danny Magrale. I'm a marketing expert. My job is in the company to promote products to the customers.\nSpeaker C: I also hope we have a pleasant work with each other.\nSpeaker A: We have some expertise from different pieces of the company. That's good.\nSpeaker A: As said, we're working on a project and the aim for that project is to create a new remote control which has to be original, trendy and user friendly.\nSpeaker A: I hope we have the expertise to create such a project.\nSpeaker A: The way we hope to achieve that is through the following methods. It consists of three phases, namely the functional design, conceptual design and detailed design.\nSpeaker A: As you can see, all of these phases consist of two parts, namely individual work part and a meeting where we will discuss our work so far.\nSpeaker A: But first, I will tell you something about the tools we have here. I already talked about the cameras and microphones, but they are not much use to us.\nSpeaker A: We will have to take advantage of these two things. They are smart boards. As you can see, you can give a presentation on them.\nSpeaker A: This one here is a white board. I will instruct you about that soon.\nSpeaker A: As you also noticed, this presentation document is in our project folder. Every document you put in this folder is possible to show that here in our meeting room.\nSpeaker A: There are available on both smart boards, but I think we will use this one for the document in the shared folder.\nSpeaker A: As you can see, this is the same toolbar as is located here. The most functions we will use will be to add a new page to go back and forward between pages and, of course, to save it every now and then.\nSpeaker A: This is the pen with which you can draw on the board. For instance, like this, everything is OK, but I have to put it on the pen.\nSpeaker A: You can write things like test or whatever you want. As you can see, you have to move it a little bit slow. It is not such a fast board. It is a smart board, but also a slow board.\nSpeaker A: You can write things, and, of course, you can also, when you click here, erase things, so we have asked left.\nSpeaker A: You can also delete an entire page, but we ask you not to do that. Just simply create a new one and start all over because we want to save all the results.\nSpeaker A: Does everyone understand this?\nSpeaker A: You can erase anything. You can erase it with the eraser, but you shouldn't delete an entire page.\nSpeaker A: You can delete a blank one, and I will delete this one now because we don't use it yet.\nSpeaker A: You can, of course, erase when you make a mistake, but don't delete entire pages.\nSpeaker A: Also, let's see, I think it's here, change the color of your pen. For instance, take a blue one and change the line with 5.\nSpeaker A: And that's what you will need for our first exercise because I'm going to ask you to draw your favorite animal.\nSpeaker A: It's also to guess to know each other because I'm asking three things for that drawing. To do it on a blank sheet with different colors, and I just showed you how to pick a color, and also with different pen widths, which I also showed you.\nSpeaker A: And favorite characteristic can be just one word.\nSpeaker A: Well, I'm not very good at drawing, but I will go first and try to draw.\nSpeaker A: Or maybe you should guess what I'm drawing.\nSpeaker B: It's a sheep.\nSpeaker D: It's a silver.\nSpeaker D: A silver.\nSpeaker D: A beaver.\nSpeaker A: Well, it's weird.\nSpeaker A: It could be everything.\nSpeaker A: Maybe when I put on this thing, it could be a turtle snail.\nSpeaker A: Or a snail.\nSpeaker A: But the snail doesn't have legs.\nSpeaker B: The turtle has.\nSpeaker A: And those are slow.\nSpeaker A: And I hope our project group will not be slow, but we will work to a good result and do it as fast as we can.\nSpeaker A: Okay.\nSpeaker A: Time for another animal. Would you like to go next?\nSpeaker C: It was forward.\nNone: That is so good.\nSpeaker B: Oh.\nSpeaker D: That helps.\nSpeaker C: To make it a little bit easier.\nSpeaker B: It's not that cute.\nSpeaker B: Or a dinosaur.\nSpeaker C: Okay.\nSpeaker C: I think it's easy to recognize.\nSpeaker C: Yeah.\nSpeaker C: The favorite characteristic is the long neck.\nSpeaker C: It can reach everything.\nSpeaker C: I can also reach a lot with this project.\nSpeaker A: That's my favorite animal.\nSpeaker A: Anything else you need to know?\nSpeaker B: I can draw with this.\nSpeaker D: Wow.\nSpeaker D: It's a mouse.\nSpeaker B: Okay.\nSpeaker B: You can guess what it is, I hope.\nSpeaker B: It's a rabbit.\nSpeaker B: Well, it's a quick, I guess.\nSpeaker B: That's my favorite animal.\nSpeaker A: Thank you.\nSpeaker D: Okay.\nSpeaker D: I've drawn a dolphin because of its intelligence, one of the most intelligent animals in our world.\nSpeaker D: I can see we have some drawing sounds.\nSpeaker A: Well, nice animals, nice words.\nSpeaker A: Sounds good.\nSpeaker A: Back to business, back to the money part.\nSpeaker A: From the finance department, I've learned that we are aiming for a selling price of 25 euros and we're hoping for a aim of 50 million euros and we're hoping to achieve that by aiming for an international market.\nSpeaker A: And the production costs will be 12 euro 50 max.\nSpeaker A: Okay, well, it's time for some discussion.\nSpeaker A: I've wrote down some examples here of what we can speak about, what's your experience with remote controls, what kind of ideas do you have to design a new remote control, maybe, for which market segments should be aimed or should be aimed for all segments.\nSpeaker A: Well, actually, I'd like to hand the word back to you.\nSpeaker A: What's your experience with remote control?\nSpeaker B: A lot of buttons.\nSpeaker D: Which you don't use.\nSpeaker D: Complex.\nSpeaker D: Complex.\nSpeaker D: Not user friendly.\nSpeaker B: You search for the buttons, which ones which.\nSpeaker C: Boring.\nSpeaker C: Boring.\nSpeaker A: It's not fun to use in remote.\nSpeaker A: Black or black.\nSpeaker D: So, black colors.\nSpeaker A: Yeah.\nSpeaker A: Or maybe we should try to make it fun.\nSpeaker D: We use batteries and last batteries and poor signal.\nSpeaker D: The angle is to use.\nSpeaker B: Different remote control.\nSpeaker B: Different remote control.\nSpeaker B: Different remote control.\nSpeaker B: We can integrate them or something.\nSpeaker D: For the use of different devices.\nSpeaker B: Your stereo and your TV and the...\nSpeaker B: That's nice.\nSpeaker C: Yeah, but then again, you still have a lot of buttons.\nSpeaker C: Yeah, that's right.\nSpeaker D: Yeah, but I think there's a possibility to put those buttons behind some kind of protection.\nSpeaker D: So, if you only get serum when you need them.\nSpeaker C: Yeah, okay.\nSpeaker C: It's possible.\nSpeaker C: So, do you only get...\nSpeaker D: No, no, no, no.\nSpeaker B: You just keep it tomorrow.\nSpeaker D: For example, you get the same size remote control you use every day.\nSpeaker D: And the usual buttons such as zapping, which is called a Dutch.\nSpeaker D: And the following is the only possible buttons to use directly.\nSpeaker C: Or the numbers.\nSpeaker D: And the numbers, of course.\nSpeaker D: But not the buttons used to search on the channels on your television.\nSpeaker D: You're going to use those the first time.\nSpeaker D: Play-balls.\nSpeaker A: So, maybe a minimalist design.\nSpeaker A: Yeah, I think so.\nSpeaker B: Of buttons.\nSpeaker B: But you should make sure that you have every button they need on it.\nSpeaker B: That's a...\nSpeaker B: Thanks for a...\nSpeaker B: Tailor text.\nSpeaker B: I don't know that.\nSpeaker B: That's a name.\nSpeaker D: Tailor text.\nSpeaker D: Thanks.\nSpeaker D: Okay.\nSpeaker A: So, you don't want to bother people with loads of buttons, but on the other hand, they need many buttons, so they don't have to get out of their seat.\nSpeaker A: Yeah.\nSpeaker C: I think the market will be all kinds of people.\nSpeaker C: All of the young people.\nSpeaker B: But if it's international, you should look in Britain, they have different things they can do with TV.\nSpeaker B: Or so that you can choose what you want to see.\nSpeaker B: I don't know if you should take that into consideration.\nSpeaker B: It just should aim for the normal TV.\nSpeaker B: Yeah, I think that's better.\nSpeaker C: I think if you're going to target a lot of people in the whole world and only Britain, then I think the cost will rise higher than 1250, I think.\nSpeaker B: Yeah, I don't know if the...\nSpeaker C: It was better to use the whole world in Britain.\nSpeaker C: Yeah.\nSpeaker B: No, it could be...\nSpeaker D: I think the main idea of this remote control is to make it user-friendly.\nSpeaker D: So I think when the customers will buy this remote control, they already have the remote control, which companies...\nSpeaker D: With the standard remote control, which comes with the television.\nSpeaker D: So it only has to have the most used buttons.\nSpeaker D: You don't have to integrate the buttons to search the channels on your television in those...\nSpeaker D: Oh, but then you have to find your other remote control...\nSpeaker B: Yeah.\nSpeaker D: Yeah, but it's impossible to accommodate all the buttons on the different televisions set on one remote control.\nSpeaker D: It's impossible.\nSpeaker D: For example, Sony Television has the opportunity to make possible for it to see on one side of the screen, teletext, on the other side, just a regular television.\nSpeaker C: I think most television nowadays do this.\nSpeaker D: Yeah, but they don't use the same signal on remote control, because you can use a Panasonic remote control on a phillips television.\nSpeaker C: Yeah, but then you have to choose the...\nSpeaker C: It's always a universal control mode.\nSpeaker C: You have to choose the code.\nSpeaker C: I'm choosing the code.\nSpeaker C: I guess which type of television you have.\nSpeaker C: There's no problem.\nSpeaker C: I think the two pages on the same screen, like teletext and normal television, that's nowadays standard.\nSpeaker D: I think most people...\nSpeaker D: People will buy their remote control because they lost the first one.\nSpeaker D: The first one is broken, so...\nSpeaker D: Perhaps they haven't got an older television, so that option is not...\nSpeaker D: Yeah, definitely optional for those people.\nSpeaker B: Then people have a new television, and if you look into the future, then they want to...\nSpeaker B: It's very beautiful.\nSpeaker D: So we should take that in consideration.\nSpeaker D: Mm-hmm.\nSpeaker A: Okay, well, any more ideas?\nSpeaker D: No.\nSpeaker D: No.\nSpeaker A: No, I think they'll come up.\nSpeaker A: Yeah.\nSpeaker A: Okay, well, we have some time.\nSpeaker A: Let's see.\nSpeaker A: One more I have to tell.\nSpeaker A: I don't think there's much left.\nSpeaker A: Nope, we're starting to close.\nSpeaker A: Our next meeting will start a little bit early, but our next meeting will start in 30 minutes.\nSpeaker A: In the meantime, there's time for some individual actions.\nSpeaker A: As you can see, the different roles have different tasks.\nSpeaker A: And there's a ping.\nSpeaker A: Is it my laptop?\nSpeaker D: Yep.\nSpeaker D: Stop the meeting.\nSpeaker A: Oh, the meeting will come up in five minutes.\nSpeaker A: Five minutes in the meeting, so we're right on schedule.\nSpeaker A: The marketing expert will take a look at the user requirements specification.\nSpeaker A: The user interface designer will work out the technical functions design.\nSpeaker A: And this was the interface designer or the interaction designer?\nSpeaker A: No, interface designer.\nSpeaker A: Okay, first guess was right.\nSpeaker A: We'll take a look at the working design.\nSpeaker D: No, the industrial designer will take a look at the working design.\nSpeaker D: Any usability interaction?\nSpeaker D: Industrial design.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Let's just use the acronyms.\nSpeaker A: And of course, specific restrictions will be sent to you for your personal coach.\nSpeaker B: So I should look at what you should be able to do with your mouse or how?\nSpeaker A: Well, those instructions will be in the email you will receive shortly, I hope.\nSpeaker B: Me too.\nSpeaker A: And of course, you have your own expertise.\nSpeaker A: Well, that was what I had to say.\nSpeaker A: Are there any more questions?\nSpeaker A: No.\nSpeaker D: No?\nSpeaker A: Okay.\nSpeaker A: Well, I think then we have to head back to our offices and start working.\nSpeaker A: I have a more question.\nSpeaker C: One question?\nSpeaker C: Where does it says we have to make a remote because I pursued.\nSpeaker C: Okay, go.\nSpeaker C: I didn't know who.\nSpeaker C: Okay, no problem.\nSpeaker C: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I'll expect everything will be much clearer with the instructions we will receive shortly.\nSpeaker A: Okay.\nSpeaker A: Well, I'll see you all in about 30 minutes then.\nSpeaker A: Okay.\nSpeaker C: Yeah.\nSpeaker C: Good.\nNone: That's very bad.\nNone: Yeah.\nNone: I can't do this, but I should do it.\nNone: I need to.\nNone: Okay.\n"}, {"meeting_id": "ES2016a", "summary": "Project Manager introduced a new remote control project. The team started to know each other by introducing their roles in this project. Project Manager then invited everyone to raise ideas about the remote. Their discussion included its versatility, anti-loss design and possible appearance, but there was no final decision on these matters. The team agreed that the remote control should be one-handed, user-friendly and globally attractive. In addition, they noticed that these ideas should be realized under the price target and balance the stability and design sense of the remote control.", "dialogue": "None: Okay.\nNone: Oh, that's not going to work.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker C: Okay.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: All right.\nSpeaker D: As you all know, we're here to create a brand new fantastic remote.\nSpeaker D: I'm Nick Debusk.\nSpeaker D: I'm the project manager.\nSpeaker D: We'll just get started with everyone kind of letting each other know who they are and what you're doing and what your role is.\nSpeaker D: Go ahead.\nSpeaker A: Okay.\nSpeaker A: I'm Corinne Weiden and I will be the marketing expert and in each of the three phases, I will have a different role in the function design phase.\nSpeaker A: I will be talking about user requirements specification and this means what needs and desires are to be fulfilled and I'll be doing research to figure this out.\nSpeaker A: In the conceptual design phase, I will be dealing with trend watching and I'll be doing marketing research on the web.\nSpeaker A: And then finally in the detailed design phase, I will be doing product evaluation and so I will be collecting the requirements and ranking all of the requirements to see how we did.\nSpeaker D: Okay.\nSpeaker C: Hi, I'm Ryan.\nSpeaker C: I'm the user interface designer.\nSpeaker C: Likewise, I have three different roles for each stage of the design.\nSpeaker C: The functional design is looking at the technical functions of remote control.\nSpeaker C: In the concept design, the user interface, how the user reacts with the product and the detailed design, sort of like the user interface design, what they might be looking for, things like fashions, what makes, how we're going to make it special.\nSpeaker C: That's about it.\nSpeaker B: I'm Manuel.\nSpeaker B: I'm the industrial designer in this project.\nSpeaker B: In the functional design phase, I'll be dealing mostly with the requirements.\nSpeaker B: We'll discuss what functions the product has to fulfill and so on and so on.\nSpeaker B: I suppose we'll work pretty much together on that one.\nSpeaker B: In the conceptual design, I'll be mostly dealing with properties and materials of our product.\nSpeaker B: The detailed design and the detailed design, I'll be concerned with the look and feel of the product itself.\nSpeaker B: Pretty much working together, obviously, on the design.\nSpeaker B: Okay.\nSpeaker D: We've got our opening, our agenda is the opening acquaintance, which we've kind of done.\nSpeaker D: Tool training, project, plan, discussion, and then closing.\nSpeaker D: Grant tool of 25 minutes we have here.\nSpeaker D: We are putting together a new remote control.\nSpeaker D: We want it to be something original.\nSpeaker D: Of course, we're not only an electronics company, but a fashion conscious electronics company.\nSpeaker D: So we want it to be trendy.\nSpeaker D: And we want it to be easy to use.\nSpeaker D: We've got the functional design, conceptual design, and detailed design, which basically is the three of you.\nSpeaker D: Well, functional design.\nSpeaker D: Do we have any ideas of maybe, let's just throw out some ideas of what kind of remote control we want to have?\nSpeaker D: Then we can go into how we're going to design it and how we're going to do the detailing\nSpeaker C: on it. Well, functional remote control is just a change of channels.\nSpeaker C: It's a main function.\nSpeaker D: So we want it to be a TV remote or do we want it to do other things besides just be a television remote?\nSpeaker C: I suppose trying to make it a universal remote can work on all sort of electrical products in one person's house.\nSpeaker C: But they all sort of have the same role changing channels, volumes, and then programming.\nSpeaker C: I think they all work on the same principles.\nSpeaker C: I don't actually know.\nSpeaker C: Tyler, but is it just infrared?\nSpeaker C: Is that sent out?\nSpeaker D: I think, yeah, universal remote.\nSpeaker D: This is my first go around with creating remote control.\nSpeaker D: So I think we're all in the same boat here.\nSpeaker C: One thing I thought of when we were in remote control is you always lose them.\nSpeaker C: Okay.\nSpeaker C: So I have those away I find them quite easily.\nSpeaker D: So we should fit you.\nSpeaker D: But our remote control up to where it has a tracking device.\nSpeaker A: Like a tracking device.\nSpeaker D: Or you get those.\nSpeaker C: Or like a...\nSpeaker C: You whistle.\nSpeaker C: I'll make a noise.\nSpeaker D: It makes a noise.\nSpeaker D: There's a button on the TV that you press.\nSpeaker A: Okay.\nSpeaker B: Do you get it?\nSpeaker C: Genuinely all remote was a sort of quite similar in their appearance.\nSpeaker C: Yeah.\nSpeaker D: It's just long.\nSpeaker D: So they're kind of like long and rectangular.\nSpeaker A: We want something crazy.\nSpeaker D: We want something new that's going to stand out.\nSpeaker D: I think so.\nSpeaker D: A modern.\nSpeaker D: So our remote should be.\nSpeaker C: It's very close to a ball.\nSpeaker A: Maybe like user friendly like a little.\nSpeaker A: Yeah.\nSpeaker A: You know where you can use both hands like a little keyboard.\nSpeaker C: People, I thought maybe because people always sent a thorough remote control about the place to one other.\nSpeaker C: If it was in a ball and you got your controls on the side.\nSpeaker B: Well there are of course certain restrictions.\nSpeaker B: You can't have it be any form and fulfill all functions at the same time.\nSpeaker B: So there are obviously some restrictions we have to apply here.\nSpeaker B: However, one question is how stable is that thing supposed to be that refers to the material pretty much?\nSpeaker B: What are we going to build the thing out of?\nSpeaker B: How sturdy is it going to be to be one to it to last longer or rather have people.\nNone: Whatever.\nSpeaker B: Okay.\nSpeaker B: So buy one.\nSpeaker B: Yeah.\nSpeaker D: So we want to be sturdy.\nSpeaker D: We wanted to hold up to somebody's child throwing it across the room or as you said people kind of throw it.\nSpeaker D: So ball shaped.\nSpeaker D: You know, if it were ball shaped, maybe kind of cast on the outside.\nSpeaker D: Okay.\nSpeaker D: So we want it to be modern, fun, sturdy.\nSpeaker D: So our form and our function.\nSpeaker D: We want it to be easy to find.\nSpeaker D: What else do we want it to do?\nSpeaker D: So we want it to be universal.\nSpeaker D: It's something that we're supposed to sell for about 25 euros.\nSpeaker D: It's goals for profits are I think somewhere around 50 million euros what they want to make on it.\nNone: So.\nSpeaker A: Also, since we're partners of the International Remote Control Association, maybe we want to make it something that would globally appeal.\nSpeaker D: Okay.\nSpeaker A: That's more on the research end, but the marketing.\nSpeaker D: The marketing, you know, how maybe marketing you could find out what is the most universally appealing remote control out there.\nSpeaker A: Yeah.\nSpeaker A: And maybe as far as design goes, maybe we could have different ones for different target audiences.\nSpeaker A: Okay.\nSpeaker A: Because maybe one won't apply to all of the countries we're targeting.\nSpeaker B: All right.\nSpeaker D: You guys have any ideas for what it should look like?\nSpeaker D: Maybe we could draw it up on the board over there.\nSpeaker D: Some ideas.\nSpeaker D: We want to be a job ball drop the ball and maybe where the buttons are located.\nSpeaker C: Or my original idea was just simply sort of a sphere where maybe you see where it's connected together.\nSpeaker C: And then when you open it out, it could be a flip, like a flip film.\nSpeaker C: And then when you fold it out the middle, maybe a hinge, that would be the strongest part of it.\nSpeaker C: And then if you did use a hinge, it was just two parts.\nSpeaker C: And then you'd have just like any buttons.\nSpeaker C: Things inside.\nSpeaker C: I think sometimes you're amongst too many buttons.\nSpeaker C: So maybe as simple as possible.\nSpeaker C: There's a few of those inside as possible.\nSpeaker C: And then there was the idea for just something maybe if it had like some kind of light or something or lights around it, it's looking a bit like some kind of Star Wars at the moment, no?\nSpeaker C: To be fair.\nSpeaker C: But yeah.\nSpeaker C: That was a sort of simple idea.\nSpeaker C: I had any note you could feel it about.\nSpeaker C: It would almost be like a ball.\nSpeaker C: So that was just the idea I had.\nSpeaker C: I don't have any of the others or the ideas.\nSpeaker B: Right.\nSpeaker B: One problem you'd get with this design is the ball is a nice idea because of its stability.\nSpeaker B: But of course, since it's a ball, it'll roll.\nSpeaker B: So you'd have to have it flat on the side of the beast down here somewhere.\nSpeaker B: Take away that part.\nSpeaker B: That's one of the big issues.\nSpeaker B: Also you risk the hinges here.\nSpeaker B: That's a problem.\nSpeaker C: They didn't have that shallow beef.\nSpeaker C: That's an idea.\nSpeaker B: That's interesting, of course.\nSpeaker B: But that's of course a weak point.\nSpeaker C: Yeah.\nSpeaker D: How would we go about making, you know, getting rid of our weak points?\nSpeaker D: What?\nSpeaker D: I mean, but we just have a flat spot on the bottom of the ball.\nSpeaker D: Not to put you on the spot, but.\nSpeaker D: What did you say your title was again?\nSpeaker D: I'm the industrial designer.\nSpeaker B: The industrial designer.\nSpeaker B: So, well, the point is that.\nSpeaker B: Well, maybe.\nSpeaker B: I don't know.\nSpeaker B: The shape is perhaps not the most ideal.\nSpeaker B: Okay.\nSpeaker B: As stable as it is, there must be a compromise between.\nSpeaker B: I suppose there's a possibility design.\nSpeaker C: I mean, I'm sure I think of like the design of others.\nSpeaker C: I can't think of anything other than the long rectangle.\nSpeaker C: For a more sort of faster ones, but.\nSpeaker C: Yeah.\nSpeaker C: There's nothing being done sort of out of.\nSpeaker C: That's not new.\nSpeaker D: It's not innovative.\nSpeaker D: It's, you know, everybody does longer remote because it's easy.\nSpeaker D: It's stable.\nSpeaker A: I'll do something.\nSpeaker A: So if.\nSpeaker D: No, go ahead.\nSpeaker A: My idea was just to have it be kind of.\nSpeaker A: Like a.\nSpeaker A: Keyboard type shape, you know, like video games.\nSpeaker A: So, but maybe.\nSpeaker A: I mean, that would be kind of big and bulky.\nSpeaker A: We could also try to do the hinge then.\nSpeaker A: So it could like flip out that way.\nSpeaker A: I don't know.\nSpeaker C: I think definitely doing something different.\nSpeaker C: Yeah.\nSpeaker C: It's a good idea.\nSpeaker C: I mean, very design something that's sort of like.\nSpeaker C: Well, suppose not always everybody's having the same, but something that would maybe fit in the hand.\nSpeaker C: Yeah, it's easy.\nSpeaker D: Something with a grip.\nSpeaker D: Yeah.\nSpeaker D: Because even I suppose even with the ball.\nSpeaker C: You still might be.\nSpeaker C: It's still.\nSpeaker D: It might not be the easiest to hold on to.\nSpeaker D: So perhaps the.\nSpeaker D: The joystick.\nSpeaker D: Yeah, the keyboard idea.\nSpeaker D: Might work better.\nSpeaker D: But again, people like to use one hand to flip and one hand to hold their soda.\nSpeaker D: So maybe.\nSpeaker D: Maybe it's.\nSpeaker C: Yeah, I think it's definitely got to be a one handed.\nSpeaker C: A one handed job.\nSpeaker D: I feel like I'm just shooting everything down here.\nSpeaker A: You're the boss.\nSpeaker B: One handed design, you also have a problem of the size, you know, from cell phones, they can be too small.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: It's small.\nSpeaker B: It probably looks better, but.\nSpeaker B: It may not be as functional.\nSpeaker D: Okay.\nSpeaker D: So.\nSpeaker D: Fortunately, we've got about five minutes here to come up with our.\nSpeaker D: Remote control idea and start rolling with it.\nSpeaker D: We've talked about our experiences with remote control and.\nSpeaker D: We've got a couple ideas.\nSpeaker D: See here.\nSpeaker D: What if we had what if we had not only.\nSpeaker D: Say we went with the ball, the ball function.\nSpeaker D: But maybe we give it sort of grips along the side.\nSpeaker D: To make it easier to hold on to.\nSpeaker D: So.\nSpeaker D: So it's easier to hold on to that way.\nSpeaker D: Of course, that will then remove some of our.\nSpeaker D: Our ball, unless this, unless this part were raised.\nSpeaker D: So.\nSpeaker D: Say the cover flips over and covers that part.\nSpeaker D: So the grip is.\nSpeaker D: No, that wouldn't work either.\nSpeaker D: But if we're going to make it flat on the bottom.\nSpeaker D: Then that eliminates our ball anyways.\nSpeaker D: So if it were.\nSpeaker D: Flat on the bottom and then had the.\nSpeaker D: Sort of grips in the side here, I guess.\nSpeaker D: And then we have the problem with the hinge.\nSpeaker D: So if we're flat on the bottom, it's not going to roll away.\nSpeaker D: It'll stay where we want.\nSpeaker B: Questions also.\nSpeaker B: I don't know.\nSpeaker B: Do you really always want to open that thing when you have to use it.\nSpeaker D: That's true.\nSpeaker B: It's probably going to lie around open all the time anyway.\nSpeaker B: So I don't know if a lid is a good idea.\nSpeaker B: The stability point of view.\nSpeaker B: It certainly is.\nSpeaker B: But also.\nSpeaker B: You have to face it and take into account the more of these things.\nSpeaker B: Break by accident.\nSpeaker B: The more we sell.\nSpeaker B: So.\nSpeaker B: Don't make.\nSpeaker D: So we don't have it flip open.\nSpeaker D: We just have a ball.\nSpeaker C: But then maybe go back to the.\nSpeaker C: Something along those things.\nSpeaker D: Okay.\nSpeaker D: So then we forget the ball.\nSpeaker D: It looks cool.\nSpeaker D: It looks cool, but it's really not.\nSpeaker D: It's not functional.\nSpeaker D: So we've got our sort of keyboard kind.\nSpeaker D: What if we flipped it around here so that it were.\nSpeaker D: Sorry, that doesn't look anything like what you had there.\nSpeaker D: So it's up and down.\nSpeaker D: You hold it this way.\nSpeaker D: It's like the rectangular again only with a couple of jutting out points.\nSpeaker D: But it's one handed.\nSpeaker B: The question is what makes those game pads functional.\nSpeaker B: I think that's pretty much the form for the hand.\nSpeaker B: So it's a round shape underneath that makes it comfy.\nSpeaker B: Yeah.\nSpeaker B: Makes it nice.\nSpeaker B: So that's the essential part except for that I think.\nSpeaker B: Yeah.\nSpeaker B: We'll not probably not get away from some longer design.\nSpeaker B: Because you also have to know which way around to point this thing.\nSpeaker B: Yeah.\nSpeaker B: Because it doesn't have a chord like joysticks do.\nSpeaker B: And the batteries go weak as well.\nSpeaker B: So after a while you have to point it towards the equipment you want to control with it.\nSpeaker B: Show which is the front, which is the back.\nSpeaker D: Is it possible to have it to where it would work with like a sensor on either side?\nSpeaker D: So that either way you're pointing it would work.\nSpeaker B: Of course the more technology you stick in that, the more it'll cost.\nSpeaker B: Of course you can do.\nSpeaker D: Okay.\nSpeaker B: I mean of course it will be evident after a while.\nSpeaker B: If you look at it it will be evident which way around to point it since you have the numbers and the buttons and stuff.\nSpeaker B: But it's about an instinctual thing.\nSpeaker B: You can just grab it and you don't have to look at it which way around to point it.\nSpeaker B: Otherwise the design of the point of putting two sensors on both sides would probably work.\nSpeaker C: Even if you designed it in a way that isn't a rectangle but still pointed in a direction that had a definite point.\nSpeaker C: So that's your thing.\nSpeaker C: Something like that instead.\nSpeaker C: And there you can know which way you can point today.\nSpeaker C: Okay.\nSpeaker A: Sorry to interrupt but we have a warning.\nSpeaker D: Before we add a time.\nSpeaker D: Okay.\nSpeaker D: Well, just to finish up should we go with this plan?\nSpeaker D: Start making some.\nSpeaker C: Are good ideas what are not.\nSpeaker D: Finish now.\nSpeaker D: And then marketing will look and see what people want.\nSpeaker D: Okay.\nSpeaker D: Project major will design a better meeting for the next time around be a little bit more prepared.\nSpeaker D: And alright.\nSpeaker D: Good meeting.\n"}, {"meeting_id": "TS3008b", "summary": "In this meeting, the team brainstormed about the look, functions and gadgets to adopt in the remote control and finally reached a decision on the overall design. The meeting started with discussion on the technical functions based on some new requirements delivered by Project Manager and on market research results presented by Marketing. In the end, the team agreed on including mainly basic functions on the remote control, adding a joystick-like gadget that was able to charge and track the device, enlarging certain buttons and making it flashy.", "dialogue": "Speaker D: My mouth is not working anymore.\nSpeaker C: I'm going to put it in.\nSpeaker C: It's going to beep beep beep.\nSpeaker B: Oh, I've got a nice little screen here over here.\nSpeaker B: I've got this big black border on every side.\nSpeaker A: OK.\nSpeaker A: OK.\nSpeaker A: Everybody ready?\nSpeaker A: I'll fix it.\nSpeaker A: Yeah, we're going with the functional design meeting.\nSpeaker A: Again, presented by Martin.\nSpeaker A: This is the agenda opening.\nSpeaker A: We've got three presentations.\nSpeaker A: And I'm going to show you some of the new project requirements that were sent to me.\nSpeaker A: And we're going to make a decision on the remote control functions.\nSpeaker A: We have 40 minutes.\nSpeaker B: Yes.\nSpeaker A: Well, this is the closing already.\nSpeaker A: So well, we start off with the first presentation.\nSpeaker A: I think we have to do it in the right order.\nSpeaker A: Maybe you don't know what the right order is.\nSpeaker A: I'm going to do it.\nSpeaker A: Maybe we should start with the technical functions.\nSpeaker C: OK.\nSpeaker C: How can I get this on the one board?\nSpeaker A: You don't file in the project document folder.\nSpeaker A: You've already done that?\nSpeaker D: Yes.\nNone: And that one.\nSpeaker A: Well, let's close this one.\nSpeaker A: We'll just open the new one there.\nSpeaker A: Oh, you're saying?\nSpeaker A: Oh, yeah.\nSpeaker A: I have to say something.\nSpeaker A: Due to some technical problems, I haven't digitized the last meeting minutes.\nSpeaker A: But I'll make sure that happens next time.\nSpeaker A: And I'll get this one in digital form too.\nSpeaker C: OK.\nSpeaker C: We're going to talk about working design.\nSpeaker C: Method of the remote control is electrical energy.\nSpeaker C: It activates a chip in the remote.\nSpeaker C: It's an electrical circuit which compose messages in the form of infrared signals to control the television.\nSpeaker C: It's nowadays very known technology.\nSpeaker C: The known technology can make the cost very low.\nSpeaker C: It's a wide tail of remote controls in the world.\nSpeaker C: And the components are very cheap.\nSpeaker C: Diode's batteries and lead lights, they're needed.\nSpeaker C: And they're everywhere available against a fair price.\nSpeaker C: It's common technology.\nSpeaker C: I could tell the circuit board, it's the most important part of the remote control.\nSpeaker C: We can use for that fiberglass with copper wires.\nSpeaker C: It can be made as fast as printing paper.\nSpeaker C: It's all very making it all the time.\nSpeaker C: And it's not very specialized technology.\nSpeaker C: I haven't come to here, but I've got some images of remote controls.\nSpeaker C: They were not very trendy or just a remote control, like everyone knows.\nSpeaker C: So I don't know why I should put it here.\nSpeaker A: Well, it's the technical side of the remote control.\nSpeaker C: Yes, but I haven't made it because of the time.\nSpeaker A: OK.\nSpeaker A: Oh, let's skip that part time.\nSpeaker A: But you don't think it's a problem to design the technical part of the remote control.\nSpeaker A: It's going to be easy.\nSpeaker A: Yes.\nSpeaker A: OK.\nSpeaker D: But nothing restricted for a user interface.\nSpeaker C: No, it's just a part of a known technology.\nSpeaker C: Remote control is nothing special nowadays.\nSpeaker A: Regardless of what type of functions we want to implement, it doesn't really matter.\nSpeaker C: I don't think so because of the old televisions, there are a few, maybe a couple of televisions with the new functions.\nSpeaker C: But is it useful to put them on a standard remote?\nSpeaker A: Well, we'll see.\nSpeaker A: We'll see later on.\nSpeaker D: Well, the technical functions.\nSpeaker D: I don't know if you've got the same pictures as I got.\nSpeaker D: No.\nSpeaker D: I got these two.\nSpeaker D: And I think we have to focus on the one hand, the expert view or the novice user.\nSpeaker D: I think it's very much depending on the user requirements.\nSpeaker D: I don't know who is.\nSpeaker A: Well, there will be some user requirements later on.\nSpeaker A: Yeah.\nSpeaker D: I think that's very important to what kind of functions there we want to put in the remote control.\nSpeaker A: Well, we'll keep this in mind and discuss it later on.\nSpeaker D: Yeah.\nSpeaker D: Well, we can put functions in it when we get user requirements.\nSpeaker D: And we can update it.\nSpeaker A: There's a big distinction between the style of remote.\nSpeaker A: We should choose one.\nSpeaker A: We should not compromise, but really choose for experts viewer or novice.\nSpeaker A: I agree.\nSpeaker A: Yeah.\nSpeaker D: Well, that's what you want trying to say.\nSpeaker D: Well, yeah.\nSpeaker D: If you want to try a huge market, if you want to reach a huge market, like LVP people, we have to choose for novice user.\nSpeaker D: But I don't know.\nSpeaker D: It's really depending on how far the remote controls are already in use.\nSpeaker A: Well, some of them, yeah.\nSpeaker A: But it will be more clear when we come to some of the new requirements.\nSpeaker A: Yeah.\nSpeaker A: Yes.\nSpeaker A: Probably.\nSpeaker D: So.\nSpeaker D: Well, there are at least basic functions, like just the channels, one, two, nine, all in an off which must be clear with a red button or something like that.\nSpeaker D: Most standard volume, of course, and a mute function.\nSpeaker D: And of course, the next and previous channel.\nSpeaker D: I think that's just basic what we need.\nSpeaker D: And from that on, we can see it's huge requirements for what we need more.\nSpeaker D: Yeah.\nSpeaker D: I just thought Yos was looking at the trendy, the trends in the market.\nSpeaker D: Yes.\nSpeaker D: And I don't know if there are any, if you put more functions, more buttons, maybe it's becoming less trendy or something like that.\nSpeaker D: You can just.\nSpeaker D: You can just find a conclusion like that.\nSpeaker D: You can keep it in mind.\nSpeaker D: Yeah.\nSpeaker D: I don't know.\nSpeaker D: I thought with less buttons, you can make a more trendier design.\nSpeaker D: Yeah, more trendier design, I think.\nSpeaker D: Sounds interesting.\nSpeaker B: Yeah.\nSpeaker D: Well, that's all I say.\nSpeaker D: Okay.\nSpeaker D: Right.\nSpeaker A: Well, then the marketing expert can.\nSpeaker A: Yes.\nSpeaker A: Tell us something about the current market.\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker B: All right.\nSpeaker B: I've done some research for functional requirements.\nSpeaker B: Yes.\nSpeaker B: The working method.\nSpeaker B: There were 100, 100 people, how do you say, watched using remote control in the usability lab?\nSpeaker B: And they also filled out a questioner with a few questions.\nSpeaker B: I've lined them up here.\nSpeaker B: Ask whether common remote control looks good or not about willingness to spend money on remote control about zapping behavior and stuff like that.\nSpeaker B: I've found some interesting things.\nSpeaker B: We do got a market.\nSpeaker B: Three out of four people claim to find remote controls ugly.\nSpeaker B: So if we make a trendier design, we sure have 75% of the market which we can reach.\nSpeaker B: Three out of four users zaps a lot, as I quoted here from the results.\nSpeaker B: Zap buttons are used for 168 times per hour.\nSpeaker B: That's quite a lot.\nSpeaker B: Relevant options are, of course, power buttons, although only used once per hour.\nSpeaker B: Channel selection, volume and buttons for text.\nSpeaker B: And the more other functions like audio settings, video settings, sound settings are not said to be very important.\nSpeaker B: It's very much used.\nSpeaker B: Furthermore, 50% says they only use 10% of the buttons on the remote control.\nSpeaker B: That doesn't say we can leave 90% off, but it sure says we shouldn't make it too complicated.\nSpeaker B: 50% also claims to have last remote control very often in the world.\nSpeaker B: An important thing here.\nSpeaker B: The most important customers, which is over 70% of our market is in the age range of 36 to 65 years old.\nSpeaker B: And elderly people, our market, are less interested in nice features, but more willingly to spend more money on remote control.\nSpeaker B: What I was thinking, wrong side, we shouldn't implement too much features on our remote control because elderly people will get last group features for a higher usability, what I was claiming in the previous meeting.\nSpeaker B: All the settings about audio settings, video settings and audio settings, which are not very often used, we could group them on one button and make them accessible in one menu button or whatever because they are used very rarely.\nSpeaker B: There are a lot of options there, so we can really make, yeah, how do you say?\nSpeaker B: We can spare buttons over there.\nSpeaker B: If you want to implement PCR and DVD options, group them in the button not to, yeah, small buttons, so they won't be very, how do you say?\nSpeaker B: Especially presence.\nSpeaker B: Yes, won't be very present.\nSpeaker B: Thank you.\nSpeaker B: And a trendy look, although 70% of the market consists of elderly people who don't really care for trendy looks or whatever, I guess it can do no harm to make a trendy for the other 30%.\nSpeaker B: That was kind of what I found.\nSpeaker B: Okay.\nSpeaker A: Well, then we, I'm going to show you some of the new project requirements and then we're going to discuss on what features we find important.\nSpeaker A: Some of the new requirements make some of your findings quite irrelevant, I think.\nSpeaker A: Because, decided to put, they have to set two additional requirements forward.\nSpeaker A: Well, I see four.\nSpeaker A: This is kind of strange.\nSpeaker A: Well, they say that the data text becomes outdated since the popularity of the Internet.\nSpeaker A: Well, I think that may be so, but, well, we can't just leave the data text button off.\nSpeaker A: No, I agree.\nSpeaker A: I agree.\nSpeaker A: So the compromise we could make is just to make one teletext button, you know, like on and off and don't make a lot of special, put a lot of special features on it to make it transparent or, you know, it's just you want, you want to be able to make use of teletext but not to play with it that much.\nSpeaker A: So we have to think of that.\nSpeaker A: The remote controls should only be used for the television.\nSpeaker A: Otherwise, the project becomes more complex, which endangers the time to market.\nSpeaker A: So maybe we should leave all DVD and VCR related features off completely.\nSpeaker B: Okay.\nSpeaker A: I don't know.\nSpeaker A: I think that's what they're trying to say.\nSpeaker A: Our current customers are within the age group of 40 plus.\nSpeaker A: New products should reach a new market with customers that are younger than 40.\nSpeaker A: So you talked about the elderly where we're willing to spend more on the remote control and we're interested, but, well, they're not relevant because we are aiming at a younger, I don't really\nSpeaker B: agree actually, to be honest. It's a very small market which we will approach then if we want to reach customers younger than 40.\nSpeaker B: It's only like 30% of the total market.\nSpeaker A: Yeah, but it's the fact that the bigger market you're talking about, we already covered it.\nSpeaker A: Our company already sells remote control to the older people, but we also want a new customer group.\nSpeaker A: That's the one we haven't covered yet.\nSpeaker A: So I think that's what the problem is.\nSpeaker A: We haven't got remote controls for you.\nSpeaker A: Maybe we can compromise a little bit.\nSpeaker A: Yeah, I think so.\nSpeaker A: Maybe not too much then.\nSpeaker A: No, no, but I think we have to just keep in mind what the older age group wants and maybe we can make a remote control that's primarily interesting for the younger group but isn't that bad for a older person either.\nSpeaker A: Yes.\nSpeaker A: Our corporate image should stay recognizable in our products.\nSpeaker A: Our product, corporate color and slogan must be implemented in a new design.\nSpeaker A: Okay, something is nice to know.\nSpeaker D: What else are slogan?\nSpeaker A: Sorry?\nSpeaker A: Yeah.\nSpeaker A: We have to look that up.\nSpeaker B: Yeah, I'll have a look.\nSpeaker A: We figured something about electronics.\nSpeaker A: Oh, okay.\nSpeaker A: I thought it might be, let's make things better or something.\nSpeaker A: Sensors and simplicity.\nSpeaker A: Okay, well, let's go back to the agenda.\nSpeaker A: So if not the three presentations we know about the new project requirements.\nSpeaker A: That means we can discuss under remote control functions.\nSpeaker A: Well, if I can make a start, I think it's becoming more clear what kind of remote control is going to be and I already talked about maybe you're talking about remote control that has the can opener underneath it.\nSpeaker A: I think we're looking for some, we're looking for a really simple remote control with only basic TV functions.\nSpeaker B: Well, to be honest, if our aim group is still 40, not older than 40, maybe that's not very, we don't really need to have a simple remote control.\nSpeaker B: I think we can implement more functions then because basically the younger people are more able to adapt to new technology and therefore will be a more.\nSpeaker A: That's why a lot of the requirements, the account managers send me, I think they are contradicting each other because they want a simpler design and no other functions in just the fee but they do aim at the younger.\nSpeaker D: But I think it's a lot of functions that aren't used.\nSpeaker D: So why should we put a function in?\nSpeaker D: Well, I think younger people are more looking for just a trendy look than more functions.\nSpeaker A: I think we can make some discussions in functions.\nSpeaker A: You have different kind of equipment in your room, like a TV and a DVD player.\nSpeaker A: You can put very detailed functions regarding the TV set on your remote control with audio settings and screen settings.\nSpeaker A: We don't want that. I think that was, that became clear.\nSpeaker A: Maybe we should put something, I know that the younger people most likely have a DVD player.\nSpeaker A: They want to control remotely.\nSpeaker D: But the functions are not in the remote control we are making.\nSpeaker A: The user requirements of the new project requirements still is not too, but maybe, I think we maybe should put some functions for other equipment on it.\nSpeaker A: But just the basic functions, maybe like wind and wind.\nSpeaker C: What do you guys want?\nSpeaker C: Not much more than that.\nSpeaker A: As far as possible.\nSpeaker B: Or we could put them behind the flip flap or whatever.\nSpeaker A: What do you think?\nSpeaker A: Should we implement functions to control other devices?\nSpeaker A: No, you don't think so.\nSpeaker A: No, the new requirements.\nSpeaker C: But you can put a play and stop.\nSpeaker A: Well, maybe it's something, most of the time these functions don't support particular devices as well as their own remote control.\nSpeaker A: A lot of DVD players have some tricky settings.\nSpeaker D: If it's too simple, they won't use the remote control.\nSpeaker D: They use their own.\nSpeaker B: For example, a VCR, that's a better example in this case.\nSpeaker B: I think on a remote control for television you don't need to be able to program the VCR to start recording at 3pm or whatever.\nSpeaker A: But you can also do it with a play, stop, rewind and fast forward.\nSpeaker A: I have to think DVD.\nSpeaker A: I know.\nSpeaker A: From my experience, a lot of DVD players, like forwarding, goes differently.\nSpeaker A: You could do 2 speed or 8 speed or 16 speed.\nSpeaker A: Sometimes a bit difficult.\nSpeaker A: Maybe if we just leave the DVD functionality.\nSpeaker A: I was thinking about putting it in.\nSpeaker A: But concerning the project requirements.\nSpeaker A: You should focus on the TV then.\nSpeaker D: Keep it simple and look more.\nSpeaker A: It's just a complementary remote and not a universal one.\nSpeaker A: If you want to keep it simple, you can make a universal remote.\nSpeaker A: It's only for television.\nSpeaker A: It should be something that is like gadget on your coffee table.\nSpeaker A: It's just when you jump on the couch, you pick up the nice remote, a simple one, just to put on the television, nothing more.\nSpeaker A: Nothing more.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker C: But what will you be able to handle a whole market?\nSpeaker C: There are dozens of remote controls.\nSpeaker A: We'll make this one trendy.\nSpeaker A: Simple.\nSpeaker A: We will come to that.\nSpeaker A: First on the functions.\nSpeaker A: So we should put zap buttons on it.\nSpeaker A: Also numbers to go to this specific channel.\nSpeaker D: Yes, definitely.\nSpeaker D: It's too much integrated.\nSpeaker A: A teletext button should be there, but just one big teletext button on and off.\nSpeaker B: Yeah, maybe two or three other options.\nSpeaker B: But not nothing more than that.\nSpeaker B: I think stop function is very useful if you have a page with consist of more pages and you are not a very quick reader.\nSpeaker B: Then I think it's very irritating if the next page shows up.\nSpeaker A: I think it becomes too difficult.\nSpeaker A: It's not a very common function.\nSpeaker A: Wow.\nSpeaker A: I use a fairly remote action.\nSpeaker B: I use it.\nSpeaker A: Yeah, maybe you do.\nSpeaker A: I've never heard of it in the first place.\nSpeaker A: We have to look at the market.\nSpeaker A: Yeah, and teletext is becoming outdated.\nSpeaker A: We just want to see what programs are on and what time it is probably.\nSpeaker A: I don't know.\nSpeaker A: I think that's kind of getting too complex for our remote.\nSpeaker A: I don't know what you're using.\nSpeaker D: I think.\nSpeaker D: Yeah, it's Sunday.\nSpeaker D: I always use it for the sucker.\nSpeaker A: Yeah, but do you like to have such a stop button?\nSpeaker A: Do you think it's a kind of a very rare and special function?\nSpeaker D: When you look, for example, a couple of weeks ago, I looked at the flights.\nSpeaker D: There are a lot of flights in one page.\nSpeaker A: But that's kind of stuff we should do on the internet right now.\nSpeaker A: Yes, you could have it.\nSpeaker A: In the use of the environment.\nSpeaker C: There are some buttons you can put it to access all the same pages.\nSpeaker A: Well, yeah, lots of new televisions can store pages, you know, and then you can just skip manually through them using.\nSpeaker A: I think we just put one teletext button on it.\nSpeaker A: Then we meet the new requirements.\nSpeaker A: We also meet the other thing.\nSpeaker A: We have to choose for the simple design, I guess.\nSpeaker A: What functions do we have to decide on?\nSpeaker A: I don't know if we specifically name all the functions we want.\nSpeaker A: We have the zebbunds, the volume.\nSpeaker A: We do make them very big, the zebbunds.\nSpeaker A: I think the plus and the minus, but that's considered to be 20 as well.\nSpeaker A: Maybe you should place them on a kind of joystick.\nSpeaker A: Yeah, something special way to zeb through the...\nSpeaker A: It has to be...\nSpeaker A: In quick, you have to use it very quickly.\nSpeaker B: If you grab the remote, your hand should be on top of the plus.\nSpeaker A: The buttons should make it possible to zeb through your channels in a rapid base.\nSpeaker B: Yes.\nSpeaker A: What should we decide on them?\nSpeaker A: I think in the case of this simple remote control, the technical aspects which weren't worked out already, but it shouldn't be a problem then.\nSpeaker C: No, it's just signals.\nSpeaker C: Maybe...\nSpeaker C: And the television does the rest.\nSpeaker A: No, okay, but we don't have to.\nSpeaker A: When we don't want to control other devices, I think it makes it even more simpler.\nSpeaker D: The batteries maybe, if you use large batteries or small batteries...\nSpeaker C: I think standard batteries.\nSpeaker A: I think we should use...\nSpeaker A: Not the watch kind of...\nSpeaker A: Well, it has to be simple when I...\nSpeaker A: The most ordinary batteries.\nSpeaker A: Which are most likely to be found somewhere in the house.\nSpeaker A: Yeah.\nSpeaker A: How much time do we have left?\nSpeaker C: I think about 20 minutes.\nSpeaker C: But in a way we have to be special.\nSpeaker A: Yeah, yeah, okay.\nSpeaker A: But that's...\nSpeaker A: If you think about that, thought about that.\nSpeaker A: How we can...\nSpeaker A: What the extra touch can be?\nSpeaker A: Maybe just the design or the shape or some gadget kind of future.\nSpeaker A: Or what's something about how we lose them?\nSpeaker A: Maybe it should be a remote control.\nSpeaker A: When you clip your hands, it makes some noise or some gadget kind of thing.\nSpeaker A: I think that's more for...\nSpeaker B: For an age range of 10 to 12 or whatever.\nSpeaker A: I don't know.\nSpeaker A: A lot of people like to have such kind of gate kind of things.\nSpeaker A: It should be like a birthday present or something that you get someone in.\nSpeaker A: That's a good one.\nSpeaker A: Or maybe it should have a big light that can flash for something on it.\nSpeaker A: Or...\nSpeaker A: Or the can open it.\nSpeaker A: Maybe it contains some future that you don't normally link to a remote control.\nSpeaker A: I think it's very important because we're going to make such a basic remote control.\nSpeaker A: We have to do something to make it special.\nSpeaker A: It's going to cost 25 euros.\nSpeaker B: I think the can opener is brilliant idea actually.\nSpeaker B: But the television and beer is not a rare combination.\nSpeaker A: Yeah, that's true.\nSpeaker A: And I think it's going to be very...\nSpeaker A: It has to be sturdy or something.\nSpeaker A: Or maybe with bouncing pads so that you can just throw it on the floor or something.\nSpeaker A: It has to be something special.\nSpeaker A: It has to...\nSpeaker A: People when they buy it, they have to think about this one less for a long time.\nSpeaker A: We're really going to use it not something you throw away next week.\nSpeaker A: No, that's true.\nSpeaker A: So maybe I think that when we decide on these types of functions, basic functions, it's very important to find something like that.\nSpeaker A: So there's a very important task for you.\nSpeaker A: Maybe we can all think about it.\nSpeaker A: Also for you, maybe it's very nice when you can be trending as a friend.\nSpeaker A: You're friendly as well.\nSpeaker A: So big buttons, flashy design, maybe some kind of gauges kind of thing.\nSpeaker A: Yeah, a swappable front or whatever.\nSpeaker A: Yeah.\nSpeaker A: Or just different colors would be...\nSpeaker A: I don't know if people also want to spend more money on fronts for their...\nSpeaker A: You might control it too.\nSpeaker A: Yeah, you never know.\nSpeaker D: Yeah.\nSpeaker D: More money for us.\nSpeaker A: And I think we have to make it quite big.\nSpeaker D: Quite big?\nSpeaker C: Yes, definitely.\nSpeaker A: Definitely.\nSpeaker A: Also because it's expensive, it wanted to be something.\nSpeaker A: It doesn't have much function.\nSpeaker A: You don't want to get it all.\nSpeaker B: It should be visible nearly anywhere in the room.\nSpeaker B: As I said during my presentation, 50%.\nSpeaker A: And it should not look nice when you put it on a table.\nSpeaker A: I think you might want to put it standard.\nSpeaker A: Yeah, it stands up.\nSpeaker A: Yeah.\nSpeaker A: You have to put it on its...\nSpeaker A: So it's like a vase or something you put on a table.\nSpeaker A: A vase?\nSpeaker A: I don't know if you put it on a table.\nSpeaker A: More like a statue.\nSpeaker A: It's like a statue or something.\nSpeaker A: Yeah, yeah, I see what you're doing.\nSpeaker D: It's like four phones.\nSpeaker A: But you also can put it somewhere...\nSpeaker B: If you do that, but I don't know if that's possible within the production cost of 12 and a half euro.\nSpeaker B: In the base, we could make a button and if you push it, the remote control itself makes noise.\nSpeaker B: That's probably stupid.\nSpeaker B: But as I found here, 50% was it 50?\nSpeaker B: That's the first time.\nSpeaker B: 50% often loses the remote control.\nSpeaker A: Yeah, but when it gets lost, how can you press the button to make it?\nSpeaker B: No, off the base.\nSpeaker D: The thing you put it in.\nSpeaker B: Oh, man, that's kind of nice.\nSpeaker A: And then also you don't even need batteries because you can make it chargeable.\nSpeaker B: Yeah, you can make it a chargeable one, yeah.\nSpeaker A: Why not?\nSpeaker A: Yeah, but, yeah, I think it might be...\nSpeaker A: Why not?\nSpeaker A: We don't...\nSpeaker A: Yeah, maybe you...\nSpeaker A: We don't know much about production costs, but when you can imagine that when you spend 25 euros on a remote control and it's basically promoting control and then the money, there must be money to spend on that kind of rechargeable...\nSpeaker B: It should only cost 12 and a half euros, of course.\nSpeaker A: Yeah, but we do it in Taiwan and so it's not going to be dead.\nSpeaker A: Yeah, okay.\nSpeaker A: I think it's a great idea.\nSpeaker D: I think it's a good idea.\nSpeaker A: Yeah.\nSpeaker A: Some kind of bag.\nSpeaker A: I've never seen that before.\nSpeaker A: You make it...\nSpeaker A: You see it with the...\nSpeaker A: The mouses now.\nSpeaker B: Yes, definitely.\nSpeaker C: Yes, but it's handy.\nSpeaker A: Well, you can recharge it, so you never have the battery problem.\nSpeaker A: That's one thing.\nSpeaker A: You can always find a lot of control about the handings.\nSpeaker C: It's got a couple of controls.\nSpeaker C: Remote controls nowadays can last two years, three years with two batteries.\nSpeaker C: Okay.\nSpeaker A: Well, maybe you could...\nSpeaker A: And when it's too costly, you could probably skip the recharge.\nSpeaker A: But you do need...\nSpeaker A: Also, you would need a battery in the base unit as well, you know?\nSpeaker B: Well, you could connect that to 220 volts, of course.\nSpeaker D: Yeah, but that's not...\nSpeaker A: Yeah, but then it's very easy to make it also rechargeable.\nSpeaker A: I don't think that is going to cause much to make it also rechargeable.\nSpeaker A: No, function in it.\nSpeaker B: On the other hand, if you don't do it, we can also make a nice bay.\nSpeaker B: I mean, it looks trendy and still put the bleep function in it.\nSpeaker B: Yeah.\nSpeaker B: I think the bay is definitely...\nSpeaker A: I think it's a good idea.\nSpeaker A: And make it...\nSpeaker A: Well, it isn't most costly remote control.\nSpeaker A: We can save on the functions.\nSpeaker A: We just put some simple buttons in it, make it big and sturdy, nothing more.\nSpeaker A: And just make sure there's some noise that it can make, or probably some kind of cheap light thing around it.\nSpeaker A: That it lights up.\nSpeaker A: Sure, why not?\nSpeaker A: It's also nice.\nSpeaker A: And as you put away, I think it's...\nSpeaker A: Yeah.\nSpeaker A: We have to...\nSpeaker A: It's an easy market.\nSpeaker A: We have to do something special.\nSpeaker A: And for 25 euros, people want something special from your remote control.\nSpeaker A: And we can deliver that in regards to the functions because we aren't going to...\nSpeaker B: With eye candy, ear candy, whatever.\nSpeaker B: Yeah, definitely.\nSpeaker A: And then when you make it a nice looking shape, and then also you got the stand up thing.\nSpeaker A: Yeah.\nSpeaker A: I think it's a good idea.\nSpeaker D: Yeah, it must be a gadget to have.\nSpeaker D: Yeah, definitely.\nSpeaker A: Well, we'll see what's possible concerning the costs, and if it's possible, we'll do that.\nSpeaker A: And we even try to save up on other stuff to make sure we can do such a thing.\nSpeaker A: And the first thing we most likeable thing to skip is the probability recharge function or something.\nSpeaker A: If that's too expensive, we won't do that.\nSpeaker A: But it would be nice.\nSpeaker A: It would be nice, yes.\nSpeaker A: It's the idea.\nSpeaker A: I know the batteries last long nowadays.\nSpeaker A: But people just think about, oh, I'll never have to buy any batteries again.\nSpeaker A: Because it's very annoying when your battery is empty.\nSpeaker A: And when you have some batteries around, and probably for two weeks, I think...\nSpeaker B: Most television is break down before the battery pack is empty.\nSpeaker B: No.\nSpeaker B: Yes.\nSpeaker A: So easy functions?\nSpeaker A: Well, I think we'll work that out, stepping, numbers on it.\nSpeaker D: Or just if it beep when the batteries are down.\nSpeaker B: Yeah, why not?\nSpeaker A: But it's also annoying.\nSpeaker A: Because as long as it keeps working, you're not very motivated to do something about it.\nSpeaker A: No, that's true.\nSpeaker A: And you don't want to have those problems.\nSpeaker A: And you won't have, if you have to rechargeable.\nSpeaker A: And you don't have to use the unit.\nSpeaker A: You can also put it on the side if people don't like it.\nSpeaker A: But you pay for it.\nSpeaker A: Yeah, but I mean, they pay for it because they think, oh, that's a great idea.\nSpeaker A: I'm going to use it.\nSpeaker A: And when time goes by and they think, oh, I'll never put them in a recharge, I don't think it lasts long enough, then they put it on the side and they can use it now and then.\nSpeaker A: And when they look...\nSpeaker A: I know for sure that everybody who buys this remote control within a couple of months, they will be in the situation that they're seeking for their remote control.\nSpeaker A: They want to see something quick and they just push the button and I think really...\nSpeaker A: Leave, leave all their days.\nSpeaker A: Yeah.\nSpeaker A: I've never...\nSpeaker A: It's so simple, but I've never seen it.\nSpeaker D: No.\nSpeaker D: I ain't going to leave it just there.\nSpeaker A: Maybe you should really...\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Now, well, how much time have you got left?\nSpeaker A: These clocks are in sync.\nSpeaker A: Well, ladies.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So we have 10 minutes or something?\nSpeaker A: Something like that.\nSpeaker B: Yeah.\nSpeaker A: Yeah, but we're done.\nSpeaker A: I think.\nSpeaker A: We've decided on the functions.\nSpeaker A: Well, there's some...\nSpeaker A: There is a closing sheet.\nSpeaker A: We have lunch break and then we have 30 minutes of individual work.\nSpeaker B: Oh.\nNone: Okay.\nSpeaker A: I'll make sure I had some problems with digitizing the first minutes.\nSpeaker A: The next minutes won't be a problem, but I'll try to make sure the first one will be in the folder too, but maybe it won't work, but you'll see.\nSpeaker A: I think these are more important than the first one.\nSpeaker D: Can you make an...\nSpeaker D: A part of folder for the minutes?\nSpeaker D: Oh, that's everything in one.\nSpeaker A: Five minutes.\nSpeaker A: Oh, five minutes.\nSpeaker A: Yeah, because I did...\nSpeaker A: The first minutes I did were a bit scratch, you know.\nSpeaker A: Then I did a second one with a nicer layout, which I could use for the other ones as well, but I think I forgot to put done under the first one, and when you go right in second, it's not working when you try to write a second paper or something.\nSpeaker A: Oh, that's fine.\nSpeaker A: Yeah.\nSpeaker A: You have to override it or something.\nSpeaker A: I don't know.\nSpeaker A: It wasn't on my pen.\nSpeaker B: Should we, by the way, draw...\nSpeaker A: What is going on?\nSpeaker B: On our nice whiteboard.\nSpeaker B: A little idea of the shape.\nSpeaker B: Yeah.\nSpeaker B: Probably.\nSpeaker B: I don't know.\nSpeaker B: Has anyone got...\nSpeaker B: A little bit detailed ideas about the shape?\nSpeaker C: I don't.\nSpeaker C: Maybe like this pen.\nSpeaker C: A bit bigger, I guess.\nSpeaker B: Yeah, bigger.\nSpeaker D: A little bit bigger, yeah, but...\nSpeaker B: The shape is nice.\nSpeaker B: It's something different.\nSpeaker B: I'm going to want to feel more about that.\nSpeaker A: I have to say, I have this again opening remote control in my head most of the time.\nSpeaker A: Or I think some...\nSpeaker A: Maybe we should...\nSpeaker A: No, that will be too.\nSpeaker A: Oh, I think we should...\nSpeaker A: We could also...\nSpeaker A: That would also be an idea, but I don't think...\nSpeaker A: I don't know if it exists already.\nSpeaker A: You should make a less-year-something design it.\nSpeaker B: Okay.\nSpeaker A: Yeah.\nSpeaker A: That would also be nice.\nSpeaker A: But then you...\nSpeaker A: 12 and a half euros.\nSpeaker A: Yeah, but then you don't have...\nSpeaker A: Yeah, but it's not something that's in the production, because it's one time, you know?\nSpeaker A: It's a single cost.\nSpeaker A: Yeah.\nSpeaker A: But then you can...\nSpeaker A: Nah, I don't think it would be more expensive, because I bought the Alesi stuff more often and even small pencil holders or something are more expensive.\nSpeaker A: Yeah.\nSpeaker A: That's true.\nSpeaker A: It would be nice idea, though.\nSpeaker A: I don't know, I think it has to be...\nSpeaker A: It has to have round forms or something.\nSpeaker A: Like something like that or something.\nSpeaker B: Something like that is very ergonomic.\nSpeaker A: And then a...\nSpeaker A: So?\nSpeaker A: And base unit underneath it is also round.\nSpeaker A: Put it in there, and it will wire on it.\nSpeaker A: Maybe...\nSpeaker A: Yeah.\nSpeaker A: I don't know if some lights...\nSpeaker A: A big...\nSpeaker A: Well...\nSpeaker A: A flashlight at the...\nSpeaker D: A flashlight at the...\nSpeaker A: Four-human program.\nSpeaker A: A flashlight at the...\nSpeaker B: Four-human program should be there, I guess, because your hands...\nSpeaker B: And some of the extra functions will be in the smaller part.\nSpeaker A: Some of the extra functions over here.\nSpeaker A: Yeah.\nSpeaker A: And the numbers on top, I guess.\nSpeaker A: And in lights, maybe a ring of...\nSpeaker A: No, no, you have to recharge functions.\nSpeaker A: Maybe on the side of it.\nSpeaker A: Maybe round...\nSpeaker A: Around...\nSpeaker A: Yeah, side of it.\nSpeaker A: On the side of it.\nSpeaker B: On the side of it.\nSpeaker B: Maybe lights also around the four-human, the plus-minus program buttons.\nSpeaker A: Yeah, but I also meant that the blinking light...\nSpeaker A: You know the ones that also blink when you try to locate your remote.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker D: Well, probably at the side, you now look at the front.\nSpeaker A: Yeah, yeah, exactly.\nSpeaker A: When you see it from the side, it would look just like that.\nSpeaker A: Yeah.\nSpeaker D: And then there is...\nSpeaker A: It's triple for lights or something.\nSpeaker D: Yeah, okay, yeah.\nSpeaker D: Something like that, yeah.\nSpeaker A: Alright, I think it's nice for one thing.\nSpeaker A: Maybe put something on top of it or you know, like this.\nSpeaker A: Looks funny.\nSpeaker A: I don't know.\nSpeaker A: No.\nSpeaker A: Or some bump.\nSpeaker B: I think that'll be too big then.\nSpeaker D: Yeah.\nSpeaker D: Too big then.\nSpeaker D: Bumpers or something?\nSpeaker A: Well, I have to think about it.\nSpeaker A: I think we're done.\nSpeaker B: Yeah.\nSpeaker B: Yes, we are.\nSpeaker B: I'm alright.\nSpeaker B: Yes, I guess it's lunch time.\nSpeaker D: Okay then.\nNone: Mm-hmm.\nSpeaker A: I don't know.\nSpeaker B: Half an hour?\nSpeaker B: I thought our next individual round was half an hour.\nSpeaker B: Yeah, I didn't know about a lunch break.\nSpeaker D: Well...\nSpeaker D: Oh, I missed it.\nNone: I missed it.\nNone: I missed it.\nNone: I missed it.\nNone: Whoo-hoo-hoo-hoo.\nNone: I missed it.\nNone: I missed it.\n"}, {"meeting_id": "IS1000a", "summary": "Project Manager introduced a new remote control project for television sets, and the team got acquainted with each other and technical devices. The remote control would be priced at 25 Euros and a production cost of 12.5 Euros. Priority will be given to standard features, such as sorting through channels, instead of more advanced internet features.", "dialogue": "None: The Mommy Or I will go No sir.\nNone: In the car.\nNone: You Good morning.\nNone: Good morning.\nNone: Hmm.\nNone: Well, I'm supposed to go home.\nNone: I think it's supposed to be like this.\nSpeaker B: Okay.\nSpeaker E: That was good.\nSpeaker B: Oh, no, I don't think it's actually...\nSpeaker B:...9 here.\nSpeaker B: And you have to adjust the length.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: Okay.\nSpeaker B: I am done.\nNone: Okay.\nSpeaker A: So, we will wait for a few minutes.\nSpeaker B: Yeah.\nSpeaker E: Oh, he goes his weather.\nSpeaker B: Well, he can put anywhere you want, actually.\nSpeaker E: Yeah, but the mic is not...\nSpeaker B: It's not a directional mic, anyway.\nSpeaker A: It's a work like this.\nSpeaker A: So, I will try to get my presentation, all right?\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker D: Go ahead.\nSpeaker D: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker A: Okay.\nSpeaker B: Okay.\nSpeaker E: Then press the...\nSpeaker E:...this, you know?\nSpeaker A: Just try.\nSpeaker B: Okay.\nSpeaker A: Oh, this is not...\nSpeaker B: I'll...\nSpeaker B:...f-a.\nSpeaker A: But don't have to be around the screen here.\nSpeaker E: I do it.\nNone: Oh.\nSpeaker E: Wow.\nSpeaker E: Amazing.\nSpeaker E: It's working.\nSpeaker A: Okay. Thank you.\nSpeaker D: I'll just flip it on the...\nNone: Okay.\nSpeaker A: So, good morning, everyone.\nSpeaker A: Welcome at the kickoff meeting of our latest project.\nSpeaker A: I hope you all have been updated about it.\nSpeaker A: Yeah.\nSpeaker A: Good.\nSpeaker A: I'm going to do a little bit of a quick introduction.\nSpeaker A: This meeting with an agenda as presented here.\nSpeaker A: After the opening, we will get acquainted to each other.\nSpeaker A: See what our roles are in this project.\nSpeaker A: So...\nSpeaker A: We have been provided with some technical tools to...\nSpeaker A:...communicate and to...\nSpeaker A:...well, learn from each other's plans, as I can say.\nNone: So...\nSpeaker A: We will also try to...\nSpeaker A:...to get acquainted to these tools.\nSpeaker A: So, they are also new to me.\nSpeaker A: I don't know whether you've worked with them before.\nSpeaker A: Then we will come to the actual project plan.\nSpeaker A: I hope it's about the new remote control we are going to.\nSpeaker A: Going to the design.\nSpeaker A: Then we will discuss...\nSpeaker A:...well, how it should be and...\nSpeaker A:...what our new project should look like.\nSpeaker A: And then, after some 25 minutes, I hope we can end this meeting.\nSpeaker A: So...\nSpeaker A:...basically, this is about a new remote control.\nSpeaker A: When you design a new project, you of course want it to be original.\nSpeaker A: We want to be distinguished.\nSpeaker A: People want to look at the shelf.\nSpeaker A: Well, that's the product I need.\nSpeaker A: So, it needs to be trendy.\nSpeaker A: I mean, trendy is what people want, so then they will buy our project.\nSpeaker A: But then, it also should work user-friendly.\nSpeaker A: And otherwise, people...\nSpeaker A:...well, it will not be rated very well in consumer articles, like that.\nSpeaker A: So, the general outline of a new project will be...\nSpeaker A:...we first go through a functional design phase.\nSpeaker A: You all get certain tasks in this phase.\nSpeaker A: And then we will meet again and discuss this functional design.\nSpeaker A: And the same holds for the two phases.\nSpeaker A: After this, the conceptual design.\nSpeaker A: And after that, a more detailed design in which the final project should get...\nSpeaker A:...is definitely the shape.\nSpeaker A: But first, we will do some two frames.\nSpeaker A: In all in front of you, you see the notebooks and not blocks.\nSpeaker A: And we have here a whiteboard.\nSpeaker A: And, well, it should work.\nSpeaker A: I've read it from some colleagues.\nSpeaker A: I should work with some kind of toolbar. I didn't find out yet how it worked.\nSpeaker A: But maybe one of you did, so...\nSpeaker E: And the documents in the shared folder.\nSpeaker A: Yes. Do we have to say something about that?\nSpeaker A: I'm not fully updated about this shared folder.\nSpeaker E: Yeah, I guess we'll have a shared folder with documents that can share.\nSpeaker A: Yes, well, we'll then find out how it works.\nSpeaker A: Yes.\nSpeaker A: Well, this seems to me as some computer program, but I didn't find it yet.\nSpeaker A: So, we'll come to that later.\nSpeaker A: So, now we will try out the whiteboard we have here.\nSpeaker A: So, I would suggest...\nSpeaker A: Each of us is going to...\nSpeaker A: Well, yes.\nSpeaker A: We try to draw on it and then...\nSpeaker A: Well, it should be smart some way.\nSpeaker A: I'm not really sure how this works.\nSpeaker A: Okay. Shall I stop?\nSpeaker C: Yes, I think for us it's just like a normal whiteboard.\nSpeaker C: But they'll be recording.\nSpeaker C: So...\nSpeaker B: Now they will record through that, in a sense, so that we just need to record the strokes.\nSpeaker C: But for us it's just like a normal whiteboard.\nSpeaker A: Okay.\nSpeaker E: But it's...\nSpeaker E: Actually, I think I can't grow with...\nSpeaker A: You...\nSpeaker A: Doesn't it work?\nSpeaker A: Maybe some... maybe Anna, maybe you can start and you can maybe find out.\nSpeaker A: To get a score, right?\nSpeaker A: I have to draw...\nSpeaker A: Why don't you draw your favorite animal on the whiteboard?\nSpeaker C: My favorite animal.\nSpeaker C: Sorry, it's a little time, but up here.\nSpeaker A: Oh, I see...\nSpeaker D: There we go.\nSpeaker A: So, draw it. We will try to guess what it is.\nSpeaker C: I'm a very bad drawer.\nSpeaker D: Weird.\nSpeaker C: You're not going to be able to guess from my drawing.\nSpeaker C: I'm not going to be a bad drawer.\nSpeaker C: There, ears, by the way.\nSpeaker C: No?\nSpeaker C: Close side.\nSpeaker C: Okay, so, like a pet animal.\nSpeaker C: Like a cat.\nSpeaker A: It's like a cat. So, I guess it's a cat.\nSpeaker C: No, not a cat though.\nSpeaker A: What is this now?\nNone: You forgot about him.\nSpeaker E: You're on the knife?\nSpeaker B: Yeah.\nSpeaker B: I think it's fine.\nSpeaker B: I just don't want to cut it off.\nSpeaker B: And this wire, no?\nSpeaker B: We need a wireless microphone.\nSpeaker B: You know?\nSpeaker B: Especially, we should next pause it.\nSpeaker B: We should take it like that.\nSpeaker A: Oh, sorry.\nSpeaker A: That's the cat.\nSpeaker A: It's not a cat.\nSpeaker A: It's a dog.\nSpeaker A: Yes.\nSpeaker A: So good. That's also kind of cat.\nSpeaker C: Oh, the dog doesn't have a tail?\nSpeaker C: It's got a tail in.\nSpeaker B: Both predators, sure.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I thought so. The dogs have a tail.\nSpeaker B: It's all for the cats.\nSpeaker C: Did you?\nSpeaker D: And he gets cast out of tail.\nSpeaker E: Yeah, I think I will go without it.\nSpeaker E: Right?\nSpeaker A: Okay.\nSpeaker B: It will still not extend, right?\nSpeaker B: It's not up to that.\nSpeaker C: Yeah.\nSpeaker C: So, what?\nSpeaker C: Favorite characteristics?\nSpeaker C: Dogs are always friendly and loyal and fun.\nNone: Oh.\nNone: What's that?\nNone: To another ski.\nSpeaker B: Of course.\nSpeaker D: Yes.\nSpeaker C: This is why you're the designer.\nSpeaker C: Am I marketing?\nSpeaker C: Yes.\nSpeaker C: Yes.\nSpeaker A: Definitely a whore.\nSpeaker A: Yes.\nSpeaker E: Very good.\nSpeaker E: So, yes, with different colors.\nSpeaker B: So, ah.\nSpeaker B: I think you can put that.\nSpeaker E: We can spend weeks.\nNone: I'm at 20.4.\nNone: Is it there?\nSpeaker D: Yes, sir.\nSpeaker D: So, no, it's zero.\nSpeaker C: A blue and black zebra.\nSpeaker C: Yes.\nSpeaker A: Again, you can meet him in Africa, I think.\nSpeaker A: Very good.\nSpeaker C: Very rare blue zebra.\nSpeaker C: Yes.\nSpeaker B: Oh, I love.\nSpeaker B: Yeah, that's true.\nSpeaker B: I have to go clear off my...\nNone:...lilence.\nNone: Don't need to talk about that.\nSpeaker B: Yes.\nSpeaker B: Sir.\nSpeaker C: You've got a lot of room here, you probably.\nSpeaker C: Rich.\nSpeaker C: It's not for them.\nSpeaker A: I hope you have some space in here, the horse of...\nSpeaker A:...Meal.\nSpeaker B: Sir, what did I draw?\nSpeaker B: We have already taken the pen.\nSpeaker C: That's a good dog.\nNone: Really?\nSpeaker C: A mouse?\nNone: Oh, no.\nSpeaker A: This looks like a cat who has been driven over.\nSpeaker A: A cat?\nSpeaker B: You always play with it.\nSpeaker E: Oh, okay.\nSpeaker E: And we should sum up its favorite characteristics, right?\nSpeaker B: Yeah, some cats would be...\nNone:...the cats.\nSpeaker A: Yes, the mooftars.\nSpeaker A: So, that's the leek cat?\nSpeaker E: Yeah.\nSpeaker E: They like to sleep, that's why you said you...\nSpeaker E:...they are like this.\nNone: It's easy to just take a rest to me.\nSpeaker A: It's quite, you know, relaxed situation.\nSpeaker B: Yeah, it's quite relaxed today.\nSpeaker B: It's like a...\nNone:...they're very good at sex for you.\nSpeaker B: That's okay.\nSpeaker E: So, you have some small legs.\nSpeaker A: Thank you, Matthew.\nNone: Oh, that's kind of bad here.\nSpeaker E: Thank you, Matthew.\nSpeaker C: It's a very big rat, very small rat.\nSpeaker B: I hope they get something.\nSpeaker A: Yes, this is certain...\nSpeaker A:...some contribution to our positive name.\nSpeaker A: So, let's see which animal has not been drawn yet.\nSpeaker A: So, you've all drawn bland animals, so why not draw an animal in water?\nSpeaker E: Okay, in the water.\nSpeaker C: I don't know what that is.\nSpeaker C: It's a bit hard to guess.\nSpeaker D: Yeah.\nSpeaker E: Put it in colors, maybe to the help.\nSpeaker B: Yes.\nSpeaker E: The cat is going to eat the fish or the...\nSpeaker E:...different bandwidth.\nSpeaker A: So...\nSpeaker C: Oh, it's a shark now.\nSpeaker E: It's a shark, yeah.\nSpeaker A: Oh, yes. Why not?\nSpeaker A: Good idea.\nSpeaker B: I could maybe shark it goes to me.\nSpeaker B: It's going to eat the cat rather than the cat eating the fish.\nSpeaker C: No, it's a swordfish.\nSpeaker E: You have something in Australia, right?\nSpeaker E: Swordfish?\nSpeaker C: Yeah.\nSpeaker C: I don't know.\nSpeaker C: Maybe.\nSpeaker C: Another same one.\nSpeaker A: I hope it still works.\nSpeaker E: I'm sorry.\nSpeaker E: So, I don't know if we need to spend time on that, actually.\nSpeaker A: Well, this tool seems to work.\nSpeaker A: Yeah, exactly.\nSpeaker A: Let's continue to do real stuff.\nSpeaker A: Our project finance thing.\nSpeaker A: When we are...\nSpeaker A:...and when you are going to design, we must keep in mind that the selling price of the product will be about 25 euros.\nSpeaker A: When designing a project, I also look at you a mile.\nSpeaker A: Yeah.\nSpeaker A: Keep in mind, people want to get the feeling this is a 25 euro project.\nSpeaker A: Peruent control?\nSpeaker E: Yeah, project.\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker A: More interesting for our company, of course, profit aim, about 50 million euro.\nSpeaker A: So, we have to sell quite a lot of this things.\nSpeaker A: We will try to get at the international market, so it will be, I think, mainly Europe and Northern America.\nSpeaker A: Oh, yeah, the seven men.\nSpeaker A: Maybe some Asian countries.\nSpeaker A: Also, important for all is the production cost.\nSpeaker A: Maximum 12 euro.\nSpeaker E: So, it's half of the selling price, if I am good in mathematics.\nSpeaker A: Yes.\nSpeaker A: I mean, we still have to make our profit.\nSpeaker A: Of course.\nSpeaker B: You all have to pay.\nSpeaker B: Excuse me.\nSpeaker B: We have to sell at least 4 million to make our own profit.\nSpeaker E: Oh, yeah, very good in mathematics.\nSpeaker A: Yes.\nSpeaker E: 4 million.\nSpeaker A: So, I think when we are working on the international market in principle, it has enough customers.\nSpeaker A: So, when we have a good product, we could meet this aim, I think.\nSpeaker A: So, that about the finance.\nSpeaker A: And now, let's have some discussion about what is a good remote control.\nSpeaker A: And, well, keep in mind this first point, it has to be original, it has to be trendy, it has to be user-friendly.\nSpeaker A: Maybe some one can mention some additional rerequisites for a good remote control.\nSpeaker B: Of course, it should have an honor, but...\nSpeaker A: Yes, it should have the expected functionality of a remote control.\nSpeaker B: Yes.\nSpeaker B: And it depends what application you are using it for.\nSpeaker B: You might need...\nSpeaker A: We are thinking television.\nSpeaker B: We are targeting the television set, so you need to record channels.\nSpeaker B: Yeah.\nSpeaker B: You need to browse the channels in upwards.\nSpeaker A: Yes, that's very handy.\nSpeaker A: I always miss it and also remote controls that you can go channel up or down instead of re-typing the number, especially when you have to load channels.\nSpeaker E: And just before starting the detailed discussion, maybe you are the marketing guy.\nSpeaker E: I'm marketing.\nSpeaker E: So, you are marketing and you are in user interface design.\nSpeaker E: So, I just wanted to be sure.\nSpeaker E: And I am the industrial designer.\nSpeaker E: Yes.\nSpeaker E: Because I don't know you very well actually.\nSpeaker D: I am not too well.\nSpeaker E: I'm happy to meet you.\nSpeaker B: And I'm not.\nSpeaker B: And I'm not.\nSpeaker B: I'm not too.\nSpeaker B: I think you know me.\nSpeaker B: Yes.\nSpeaker E: Just on your web page, but not face to face.\nSpeaker E: So...\nSpeaker A: So, some other very important things to do.\nSpeaker A: Well, to specify in this first phase of the project, the browse function.\nSpeaker A: Yeah.\nSpeaker B: And you need the user once like changing the volume, changing the channel and then you, today we have teletext and all those things.\nSpeaker B: Tomorrow you might have some more functions which might come to the...\nSpeaker E: Like what?\nSpeaker E: Like internet on the...\nSpeaker B: You know we are looking for television things or IP.\nSpeaker B: For example, personal video recorder and all those stuff are coming up.\nSpeaker C: But we can't really design for something that hasn't been invented yet.\nSpeaker B: It's coming up actually.\nSpeaker B: The personal video recorder and all those things.\nSpeaker B: Well, I think...\nSpeaker E: Actually, yes.\nSpeaker A: You should, I think, think this over especially what functionality.\nSpeaker E: Of course, and first, before designing the...\nSpeaker E: Well, thinking about the functionalities, we need to know what are the user requirements.\nSpeaker E: Then if they need internet, then we would be able to propose something with the TV over IP.\nSpeaker E: But before...\nSpeaker C: Ninety percent of the time, ninety-nine percent of the time people will be using the main functions, the volume, the different channels.\nSpeaker C: So we can have all the fancy things as well.\nSpeaker C: But the main controls need to be very obvious and very easy to use.\nSpeaker A: Keep in mind, it's a 25 euro unit.\nSpeaker A: So the very fancy stuff, we can leave it out, I think.\nSpeaker A: So 25 euro, you expect a quite, well, normal, but good functioning user-friendly remote control.\nSpeaker B: In that case, you always hook up with someone who is providing that and you know...\nSpeaker B: You sell that product as well as your product with them.\nSpeaker B: So try and get TV, many things.\nSpeaker B: We're on to designing you one.\nSpeaker B: Yeah.\nSpeaker B: No, it's okay.\nSpeaker B: Yeah, I understand.\nSpeaker B: So we need some numbering buttons, some teletext things.\nSpeaker B: And then...\nSpeaker E: The main is...\nSpeaker A: About the buttons that will be on it.\nSpeaker A: I think we can discuss that in the next meeting.\nSpeaker A: Yeah.\nSpeaker A: I would like to get this wrapped up and go to the end of this meeting.\nSpeaker B: Okay.\nSpeaker B: Okay, we are all ready.\nSpeaker A: Okay.\nSpeaker A: So you know now the basic things.\nSpeaker A: Yeah.\nSpeaker A: And...\nSpeaker A: Well, just for the next meeting...\nNone: Yeah.\nSpeaker A: You will work on eight designs.\nSpeaker A: Keep it general.\nSpeaker A: I mean, so we will be still flexible with maybe adding some functions.\nSpeaker A: So...\nSpeaker A: You will be working on technical function design.\nSpeaker B: Yeah, sure.\nSpeaker A: And...\nSpeaker A: When you...\nSpeaker A: Well, think about requirements.\nSpeaker A: Does it need internet?\nSpeaker A: Or do we stay at...\nSpeaker A: Basic...\nSpeaker A: Basic television interface?\nSpeaker A: So...\nSpeaker A: I think we should now go work at this.\nSpeaker A: And you will be informed to find email and kind of communication.\nSpeaker A: Yep.\nSpeaker A: Okay.\nSpeaker A: So the next meeting will be in 30 minutes.\nSpeaker A: Keep in mind.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "Bro019", "summary": "The meeting covered the issues with different kinds of mics, progress on the Aurora project, effectiveness of spectral subtraction, adding parameter to spectral coefficient, and studying energy in the data, along with a short discussion on student housing. The team thought that the LDA was doing well for the Aurora project, where the team had made good progress. The linear normalization was not as good. For spectral subtraction, the team discussed recreating data with FFT of combining the Weiner filtering with the signal subspace. The team had not made much progress on using spectral coefficients for voice-unvoice. The current approach was to study the variance, so other ideas were suggested. The meeting ended with a short discussion on student housing and a graduate student's interest in studying energy.", "dialogue": "Speaker G: We're on.\nSpeaker G: 24th?\nSpeaker C: Yeah.\nSpeaker C: Joggy's a mic, wireless?\nSpeaker C: Yes.\nSpeaker C: Wireless headset?\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker D: Yeah, for you at this.\nSpeaker G: Yeah, we abandoned the lapel because they sort of were not too hot, not too cold.\nSpeaker G: They were far enough away that you got more background noise and so forth, but they weren't so close that they got quite the really good.\nSpeaker G: They didn't, I'm saying the right, they were not so far away that they were really good representative distant mics.\nSpeaker G: On the other hand, they were not so close that they got rid of all the interference, so it didn't seem to be a good point to them.\nSpeaker G: On the other hand, if you only had to have one mic in some ways, you could argue that lapel was a good choice precisely because it's in the middle.\nSpeaker G: There's some kinds of junk that you get with these things that you don't get with lapel, little mouth clerks, and breaths, and so forth.\nSpeaker G: They're worse with these than with lapel, but given the choice, we, there seemed to be very strong opinions for getting rid of lapels.\nSpeaker G: So...\nSpeaker C: You mic numbers.\nSpeaker D: Your mic number is written on the back of that unit there?\nSpeaker D: Oh, yeah.\nSpeaker D: And then the channel number is usually one less than that.\nSpeaker D: Oh, okay.\nSpeaker D: It's one less than what's written on the back.\nSpeaker D: Okay.\nSpeaker D: So you should be zero actually.\nSpeaker G: Yep.\nSpeaker G: Or your channel number.\nSpeaker G: You should do a lot of talking so we get a lot more of your pronunciations.\nSpeaker D: So what we usually do is we typically will have our meetings and then at the end of the meetings we'll read the digits.\nSpeaker D: Everybody goes around and reads the digit session.\nSpeaker D: The bottom of their form.\nSpeaker D: Our 19?\nSpeaker D: Our 19.\nSpeaker D: Yeah, we're a succession of our 19.\nSpeaker G: If you say so.\nSpeaker G: Okay, do we have any kind of agenda?\nSpeaker G: What's going on?\nSpeaker G: I guess...\nSpeaker G: So...\nSpeaker G: So they'll see here for the summer, right?\nSpeaker G: So one thing is to talk about a kickoff meeting maybe.\nSpeaker G: And then just, I guess, progress reports individually.\nSpeaker G: And then plans for where we go between now and then pretty much.\nSpeaker D: I could say a few words about some of the compute stuff that's happening around here.\nSpeaker D: So the people in the group know.\nSpeaker G: Okay.\nSpeaker G: Why don't you start with that?\nSpeaker D: So we just put in an order for about 12 new machines to use as sort of a compute farm.\nSpeaker D: And we ordered to send blade 100s.\nSpeaker D: And I'm not sure exactly how long it'll take for those to come in.\nSpeaker D: But in addition, we're running...\nSpeaker D: So the plan for using these is we're running P-Make and Customs here in Andreas has sort of gotten that all fixed up and up to speed.\nSpeaker D: And he's got a number of little utilities that make it very easy to run things using P-Make and Customs.\nSpeaker D: You don't actually have to write P-Make scripts and things like that.\nSpeaker D: The simplest thing.\nSpeaker D: And I can send an email around or maybe I should do an FAQ on the website about it or something.\nSpeaker D: But an email that points to the FAQ.\nSpeaker D: There's a command that you can use called Run command, Run-Command, Run-Hipin command.\nSpeaker D: And if you say that and then some job that you want to execute, it will find the fastest currently available machine and export your job to that machine.\nSpeaker D: And run it there and it'll duplicate your environment.\nSpeaker D: So you can try this as a simple test with the LS command.\nSpeaker D: So you can say Run-Command, LS, and it'll actually export that.\nSpeaker D: LS command to some machine in the institute and do an LS on your current directory.\nSpeaker D: So substitute LS for whatever command you want to run.\nSpeaker D: And that's a simple way to get started using this.\nSpeaker D: And so soon when we get all the new machines up, then we'll have lots more compute to use.\nSpeaker D: Now one of the nice things is that each machine that's part of the P-Make and Customs network has attributes associated with it.\nSpeaker D: Attributes like how much memory the machine has, what its speed is, what its operating system.\nSpeaker D: And when you use something like Run-Command, you can specify those attributes for your program.\nSpeaker D: For example, if you only want your thing to run under Linux, you can give it the Linux attribute.\nSpeaker D: And then it will find the fastest available Linux machine and run it on that.\nSpeaker D: So you can control where your jobs go, to some extent, all the way down to an individual machine.\nSpeaker D: Each machine has an attribute, which is the name of itself.\nSpeaker D: So you can give that as an attribute and it'll only run on that.\nSpeaker D: And if there's already a job running on some machine that you're trying to select, your job will get queued up.\nSpeaker D: And then when that resource, that machine becomes available, your job will yet exported there.\nSpeaker D: So there's a lot of nice features to it and it kind of helps to balance the load of the machines.\nSpeaker D: And right now, Andreas and I have been the main ones using it.\nSpeaker D: And where the SRI recognizer has all this P-Make custom stuff built into it.\nSpeaker G: So as I understand here, he's using all the machines and you're using all the machines.\nSpeaker G: Yeah, exactly.\nSpeaker D: Yeah, you know, I sort of got started using the recognizer just recently and I fired off a training job and then I fired off a recognition job and I get this email about midnight from Andreas saying, are you running two trainings simultaneously?\nSpeaker D: My jobs are not getting run.\nSpeaker D: So I had to back off a little bit.\nSpeaker D: But as soon as we get some more machines, then we'll have more compute available.\nSpeaker D: So that's just a quick update.\nSpeaker D: What we've got.\nSpeaker F: I have a question about the parallelization.\nSpeaker F: So let's say I'm like a thousand little jobs to do.\nSpeaker F: How do I do it with run command?\nSpeaker D: You could write a script, which called run command on each subjob.\nSpeaker D: Right.\nSpeaker D: But you probably want to be careful with that because you don't want to saturate the network.\nSpeaker D: So you should probably not run more than say ten jobs yourself at any one time.\nSpeaker D: Just because then it would keep other people.\nSpeaker D: Well, it's not that so much as that, you know, if everybody ran 50 jobs at once, then it would just bring everything to a halt and people's jobs would get delayed.\nSpeaker D: So it's sort of a sharing thing.\nSpeaker D: So you should try to limit it to sometime, some number around ten jobs at a time.\nSpeaker D: So if you had a script, for example, that had a thousand things that needed to run, you'd somehow need to put some logic in there.\nSpeaker D: If you were going to use run command to only have ten of those going at a time, and then when one of those finished, you'd fire off another one.\nSpeaker G: I remember I figured whether it was when the Rutgers or Hopkins workshop, I remember one of the workshops I was at there where I was really excited because I got 25 machines and there's some kind of p-make-like thing that sent things out.\nSpeaker G: So all 25 people were sending things to all 25 machines.\nSpeaker G: And things were a lot less efficient than if you just used your own machine.\nSpeaker G: Yeah, exactly.\nSpeaker D: It was a very cool thing.\nSpeaker D: Yeah, you have to be a little bit careful.\nSpeaker D: But you can also, if you have that level of parallelization, and you don't want to have to worry about writing the logic in a Perl script to take care of that, you can use p-make.\nSpeaker D: And you're basically right to make a file that, you know, your final job depends on these one thousand things.\nSpeaker D: And when you run p-make on your make file, you can give it the dash capital J and a number, and that number represents how many machines to use at once.\nSpeaker D: And it'll make sure that it never goes above that.\nSpeaker D: Okay.\nSpeaker E: So it's not systematically queued.\nSpeaker E: I mean, all the japs are running.\nSpeaker E: If you launch 20 japs, they're all running.\nSpeaker E: It depends.\nSpeaker D: Because if you run command that I mentioned before, doesn't know about other things that you might be running.\nSpeaker D: So it would be possible to run a hundred run jobs at once.\nSpeaker D: And they wouldn't know about each other.\nSpeaker D: But if you use p-make, then it knows about all the jobs that it has to run.\nSpeaker D: And it can control how many are runs simultaneously.\nSpeaker G: So run command doesn't use p-make?\nSpeaker D: It uses export under-lyingly.\nSpeaker D: But it's meant to be run one job at a time.\nSpeaker D: So you could fire off a thousand of those.\nSpeaker D: And it doesn't know any one of those doesn't know about the other ones that are running.\nSpeaker G: So why would one use that rather than p-make?\nSpeaker D: Well, if you have, like for example, if you didn't want to write a p-make script, and you just had an HTK training job that you know is going to take six hours to run.\nSpeaker D: And somebody's using the machine you typically use.\nSpeaker D: You can say run command and your HTK thing.\nSpeaker D: And it'll find another machine, the fastest currently available machine, and run your job.\nSpeaker G: Now does it have the same sort of behavior as p-make?\nSpeaker G: Which is that, you know, if you run something and somebody's machine, they come in and hit a key, then it's...\nSpeaker D: Yes, yeah.\nSpeaker D: There are...\nSpeaker D: Right.\nSpeaker D: So some of the machines at the institute have this attribute called NOEVICT.\nSpeaker D: And if you specify that in one of your attribute lines, then it'll go to a machine which your job won't be evicted from.\nSpeaker D: But the machines that don't have that attribute, if a job gets fired up on that, which could be somebody's desktop machine, and they were at lunch, they come back from lunch, and they start typing on the console, then your machine will get evicted...\nSpeaker D: Your job will get evicted from their machine and be restarted on another machine automatically.\nSpeaker D: So which can cause you to lose time, right?\nSpeaker D: If you had a two hour job and you got halfway through, and then somebody came back to their machine and got evicted.\nSpeaker D: So if you don't want your job to run on a machine where it could be evicted, then you give it that minus the attribute, you know, NOEVICT.\nSpeaker D: And it'll pick up a machine that it can't be evicted.\nSpeaker G: What about...\nSpeaker G: Remember, it was used to be an issue, maybe it's not anymore, that if you...\nSpeaker G: If something required...\nSpeaker G: If your machine required somebody hitting a key in order to evict things around it, so you could work, but if you were logged into it from home, and you weren't hitting any keys because you were home.\nSpeaker D: Yeah, I'm not sure how that works.\nSpeaker D: It seems like Andreas did something for that.\nSpeaker D: Okay, we can ask him.\nSpeaker D: Yeah, I don't know whether monitors, the keyboard, or it actually looks at the console, TTY, so maybe if you echoed something to the...\nSpeaker G: You probably wouldn't door it in early though, you're right.\nSpeaker G: You probably wouldn't order it in early.\nSpeaker G: I mean, you're sort of...\nSpeaker G: You're at home and you're trying to log in, it takes forever to even log in, and then you probably go screw this.\nSpeaker D: Yeah, so...\nSpeaker D: Yeah, I'm not sure about that one.\nSpeaker G: Yeah.\nSpeaker G: Okay.\nSpeaker C: I knew a little orientation about this environment and how to run some jobs here, because I never did anything so far for these exhibitions.\nSpeaker C: Maybe I'll ask you after the meeting.\nSpeaker D: Yeah, and also, Stefan's a really good resource for that feature.\nSpeaker D: Okay, I'm sure.\nSpeaker D: Especially with regard to the Aurora stuff.\nSpeaker D: He knows that stuff better than I do.\nSpeaker H: Okay.\nSpeaker G: Well, when we...\nSpeaker G: Sinales and Sier...\nSpeaker G: I'm been one of these.\nSpeaker G: That one, you tell us what's up with you.\nSpeaker G: Which would be not bad, hopefully.\nSpeaker C: Yeah.\nSpeaker C: So...\nSpeaker C: I don't know, shall I start from...\nSpeaker C: I don't know, how...\nSpeaker C: Okay, I think I'll start from the post...\nSpeaker C: Aurora submission maybe.\nSpeaker C: Yeah.\nSpeaker C: Yeah, after the submission, what I've been working on really was to take other submissions and then...\nSpeaker C:...over their system, what they submitted, because we didn't have any speech and enhancement system.\nSpeaker C: So I tried...\nSpeaker C:...and...\nSpeaker C:...and then I found that...\nSpeaker C:...when you were combined with LD8, you used a form of movement over there.\nSpeaker D: Are you saying LDA?\nSpeaker C: Yeah, LDA.\nSpeaker C: Yeah.\nSpeaker C: So just the LDF filters.\nSpeaker C: I just plug in...\nSpeaker C: I just take the capsule coefficients coming from their system and then plug in LD on top of that.\nSpeaker C: But LDF filters that I used was different from what we submitted in the proposal.\nSpeaker C: What I did was I took the LDF filters designed using clean speech.\nSpeaker C: Mainly because the speech is already cleaned up after the enhancement.\nSpeaker C: So instead of using this...\nSpeaker C:...narrow band LDF filters that we submitted, I got new filters.\nSpeaker C: So that seems to be giving...\nSpeaker C:...improving over their system slightly, but not very significant.\nSpeaker C: And that was...\nSpeaker C:...showing any improvement over...\nSpeaker C:...finally by plugging in an LD8.\nSpeaker C: And so then after that I added the online normalization also on top of that.\nSpeaker C: And there also, I found that I have made some changes to their time constant that I used.\nSpeaker C: Because it has a mean and variance update time constant, which is not suitable for the enhanced speech in whatever we tried on the proposal one.\nSpeaker C: But I didn't play with that time constant a lot.\nSpeaker C: I just found that I have to reduce the value...\nSpeaker C: I mean I have to increase the time constant or reduce the value of the update value.\nSpeaker C: That's all I found so I had to...\nSpeaker C: Yeah, and the other thing what I tried was...\nSpeaker C: I just took the baseline and then ran it with the endpoint information.\nSpeaker C: Just the Aurora baseline to see that how much the baseline itself improves...\nSpeaker C:...by just supplying the information of the speech and non-speech.\nSpeaker C: And I found that the baseline itself improves by 22% by just giving the...\nSpeaker G: I can back up a second.\nSpeaker G: I missed something.\nSpeaker G: I guess my line wondered that when you added the online normalization and so forth...\nSpeaker G:...things got better again.\nSpeaker G: No, no, no.\nSpeaker C: Things didn't get better with the same time constant that we used.\nSpeaker G: No, no, with a different time constant.\nSpeaker C: With a different time constant, I found that...\nSpeaker C:...I mean I didn't get an improvement over not using online normalization.\nSpeaker C: Oh, no, I didn't.\nSpeaker C: Because I found that I would have to change the value of the update factor.\nSpeaker C: But I played with...\nSpeaker C:...play quite a bit to make it better than...\nSpeaker C: Okay.\nSpeaker C: So it's still not with online normalization didn't give me any improvement.\nSpeaker C: And...\nSpeaker C: So...\nSpeaker C: So I just stopped there with the speech announcement.\nSpeaker C: The other thing what I tried was adding the endpoint information to the baseline...\nSpeaker C:...and that itself gives like 22% because the second...\nSpeaker C:...the new phase is going to be with the end-pointed speech.\nSpeaker C: And just to get a feel of how much the baseline itself is going to change...\nSpeaker C:...by adding this endpoint information.\nSpeaker D: So people won't even have to worry about doing speech non-speech.\nSpeaker C: Yeah, that's what the feeling is like.\nSpeaker C: They're going to give the endpoint information.\nSpeaker G: I guess the issue is that people do that anyway.\nSpeaker G: Everybody does that.\nSpeaker G: I mean, I wanted to see given that you're doing that...\nSpeaker G:...what are the best features that you have.\nSpeaker G: I see.\nSpeaker G: So...\nSpeaker G: I mean, clearly they're interact.\nSpeaker G: So I don't know that I entirely agree with it.\nSpeaker G: But it might be...\nSpeaker G: In some ways it might be better to...\nSpeaker G:...rather than giving the endpoints to have a standard that everybody uses...\nSpeaker G:...and then interacts with.\nSpeaker G: But, you know, it's still something reasonable.\nSpeaker D: So are people supposed to assume that there is...\nSpeaker D:...are people not supposed to use any speech outside of those endpoints?\nSpeaker D: Or can you then...\nSpeaker D: No, no, that's not.\nSpeaker C: So I think each outside of it for estimating background noise.\nSpeaker C: Exactly.\nSpeaker C: I guess that is what the consensus is.\nSpeaker C: You will...\nSpeaker C: You will be given the information about the beginning and the end of speech.\nSpeaker C: But the whole speech is available to you.\nSpeaker G: Okay.\nSpeaker G: So should make the spectral subtraction style things work even better...\nSpeaker C:...because you don't have the mistakes.\nNone: Yeah.\nSpeaker C: Okay.\nSpeaker C: So that...\nSpeaker C: That baseline itself...\nSpeaker C: I mean, it improves by 20 to 1.\nSpeaker C: And I found that in one of the speech that carcases...\nSpeaker C:...proves by just 50% by just putting the endpoints.\nSpeaker C: Wow.\nSpeaker C: And you don't need any speech in answer.\nSpeaker C: So the baseline itself improves by 50%.\nSpeaker C: Yeah, by 50%.\nSpeaker G: Yeah, so it's going to be harder to...\nSpeaker C: Yeah.\nSpeaker C: Big that actually.\nSpeaker C: Yeah.\nSpeaker C: So that is when the qualification criteria was reduced...\nSpeaker C:...from 50% to something like 25% for well-matched.\nSpeaker C: I think they have actually changed the qualification criteria now.\nSpeaker C: And yeah, I guess after that...\nSpeaker C:...I just went home for...\nSpeaker C: I just had a vacation for...\nSpeaker C: Sorry.\nSpeaker C: Okay.\nSpeaker G: That's good.\nSpeaker G: That's good.\nSpeaker C: Yeah.\nSpeaker C: And I came back and I started working on...\nSpeaker C:...some other speech enhancement algorithm.\nSpeaker C: I mean, so from the submission...\nSpeaker C:...you are found that people have tried to...\nSpeaker C:...spectral subtraction and venerable filtering.\nSpeaker C: These are the main approaches...\nSpeaker C:...what people have tried.\nSpeaker C: So just to...\nSpeaker C:...just to fill the space with some few more speech enhancement...\nSpeaker C:...algues them to see whether it...\nSpeaker C:...I've been working on this...\nSpeaker C:...signal subspace approach for speech enhancement...\nSpeaker C:...where you...\nSpeaker C:...take the noise-y signal and decompose into signal...\nSpeaker C:...and the noise subspace...\nSpeaker C:...and then try to estimate the clean speech from the signal person...\nSpeaker C:...nois subspace.\nSpeaker C: And so I've been actually running some...\nSpeaker C:...so far I've been trying it only on MATLAB...\nSpeaker C:...to test whether it was first or not.\nSpeaker C: And then I'll put it to see...\nSpeaker C:...then update it with the repository once I've...\nSpeaker C:...finally giving you some positive result.\nSpeaker G: So you said one thing...\nSpeaker G:...I wanted to jump on for a second.\nSpeaker G: So now you're getting tuned into the repository...\nSpeaker G:...thing that he has here and so we'll have a...\nSpeaker G:...simple place for this stuff.\nSpeaker G: It's cool.\nSpeaker G: So maybe just briefly...\nSpeaker G:...you could remind us about the related experience...\nSpeaker G:...because you did some stuff...\nSpeaker G:...you talked about last week I guess...\nSpeaker G:...where you were also combining something...\nSpeaker G:...both of you I guess we're combining something...\nSpeaker G:...from the telecom system with...\nSpeaker G:...I know whether it was system one or system two...\nSpeaker E:...it was system one.\nSpeaker E: So the main thing that we did is...\nSpeaker E:...just to take the spectrosuppraction from the front silicon...\nSpeaker E:...which provide us some speech samples...\nSpeaker G:...that are with noise removed.\nSpeaker G: So let me just stop you there.\nSpeaker G: So then one distinction is that...\nSpeaker G:...you were taking the actual front telecom features...\nSpeaker G:...and then applying something...\nSpeaker C: No, there was a slight difference.\nSpeaker C:...I mean, these are extractors at the handset.\nSpeaker C: Yeah.\nSpeaker C: Because they had another back end blinding...\nSpeaker G: Yeah, but that's what I mean.\nSpeaker G: Sorry, I'm not being clear.\nSpeaker G: What I meant was you had something like Kepstra or something.\nSpeaker G: And so one difference is that I guess you were taking...\nSpeaker G:...spec graph.\nSpeaker E: Yeah.\nSpeaker E: But I guess it's exactly the same thing...\nSpeaker E:...because on the handset they just...\nSpeaker E:...applied the Wiener filter...\nSpeaker E:...and then compute Kepstra features.\nSpeaker C: Yeah, the Kepstra...\nSpeaker C:...the difference is like...\nSpeaker C:...there may be a slight difference in the way...\nSpeaker C:...because they use exactly the baseline system...\nSpeaker C:...for computing the Kepstra once you have the speech.\nSpeaker C: If we are using our own code...\nSpeaker C:...that could be the only difference.\nSpeaker C: But you got some sort of different results...\nSpeaker G:...I'm trying to understand that.\nSpeaker E: Yeah, I think we should...\nSpeaker E:...have a table with all the results...\nSpeaker E:...because I don't know...\nSpeaker E:...I don't exactly know what are your results.\nSpeaker E: Okay.\nSpeaker E: Yeah, but so we did this...\nSpeaker E:...and another difference, I guess, is that we just applied...\nSpeaker E:...properse our one system after this...\nSpeaker E:...with our modification to reduce the delay of the ADA filters.\nSpeaker E: And other slight modifications...\nSpeaker E:...but it was the full proposal one.\nSpeaker E: In your case, you tried...\nSpeaker E:...only a little bit...\nSpeaker E:...just putting it the ADA and maybe...\nSpeaker E:...on the optimization.\nSpeaker C: After that, I don't know the situation.\nSpeaker E: So we just tried directly to...\nSpeaker E:...just keep the system as it was.\nSpeaker E: And...\nSpeaker E:...when we plug the spectra-attraction...\nSpeaker E:...it improves significantly.\nSpeaker E: But what seems clear also is that...\nSpeaker E:...we have to retune...\nSpeaker E:...the time constants of the online organization...\nSpeaker E:...because if we keep the value that was submitted...\nSpeaker E:...it doesn't help at all.\nSpeaker E: You can remove online organization...\nSpeaker E:...or put it, it doesn't change anything.\nSpeaker E: As long as you have the spectra-attraction.\nSpeaker E: But you can still find...\nSpeaker E:...some kind of optimum somewhere...\nSpeaker E:...and we don't know where exactly.\nSpeaker E: So it sounds like...\nSpeaker G:...you should look at some tables of results...\nSpeaker G:...or something and see where...\nSpeaker G:...where they were different when we can learn from it.\nSpeaker C: Without any change.\nSpeaker C: Yeah.\nSpeaker E: Well, we change it.\nSpeaker E: We have to have...\nSpeaker E:...alie filters.\nSpeaker E: There are other things that we...\nSpeaker E:...finally were shown to improve.\nSpeaker E: So like the 64-hz cutoff...\nSpeaker E:...it doesn't seem to hurt on TI digits, finally.\nSpeaker E: Maybe because of other changes.\nSpeaker E: Well, there are some minor changes.\nSpeaker E: And right now if we look at the results...\nSpeaker E:...it's always better than...\nSpeaker E:...it seems always better than the French Telecom for mismatch.\nSpeaker E: I mismatch.\nSpeaker E: And it still slightly worse for well-matched.\nSpeaker E: But this is not significant.\nSpeaker E: But the problem is that it's not significant.\nSpeaker E: But if you put this in the spreadsheet...\nSpeaker E:...it's still worse, even with very minor...\nSpeaker E:...even if it's only slightly worse for well-matched.\nSpeaker E: And significantly better for HM.\nSpeaker E: But well, I don't think it's important.\nSpeaker E: Because when they will change their mid-track...\nSpeaker E:...mainly because of when you plug the frame dropping...\nSpeaker E:...in the baseline system...\nSpeaker E:...it will improve a lot HM and MM.\nSpeaker E: So I guess what will happen?\nSpeaker E: I don't know what will happen.\nSpeaker E: But the different contribution I think for a different test set will be more...\nSpeaker C: HM and MM will also go down significantly in the spreadsheet.\nSpeaker C: But the well-matched may still...\nSpeaker C:...I mean, the well-matched may be the one which is least affected by adding the endpoint information.\nSpeaker C: So the MM and HM are going to be hugely affected.\nSpeaker C: Yeah.\nSpeaker C: But everything is like...\nSpeaker C:...that's how they reduce the qualification to 25% or something.\nSpeaker G: But are they changing the waiting?\nSpeaker C: No, I guess they're going ahead with the same rating.\nSpeaker G: I don't understand that.\nSpeaker G: I guess I have been part of the discussion.\nSpeaker G: So it seems to me that the well-matched condition is going to be unusual.\nSpeaker G: In this case, unusual.\nSpeaker G: Because you don't actually have good matches ordinarily for what any particular person's car is like.\nSpeaker G: Or it seems like something like the middle one is more natural.\nSpeaker G: So I don't know why the well-matched is...\nSpeaker C: Yeah, but actually the well-matched...\nSpeaker C: I mean, the well-matched condition is not like the warning PI-Tits where you have all the training conditions exactly like replicated in that testing condition.\nSpeaker C: Also, this is not calibrated by SNR or something. The well-matched has also some mismatching match.\nSpeaker G: The well-matched has mismatch.\nSpeaker C: Also some slight mismatches.\nSpeaker C: Unlike the PI-Tits where it's perfectly matched because it's artificially added now.\nSpeaker C: But this is natural recording.\nSpeaker G: So remind me of what well-matched you've told many times.\nSpeaker C: The well-matched is defined like it's 70% of the whole database is useful training and 30% of the full testing.\nSpeaker E: So it means that if the database is large enough, it's matched.\nSpeaker E: Because in each set you have a range of conditions.\nSpeaker G: So I mean, yeah, unless they deliberately chose it to be different, which they didn't because they wanted to be well-matched.\nSpeaker G: It is pretty much so it's sort of saying that you're in heat though.\nSpeaker G: It's not guaranteed.\nSpeaker C: Because the main reason for the mismatch is coming from the amount of noise and the silence frames and all those present in the database.\nSpeaker G: Again, if you have enough, if you have enough.\nSpeaker G: So it's sort of like if you're saying, okay, so much as you train your dictation machine for talking into your computer, you have a car.\nSpeaker G: And so you drive it around a bunch and record noise conditions or something.\nSpeaker G: And then I don't think that's very realistic.\nSpeaker G: I guess they're saying that if you were a company that was selling the stuff commercially, that you would have a bunch of people driving around in a bunch of cars and you would have something that was roughly similar.\nSpeaker G: And maybe that's the argument, but I'm not sure I'd buy it.\nSpeaker G: So, well, it's gone.\nSpeaker E: Yeah, we are also playing in trying to put other spectrosuppraction in the code.\nSpeaker E: It would be a very simple spectrosuppraction on the male energies, which I already tested, but without the frame dropping actually.\nSpeaker E: And I think it's important to have frame dropping.\nSpeaker D: Is it a spectrosuppraction typically done on the after the male scaling or is it done on the FFT bands?\nSpeaker D: Does it matter?\nSpeaker E: I don't know. Well, both cases.\nSpeaker E: Yeah, some of the proposes, I were doing this on the FFT bands, others on the male energies.\nSpeaker E: You can do both, but I cannot tell you which one might be better.\nSpeaker C: I guess if you want to reconstruct the speed, it may be a good idea to do it on FFT bands.\nSpeaker C: But for speech recognition, it may not be very different if you do it on male warmth or whether you do it on FFT.\nSpeaker C: So, you're going to do a linear waiting anyway after that.\nSpeaker E: Well, it gives something different, but I don't know what are the pros and cons of both.\nSpeaker C: The other thing is, when you're putting in a speech enhancement technique, is it like one stage speech enhancement?\nSpeaker C: Because everybody seems to have two stages of speech enhancement and all the proposes, which is giving them some improvement.\nSpeaker C: I mean, they just do the same thing again once more.\nSpeaker C: And so, there's something that is good about doing it, when cleaning it up once more.\nSpeaker E: Yeah, it might be. So, maybe in my implementation, I should have so tried to inspire me for this kind of thing.\nSpeaker G: Well, the other thing would be to combine what you're doing.\nSpeaker G: Maybe one or the other things that you're doing would benefit from the other happening first.\nSpeaker G: So, if he's doing a signal subspace thing, maybe it would work better if you'd already done some simple spectrosituaction or maybe the other way around.\nSpeaker C: So, the thing about combining the Venerful drink with signal subspace, just to see some such permutation combination to see whether it really helps.\nSpeaker E: Yeah.\nSpeaker G: How is, I guess I'm ignorant about this.\nSpeaker G: I mean, since Vener filter also assumes that you're adding together the two signals, how is that different from signal subspace?\nSpeaker C: Yeah. The signal subspace approach has actually an inbuilt Vener filtering in it.\nSpeaker C: Okay.\nSpeaker C: Yeah, it is like a KL transform followed by a Vener filter.\nSpeaker C: Oh, okay. So, it definitely has the K.\nSpeaker C: So, the advantage of combining two things is mainly coming from the signal subspace approach doesn't work very well if the SNR is very bad.\nSpeaker C: I see. It's very poorly with that. It's very bad for SNR conditions and in colored noise.\nSpeaker G: So, essentially you could do a simple spectral subtraction followed by a KL transform followed by a Vener filter.\nSpeaker C: It's a cascade of.\nSpeaker G: Yeah, and generally, that's right. You don't want to authorize if the things are noisy, actually.\nSpeaker G: That was something that, here Vener, we're talking about with the multi-band stuff, that if you're converting things to from bands, groups of bands into capture co-efficient, you know, local sort of local capture co-efficient, that it's not that great to do it if it's noisy.\nSpeaker H: Yeah. Okay. Yeah.\nSpeaker C: So, that's one reason maybe it would combine.\nSpeaker C: Something to improve SNR a little bit. Yeah. First stage and then do something and the second stage it should take it.\nSpeaker E: Well, if you're prompt about, about the color noise.\nSpeaker C: Oh, the colored noise. The color noise, the signal subspace approach has, I mean, it actually depends on inverting the matrix.\nSpeaker C: So, it's the co-edits matrix of the noise. So, if it is not positive definite, when it has it, it doesn't behave very well, if it is not positive definite.\nSpeaker C: It works very well with white noise because we know for sure that it has a positive.\nSpeaker G: So, the spectral subtraction had noise.\nSpeaker C: So, the way they get around is like they do an inverse filtering first of the colored noise and they make the noise white.\nSpeaker C: And then finally when you reconstruct the speech back, you do this filtering again.\nSpeaker G: I was only half-caring. You sort of do this spectral subtraction that also gets through.\nSpeaker G: And then you add a little bit of noise, noise addition. I mean, that's sort of what J-Rest it does in a way.\nSpeaker G: If you look at what J-Rest is doing, essentially it's equivalent to adding a little noise.\nSpeaker G: Or you get rid of the effects of noise.\nSpeaker E: Yeah, so there is this. And maybe we find some people also that agree to maybe work with us.\nSpeaker E: And they have implementation of VTS techniques.\nSpeaker E: So, it's a vector autolo series that are used to model the transformation between clean capstras and noisy capstras.\nSpeaker E: So, well, if you take the standard model of channel plus noise, it's a non-linear transformation in the capstras domain.\nSpeaker E: And there is a way to approximate this using first order or second order telor series.\nSpeaker E: And it can be used for getting rid of the noise and channel effect.\nSpeaker E: Who is doing the working in the capstras domain?\nSpeaker E: So, there is one guy in Granada and another in Lucent that I met at ICASP.\nSpeaker G: Who is the guy in Granada?\nSpeaker B: Jos\u00e9 Carlos Segura.\nSpeaker C: This VTS has been proposed by CME.\nSpeaker B: Yeah, yeah, yeah. Original date was from CME.\nSpeaker E: Well, it's again a different thing that could be tried.\nSpeaker G: Yeah. Yeah, so anyway, you're looking general, standing back from it, looking at ways to combine one form or another of noise removal with these other things we have.\nSpeaker G: Looks like a worthy thing to do here.\nSpeaker E: But for sure, let's re-check everything else and re-optimize the other things.\nSpeaker E: For sure, the online normalization may be the LDA filter.\nSpeaker G: Well, it seems like one of the things to go through next week when Harry's here because Harry will have his own ideas to, I guess, not next week, week and a half.\nSpeaker G: We sort of go through these alternatives, what we've seen so far and come up with some game plans.\nSpeaker G: So, I mean, one way would...here's some alternate visions.\nSpeaker G: I mean, one would be you look at a few things very quickly.\nSpeaker G: You pick on something that looks like it's promising and everybody works really hard on the same different aspects of the same thing.\nSpeaker G: Another thing would be to pick two plausible things and you know, have two working things for a while until we figure out what's better.\nSpeaker G: But, you all have some ideas on that too.\nSpeaker C: The other thing is to most of this speech and management techniques have reported research on smaller capital rate assets.\nSpeaker C: But we're going to address this Wall Street Journal in the next stage, which is also going to be in our seat.\nSpeaker C: So, very few people have reported something on using some continuous speech at home.\nSpeaker C: So, there are some...I was looking at some literature on speech and management apply to large vocabulary.\nSpeaker C: Speical subtraction doesn't seem to be the thing to do for large vocabulary.\nSpeaker C: There are always people who have shown improvement with linear filtering and maybe subspace up whichever special subtraction you have.\nSpeaker C: But, we have to use simple spectrum to make it...how to do some optimization.\nSpeaker G: So, they're making...somebody's generating Wall Street Journal without it or artificially out of knowing or something.\nSpeaker G: Sort of like what they do with TI digits.\nSpeaker C: I guess, I guess, the inter-fascism.\nSpeaker G: And then they're generating HTK scripts.\nSpeaker C: Yeah, I know that there's no...I don't know whether they have converged on HTK or using some...\nSpeaker G: It's a Mississippi State, maybe.\nSpeaker G: Yeah, so that'll be a little task in itself.\nSpeaker G: Well, we've...yeah, it's true for the additive noise.\nSpeaker G: Artificial added noise, we've always used smoke vocabulary too.\nSpeaker G: There's been noisy speech, the style of artificial vocabulary that we've worked with in broadcast news.\nSpeaker G: So, we did broadcast news evaluation and some of the focus conditions were noisy.\nSpeaker G: But we didn't do spectrosotraction.\nSpeaker G: We were doing our funny stuff.\nSpeaker G: We were doing multi-stream and so forth.\nSpeaker G: But, you know, stuff we did helped.\nSpeaker G: I mean, it did something.\nSpeaker G: Now, we have this meeting data, the stuff we're recording right now.\nSpeaker G: And that we have for the, the quote-unquote noisy data there is just noisy and reverberant actually.\nSpeaker G: It's the far-failed mic.\nSpeaker G: And we have the digits that we do at the end of these things.\nSpeaker G: And that's what most of our work has been done with that, with connected digits.\nSpeaker G: But, we have recognition now with some of the continuous speech, like vocabulary continuous speech, using switchboard, so a switchboard recognizer.\nSpeaker G: No training from this, just plainly switched words.\nSpeaker G: But, this is the switchboard thing that's what we're doing.\nSpeaker G: Now, there's some adaptation though that Andreas has been playing with.\nSpeaker G: But, we're hoping, actually Dave and I were just talking earlier today about maybe at some point, not that distant future trying some of the techniques that we've talked about on some of the electrical vocabulary data.\nSpeaker G: I mean, I guess no one has done, yet done test one on the distant mic using the SRR recognizer.\nSpeaker G: I don't know.\nSpeaker G: How did I know of it?\nSpeaker G: Yeah.\nSpeaker G: It's a very scared.\nSpeaker G: I see a little smoke coming up from the CPU.\nSpeaker G: You were trying to do it.\nSpeaker G: But, yeah.\nSpeaker G: But, you're right.\nSpeaker G: That's a real good point, that we don't know what, if any of these, I guess that's what they're pushing that in the evaluation.\nSpeaker G: But, it's good.\nSpeaker G: Anything else going on?\nSpeaker G: From the day of the century?\nSpeaker B: I don't have good results with the, including the new parameters, I don't have good results.\nSpeaker B: Are similar or a little bit worse?\nSpeaker G: Yeah, so you probably need to back up a bit, so.\nSpeaker B: I tried to include another new parameter to the traditional parameter, the catch-strong coefficient, that like the autocorrelation, the R0, and R1 over R0, and another estimation of the variance of the difference of the spectrum of the signal, and the spectrum of time after the filter bank.\nSpeaker C: Of course, I should have.\nSpeaker B: Anyway, you have the spectrum of the signal, and the other side you have the output of the filter bank.\nSpeaker B: You can extend the coefficient of the filter bank and obtain an approximation of the spectrum of the signal.\nSpeaker B: I do the difference.\nSpeaker B: I found the difference at the variance of this difference, because we think that if the variance is high, maybe you have noise.\nSpeaker B: If the variance is small, maybe you have a speech.\nSpeaker B: The idea is to find another feature for discriminating between voice sound and voice sound.\nSpeaker B: We try to use this new feature.\nSpeaker B: I need to change the window size of the analysis window size to have more information.\nSpeaker B: This is the 2.5 milliseconds.\nSpeaker B: I did the time of experiment to include this feature directly with the other feature, and to try to use a neural network to select voice and voice silence and to concate this new feature.\nSpeaker B: The result with the neural network, I have more or less the same result.\nSpeaker B: Sometimes it's worse, sometimes it's a little bit better, but not significant.\nSpeaker C: Is it with the IDG?\nSpeaker B: No, I work with the Italian and Spanish, basically.\nSpeaker B: If I use the neural network and use directly the feature, the result are worse.\nSpeaker B: But listen, Harry.\nSpeaker G: I really wonder, though.\nSpeaker G: We've had this discussion before, and one of the things that struck me was that about this line of thought that was particularly interesting to me was that whenever you condense things in an irreversible way, you throw away some information.\nSpeaker G: That's mostly viewed as a good thing, the way we use it because we want to suppress things that will cause variability for a particular phonetic units.\nSpeaker G: But you do throw something away.\nSpeaker G: The question is, can we figure out if there's something we've thrown away that we shouldn't have?\nSpeaker G: When they were looking at the difference between the filter bank and the FFT that was going into the filter bank, I was thinking, oh, they're picking on something, they're looking on it to figure out noise or voice property, whatever, that's interesting, maybe that helps to drive the thought process of coming up with the features.\nSpeaker G: But for me, the interesting thing was, well, but is there just something in that difference, which is useful?\nSpeaker G: So another way of doing it maybe would be just to take the FFT, power spectrum, and feed it into a neural network.\nSpeaker G: And then use it in combination or alone or whatever.\nSpeaker A: With what target?\nSpeaker G: No, we need to. No, just the same way we're using the filter bank.\nSpeaker G: Exactly the same way we're using the filter bank.\nSpeaker G: I mean, the filter bank is good for all the reasons that we say it's good, but it's different.\nSpeaker G: And maybe if it's used in combination, it will get at something that we're missing.\nSpeaker G: And maybe using KLT or adding probabilities, I mean, all the different ways that we've been playing with, that we would let the essentially let the neural network determine what is it that's useful that we're missing here.\nSpeaker E: Yeah, but there's so much variability in the power spectrum.\nSpeaker G: Well, that's probably why it would be unlikely to work as well by itself, but it might help in combination.\nSpeaker G: But I have to tell you, I can't remember the conference, but I think it's about 10 years ago.\nSpeaker G: I remember going to one of the speech conferences and I saw within very short distance of one another, a couple different posters that showed about the wonders of some auditory inspired friend end or something.\nSpeaker G: And a couple posters away with somebody who compared one to just putting in the FFT and the FFT did slightly better.\nSpeaker G: So I mean, it's true there's lots of variability, but again, we have these wonderful statistical mechanisms for quantifying that variability and doing something reasonable with it.\nSpeaker G: So it's the same argument that's gone both ways about we have these data driven filters, LDA, and on the other hand, if it's data driven, it means it's driven by things that have lots of variability and that are necessarily not necessarily going to be the same in training and test.\nSpeaker G: So in some ways it's good to have data driven things in some ways it's bad to have data driven things. So part of what we're discovering is ways to combine things that are data driven and are not.\nSpeaker G: So anyways, it's just a thought that if we had that, maybe it's just a baseline, which would show us what are we really getting out of the filters, or maybe probably not by itself in combination, you know, maybe there's something to be gained from it.\nSpeaker G: And let the, what, you know, you've only worked with this for a short time, maybe in a year or two you would actually come up with the right set of things to extract from this information, but maybe the neural net and the HMMs could figure it out quicker than you.\nSpeaker G: So it's just a thought.\nSpeaker C: I will try to do that. Yeah. But one one one one thing and so like what before we started using this VAD name is earlier. What we did was like I guess most of you know about this adding this additional speech silence beat to the kept stream and training that you want that.\nSpeaker C: That is just a binary feature and that seems to be improving a lot on the speech that color is a lot of. But not much on the TIT. So adding an additional feature to this discriminated between speech and not.\nSpeaker C: I'm sorry. We actually added an additional binary feature to the kept stream just the baseline.\nSpeaker C: Yeah, but in the case of the idea that they actually he was anything because there was anything that could discriminate between speech.\nSpeaker C: But what Italian was like.\nSpeaker C: Very good. The huge improvement.\nSpeaker E: But in the question is even more is within speech can we get some features.\nSpeaker E: I was dropping information that can might be useful within speech and maybe to distinguish between voice sound and voice sounds.\nSpeaker G: And it's particularly more relevant now since we're going to be given the end points. So.\nSpeaker G: Yeah.\nSpeaker G: So.\nSpeaker C: So it was a paper in I guess this I guess about the extra adding some higher order information from the kept stream coefficient.\nSpeaker C: I'm a chemilins or something.\nSpeaker C: It was taking the.\nSpeaker C: Maybe.\nSpeaker C: I think he was showing a person something on.\nSpeaker G: Yeah, but again, you could argue that that's exactly what the neural network does.\nSpeaker G: So the neural network is in some sense equivalent to computing higher order moments.\nSpeaker G: So let's do it very specifically.\nSpeaker F: You want to talk about.\nSpeaker F: Yeah.\nSpeaker F: So.\nSpeaker F: I told you I was getting prepared to take this qualifier exam.\nSpeaker F: So basically that's just trying to propose.\nSpeaker F: You're following years of your PhD work trying to find a project to define and work on.\nSpeaker F: So I've been looking into doing something about speech recognition using acoustic events.\nSpeaker F: So the idea is you have all these these different events.\nSpeaker F: For example, voicing.\nSpeaker F: Faciality are coloring.\nSpeaker F: Building robust primary detectors for these acoustic events and using the outputs of these robust detectors to do speech recognition.\nSpeaker F: And these primary detectors will be inspired by multi band techniques doing things similar to Larry Saul's work on graphical models to detect these acoustic events.\nSpeaker F: And so I've been thinking about that and some of the issues that I've been running into are exactly what kind of acoustic events I need, what acoustic events will provide a good enough coverage to in order to do the later recognition steps.\nSpeaker F: And also once I decide a set of acoustic events, how do I get labels training data for these acoustic events.\nSpeaker F: And then later on down the line, I can start playing with the models themselves, the primary detectors.\nSpeaker F: So I kind of see like after after building the primary detectors I see myself taking the outputs and feeding them in sort of tandem style to to a Gaussian mixture of HMM back end and doing recognition.\nSpeaker F: So that's just generally what I've been getting at.\nSpeaker G: But by the way, the voice down voice version of that for instance could tie right into what Cameron was looking at.\nSpeaker G: So you know, if you if a multi band approach was helpful as I think it is, it seems to be helpful for determining voice down voice might be another.\nSpeaker F: Oh, it looks okay. Yeah, and so this this past week I've been looking a little bit into traps and doing doing traps on these events too.\nSpeaker F: I'm seeing it that's possible. Other than that, I was kicked out of my house living there for four years.\nSpeaker F: Oh, yeah. So that was cardboard box on the street now. Something like that.\nSpeaker F: Yeah.\nSpeaker C: So did you find a place? No, I ordered. Yesterday I called up a lady who will have a vacant room from May 30th and she's actually been doing to more people.\nSpeaker C: So she would get back to me on Monday. That's that's only thing I have. And Diane has a few more houses. She's going to take some pictures and send me after I go back.\nSpeaker C: Oh, so you're not down here permanently yet? No. I'm going back to a GI today.\nSpeaker G: Okay. And then you're coming back.\nSpeaker C: I mean, I planned to be here on 31st. 31st. Okay. There's a house available.\nSpeaker G: Well, I mean, if they're available and they'll be able to get you something, so worse comes to worse.\nSpeaker G: We'll put you up in the hotel for a while until you.\nSpeaker A: If you're in an investment situation, you need a place to stay. You could stay with me for a while.\nSpeaker A: I've got a spare bedroom right now.\nSpeaker C: Oh, thanks.\nSpeaker C: So maybe he needs.\nSpeaker F: My car board box is actually a bedroom.\nSpeaker G: Two bedroom car board box.\nSpeaker G: That's great.\nSpeaker G: You want to say anything about you actually been last week you were doing this stuff with Pierre.\nSpeaker G: You were mentioning is that something worth talking about here?\nSpeaker A: Well, I don't think it directly relates. Well, I was helping have speech researcher named Pierre Devenny.\nSpeaker A: And he wanted to look at how people respond to formative changes, I think.\nSpeaker A: So he created a lot of synthetic audio files of vowel to vowel transitions.\nSpeaker A: And then he wanted a psychoacoustic spectrum.\nSpeaker A: And he wanted to look at how the energy is moving over time in that spectrum.\nSpeaker A: And compare that to the listener tests.\nSpeaker A: And so I gave him a PLP spectrum.\nSpeaker A: And he wanted to track the peak.\nSpeaker A: So he could look at how they're moving.\nSpeaker A: And then he took the PLP LPC coefficients.\nSpeaker A: And I found the roots.\nSpeaker A: This was something that's defied and suggested.\nSpeaker A: I found the roots of the LPC polynomial to track the peaks and the PLP LPC spectrum.\nSpeaker C: Is that a line spectral pairs?\nSpeaker G: It's root LPC.\nSpeaker C: So instead of the log, you took the root square, you could be good or something.\nSpeaker G: So it's taking to find the roots of the LPC polynomial.\nSpeaker C: Polynomial.\nSpeaker C: Yeah, is that the line spectral?\nSpeaker G: So it's like line spectral pairs.\nSpeaker G: I think what they call line spectral pairs, they push it towards the unit circle, don't they?\nSpeaker G: But what we'd used to do, when I did synthesis at National Semican Bacter 20 years ago, the technique we were playing with initially was technique LPC polynomial and finding the roots.\nSpeaker G: And it wasn't PLP because he didn't invent it yet.\nSpeaker G: But it was just LPC and we found the roots of the polynomial.\nSpeaker G: And when you do that, sometimes, there are most people that call formats, sometimes they're not.\nSpeaker G: So it's a little, a format tracking with it could be a little tricky because you get these funny values.\nSpeaker G: So you just get a few roots, two or three.\nSpeaker G: You get these complex pairs.\nSpeaker G: It depends on the order that you're doing.\nSpeaker A: Right, so if every root that's, since it's a real signal, the LPC polynomial is going to have real coefficients.\nSpeaker A: So I think that means that every root that is not a real root is going to be a complex pair, a complex value and it's conjugate.\nSpeaker A: So for each, and if you look at that on the unit circle, one of these, one of the members of the pair will be a positive frequency.\nSpeaker A: One will be a negative frequency, I think.\nSpeaker A: So I'm using an a-thorded polynomial and I'll get three or four of these pairs.\nSpeaker A: Which gives me three or four P positions.\nSpeaker G: This is from synthetic speech?\nSpeaker G: That's right.\nSpeaker G: Yeah, so if it's from synthetic speech, then maybe it'll be cleaner.\nSpeaker G: I mean, for a real speech, then what you end up having is, I guess, a funny little things that don't exactly fit your notion of formats all that well.\nSpeaker G: But mostly they do.\nNone: Yeah.\nSpeaker G: And what we were doing, which is not so much looking at things, it was okay because it was just a question of quantization.\nSpeaker G: We were just, you know, a story.\nSpeaker G: It was, we were doing a stored speech quantization.\nSpeaker E: But in your case, actually you have peaks that are not at the form of positions.\nSpeaker E: But there are lower in energy when they are much lower.\nSpeaker D: If this is synthetic speech, can't you just get the formats directly?\nSpeaker A: I mean, how was the speech created?\nSpeaker A: It was created from a synthesizer.\nSpeaker A: And was an performance synthesizer.\nSpeaker G: It may have been, but maybe he didn't have control of it or something.\nSpeaker A: In fact, we could get format frequencies out of the synthesizer as well.\nSpeaker A: And one thing that the LPC approach will hopefully give me in addition is that I might be able to find the...\nSpeaker A: the bandwidth of these HOMPS as well.\nSpeaker A: I'm still fine suggested looking at each complex pair as a second order.\nSpeaker A: I are filter.\nSpeaker A: But I don't think there's a really good reason not to get the format frequencies from the synthesizer instead.\nSpeaker A: Except that you don't have the psychoacoustic modeling in that.\nSpeaker A: Yeah.\nSpeaker G: So the actual...\nSpeaker G: So you're not getting the actual formats per se.\nSpeaker G: You're getting something that is a festrongly affected by the PLP model.\nSpeaker G: And so it's more psychoacoustic.\nSpeaker G: Oh, I see.\nSpeaker G: It's sort of a different thing.\nSpeaker G: That's sort of the point.\nSpeaker G: But yeah, ordinarily, in a form of synthesizer, the bandwidths as well as the form and centers are...\nSpeaker G: I mean, that's somewhere in the synthesizer that was put in.\nSpeaker G: But yeah, you view each complex pair as essentially a second order section.\nSpeaker G: Which has a band center bandwidth.\nSpeaker G: And...\nNone: Okay.\nSpeaker G: So, you're going back today and then back in a week, I guess.\nSpeaker G: Yeah.\nSpeaker G: Yeah.\nNone: Great.\nSpeaker G: Welcome.\nSpeaker D: I guess we should do digits quickly.\nSpeaker D: Oh, yeah, digits.\nSpeaker G: I don't forget.\nSpeaker G: I don't forget our daily digits.\nSpeaker G: I'm not bad.\nSpeaker G: Sure.\nSpeaker G: Transcript.\nSpeaker G: L-142.\nSpeaker G: 1-975-336030.\nSpeaker G: 0-814-187845.\nSpeaker G: 875-54232.\nSpeaker G: 1-187-3211.\nSpeaker G: 0-841-6505-65.\nSpeaker G: 2519-2774.\nSpeaker G: 708-484.\nSpeaker G: 1-732-964-172.\nSpeaker D: Transcript.\nSpeaker D: L-143.\nSpeaker D: 897-4-9168-6219.\nSpeaker D: 489-9488-61.\nSpeaker D: 1-619-2592-6839.\nSpeaker D: 848-0-0-607.\nSpeaker D: 6-515-262178.\nSpeaker D: 754-292-5498.\nSpeaker D: 7-427-832-97232.\nSpeaker D: 1-667-5262.\nSpeaker F: Transcript.\nSpeaker F: L-144.\nSpeaker F: 9-721-75-6322.\nSpeaker F: 6-84-320-282.\nSpeaker F: 0-91864-1042.\nSpeaker F: 577-718592.\nSpeaker F: 7-695-8549-70.\nSpeaker F: 699-453-872.\nSpeaker F: 9-345-722-90342.\nSpeaker F: 210-829-717.\nSpeaker A: Reading transcript.\nSpeaker A: L-145-589-951490.\nSpeaker A: 0-71143-295.\nSpeaker A: 736-877-160.\nSpeaker A: 3-004-65779-7.\nSpeaker A: 3-127-3259.\nSpeaker A: 11715-035-1050-2175.\nSpeaker A: 1665-6228-4689.\nSpeaker C: Transcript.\nSpeaker C: L-146-913307-464.\nSpeaker C: 463-773-1244.\nSpeaker C: 636-930-2728.\nSpeaker C: 276-875-224.\nSpeaker C: 976-93896.\nSpeaker C: 11913-4812-3442.\nSpeaker C: 619-330-35.\nSpeaker C: 794-278-5652.\nSpeaker B: Transcript.\nSpeaker B: L-140-353-184012.\nSpeaker B: 456-334-0634.\nSpeaker B: 525-922775.\nSpeaker B: 718-275284.\nSpeaker B: 0-478-604049.\nSpeaker B: 9807-5516-0071.\nSpeaker B: 0-16-059-6052.\nSpeaker B: 518-457-7197.\nSpeaker E: Transcript.\nSpeaker E: L-141-622-760-042.\nSpeaker E: 9-537-89365-4.\nSpeaker E: 2589-8515-1617-3088-9492.\nSpeaker E: 977-9.\nSpeaker E: 923-929-0056-0901.\nSpeaker E: 7985-0505.\nSpeaker E: 925-4443-95.\nSpeaker E: 8984-3042-3297.\n"}, {"meeting_id": "Bro025", "summary": "The group discussed the current state of their work, which was coming close to a conclusion. They were putting together their final model, including the various techniques they had explored. Their performance on the Aurora tasks was second and very close to those in first place. The professor thought that further tweaking and incorporation of a neural network would improve their scores. The team discussed which method to use for noise suppression, which had not been decided yet. They also delved into a more detailed discussion of the VAD and latency. Their method for noise estimation would add a notable delay to the model. Finally, the team delved further into how to finish up the model.", "dialogue": "Speaker D: Alright, we're off.\nSpeaker E: Just test test.\nSpeaker E: Just test me.\nSpeaker E: Yeah.\nNone: Okay.\nNone: Okay.\nSpeaker E: So, there's two sheets of paper in front of us.\nSpeaker E: This is the arm wrestling?\nSpeaker A: Yeah, we formed a coalition, actually.\nSpeaker A: We already made it into one.\nSpeaker E: Oh, good.\nSpeaker E: Excellent.\nSpeaker E: That's the best thing.\nSpeaker E: Tell me about it.\nSpeaker C: So, it's well, it's spectacular.\nSpeaker C: Subjection or we know, finish, right?\nSpeaker C: We're bending on the big bits.\nSpeaker C: Each risk where I'm at.\nSpeaker C: Perfect.\nSpeaker C: And then with the estimation of knowing depending on the arm, it's moving at a time.\nSpeaker C: It's working out frequency.\nSpeaker C: It's very simple.\nSpeaker C: And the best result.\nSpeaker C: But you see, you're on 50 bands with a winner filter.\nSpeaker C: And there is no noise addition.\nSpeaker C: It's good because it's difficult.\nSpeaker C: Are you looking at one or two?\nSpeaker C: Yeah, so the sheet that gives you 3.66.\nSpeaker C: The second sheet, this, about the same, it's the same idea, but it's working on med bands.\nSpeaker C: And it's a spectrosupportion instead of a winner filter.\nSpeaker C: And there is also a noise addition after cleaning up the main bits.\nNone: Well, there is a term.\nNone: Let me know.\nSpeaker E: Yeah, I mean, it's actually very similar.\nSpeaker E: I mean, if you look at databases, the one that has the smallest.\nSpeaker E: The one that has the smallest overall number is actually better on the finish and Spanish.\nSpeaker E: But it is worse on the, I mean, on the DI digits.\nSpeaker E: So, probably doesn't matter that much either way.\nSpeaker E: But when you say unified, do you mean it's one piece of software now?\nSpeaker C: So no, we are already setting up software.\nNone: Should be ready.\nSpeaker D: So, what's happened?\nSpeaker D: I think I've missed something.\nSpeaker E: Okay, so a week ago, maybe you weren't around when you and I came close here.\nSpeaker E: You didn't go to your night?\nSpeaker E: Yeah, I didn't.\nSpeaker E: Oh, okay, so yeah, let's summarize.\nSpeaker E: And then if I summarize, somebody can tell me if I'm wrong, which will also be possibly helpful.\nSpeaker E: What if I just press here?\nSpeaker E: I hope this is still working.\nSpeaker E: We looked at, after coming back from Qualcomm, we had very strong feedback.\nSpeaker E: And I think it was, he didn't think there was an eye opinion also that, you know, we sort of spread out to look at a number of different ways of doing the only suppression.\nSpeaker E: But given limited time, it was sort of time to choose one.\nSpeaker E: And so the retail series hadn't really worked out that much.\nSpeaker E: The subspace stuff had not worked with so much.\nSpeaker E: So it sort of came down to spectral subtraction versus refiltoring.\nSpeaker E: We had a long discussion about how they were the same and how they were completely different.\nSpeaker E: And I mean, fundamentally, the same sort of thing, but the math is a little different so that there's an exponent difference in, you know, what's the ideal filtering, depending on how you construct the problem.\nSpeaker E: And I guess after that meaning sort of made more sense to me, because if you're dealing with powerspectra, then how are you going to choose your error?\nSpeaker E: And typically you'll do something like variance. So that means there'll be something like the square of the power spectra.\nSpeaker E: Whereas when you're doing the, looking at it the other way, you're going to be dealing with signals and you're going to have to looking at power of noise power that you're trying to reduce.\nSpeaker E: So there should be a difference of, you know, conceptually of a factor too in the exponent.\nSpeaker E: But there's so many different little factors that you adjust in terms of over subtraction and so forth that arguably you're, and the choice of do you operate on the Melvans or the operate on the FFT beforehand.\nSpeaker E: There are so many other choices to make that are almost, well, if not independent, certainly in addition to the choice of whether you, the respective subtraction or when you're filtering, that again, we sort of felt the gang should just sort of figure out which it is they want to do.\nSpeaker E: And then let's pick it, go forward with it. So that was, that was last week. And we said, take a week, go arm wrestle, you know, figure it out.\nSpeaker E: And the joke there was that each of them had specialized in one of them. And so they, so instead they went to your somebody and bonded and they can go to a single, single piece of software.\nSpeaker E: Another, another victory for international collaboration.\nSpeaker D: So so you guys have combined for you're going to be combining software.\nSpeaker A: Well, the piece of software has like plenty of options like you can pass command line arguments. So depending on that, it becomes either spec with subtraction or we're filtering.\nSpeaker E: Well, that's fine. But the important thing is that there is a piece of software that you that we all will be using. Yeah.\nSpeaker C: Yeah, it's just one piece of software.\nSpeaker C: And we want to make optimize parameters. Sure. But still so there is a piece of software.\nSpeaker C: How is how good is that? I don't have a sense of. It's just one person. Best system. It's between. We are single.\nSpeaker E: Yeah. But compared to the last evaluation, which we started for before, but we were considerably far behind. And the thing is this doesn't have neural net in yet, for instance.\nSpeaker E: So it's so it's it's not using our full ballot bag of tricks, if you will. And it is very close performance to the best thing that was there before.\nSpeaker E: But you know, looking at another way, maybe more importantly, we didn't have any explicit noise handling stationary dealing with we didn't explicitly have anything to deal with stationary noise. And now we do.\nSpeaker D: So will the neural net operate on the output from either the weener filtering or the spectral subtraction? Well, so so arguably what we should do, I mean, I gather you have, it sounds like you have a few more days of nailing things down with the software and so on.\nSpeaker E: But but arguably what we should do is, even though the software can do many things, we should for now pick a set of things, these things, I would guess.\nSpeaker E: And not change that and then focus on everything that's left. And I think you know that our goal should be by next week when he comes back to really just to have a firm path for the for the time he's gone of what things will be attacked.\nSpeaker E: So I would I would think that what we would want to do is not thoughts with this stuff for a while because what will happen is we'll change many other things in the system. And then we'll probably want to come back to this and possibly make some other choices.\nSpeaker E: But just conceptually where does the neural net do you want to run it on the output of the spectrally subtraction? Well, depending on its size, well, one question is, is it on the server side or is it on the terminal side?\nSpeaker E: If it's on the server side, we probably don't have to worry too much about size. So that's kind of an argument for that. We do still, however, have to consider its latency. So the issue is is, for instance, could we have a neural net that only looked at the past?\nSpeaker E: What we've done in the past is to use the neural net to transform all of the features that we use. So this is done early on. This is essentially, I guess it's more or less like a speech enhancement technique here.\nSpeaker E: We're just kind of creating new, if not new speech, at least new FFTs that have, which could be turned into speech, that have some of the noise removed. After that, we still do a mess of other things to produce a bunch of features.\nSpeaker E: And then those features are not now currently transformed by the neural net. And then the way that we had it in our proposal 2 before, we had the neural net transform features and we had the untransformed features, which, I guess you actually did linearly transform with KLT, but to our Thogon-Lyzen, but they were not processed through a neural net.\nSpeaker E: And Stefan's idea with that, as a recall, was that you'd have one part of the feature vector that was very discriminant in another part that wasn't, which would smooth things a bit for those occasions when the testing set was quite different than what you'd change your discriminant features for.\nSpeaker E: So all of that is still seems like a good idea. The thing is now we know some other constraints. We can't have unlimited amounts of latency. That's still being debated by people in Europe, but no matter how they end up there, it's not going to be unlimited amounts, so we have to be a little conscious of that.\nSpeaker E: So there's the neural net issue, there's the VAD issue, and there's the second stream thing. And I think last time we agreed that those are the three things that have to get focused on.\nSpeaker D: What was the issue with the VAD?\nSpeaker E: Well, better ones are good.\nSpeaker D: And so the default boundaries that they provide are okay, that's not all that great. I guess they still allow 200 milliseconds on either side, is that for the audience?\nSpeaker C: So the VAD issue is outside the beginning.\nSpeaker C: The speech pose, which is sometimes the speech that comes, you have to pose this down to 1.25 seconds. Wow! More than one second.\nSpeaker C: And it seems to us that this way of dropping the beginning and end is not... we can do better. Because this way of dropping the frames that improve for the baseline by 14% and so we already showed that we are currently in the end improve by 100% percent.\nSpeaker D: And so the top of the VAD that they provide? No.\nSpeaker C: Oh, okay. There's this 14.\nSpeaker C: And if we just take only the VAD probably is computed on the clean signal and apply them on the far-being test, the nuances, then results are much better.\nSpeaker C: In some case, the VAD is zero already, right?\nSpeaker D: So it means that there is still... how much latency does the VAD and... is the signal again?\nSpeaker C: Right now it's under undeniable, so it's 14 milliseconds plus the rank ordering should be...\nSpeaker A: So we have another 10 frames. The rank ordering, I'm sorry.\nSpeaker A: There's this... there's this... the... the filtering of the probabilities on the...\nSpeaker E: We don't think the media is going to be... we have 11.\nSpeaker E: So, yeah, I was just noticing on this that it makes reference to delay. So what's the... if you ignore... the VAD is sort of in parallel, isn't it?\nSpeaker E: With the... I mean, the additive with the... the LDA in the winter filtering.\nSpeaker A: Yeah, so what happened right now we removed that delay of the LDA.\nSpeaker A: So we're... I mean, if... so which is like... if we reduce the delay of VAD... so the... the final delay is now... is determined by the delay of the VAD.\nSpeaker A: Because the LDA doesn't have any delay. So if you reduce the delay of the VAD, I mean, it's like... effectively reducing the delay.\nSpeaker D: How much delay was on the LDA?\nSpeaker A: So the LDA and the VAD both had 100 milliseconds delay. So when they were in parallel, so which means you pick either one of them.\nSpeaker A: The biggest order. So right now that LDA delay is removed.\nSpeaker E: And there didn't seem to be any penalty for that. But there didn't seem to be any penalty for making it causal.\nSpeaker A: Oh no, it actually made it like 0.1% better or something.\nSpeaker E: It's just winter filter is 40 milliseconds today.\nSpeaker A: Yeah, so that's the one which Stefan was discussing like...\nSpeaker A: The smoothing. Yeah, you smoothed it and then delay the decision by.\nSpeaker E: Right, okay. So that's really not bad. So we may in fact... we'll see what they decide.\nSpeaker E: We may in fact have the latency time available to have an Earl Madden. I mean, it sounds like we probably will.\nSpeaker E: So that'd be good. Because it certainly always helped us before.\nSpeaker D: What amount of latency are you thinking about when you see it?\nSpeaker E: Well, they're disputing it. They're saying one group is saying 130 milliseconds and another group is saying 250 milliseconds.\nSpeaker E: 250 is what it was before actually. So some people are lobbying to make it shorter.\nSpeaker D: Were you thinking of the 250 or the 137th position?\nSpeaker E: Well, when we find that out, it might change exactly how we do it is all.\nSpeaker E: I mean, how much effort do we put into making it causal?\nSpeaker E: I mean, I think the neural net will probably do better if it looks a little bit of the future.\nSpeaker E: But it will probably work to some extent to look only at the past.\nSpeaker E: And we limited machine and human time and effort and how much time should we put into that.\nSpeaker E: So it'll be helpful if we find out from the standards, folks, whether they're going to restrict that or not.\nSpeaker E: But I think at this point our major concern is making the performance better.\nSpeaker E: And if something has to take a little longer in latency in order to do it, that's a secondary issue.\nSpeaker E: But if we get told otherwise, then we may have to clamp down a bit more.\nSpeaker A: So one difference that was there is like we tried computing the delta and then doing the frame dropping.\nSpeaker A: The earlier system was do the frame dropping and then compute the delta on the.\nSpeaker E: So this kind of an adult. Yeah. Oh, so that's fixed in us.\nSpeaker A: Yeah. So we have now delta and then so the frame dropping is the last thing that we do.\nSpeaker A: So we have what we do is we compute the silence probability converted into that binary flag.\nSpeaker A: And then in the end you up sampled it to match the final features number of.\nSpeaker A: It seems to be helping on the well match condition.\nSpeaker A: So that's why this improvement I got from the last result.\nSpeaker A: So and it actually reduced a little bit on the high mismatch.\nSpeaker A: So the final weight is it's better because the well match is the weighted more than.\nSpeaker E: So I mean you were doing a lot of changes. Did you happen to notice how much.\nSpeaker E: Change was due to just this frame dropping problem.\nSpeaker A: You had something on it, right?\nSpeaker C: Just a friend dropping problem. Yeah, but it's difficult. Sometimes we would change to things together.\nSpeaker E: But it's around maybe it's less than my personal.\nSpeaker E: Yeah.\nSpeaker E: But like we're saying there's four or five things like that.\nSpeaker E: Pretty shocked. Soon you're talking real improvement.\nSpeaker C: And the proposal that you're not that flexible. So working on that.\nSpeaker B: Mm hmm.\nSpeaker E: Oh, that's a real good point.\nSpeaker C: You can be the same guy.\nSpeaker E: Might be hard if it's at the server side, right?\nSpeaker C: Well, we can be the friend dropping server side or we can't just be careful.\nSpeaker D: Okay.\nSpeaker D: You have.\nSpeaker D: So when you maybe I don't quite understand how this works, but.\nSpeaker D: Couldn't you just send all of the frames, but mark the ones that are supposed to be dropped.\nSpeaker D: Because you have a bunch more bandwidth, right?\nSpeaker E: Well, you could, you know, I mean, it it always seemed to us that it would be kind of nice to in addition to reducing insertions actually use a plus bandwidth.\nSpeaker E: But nobody seems to care about that in this evaluation.\nSpeaker D: That's why the net use.\nSpeaker D: If the net's on the server side, then it could use all of the frames.\nSpeaker A: Yes, it could be like, you mean you just transmit everything and then finally drop the frames after the neuralite, right?\nSpeaker A: Yeah, that's one thing which you could even mark them.\nSpeaker A: Yeah, right now we have the server.\nSpeaker A: Right now what we did is like we just we just have this additional bit which goes along the features saying it's currently it's a speech or a non speech.\nSpeaker A: So there is no frame dropping till the final features like including the deltas are computed.\nSpeaker A: And after the delta's are computed, you just pick up the ones that are marked silence and then drop them.\nSpeaker E: So be more or less the same thing with the neural net, I guess.\nSpeaker A: So that's what that's what this is doing right now.\nSpeaker A: Yeah.\nSpeaker E: Okay.\nSpeaker E: So what's that's good set of work that.\nSpeaker A: Just one more thing like should we do something more for the noise estimation because we still.\nSpeaker E: Yeah, I was wondering about that. I hate to written that down there.\nNone: Actually, I did this experiment.\nSpeaker C: We just used in frames.\nSpeaker C: We take the first 15 frames of the judgements.\nSpeaker E: Yeah.\nSpeaker C: I tried just breaking the.\nSpeaker C: But of course, I didn't play.\nSpeaker E: Yeah, well, it's not surprising to be worse first time, but it does seem like some compromise between always depending on the first 15 frames and always depending on pause is a good idea.\nSpeaker E: Maybe you have to wait the estimate from the first 15 frames more heavily than was done in your first attempt.\nSpeaker E: Yeah.\nSpeaker E: I mean, do you have any way of assessing how well or how poorly the noise estimation is currently doing?\nSpeaker C: Yeah.\nSpeaker A: Was there any experiment with because I did the only experiment what I tried was I used the channel zero-watt for the noise estimation and frame dropping.\nSpeaker A: So I don't have a split like which one help more.\nSpeaker A: So it was the best result I could get.\nSpeaker E: So that's something you could do with this final system, right?\nSpeaker E: Just do this everything that is in this final system except use the channel zero for the noise estimation.\nSpeaker A: Yeah.\nSpeaker E: And then see how much better it gets.\nSpeaker E: If it's essentially not better than it's probably not worth.\nSpeaker A: Yeah, but the Ginters argument is slightly different. It's like even if I use a channel zero-watt, I'm just averaging the the pause spectrum.\nSpeaker A: But the Ginters argument is like if it is a non stationary segment, then he doesn't update the noise spectrum.\nSpeaker A: So it's like it tries to capture only the stationary part.\nSpeaker A: And so the averaging is like different from updating the noise spectrum only during stationary segments.\nSpeaker A: So the Ginters was arguing that even if you have a very good VAD averaging it like over the whole thing is not a good idea.\nSpeaker A: Because you are averaging the stationary and the non stationary and finally you end up getting something which is not really the same.\nSpeaker A: Anyway, you can't remove the stationary part from the signal.\nSpeaker A: No, using these messages.\nSpeaker A: Yeah, so you just update only the stationary components.\nSpeaker A: So that's still slight difference from what Ginters is trying in.\nSpeaker E: Well, yeah. And also there's just the fact that although we're trying to do very well in this evaluation, we actually would like to have something that worked well in general.\nSpeaker E: And relying on having 15 frames at the front or something is pretty.\nSpeaker E: I mean, you might not.\nSpeaker E: So it certainly be more robust to different kinds of input if you had at least some updates.\nSpeaker E: Well, what do you guys see as being what you would be doing in the next week given what's happened?\nSpeaker A: We have the VAD.\nSpeaker A: Was that VAD?\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker D: So I don't remember what you said.\nSpeaker D: The answer to my question earlier.\nSpeaker D: Were you trained in that on after you've done the spectral subtraction or the different net?\nSpeaker A: Yeah, which is a new that's a neural it is some of the VAD net.\nSpeaker A: So that VAD was trained on the noisy features.\nSpeaker A: So right now we have like we have the cleaned up features so we can have a better VAD by turning the net on the clean up speeches.\nSpeaker A: But we need a VAD for noise estimation also.\nSpeaker D: Can you use the same net to do both for can use the same net that you that I was talking about to the VAD?\nSpeaker A: It actually comes at the very end.\nSpeaker A: So the net the final net I mean which is the feature net so that actually comes after a chain of like LDA plus everything.\nSpeaker A: So it's like it takes a long time to get a decision out of it and I can actually do it for final frame dropping but not for the V.\nSpeaker A: Noise estimation.\nSpeaker E: See the idea is that the initial decision to that that you're in silence or speech happens pretty quickly.\nSpeaker D: Is that used by somebody's own?\nSpeaker D: Yeah, that sort of fed forward and you say well flush everything it's not speech anymore.\nSpeaker D: I thought it would be used for doing frame dropping.\nSpeaker E: It is used.\nSpeaker E: Yeah, it's only used well it's used for frame dropping.\nSpeaker E: It's used for end of utterance because you know there's if you have more than 500 milliseconds of of non speech then you figure it's end of utterance.\nSpeaker E: Something like that so.\nSpeaker C: I see.\nSpeaker E: Yeah, so probably the VAD and maybe testing out the noise estimation a little bit and keeping the same method but seeing if the noise estimation could be improved.\nSpeaker E: It's sort of related issues.\nSpeaker E: It probably makes sense to move from there and then later on in the month I think we want to start including the neural at the end.\nSpeaker E: Okay, anything else?\nSpeaker E: Didn't fall.\nSpeaker E: Our effort would have been devastated.\nSpeaker E: So Henik is coming back next week?\nSpeaker E: No, no, he's dropped into the US.\nSpeaker E: So the idea was that we'd sort out where we're going next with this work before he left on his next trip.\nSpeaker E: Good.\nSpeaker E: Very you just got through your qual so I don't know if you have much to say but...\nSpeaker B: Now just looking into some of the things that John O'Halla and Henik gave as feedback as a starting point for the project.\nSpeaker B: In my proposal I was thinking about starting from a set of phonological features or a subset of them.\nSpeaker B: It might not be necessarily a good idea according to John. He said these phonological features are sort of figments of imagination also.\nSpeaker E: In conversational speech in particular, I think you can put them in pretty reliably in synthetic speech but we don't have too much trouble recognizing synthetic speech since we created in the first place.\nSpeaker B: Yeah, so a better way would be something more data-driven just looking at the data and seeing what's similar and what's not similar.\nSpeaker B: I'm taking a look at some of Sun Gita's work on traps.\nSpeaker B: She did something where the traps she clustered the temporal patterns of certain phonemes in average over many, many contexts and some things tended to cluster.\nSpeaker B: So I've stopped Constance clustered very well. Silence was by its own self and Vocalic was clustered.\nSpeaker B: So those are interesting things.\nSpeaker D: Now you're sort of looking to try to gather a set of these types of features.\nSpeaker B: Right, yeah. See where I could start off from a set of small features and continue to iterate and find a better set.\nSpeaker E: Okay, well, short meeting. It's okay. So next we hopefully will get Tina Kear to join us.\nSpeaker D: Digits.\nSpeaker E: Digits.\nSpeaker E: Okay, let me go ahead and get my glasses on so I can see. Okay. Transcript L-327.\nSpeaker E: 821-067-40000-800-4176-1281-630-224-1912-650-869-4624-8919-1485-6845-388-383-80-4518-829135-0-234-44812-2\nSpeaker C: Transcript L-328-9060-3955-984-3165-343-114-4166-4194-33-7655-505-754-075-666-8 630-5487-701-812-831-5734-8703-68\nSpeaker D: Transcript L-329-9977-30368-7627-1700-0996-9388-96987-099 2126-937206-672-308-949-8032-631-489-6444-2669-318-791-3247\nSpeaker A: Transcript L-330-2360-9593-546-348-6675-3704-3844-866-3675-8705-5739-150-9016-22-409-277-701-9206-8967-896701-9206-8967-660 164-191-2428\nSpeaker B: Transcript L-331-3788-9100-623-270-385-138-183795-096-6610-197-541-2420-41-11-11-197-142-121-107-8 096-761-0197-541-240-41-5612-309-1-6057-6528-341-164739-4528-8507 Okay, and we're up.\n"}, {"meeting_id": "Bro024", "summary": "The team discussed SRI experiments, multi-band approaches, latency, VAD, and noise addition. The mean subtraction approach for SRI that relied on a twelve second interval resulted in a zero point three percent improvement over the six second model, not necessarily significant. The team thought that training the model on different speeds of speech would be helpful to add to this progress. Grad A introduced the team to her formal proposal for the qualifier exam. The work focused on intermediate classifiers and multi-band graphical models for speech recognition. The team then discussed noise removal as well The model PhD B was working on did not do well in non-stationary noise environments. In general, however, the model had improved. The VAD was labeling speech as silence segments, which could be attributed to alignment issues. The team thought that noise addition techniques might help the model.", "dialogue": "Speaker I: And we're on.\nSpeaker G: Okay.\nSpeaker G: When I close the door, I get it, uh, stuff in it all.\nSpeaker I: Hey Dave, you go ahead and turn on that stuff on.\nSpeaker G: So that's the virtual stuff.\nSpeaker I: You see for recording or?\nSpeaker I: Yeah, learning spots.\nSpeaker I: It's got like 16 channels.\nSpeaker I: The quality is quite good though.\nSpeaker I: Yeah, it's up to 30, pretty good.\nSpeaker G: So, uh, yeah, it's just, I must have.\nSpeaker A: Let's go ahead and start.\nSpeaker A: Okay.\nSpeaker A: So, yeah, this past week I've been mainly occupied with, um, getting some results from the SRI system trained on this short Hub 5 training set for the mean subtraction method.\nSpeaker A: I've done some tests last night, but, um, the results are suspicious. Um, it's, um, because they're the baseline results are worse than, um, Andreas, the results Andreas got previously.\nSpeaker A: And it could have something to do with, um, that's on digits.\nSpeaker A: That's on digits.\nSpeaker A: It could, it could have something to do with, um, down sampling. That's, that's worth looking into.\nSpeaker A: Um, and, um, a part of that, I guess, the main thing I have to talk about is, um, where I'm planning to go over the next week.\nSpeaker A: So, I've been working on integrating this mean subtraction approach into the smart com system.\nSpeaker A: And there's this question of, well, so, um, in my test before with HDK, I found it worked, it worked the best with about 12 seconds of data used to estimate the mean, but we'll often have less in the smart com system.\nSpeaker A: Um, so I think we'll use as much data as we have at a particular time, and we'll, we'll concatenate utterances together, um, to get as much data as we possibly can from the user.\nSpeaker A: But, um, there's a question of how to set up the models. So, um, we could train the models.\nSpeaker A: If we think 12 seconds is ideal, we could train the models using 12 seconds to calculate the mean to mean subtract the training data, or we could, um, use some other amount.\nSpeaker A: So, like I did an experiment where I, um, was using six seconds in test.\nSpeaker A: Um, but I tried 12 seconds in train, and I tried, um, the same in train. I tried six seconds in train, and six seconds in train was about 0.3% better.\nSpeaker A: Um, and, um, it's not clear to me yet, whether that's something significant. So, I want to do some tests and, um, actually make some plots of, um, for a particular amount of data in test, what happens if you vary the amount of data in train.\nSpeaker G: I don't know if you'd follow the stuff, but this is, uh, a, uh, long term, long term window, FFTs. Yeah, we'll be talking about it.\nSpeaker A: So, I was, I actually ran the experiments mostly. You know, I was, I was hoping to have the plots with me today. I just didn't get to it. But, um, yeah, I would be curious about people's feedback on this, because I'm, I think there are some, I think this is kind of like a bit of a tricky engineering problem, trying to figure out what's the optimal way to set this up.\nSpeaker A: So, um, I'll try to make the plots and then put some post script up on my, on my web page, and I'll mention my status report if you want to take a look.\nSpeaker G: You can clarify something for me. You're saying 0.3%. You take a 0.3% hit when the training and testing links aren't don't match or something. Is that what it is?\nSpeaker A: Well, I don't think it's just for any mismatch. Yeah. Take a hit. In some cases, it might be better to have a mismatch. Yeah. Like, I think I saw something like, like if you only have two seconds in test, or, um, maybe it was something like four seconds, you actually do a little better if you, um, train on six seconds. And if you train on four seconds.\nSpeaker A: Um, but the case that with the 0.3% hit was using six seconds in test, um, comparing train on 12 seconds versus train on six seconds, which was worse, the train on 12 seconds.\nSpeaker G: Okay, but 0.3% from what to what? That's 0.3%.\nSpeaker A: Um, the, the accuracies went from it was something vaguely like 95.6 accuracy, um, improved to 95.9. What I, what I, 4.4 to 4.1.\nSpeaker G: Okay. So, yeah. So about about an 8%, uh, 78% relative. Okay.\nSpeaker G: Um, yeah. Well, I think, you know, if you're going for an evaluation system, you'd care, but if you were doing a live system of people are actually using nobody would notice, I think the thing is to get something that's practical.\nSpeaker A: That's interesting. I see a point. I guess I was thinking of it as, um, an interesting research problem. Yeah.\nSpeaker A: I was thinking for the ASRU paper, we could have a section saying for smart com, we, we tried this approach in an interactive system, which I don't think has been done before.\nSpeaker A: And, and then there was two research questions from that. And one is that does it still work if you just use the past history.\nSpeaker A: And the other was this question of, um, that was just talking about now. So I guess that's why I thought it was interesting.\nSpeaker G: So, um, the time FFT, short time Keptstrom calculation, uh, mean, mean calculation work that people have in commercial systems, they do this all the time.\nSpeaker G: They, they calculated from previous utterances. Yes. But, but, uh, as you say, there hasn't been that much of this long, long time, uh, specter work.\nSpeaker G: Oh, oh, okay. So that's, that's, that's standard. Uh, pretty common. Yeah. Okay. Um, but, uh, yeah. So it is interesting.\nSpeaker G: I mean, there's two sides to these really small, uh, gradations and performance. Um, I mean on the one hand, in a practical system, if something is, uh, 4.4% error, 4.1% error, people won't really tell it to be able to tell the difference.\nSpeaker G: On the other hand, when you're doing, uh, research, you may, you might find that the way that you build up a change from a 95% accurate system to a 98% accurate system is through 10 or 12 little things that you do that each are 0.3%.\nSpeaker G: So, so they, they, it's, I don't mean to say that they're, they're irrelevant. Uh, they are relevant. But, um, for a demo, you won't see it. Right. Okay.\nSpeaker A: And, um, let's, let's see. Um, okay. And then there's another thing I want to start looking at. Um, with is the choice of the analysis window length. So I've just been using two seconds just because that's what Carlos did before.\nSpeaker A: Um, I wrote to him asking about how he chose the two seconds and it seemed like he chose it a bit informally. So, um, with it with the HDK setup, I should be able to do some experiments. Um, just varying that length, say between one and three seconds in a few different reverberation conditions.\nSpeaker A: Say this room and also a few of the artificial impulse responses we have for reverberation, just making some plots and seeing how they look. And, um, so with the sampling rate I was using one second or two seconds or four seconds is a power of two.\nSpeaker G: Um, number of samples and, um, I'll, I'll do for the ones in between. I guess I'll just zero pad. I guess one thing that might also be an issue. Um, it's part of what you're doing is you're getting a spectrum over a bunch of different kinds of speech sounds.\nSpeaker G: Um, and so it might matter how fast someone was talking. Oh, you know, if, if, if there's a lot of phones in one second, maybe you'll get a really good sampling of all these different things. And, uh, on the other hand, someone's talking slowly, maybe you need more.\nSpeaker G: So I don't know if you have some samples of faster or slower speech, but it might make a difference. I don't know.\nSpeaker A: Yeah, I don't, I don't think the TI digits data that I have, um, it would be appropriate. Yeah.\nSpeaker A: What would you, what about if I fed it through some kind of, um, speech processing algorithm that changed the speech rate?\nSpeaker G: Yeah, but then you'll have the degradation of, uh, whatever you do, uh, add it onto that. But maybe, yeah, maybe if you get something that sounds that does a pretty good job at that.\nSpeaker A: Yeah. Well, I just, if you think it's worth looking into, I mean, it is getting a little away from reverberation.\nSpeaker G: Um, yeah. It's just that you're making a choice. I was thinking more from the system aspect. If you're making a choice for smart com, that, that, that it might be that it's.\nSpeaker G: The optimal number could be different. Right. Yeah. Could be.\nSpeaker A: And then the third thing, um, I was, um, very explained LDA filtering to me yesterday. And so, um, Mike, Sharon is thesis, um, did a series of experiments, um, training LDA filters in different conditions.\nSpeaker A: And you were interested in having me repeat this for, for this mean subtraction approach. Is that right? Or for these long analysis windows, I guess is the right way to put it?\nSpeaker G: I guess the, the issue I was, the general issue I was bringing up was that if you're, have a moving, moving window, uh, a set of weights, times things that, uh, move along, shift along in time, that you have, in fact, a linear time and varying filter.\nSpeaker G: And you just happen to have picked a particular one by setting all the weights to be equal. And so the issue is what are some other filters that you could use in that sense of filter.\nSpeaker G: And, um, as I was saying, I think the simplest thing to do is not to train anything, but just to do some sort of, uh, hamming or handing, right, kind of window, kind of thing, just sort of the, the emphasize the journey.\nSpeaker G: So I think that would sort of be the first thing to do. But then, yeah, the LDA is, is interesting because it would sort of say, well, suppose you actually trained this up to do the best you could by some criterion.\nSpeaker G: What would the filter look like that? And, um, that sort of we're doing in this, uh, Aurora stuff. And, uh, it's still not clear to me in the long run whether the best thing to do would be to do that or to have some stylized version of the filter that looks like these things you've trained up because, um, you always have the problem that it's trained up for one condition isn't quite right for another. So, uh, that's, that's why that's why the rest of the filter is actually ended up lasting a long time.\nSpeaker G: People still using it quite a bit because you don't change it. So it doesn't get any worse.\nSpeaker A: Okay, so, um, actually, I was just thinking about what I was asking about earlier, which is about having less than, say, 12 seconds in the smart calm system to do the mean subtraction.\nSpeaker A: You said in systems where you kept still mean subtraction, they can catenate utterances. And do you know how they address this issue of testing versus training?\nSpeaker D: I think what they do is they do it always online. I mean that you just take what you have from the past that you calculate some mean of this and subtract some mean.\nSpeaker D: Okay. Um, then you can, you can increase your window. Why do I get, why do I get more samples?\nSpeaker A: Okay. Um, and so, so in that, in that case, what, what do they do when they're performing the capital mean subtraction on the training data?\nSpeaker A: So because you'd have hours and hours of training data. So do they cut it off and start over at intervals or.\nSpeaker D: So do you have, you mean you have files which are hours of hours alone or?\nSpeaker D: Well, no, I guess not. I mean, usually you have in the training set, you have similar conditions. I mean, file links are, I guess, the same order or the same size for test data.\nSpeaker A: Okay. But it's okay. So if someone's interacting with the system, Morgan, Morgan said that you would tend to chain utterances together.\nSpeaker G: Well, I think what I was, I thought what I was saying was that, um, at any given point, you are going to start off with what you had from before.\nSpeaker G: From, and so if you're splitting things up into utterances, so for instance, in a dialogue system, where are you going to be asking for some information?\nSpeaker G: There's some initials of something. And, you know, the first time out, you might have some general average, but you don't have very much information yet.\nSpeaker G: But after they've given one utterance, you've got something. You can compute your mean capture from that and then can use it for the next thing that they say, so that, you know, the performance should be better that second time.\nSpeaker G: And I think the heuristics of exactly how people handle that and how they handle their training. I'm sure very from place to place, but I think the ideally it seems to me anyway that you, you would want to do the same thing in training as you do in test.\nSpeaker G: But that's just a prejudice. And I think anybody, and this with some particular task, we experiment.\nSpeaker A: I guess the question I had was, um, amount of data was the amount of data that you give it to, um, update this estimate because say you, if you have say 5,000 utterances in your training set, um, and you keep the mean from the last utterance by the time it gets to the 5,000 utterances.\nSpeaker G: So those are all different people with different, I mean, in, so for instance, in the telephone task, these are different phone calls, so you don't want to chain it together from a different phone call.\nSpeaker G: Okay, so, so they would, so it's within speaker, within phone call, if the dialogue system, it's within whatever this characteristic you're trying to get rid of is expected to be consistent over.\nSpeaker A: Right, and, right, okay, so you, in so in training, you would start over at every new phone call or at every new speaker. Yeah, okay.\nSpeaker G: Now, you know, maybe do something from the others just because at the beginning of a call, you don't know anything.\nSpeaker G: So you might have some kind of general thing that's your best guess to start with.\nSpeaker G: So I, you know, a lot of these things are proprietary, so we're doing a little bit of guesswork here.\nSpeaker G: And what do, what do people do who really face these problems in the field? Well, they have companies, they don't tell other people exactly what they do.\nSpeaker G: But, but I mean, when you, the hints that you get from what they, when they talk about it, or that they do, they all do something like this.\nSpeaker A: Right, okay, I see. Because I, so this smart contest, first of all, it's this TV and movie information system.\nSpeaker G: Yeah, but you might have somebody who's using it. And later you might have somebody else using it.\nSpeaker A: Right. I see. I was about to say. So if you ask it, what, what movies are on TV tonight?\nSpeaker A: If I look at my wristwatch, when I say that, it's about two seconds.\nSpeaker A: Yeah. So when I currently have the mean subtraction, set up the analysis windows two seconds.\nSpeaker A: So what you just said about what do you start with raises a question of what do I start with then? I guess it, because...\nSpeaker G: Well, okay. So in that situation, though, maybe it's a little different there is, I think you're talking about there's only one...\nSpeaker G: It, it, it also depends. We're getting a little off track here.\nSpeaker G: Oh, right. But, but, but, there's been some discussion about whether the work we're doing in that project is going to be for the kiosk or for the mobile or for both.\nSpeaker G: And I think for this kind of discussion, it matters. If it's in the kiosk, then the physical situation is the same.\nSpeaker G: It's going to, you know, the exact interaction with the microphones is going to differ depending on the person and so forth, but at least the basic acoustics are going to be the same.\nSpeaker G: So if it's really in one kiosk, then I think that you could just chain together and, you know, as much as much speech as possible to, because what you're really trying to get at is the reverberation characteristic.\nNone: Yeah.\nSpeaker G: But in, in the case of the mobile, presumably the acoustics is changing all over the place.\nSpeaker G: Right.\nSpeaker G: And in that case, you probably don't want to have it be endless, because you want to have some sort of, it's a question of how long do you think it's, you can get an approximation to a stationary something that it's not really stationary.\nSpeaker G: Right.\nSpeaker A: Right.\nSpeaker A: I, I guess I just started thinking of another question, which is for the very first frame, what, what do I do if I, if I take, if I use that frame to calculate the mean, then I'm just going to get nothing.\nSpeaker A: Right.\nSpeaker A: Right. So I should probably have some kind of default, mean for the first couple of frames.\nSpeaker A: Okay.\nSpeaker G: Or subtract nothing.\nSpeaker A: I mean, it's, or subtract nothing.\nSpeaker A: And that's, I guess that's something that's, people have figured out how to deal with in capstone mean subtraction as well.\nSpeaker G: Yeah, people do something.\nSpeaker G: They, they, they have some, in, in capstone mean subtraction for short term window analysis windows, as is usually done.\nSpeaker G: You're trying to get rid of some very general characteristic.\nSpeaker G: And so, if you have any other information about what a general kind of characteristic would be, then you can do it there.\nSpeaker I: You can also reflect the data.\nSpeaker I: So you take, you know, I'm not sure how many frames you need, but if you take that many from the front, flip it around.\nSpeaker I: Yeah, that's the negative values.\nSpeaker G: Yeah.\nSpeaker G: The other thing is that, and I remember BBN doing this is that if you have a multi pass system, if the first pass takes, it takes most of the computation.\nSpeaker G: The second and the third pass could be very, very quick, just looking at a relatively small, small space of hypotheses.\nSpeaker G: Then you can do your first pass without any subtraction at all.\nSpeaker G: And then your second pass eliminates those, most of those hypotheses by, by having improved, improved version of the analysis.\nSpeaker A: So, so that was all I had.\nSpeaker A: Yeah.\nSpeaker F: Okay, so for the past week or two, I've been just writing my formal thesis proposal. So I'm taking this qualifier exam, it's coming up in two weeks.\nSpeaker F: I finished writing a proposal and submit to the committee.\nSpeaker F: And should I explain more about what I'm proposing to do this?\nSpeaker F: Brief.\nSpeaker F: Okay. So briefly, I'm proposing to do a new approach to speech recognition, a combination of multi band ideas and ideas about the acoustic phonetic approach to speech recognition.\nSpeaker F: So I will be using these graphical models that implement the multi band approach to recognize a set of intermediate categories that might involve things like phonetic features or other feature things that are more closely related to the acoustic signal itself.\nSpeaker F: And the hope in all this is that by going multi band and by going into these intermediate classifications that we can get a system that's more robust to unseen noises and situations like that.\nSpeaker F: And so some of the research issues involved in this are what kind of intermediate categories do we need to classify? Another one is what other types of structures in these multi band graphical models should we consider in order to combine evidence from the sub-ant.\nSpeaker F: And the third one is how do we merge all the information from the individual multi band classifiers to come up with word recognition or from recognition.\nSpeaker F: So basically that's what I've been doing two weeks.\nSpeaker F: I got two weeks to brush up on presentation.\nSpeaker G: So I thought you were finishing your thesis in two weeks.\nSpeaker I: Oh that too.\nSpeaker F: Are you going to do any dry runs for your finger?\nSpeaker F: Yes.\nSpeaker F: I'm going to do some.\nSpeaker F: Would you be interested?\nSpeaker F: Sure.\nSpeaker F: I hope that.\nSpeaker I: Is that it?\nSpeaker I: That's it.\nSpeaker I: Okay.\nSpeaker I: Let's see.\nSpeaker I: So we've got 40 minutes left.\nSpeaker I: It seems like there's a lot of material.\nSpeaker I: Any suggestions about where we should go next?\nNone: Yeah.\nSpeaker C: Actually, most of these in our last meeting with Ginter.\nSpeaker C: But I'll just.\nSpeaker C: So the last week I showed some results with only speech thread curve, which was like some 56% and I didn't.\nSpeaker C: I mean, I found out the results.\nSpeaker C: I mean, I wasn't getting that results on the TI digit.\nSpeaker C: So I was like looking into what is wrong with the TI digits.\nSpeaker C: Why I was not getting it and I found that the noise estimation is the reason for the TI digits to perform worse than the baseline.\nSpeaker C: So I actually, I mean, the first thing I did was I just scaled the noise estimate by a factor which is less than 1 to see if that.\nSpeaker C: Because I found that a lot of zeros in the spectrogram for the TI digits when I use this approach.\nSpeaker C: So the first thing I did was I just scaled the noise estimate and I found.\nSpeaker C: So the results that I have shown here are the complete results using the new, the new technique is nothing but the noise estimate scale by a factor of 0.5.\nSpeaker C: So it's just an id hoc.\nSpeaker C: I mean, some intermediate result because it's not optimized for anything.\nSpeaker C: So the results that trend, the only trend I could see from those results was like the current noise estimation or the noise composition scheme is working good for like the car noise type of thing.\nSpeaker C: Because I've the only very good result in the TI digits is the noise car noise condition for the test A which is like the best I could see that for any non stationary noise like Babel or subway or any street some restaurant noise.\nSpeaker C: It's like it's not performing very well.\nSpeaker C: So that's the first thing I could make out from this stuff.\nSpeaker D: I think what is important to see is that there is a big difference between the training modes.\nSpeaker D: If you have clean training, you get also a 50% improvement.\nSpeaker D: But if you have multi-condition training, you get only 20%.\nSpeaker C: And in that 20% is very inconsistent across different noise conditions. So I have like a 45% for car noise and then there's a minus 5% for the Babel and there is a 33% for the station.\nSpeaker C: And so it's not actually very consistent across.\nSpeaker C: So the only correlation between the speech.car and this performance is the stationarity of the noise that is there in these conditions and the speech.car.\nSpeaker C: So the overall result is like in the last page which is like 47% which is still very imbalanced because I have like 56% on speech.car and 35% on the TI digits.\nSpeaker C: And the 56% is like comparable to what the French telecom gets but 35% is way off.\nSpeaker G: So I can fuse but looking at the second page and it says 50% looking in the lower right hand corner, 50% relative performance.\nSpeaker D: For the clean training.\nSpeaker G: Is that if you look at 50% improvement?\nSpeaker C: That's for the clean training and the noise you're testing for the i digits.\nSpeaker G: So it's improvement over the baseline melcapstrom. But the baseline melcapstrom under those training doesn't do as well.\nSpeaker G: I'm trying to understand why it's 80% that's an accuracy number I guess right?\nSpeaker G: So that's not as good as the one up above.\nSpeaker G: But the 50 is better than the one up above so I'm confused.\nSpeaker C: Actually the noise composition whatever we are putting in it works very well for the high mismatched condition.\nSpeaker C: When it's consistent in the speech.car and in the clean training also it gives but this 50% is that the high mismatched performance.\nSpeaker C: It's equivalent to the high mismatched performance.\nSpeaker I: So since the high mismatched performance is much worse to begin with, it's easier to get it.\nSpeaker C: So by putting this noise.\nSpeaker E: Yeah if we look at the figures on the right we see that the reference.\nSpeaker C: The reference drops like a way of life.\nSpeaker E: Oh my gosh.\nSpeaker G: This is T.I. digits?\nSpeaker C: Yeah.\nSpeaker C: Yeah it's not written anywhere.\nSpeaker C: It's the first spreadsheet is T.I. digits.\nSpeaker G: How does clean training do for the car?\nSpeaker C: The car still that's still consistent.\nSpeaker C: I mean I get the best performance in the case of car which is the third column in the A condition.\nSpeaker G: No I mean this is added noise.\nSpeaker G: I mean this is T.I. digits.\nSpeaker G: I'm sorry I'm in the multi-language.\nSpeaker C: That's the next spreadsheet.\nSpeaker C: So that is the performance for Italian, Finnish and Spanish.\nSpeaker G: Training condition.\nSpeaker G: Oh right so clean it corresponds to the high mismatched.\nSpeaker G: And increase.\nSpeaker C: That's increase improvement.\nSpeaker C: That percentage increases the percentage improvement of the baseline.\nSpeaker G: Which means decrease in order right?\nSpeaker G: Yep.\nSpeaker G: Okay so percentage increase means decrease.\nSpeaker D: Okay.\nSpeaker D: There was a very long discussion about this on the answer of meeting.\nSpeaker D: How to calculate it.\nSpeaker D: I guess you are using finally this key which is that in the spreadsheet.\nSpeaker C: I'm not changing anything in there.\nSpeaker C: So yeah so all the HM numbers are very good.\nSpeaker C: And the percentage they are better than what the French League of Cats.\nSpeaker C: But the only number that's still I mean which Stefan also guarding is resolved was that medium mismatch of the Finnish which is a very strange situation where we use the we change the proto for initializing the HM.\nSpeaker C: I mean this is basically because it gets stuck in some local minimum in the training.\nSpeaker C: That's 75.79 in the Finnish mismatch which is that the 11.96 what do you see?\nSpeaker G: We have to jiggle it somehow.\nSpeaker C: Yeah so we start with the difference proto and it becomes 88 which is like some 50% improvement.\nSpeaker C: We'll start with a different one.\nSpeaker C: Different prototype which is like a different initialization for the transition probabilities.\nSpeaker C: The right now the initialization is to stay more in the current state which is 0.4.6 right?\nSpeaker C: Yeah.\nSpeaker C: And if you change it to 0.5.5 which is equal the voltage for transition and self-loop when it becomes 88%.\nSpeaker I: So that involves mucking with the back end?\nSpeaker I: Yeah we can't do it.\nSpeaker H: Yeah.\nSpeaker D: It is well known this medium-match condition of the Finnish data is very strange effects.\nSpeaker C: It has a very few words also it's very very small said actually.\nSpeaker D: There is a lot of utterances with music and the background.\nSpeaker C: It has a musicals.\nSpeaker C: I mean very audible music like you can.\nSpeaker G: So maybe for that one you need a much smarter V80.\nSpeaker C: So that's about the results.\nSpeaker C: The summary is like okay so the other thing I tried was which I explained in the last meeting is using the channel 0 for both dropping and estimating the noise.\nSpeaker C: And that's like just to get a feel of how good it is.\nSpeaker C: And it gets the 56% improvement in the speech that car becomes like 67% like 10% better.\nSpeaker C: But that's not a cheating experiment.\nSpeaker D: But the 47.9% which you have now sets already a remarkable improvement in comparison to the first proposal.\nSpeaker C: Yeah so we had 44% in the first proposal.\nSpeaker C: So the major improvement that we got was in all the high mismatch cases because all those numbers were in 60s and 70s because we never had in our compensations.\nSpeaker C: So that's where the biggest improvement came up not much in the well match and the medium match and the iDG it's also right now.\nSpeaker C: So this is still a 3 or 4% improvement over the first proposal.\nSpeaker G: Yeah so that's good.\nSpeaker G: So we can improve the noise estimation.\nSpeaker D: Yeah I started thinking about that.\nSpeaker D: I mean I discovered the same problem when I started working on this Aurora task almost two years ago that you have the problem with this multi.\nSpeaker D: At the beginning we had only this multi-conditioned training of the t iDG and I found the same problem just taking what we were used to use.\nSpeaker D: I mean some type of spectral subtraction you get even worse results in the basis.\nSpeaker D: Yeah I tried to find an explanation for it.\nSpeaker C: Yes Stefan also has the same experience of using the spectral subtraction right.\nSpeaker C: So here I mean I found that if I change the noise estimate I could get an improvement.\nSpeaker C: So something which I can actually pursue is the noise estimate.\nSpeaker D: Yeah I think what you do is when you have this multi-conditioned training mode then you can train models for the speech, for the words as well as for the pauses where you really have all information about the noise available.\nSpeaker D: It was surprising at the beginning it was also surprising to me that you get really the best results when doing it this way.\nSpeaker D: I mean in comparison to any type of training on clean data and any type of processing but it was so it seems to be the best what we can do in this moment with this multi-conditioned training.\nSpeaker D: And when we now start introducing some noise reduction technique we introduce also some how artificial distortions.\nSpeaker D: And this artificial distortions I have the feeling that they are the reason why we have the problems in this multi-conditioned training.\nSpeaker D: I mean the HMM suit chains they are based on gossians and modeling gossians.\nSpeaker D: And can I move a little bit with this?\nSpeaker D: And if we introduce now this spectral subtraction or wiener filtering stuff.\nSpeaker D: So usually what you have is maybe I'm showing now an adulope, maybe first time.\nSpeaker D: So usually in clean condition you have something which looks like this and if it is noisy it is somewhere here.\nSpeaker D: And then you try to subtract it or wiener filter or whatever.\nSpeaker D: And what you get is you have always these problems that you have these zeros in there.\nSpeaker D: And you have to do something if you get these negative values.\nSpeaker D: I mean this is your noise estimate and you somehow subtract it or do whatever.\nSpeaker D: And then you have, and then I think what you do is you introduce some artificial distribution in this model.\nSpeaker D: I mean you train it also this way but somehow there is no longer a gossian distribution.\nSpeaker D: It is somehow a strange distribution which we introduce with this artificial distortions.\nSpeaker D: And I was thinking that that might be the reason why you get these problems especially in the multi-conditioned training.\nSpeaker C: The models are not complex enough to absorb that additional variability that you are introducing.\nSpeaker E: I also have the feeling that the reason why it doesn't work is that the models are not complex enough.\nSpeaker E: Because I actually always had a good experience with spectral subtraction, just a straight spectral subtraction algorithm when I was using neural networks, big neural networks which maybe are more able to model strange distributions.\nSpeaker E: But yeah, then I tried exactly the same spectral subtraction algorithm on these Aurora tasks and it simply doesn't work. It's even the Earths.\nSpeaker G: We probably should at some point here try the tandem, the system two kind of stuff with this with the spectral subtraction for that reason.\nSpeaker G: Because again it should do a transformation to the main war maybe. It looks more gossian.\nSpeaker D: Yeah, I was just yesterday when I was thinking about it.\nSpeaker D: What we could try to do about it, I mean if you get at this situation that you get this negative failure, you simply set it to zero or to a constant or whatever.\nSpeaker D: If we would use some random generator which has a certain distribution, not a certain special distribution, we have to think about it.\nSpeaker D: And so we introduce again some natural behavior in this trajectory.\nSpeaker C: Very different from speech.\nSpeaker D: I mean it shouldn't confuse them. Yeah, similar to what you see really in the real noisy situation or in the clean situation but somehow a natural distribution.\nSpeaker G: This nots again sort of the idea of the additive thing as we head in the J stuff. Basically if you have random data in the time domain, then we look at this spectrum, it's going to be pretty flat.\nSpeaker G: So just add something everywhere rather than just in those places. It's just a constant.\nSpeaker D: I think it's just especially in this segment I mean you introduce some very artificial behavior.\nSpeaker G: Yeah, we see if you add something everywhere it has almost no effect up on top. And it has significant effect down there.\nSpeaker C: That's true. Those regions are the cost for this big radiation. Those negative values and whatever you get.\nSpeaker D: We could think how what we could try. It was just an idea.\nSpeaker G: I think it was noisy. People should just speak up.\nSpeaker E: Look at the French Silicon proposal. They use some kind of noise addition. They have a random number generator right.\nSpeaker E: They add noise on the trajectory of the low energy.\nSpeaker E: Cicero low energy.\nSpeaker E: But I don't know much effect.\nSpeaker E: So it is similar to what I think because they have to log energy.\nSpeaker E: Yeah. And then just generate random number. They have some kind of mean and variance.\nSpeaker E: And they add this number to the log energy simply.\nSpeaker C: So the log energy after the cleaning up. So they add a random noise to it.\nSpeaker G: To the just the energy or to the to the male filter only to the log energy.\nSpeaker G: So because I mean I think this is most interesting for the male filters.\nSpeaker G: Right. Or FFTs one of the other.\nSpeaker D: But they do not apply filtering of the low energy or what?\nSpeaker D: Like like like spectral subtraction.\nSpeaker C: No, their filter is in the time domain.\nSpeaker C: Yeah. So they filter the time signal and then what are they calculate from this low energy?\nSpeaker C: Yeah. And after that it is almost the same as the baseline system.\nSpeaker C: And then the final log energy that they get that that they add some random noise.\nSpeaker G: Yeah, but again it's just log energy as opposed to.\nSpeaker C: Yeah. So it's not the right. Yeah. It's not the male filter bank output.\nSpeaker C: This is a log energy computed from the time domain signal or from the male filter banks.\nSpeaker E: So maybe it's just a way to decrease the importance of this particular parameter in the world feature vector.\nSpeaker E: If you add noise to one of the parameters you want to flat-ribute.\nSpeaker C: The variance here reduced.\nSpeaker G: So it could reduce the dependence on the amplitude.\nSpeaker D: So maybe.\nSpeaker C: So. So the other thing is just looking at a little bit on the delay issue where the delay of the system is like 180 millisecond.\nSpeaker C: So I just just tried another system.\nSpeaker C: I mean another filter which I've like shown at the end, which is very similar to the existing filter.\nSpeaker C: Only thing that the phase is like a totally non-linear phase because it's a it's not a symmetric filter anymore.\nSpeaker C: This is for the all the A. Yeah. So this like so this makes the delay like zero for the LDA because it's completely causal.\nSpeaker C: So I got actually just the results for the Italian for that and that's like so the 51.09 has become 48.06 which is like 3% relative degradation.\nSpeaker C: So I have like the 51.09.\nSpeaker C: So I know how it fares for the other conditions. So it's like a 3% relative degradation.\nSpeaker D: But is there a problem of the 180 millisecond?\nSpeaker D: Well, I mean I talked about it with Ufine.\nSpeaker G: So basically our position is that we shouldn't be unduly constraining the latency at this point because we're all still experimenting with trying to make the performance better in the presence of noise.\nSpeaker G: There is a minority in that group who is arguing who are arguing for having a further constraining of the latency.\nSpeaker G: So we're just continuing to keep aware of what the tradeoffs are and you know what do we gain from having longer or shorter latencies.\nSpeaker G: But since we always seem to at least get something out of longer latencies not being so constrained we're tending to go with that if we're not told we can't do it.\nSpeaker I: Where was the smallest latency of all the systems last time?\nSpeaker G: Well, the French Telecom was was was very short latency and they had very good result of 35.\nSpeaker G: So it's possible to get very short latency but again we're the the approaches that we're using are ones that.\nSpeaker G: I was just curious about where we are.\nSpeaker D: But I think this 30 milliseconds it did not include the CDLT calculation.\nSpeaker C: This is included now.\nSpeaker C: If they include the Delta it will be additional 40 milliseconds.\nSpeaker C: Yeah, they're using a nine point window which is like a four on either side which is like.\nSpeaker C: They didn't include that.\nSpeaker E: Where does the compression compression and decoding delay comes from?\nSpeaker C: That's the way the the frames are packed like you have to wait for one more frame to pack because it's the CRC is computed for two frames all years.\nSpeaker G: Well, they would need that 40 milliseconds also.\nSpeaker C: No, they actually change the compression scheme altogether.\nSpeaker C: So they have their own compression and decoding scheme and they know what they have but they have coded zero delay for that because they know they changed it.\nSpeaker C: The compression they have their own CRC their own error correction mechanism so they don't have to wait more one more frame to know whether the current frame is in error.\nSpeaker C: So they change the whole thing so that there's no delay for that compression and part also.\nSpeaker C: Even you have reported actually zero delay for the compression.\nSpeaker C: I don't know maybe you also have some different.\nSpeaker D: I think I used this scheme as it was before.\nSpeaker I: Okay, we've got 20 minutes.\nSpeaker E: Did you want to go next?\nSpeaker E: I can go next, you have.\nSpeaker E: Oh, it's a moment.\nSpeaker E: Yeah, you have to take my man.\nSpeaker G: I think I'm confused.\nSpeaker E: All right, so you have one sheet.\nSpeaker E: This one is you don't need it.\nSpeaker E: So you have to take the wall.\nSpeaker E: The five there should be five sheets.\nSpeaker G: Okay, I have four now because I left one with Dave because I thought I was dropping one off and passing the other's on.\nSpeaker G: So no, we're not.\nNone: Thanks.\nSpeaker E: Maybe there's not enough.\nSpeaker E: Yeah, this.\nSpeaker E: So yeah, there are two figures showing actually the performance of the current VAD.\nSpeaker E: So it's a new run network based on VLP parameters, which estimate seconds probabilities.\nSpeaker E: And then I just put a median filtering on this to smooth the probabilities.\nSpeaker E: I didn't use the scheme that's currently in the proposal because I don't want to.\nSpeaker E: In the proposal, well, in the system, we want to add like speech frame before every word and a little bit of a couple of frames after also.\nSpeaker E: But to estimate the performance of the VAD, we don't want to do that because it would artificially increase the.\nSpeaker E: The false alarm rate of speech detection.\nSpeaker E: So there is normally a figure for a finish and one for Italian.\nSpeaker E: And maybe someone has two for Italian because I'm missing one figure there.\nSpeaker E: No.\nSpeaker E: Yeah, so one surprising thing that we can notice first is that apparently the speech miss rate is higher than the false alarm rate.\nSpeaker D: So what is the little curve?\nSpeaker E: Yeah, there are two curves.\nSpeaker E: One curves for the close talking microphone, which is the lower curve.\nSpeaker E: One is for the distant microphone, which has more noise.\nSpeaker E: It's logical that the performance works.\nSpeaker E: So as I was saying, the miss rate is quite important, which means that we tend to label speech as silence.\nSpeaker E: I didn't analyze further yet, but I think it's maybe due to the fricative sounds, which maybe in noisy condition, maybe labeled silence.\nSpeaker E: And it may also be due to the alignment because, well, the reference alignment, because right now I just used an alignment obtained from a system train on channel zero.\nSpeaker E: I checked it a little bit, but there might be alignment errors.\nSpeaker E: Like the fact that the model tend to align their first state on silence and their last state on silence also.\nSpeaker E: So the reference alignment would label as speech some silence frame before speech and after speech.\nSpeaker E: This is something that we already noticed before.\nSpeaker E: So this code also explained the miss rate, maybe.\nSpeaker D: And this curve has the average over low database law?\nSpeaker E: Yeah, right.\nSpeaker E: And the different points of the curve are for five thresholds on the probability from point three to point seven.\nSpeaker E: So the threshold is the first threshold on the probability that puts the values to zero or one.\nSpeaker E: And then the median filtering.\nSpeaker C: So the median filtering is fixed.\nSpeaker E: You just change the threshold.\nSpeaker E: So going from channel zero to channel one, almost double the error rate.\nSpeaker E: Well, so it's a reference performance that we can know if you want to work on the VAD.\nSpeaker E: We can work on this basis.\nSpeaker F: Okay.\nSpeaker F: Is this VAD MLP?\nSpeaker F: Yeah.\nSpeaker E: How big?\nSpeaker E: It's a very big one.\nSpeaker E: I don't remember.\nSpeaker C: 350 inputs, 1000 hidden inputs and 12 outputs.\nSpeaker G: Middle sized one.\nSpeaker H: Yeah.\nSpeaker E: I don't know if you have questions about that or suggestions.\nSpeaker E: It seems the performance seems worse on finish.\nSpeaker E: Well, it's not trained on finish.\nSpeaker G: What's the train?\nSpeaker C: I mean, the MLP is not trained on finish.\nSpeaker C: Right.\nSpeaker C: What's the train?\nSpeaker C: Oh, sorry.\nSpeaker C: It's Italian.\nSpeaker C: Yeah, it is.\nSpeaker C: Oh, it's train.\nSpeaker C: Yeah.\nSpeaker H: Okay.\nSpeaker E: And also there are funny noises on finish more than on Italian.\nSpeaker E: I mean, like music.\nSpeaker E: Yeah, that's true.\nSpeaker E: So, yeah, we were looking at this.\nSpeaker E: But for most of the noises, noises are...\nSpeaker E: I don't know if we want to talk about that.\nSpeaker E: But the car noises are below like 500 Earth.\nSpeaker E: And we were looking at the music utterances.\nSpeaker E: And in this case, the noise is smaller about 2000 Earths.\nSpeaker E: Well, music energy is very low, apparently, from 0 to 2000 Earths.\nSpeaker E: So maybe just looking at this frequency range from 500 to 2000 would improve somewhat the VD.\nSpeaker H: Yeah.\nSpeaker C: So the world...\nSpeaker C: Some parameters we wanted to use or something.\nSpeaker D: So is the training based on this label files which you take as reference?\nSpeaker D: Well, it trains the neural net.\nSpeaker E: No, it's not.\nSpeaker E: It was trained on some alignment obtained.\nSpeaker E: For the Italian data, I think we trained a neural network with embedded training.\nSpeaker E: So, re-estimation of the alignment using the neural net work, yes.\nSpeaker E: Right?\nSpeaker C: Yeah, we actually trained on the Italian training part when we had another system.\nSpeaker E: So it was a phonetic classification system for the Italian or data.\nSpeaker E: For the other data that it was trained on, it was different. Like for the IDGETs, you used a word, a previous system that you added.\nSpeaker E: Yeah, yeah, that's true.\nSpeaker E: So the alignments from the different database that are used for training came from different system.\nSpeaker E: So system, then we put them together and you put them together and train the real.\nSpeaker E: Yeah.\nNone: Yeah.\nSpeaker E: But did you use channel...\nSpeaker E: Did you align channel one or so?\nSpeaker C: I just took the entire Italian training part.\nSpeaker C: So it was both channel zero plus channel one.\nSpeaker E: So the alignments might be wrong on channel one.\nSpeaker E: You know what?\nSpeaker E: It's possible.\nSpeaker E: So we might...\nSpeaker E: Yeah.\nSpeaker E: We can do a real alignments.\nSpeaker E: So these ones to retrain these alignments, which should be better because they come from close to working.\nSpeaker E: Yeah, that was my idea.\nSpeaker D: I mean, if it's not the same labeling, which is taking the spaces.\nSpeaker D: Yeah, possible.\nSpeaker D: Yeah.\nSpeaker C: I mean, so the system, so the VAD was trained on maybe different set of labels for channel zero and channel one.\nSpeaker C: Because the alignments were different for...\nSpeaker C: Certainly different because they were independently trained.\nSpeaker C: We didn't copy the channel zero alignments to channel one.\nSpeaker H: Yeah.\nSpeaker C: But for the new alignments, what you generated, you just copy the channel zero to channel one, right?\nNone: Yeah.\nSpeaker E: And actually, when we look at the VAD for some utterance, it's almost perfect.\nSpeaker E: I mean, just drop one frame, the first frame of speech.\nSpeaker E: So there are some utterances where it's almost 100% VAD performance.\nSpeaker E: But...\nSpeaker E: Yeah.\nSpeaker E: So the next thing is...\nSpeaker E: I have the spreadsheet for three different systems.\nSpeaker E: But for this, you only have to look right now on the speech data curve performance because I didn't test...\nSpeaker E: So I didn't test the spectrosuppraction on the IDGT yet.\nSpeaker E: So you have three sheets.\nSpeaker E: One is the proposal one system.\nSpeaker E: Actually, it's not exactly a proposal one.\nSpeaker E: It's the system that Sony will just describe.\nSpeaker E: But with Wiener filtering from...\nSpeaker E: Franz Telekom included.\nSpeaker E: So this gives like 57.7% error rate reduction on the speech data curve data.\nSpeaker E: And then I have two sheets where it's for a system where...\nSpeaker E: So it's again the same system, but in this case, we have spectrosuppraction with a maximum overestimation factor of 2.5.\nSpeaker E: There is smoothing of the gain trajectory with some kind of low pass filter, which has 40 milliseconds latency.\nSpeaker E: And then after subtraction, I add a constant to the energies.\nSpeaker E: And I have two cases where the first cases were the constant is 25 dB below the mean speech energy and the other is 30 dB below.\nSpeaker E: For this two system, we have like 55.5% improvement and 58.1.\nSpeaker E: So again, it's around 56.57.\nSpeaker G: As I know, the TI digit number is exactly the same for this last two.\nSpeaker E: Yeah, because I didn't...\nSpeaker E: For the Franz Telekom spectrosuppraction, including in our system, the TI digit number are the right one, but not for the other system.\nSpeaker E: Because I didn't test it yet, this system, including with spectrosuppraction on the TI digit data.\nSpeaker E: I just tested it on speech.com.\nSpeaker E: Ah, so that means the only thing...\nSpeaker E: You have two.\nSpeaker G: You just should look at that 58.09%.\nSpeaker G: Okay, good.\nSpeaker C: So by reducing the noise, the addition threshold to like minus 30 dB is like...\nSpeaker C: You are like reducing the flow of the...\nSpeaker C: No, it's a region.\nSpeaker E: It's a flow of slower.\nSpeaker G: I'm sorry, so when you say minus 25 or minus 30 dB with respect to what?\nSpeaker E: To the average speech energy, which is estimated on...\nSpeaker G: Okay, so basically you're creating a signal-to-noise ratio of 25 or 30 dB.\nSpeaker D: But I think what you do is...\nSpeaker D: When you have this...\nSpeaker D: After you subtract it, I mean, then you get something with this...\nSpeaker D: When you set the values to zero, and then you simply add an additive constant again.\nSpeaker D: So you shifted somehow, this whole curve is shifted again.\nSpeaker G: But did you do that before the thresholding to zero?\nSpeaker E: It's after the thresholding.\nSpeaker E: Oh, so you really want to do it before, right?\nSpeaker E: Maybe you might do it before.\nSpeaker G: Yeah, because then you would have lots of that phenomenon.\nSpeaker G: Yeah.\nSpeaker G: I think.\nSpeaker E: But still, when you do this and you take the log after that, it reduces the variance.\nSpeaker E: Right.\nSpeaker G: Yeah, that will reduce the variance that will help.\nSpeaker G: But maybe if you did it before, you get lots of these funny looking things.\nSpeaker G: He's trying.\nSpeaker C: But before, it's like adding this...\nSpeaker G: Right at the point where you're doing the subtraction.\nSpeaker G: Okay.\nSpeaker G: Essentially, you're adding a constant into everything.\nSpeaker D: But the way you step on it, it is exactly the way I've implemented it, the fold.\nSpeaker G: Oh, you better do it different than that.\nSpeaker G: Just use a set it for a particular signal-to-noise ratio.\nSpeaker G: Set you what?\nNone: Yeah.\nSpeaker D: I made a similar investigation lecture for that year.\nSpeaker D: Just adding this constant and looking...\nSpeaker D: How do you paint it?\nSpeaker D: Is it on the value of the constant?\nSpeaker D: Yeah.\nSpeaker D: So, those students are more to give on average the best result.\nSpeaker D: Yeah.\nSpeaker D: So, a range of a signal-to-noise ratio.\nSpeaker H: Oh, it's clear.\nSpeaker E: I should have given other results.\nSpeaker E: So, it's clear.\nSpeaker E: When you don't unknow, it's much worse, like around 5% worse, I guess.\nSpeaker E: And if you add too much noise, it gets worse.\nSpeaker E: So, and it seems that right now, this is a constant that does not depend on anything that you can learn from the utterance.\nSpeaker E: It's just a constant noise addition.\nSpeaker E: And I think...\nSpeaker G: I'm confused.\nSpeaker G: I thought you were saying it doesn't depend on the utterance, but I thought you were adding an amount that was 25 dB down from the signal-to-noise ratio.\nSpeaker E: Yeah. So, the way I did that, I just measured the average speech energy of the old Italian data.\nSpeaker E: Oh!\nSpeaker E: Then I use this as mean speech energy.\nSpeaker G: Oh, it's just a constant amount.\nSpeaker E: And overall...\nSpeaker E: I observe that for Italian and Spanish, when you go to 30 and 25 dB, it's good.\nSpeaker E: It stays in this range.\nSpeaker E: Well, the performance of this algorithm is quite good.\nSpeaker E: But for Finnish, you have a degradation already when you go from 35 to 30 and then from 30 to 25.\nSpeaker E: And I have the feeling that maybe it's because just Finnish has a mean energy that's lower than the other databases.\nSpeaker E: And due to this, the threshold should be...\nSpeaker E: Yeah.\nSpeaker E: The noise addition should be lower.\nSpeaker G: But in the real thing, you're not going to be able to measure what people are doing over half an hour or an hour or anything, right?\nSpeaker G: So, you have to come up with this number from something else.\nSpeaker D: But you're not doing it now, like, which dependent or?\nSpeaker E: It's not. It's just something that's fixed.\nSpeaker G: What he is doing language dependent is measuring what that number references that he comes down to 25.\nSpeaker E: No, because I did it.\nSpeaker E: I started working on Italian.\nSpeaker E: I obtained this average energy.\nSpeaker E: Yeah.\nSpeaker E: Then I used this one.\nSpeaker C: For all languages.\nSpeaker C: Yeah.\nSpeaker G: Okay.\nSpeaker G: So it's sort of arbitrary.\nSpeaker E: I mean, so if you think, yeah.\nSpeaker E: Yeah. So the next thing is to use this as maybe initialization and then use something online.\nSpeaker E: It's an important thing.\nSpeaker E: I expect improvement at least on Finnish because the way...\nSpeaker E: Well, for Italian and Spanish, this value works good, but not necessarily for Finnish.\nSpeaker E: But unfortunately, there is, like, this 40 millisecond latency.\nSpeaker E: Yeah. So I will try to so much reduce this.\nSpeaker E: I already know that if I completely remove this latency, so it...\nSpeaker E: There is 3% hit on Italian.\nSpeaker C: This latency...\nSpeaker C: Sorry.\nSpeaker D: Yeah.\nSpeaker D: Yes, moving was over this, sort of, say, the factor of the Wiener.\nSpeaker D: What was it?\nSpeaker D: This moving, it was over the subtraction factor, sort of, say.\nSpeaker E: It's the smoothing over the gain of the subtraction aggregate.\nSpeaker D: And you are looking into the future, into the past.\nSpeaker E: Right.\nSpeaker E: So you do smoot this.\nSpeaker D: Did you try simply to smooth...\nSpeaker D: To smooth through is the envelope?\nSpeaker E: No, I did not.\nSpeaker D: Because it means you should have a similar effect if you...\nSpeaker D: I mean, you have now several stages of moving, so to say you start up.\nSpeaker D: As far as I remember, you smooth somehow.\nSpeaker D: The envelope, you smooth somehow the noise estimate and later on, you smooth also this subtraction factor.\nSpeaker E: No, it's...\nSpeaker E: It's just a gain that smooth actually.\nSpeaker E: Actually, I do all the smoothing.\nSpeaker E: Oh, it was...\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: No, in this case, it's just a gain.\nSpeaker E: But the way it's done is that...\nSpeaker E: For low gain, there is this non-linear smoothing actually.\nSpeaker E: For low gains, I use the smoothed version.\nSpeaker E: But for high gain, I don't smooth.\nSpeaker D: It just...\nSpeaker D: The experience shows if you do the...\nSpeaker D: The best is to do the smooths moving as early as possible.\nSpeaker D: So when you start up, I mean, you start up with the...\nSpeaker D: Somehow with the noisy envelope.\nSpeaker D: But the best is to smooth this somehow.\nSpeaker E: Yeah, I could try this.\nSpeaker C: So before estimating the SNR itself smoothed envelope.\nSpeaker E: But yeah.\nSpeaker E: Then we need to find a way to smooth less also when there is high energy.\nSpeaker E: Because I noticed that it helps a little bit to smooth more during low energy portions.\nSpeaker E: Yes.\nSpeaker E: Less during speech because if you smooth...\nSpeaker E: Then you kind of distort the speech.\nSpeaker E: Right.\nSpeaker D: Yeah, I think when...\nSpeaker D: You could do it in this way that you say if you...\nSpeaker D: You have somehow a noise estimate.\nSpeaker D: If you say with my envelope, I'm close to this noise estimate.\nSpeaker D: Then you have a bad signal to a noise ratio and then you would like to have a stronger smoothing.\nSpeaker D: So you could...\nSpeaker D: Yeah, you could base it on your estimation of the signal to noise ratio on your actual.\nSpeaker C: Or some...\nSpeaker E: Yeah, but I don't trust the current value.\nSpeaker C: Yeah, not right now.\nSpeaker H: The value later will much better.\nSpeaker E: I think that's it.\nSpeaker D: To summarize the performance of the speech that car results as similar to the euros.\nSpeaker C: Yeah, so the 5080 is like the bad...\nSpeaker C: You have 56,000 euros.\nSpeaker C: Yeah, that's true.\nSpeaker D: And depending on this, the additive constant is slightly better.\nSpeaker E: It's better, of course.\nSpeaker E: And yeah, the condition where it's better than your approach, it's just because maybe it's better on well matched and that the weight on well matched is...\nSpeaker C: Yeah, you got a...\nSpeaker E: If you don't weigh differently, the different condition, you can see that your...\nSpeaker E: Well, the two-stage winner filtering is maybe better or...\nSpeaker E: It's better for I miss match, right?\nSpeaker C: Yeah, it's better for I miss match.\nSpeaker C: So the overall... Yeah, it was for the well matched condition.\nSpeaker I: So we need to combine these two.\nSpeaker C: That's the best thing is like the French telecom system is optimized for the well matched condition.\nSpeaker C: So they know that the weighting is good for the well matched.\nSpeaker C: So everywhere the well matched performance is very good for French telecom.\nSpeaker C: We also have to do something similar.\nSpeaker G: Our tradition here has always been to focus on this match.\nSpeaker D: Is this more interesting?\nSpeaker D: My body was a tool.\nSpeaker D: For a started working on this world.\nSpeaker I: Yeah.\nSpeaker I: Okay, Carmen, do you...\nSpeaker B: I only say that this is the summary of all the BTS' experiment.\nSpeaker B: And say that the result in the last...\nSpeaker B: For Italian, the last experiment for Italian are bad.\nSpeaker B: I make a mistake when I write.\nSpeaker B: Obviously, I copy one of the bad results.\nSpeaker B: So you...\nSpeaker B: Yeah.\nSpeaker B: You know, this...\nSpeaker B: Well, if we put everything, we improve a lot, compared to usual BTS, but the final result are not still good like the winner filter, for example.\nSpeaker B: I don't know, maybe it's possible.\nSpeaker B: It's somewhere to have the same result, I don't know exactly.\nSpeaker B: Because I have...\nSpeaker C: You have a better result.\nSpeaker B: What result in medium mismatch?\nSpeaker C: I have some results that are good for the high mismatch.\nSpeaker B: Yeah.\nSpeaker B: Or something.\nSpeaker B: I'm more or less similar, but are worse.\nSpeaker B: And still, I don't have the result for K-digit.\nSpeaker B: The program is running.\nSpeaker B: Maybe for this weekend, I will have the result for K-digit.\nSpeaker B: And I can't complete it.\nSpeaker B: I'm going to write this.\nSpeaker H: Right.\nSpeaker B: One thing that I know are not here in this result, but I spoke in before with Sunil, I improved my result using clean LDA filter.\nSpeaker B: If I use the LDA filter that are running with the noise speed, that hurts my result.\nSpeaker G: So what are these numbers here?\nSpeaker G: Are these with the cleaner with the noise?\nSpeaker B: With the noise, I have worse results that I didn't use.\nSpeaker B: Maybe because with this technique, we are using really clean speed.\nSpeaker B: The speed representation that goes to the HTK is really clean speed because it's from the dictionary.\nSpeaker B: They could move.\nSpeaker B: Maybe for that.\nSpeaker B: Because I think that it's an experiment using the two LDA filter, a noisard, doesn't matter to me.\nSpeaker E: Yeah, I did that, but it doesn't matter on the speech that car, but it matters a lot on the idgis.\nSpeaker E: It's better when you use the clean filter.\nSpeaker E: Yeah, it's much better when you use the clean derived LDA filter.\nSpeaker H: Yeah, it's clean.\nSpeaker C: But yeah, Sunil, you know, my result is with the noisard.\nSpeaker C: No, it's with the noisard.\nSpeaker C: It's not the clean LDA.\nSpeaker C: In the front sheet, I have like the summary.\nSpeaker E: And your result is with the clean LDA.\nSpeaker C: And in your case, it's all noisy.\nSpeaker E: All noisy.\nSpeaker E: Yeah, but I'm serving my case.\nSpeaker E: At least on speech that car, it doesn't matter, but the idgis is matters.\nSpeaker E: Like two or three percent absolute.\nSpeaker E: Absolutely.\nSpeaker E: So you really might want to.\nSpeaker G: Yeah, I would like to look at it.\nSpeaker G: Yeah, I could be sizable right there.\nSpeaker D: Maybe you're living in about two weeks.\nSpeaker D: Yeah.\nSpeaker D: So I mean, if I would put on the head of a project manager, I would say I'm the most.\nSpeaker D: So much time left.\nSpeaker D: So I mean, what I would do is I would pick the best consolation which you think.\nSpeaker D: And create all the results for the whole database that you get to the final number.\nSpeaker D: So no, didn't.\nSpeaker D: Maybe also to write somehow a document where you describe your approach.\nSpeaker B: I was thinking to do that this week.\nSpeaker G: I'll borrow the head back and agree.\nSpeaker B: Yeah, that's the.\nSpeaker G: Right.\nSpeaker G: In fact, actually, I guess the Spanish government requires that anyway they want some kind of report from everybody who's in the program.\nSpeaker G: So of course, we'd like to see it too.\nSpeaker I: So what do you think we should do the digits or skip it or?\nSpeaker G: We have them now.\nSpeaker G: Why don't we do it?\nSpeaker G: Just take a minute.\nSpeaker G: Oh, sorry.\nSpeaker I: So I guess I'll go ahead.\nSpeaker I: It was mentioned.\nSpeaker I: 5458895\nSpeaker B: Transcript L-303-136084-010 10401-238-4673-7474-3061536-3911-321472417-818018387 7358894381 083734619\nSpeaker C: Transcript L-3041164800349 821974208 6291-26456761-482749558 6101109292 6703449512 0380274696 819163804\nSpeaker D: This is transcript L-3052421872083 310896351060801834788 91917067026595916040016661048985 29053636371425227031\nSpeaker E: Transcript L-3025785876209 7251630444 8930628276 87297015 83840504347 892661195736 623105907\nSpeaker A: 71444222 8344222\nNone: 8344222 8344237404222 6571 54608550658916864078 012947781 94223648053\nSpeaker G: Transcript L-301 76698879925 2149198989 9411349543 8395604911 176734304 6469388185 12938212 175820090255\nSpeaker F: Transcript L-282 6730000596 733051850 0457555828 291515814306692073 8638177386 158196395967 5642534762\nSpeaker I: 8396 476\nNone: 642R 71 7\nNone: Thank you.\n"}, {"meeting_id": "Bro018", "summary": "The meeting began with an update on the voice-unvoice detection. The professor suggested the solution might just be feeding the log magnitude of the spectrum into a simple neural net. The team then had a brief discussion about some confusing formula on the France Telecom proposal. Following this, the team talked about how to pick out the most important acoustic events. Then, Grad A introduced the team to phonological matchings that he was working on using Support Vector Machines. The goal was to map MFCC's to phonological features. The team concluded by discussing a potential bug which led to a significant difference between PLP and mel cepstrum and the FFT method that they were exploring.", "dialogue": "Speaker C: Okay, so, it's not here, so.\nSpeaker G: Yeah, I would try to explain the thing that I did this week, during this week.\nSpeaker G: Well, you know that I began to work with a new feature to the text voice and voice, where I tried to MLP to with this new feature and the feature from the bass system.\nSpeaker E: The Melcapsum?\nSpeaker G: No, exactly the Melcapsum, the new bass system, the new bass system.\nSpeaker G: Oh, the Aurora system.\nSpeaker G: Yeah, the Aurora system with the new filter, the idea, something like that.\nSpeaker G: And I tried to MLP one that only have three outputs, voice and voice and silence, and other one that have 56 outputs, the probabilities that they have.\nSpeaker G: And I tried to do some experiment of recognition with that.\nSpeaker G: And I only have a result with the MLP with the three outputs.\nSpeaker G: And I put together the 15 features and the three MLP outputs.\nSpeaker G: And while the results are a little bit better but more or less similar.\nSpeaker C: I'm slightly confused.\nSpeaker C: What feeds the three output net?\nSpeaker G: Voice and voice and silence.\nSpeaker C: No, no, what feeds it? What features does it say?\nSpeaker G: The input, the input are the 15, the 15 basses feature with the new code.\nSpeaker G: And the other three features are the variance of the different between the two spectrum.\nSpeaker G: The variance of the autocorrelation function, I said that the first point, because half the height value is 0 and also R0.\nSpeaker G: The first coefficient of the autocorrelation function, that is like the energy with the three features.\nSpeaker C: You wouldn't do like R1 over R0 or something like that.\nSpeaker C: Usually for voice and voice, you do something, you do energy but then you have something like spectral slope, which is you get like R1 over R0 or something like that.\nSpeaker G: What are the R's?\nSpeaker B: Are the correlations?\nSpeaker G: No, autocorrelations, yes, the variance of the autocorrelation function that used that.\nSpeaker G: Well, that's the variance.\nSpeaker C: But if you just say what is, I mean, the first order, one of the different between voice and voice and silence is energy.\nSpeaker C: But the other one is the spectral shape.\nSpeaker C: The spectral shape, yes.\nSpeaker C: Yeah, and so R1 over R0 is what you typically use for that.\nSpeaker G: No, I don't use that. I can use.\nSpeaker C: No, I'm saying that's what people usually typically use.\nSpeaker C: See, because this is just like a single number to tell you does the spectrum look like that or does it look like that?\nSpeaker C: So if it's low energy but the spectrum looks like that or like that, it's probably silence.\nSpeaker C: But if it's low energy and the spectrum looks like that, it's probably unvoiced.\nSpeaker C: Yeah.\nSpeaker C: So if you just had to pick two features to determine voice and voice, you pick something about the spectrum like R1 over R0 and R0.\nSpeaker C: Or you know, you have some other energy measure and like in the old days people did like zero crossing counts.\nSpeaker C: Yeah, yeah.\nSpeaker G: Right.\nSpeaker G: I can also use this.\nSpeaker B: Yeah.\nSpeaker G: Because the result are really better but we have in a point that everything is more or less similar.\nSpeaker G: But not quite the...\nSpeaker C: Right, but it seemed to me that what you were getting at before was that there is something about the difference between the original signal or the original FFT.\nSpeaker C: And what the filter, which is what...\nSpeaker C: And the variance was one take on it.\nSpeaker C: Yeah, I used this too.\nSpeaker C: Right, but it could be something else because suppose you didn't have anything like that.\nSpeaker C: Then in that case, if you have two nets, and this one has three outputs, and this one has whatever, 56 or something.\nSpeaker C: If you were to sum up the probabilities for the voice and for the unvoiced and for the silence here, we found in the past you would do better at voice and voice silence than you do with this one.\nSpeaker C: So just having the three output thing doesn't really value anything.\nSpeaker C: Yeah.\nSpeaker C: The issue is what you feed it.\nSpeaker G: Yeah, I have...\nSpeaker G: Yeah.\nSpeaker C: So you're saying...\nSpeaker E: So you can take the features that go into the voice and voice silence net and feed those into the other one as additional inputs rather.\nSpeaker C: Yeah.\nSpeaker C: Well, that's another way.\nSpeaker C: That wasn't what I was saying, but yeah, that's certainly another thing to do.\nSpeaker C: Now, I was just trying to say, if you bring this into the picture over this, what more does it buy you?\nSpeaker C: And what I was saying is that the only thing that I think that it buys you is based on whether you feed it something different.\nSpeaker C: And something different and something fun to men away.\nSpeaker C: And so the kind of thing that she was talking about before was looking at something...\nSpeaker C: Something about the difference between the log FFT log power and the log magnitude FF spectrum in the filter bank.\nSpeaker C: Yeah.\nSpeaker C: And so the filter bank is chosen, in fact, to sort of integrate out the effects of pitch.\nSpeaker C: And she's saying, you know, trying...\nSpeaker C: So the particular measure that she chose was the variance of this difference.\nSpeaker C: But that might not be the right number.\nSpeaker C: Maybe.\nSpeaker C: I mean, maybe there's something about the variance that's not enough or maybe there's something else that one could use.\nSpeaker C: I think that for me, the thing that struck me was that you want to get something back here.\nSpeaker C: So here's an idea.\nSpeaker C: What about if you skip all the really clever things and just fed the log magnitude spectrum into this?\nSpeaker G: I'm sorry.\nSpeaker C: This is...\nSpeaker C: You have the log magnitude spectrum.\nNone: Yeah.\nSpeaker C: And you were looking at that and the difference between the filter bank and computing the variance.\nSpeaker C: That's a clever thing to do. What if you stop being clever?\nSpeaker C: And you just took this thing in here because it's a neural net and neural nets are wonderful and figure out what they can...\nSpeaker C: What they most need from things.\nSpeaker C: Yeah.\nSpeaker C: I mean, that's a good act.\nSpeaker C: So I mean, you're trying to be clever and say, what's the statistic that we should get about this difference?\nSpeaker C: But in fact, you know, maybe just feeding this in or feeding both of them in.\nSpeaker C: In other words, saying let it figure out what is the interaction.\nSpeaker C: Especially if you do this over multiple frames, then you have this over time and both kinds of measures and you might get something better.\nSpeaker E: So don't do the division but let the net have everything.\nSpeaker B: That's another thing you could do, yeah.\nSpeaker C: I mean, it seems to me if you have exactly the right thing, then it's better to do it without the net because otherwise you're asking the net to learn this thing.\nSpeaker C: You know, say if you wanted to learn how to do multiplication.\nSpeaker C: I mean, you could feed two numbers that you wanted to multiply into a net and have a bunch of nonlinearities in the middle and train it to get the product at the output.\nSpeaker C: And it would work, but it's kind of crazy because we know how to multiply and you'd be much lower error usually if you just multiply it out.\nSpeaker C: But suppose you don't really know what the right thing is.\nSpeaker C: And that's what these sort of dumb machine learning methods are good at.\nSpeaker E: How long does it take Carmen to train up on a business?\nSpeaker G: Oh, not too much. One day useless.\nSpeaker C: Yeah, it's probably worth it.\nSpeaker A: What are your frame error rates?\nSpeaker G: 56, the frame error rate 56 AC.\nSpeaker G: Is that maybe that's accuracy?\nSpeaker A: The accuracy.\nSpeaker A: 56% accurate for voice unvoiced.\nSpeaker G: No, yes.\nSpeaker G: I don't remember for voice unvoiced.\nSpeaker G: Maybe for the other one.\nSpeaker G: Yeah, for voice unvoiced, hopefully it'd be a lot better.\nSpeaker G: Maybe for voice unvoiced.\nSpeaker G: This is for the other one.\nSpeaker G: I can't say that.\nSpeaker G: But I think that 55 was for the when the output are the 56 foam.\nSpeaker G: That I look in the with the other.\nSpeaker G: The other MLP that we have are more or less the same number.\nSpeaker G: It's a little bit better, but more or less the same.\nSpeaker C: I think at the frame level for 56, I was the kind of number we were getting for.\nSpeaker C: Reduce bandwidth.\nSpeaker G: I think that for the other one for the three output is 62, 63 more or less.\nSpeaker G: That's all?\nSpeaker C: Yeah.\nSpeaker G: That's pretty bad.\nSpeaker G: Yeah, because it's noise also.\nSpeaker G: And we have.\nSpeaker C: I know.\nSpeaker C: But even in training still.\nSpeaker C: Well, actually, so this is a test you should do then.\nSpeaker C: If you're getting 56% over here, that's a noise also, right?\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: Okay.\nSpeaker C: If you're getting 56 here, try adding together the probabilities of all of the voice phones here and all the unvoiced phones and see what you get then.\nSpeaker C: I bet you get better than 63.\nSpeaker G: Well, I don't know.\nSpeaker G: I think that we have to resolve more.\nSpeaker G: Maybe.\nSpeaker G: I don't know.\nSpeaker G: I'm not sure, but I remember it's a sensitivity.\nSpeaker G: I can't solve that.\nSpeaker C: Okay, but that's a good checkpoint.\nSpeaker C: You should do that anyway.\nSpeaker C: Given this regular old net that's just what you're using for other purposes, add up the probabilities of the different subclasses and see how well you do.\nSpeaker C: Anything that you do over here should be at least as good as that.\nSpeaker G: How do you do that?\nSpeaker E: The targets for the neural net.\nSpeaker E: They come from forced alignments.\nSpeaker G: Timmit.\nSpeaker G: Oh, this is trained on Timmit.\nSpeaker C: Yeah.\nSpeaker G: Okay.\nSpeaker G: Yeah, this is for Timmit.\nSpeaker C: But noisy Timmit?\nSpeaker G: No, it's a Timmit.\nSpeaker G: We have no system with the noise of the TID digits.\nSpeaker G: And now we have another noise system, it also with the noise of Italian database.\nSpeaker B: Yeah.\nSpeaker C: Well, there's going to be, it looks like there's going to be a noisy, some large vocabulary noisy stuff too.\nSpeaker C: Something that's preparing.\nSpeaker C: Yeah.\nSpeaker C: I forget it will be resource management, Wall Street Journal, something.\nSpeaker C: Some red task, actually, that there.\nSpeaker C: Or what?\nSpeaker C: Or a raw?\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: So the issue is whether people make a decision now based on what they've already seen or they make it later.\nSpeaker C: And one of the arguments for making it later is let's make sure that whatever techniques that we're using work for something more than connected digits.\nSpeaker B: So.\nSpeaker E: When are they planning, when would they do that?\nSpeaker C: I think in the summer sometime.\nSpeaker B: Okay.\nSpeaker G: This is the work that I did in this date.\nSpeaker G: And also, I, Henik, last week, say that if I have time, I can't begin to study seriously the French telecom proposal to look at the code and something like that.\nSpeaker G: To know, is that what they are doing?\nSpeaker G: Because maybe that we can have some ideas.\nSpeaker G: But not only to read the proposal, look carefully what they are doing with the program and something like that.\nSpeaker G: And I begin to work also in that.\nSpeaker G: But the first thing that they don't understand is that they are using the lock energy.\nSpeaker G: That this quiet, I don't know why they have some constant in the expression of the law energy.\nSpeaker G: I don't know what that means.\nSpeaker E: They have a constant.\nNone: Yeah.\nSpeaker C: At the front, it says, a lock energy is equal to the rounded version of 16 over the log of two.\nSpeaker B: Well, this is natural log and maybe something to do with the fact that this is, I have no idea.\nSpeaker B: Yeah, that's what I was thinking.\nSpeaker B: But there's a 64.\nSpeaker G: Because maybe the threshold that they are using and the basis of this value, I don't know exactly.\nSpeaker G: Because I thought maybe they have a meaning, but I don't know what the meaning of this value is.\nSpeaker G: Yeah, that's pretty funny.\nSpeaker E: The number inside the log and raising it to 16 over log is two.\nSpeaker B: Yeah, I'm right.\nSpeaker E: I have to do with the 64.\nSpeaker C: For you know, the 16, the natural log of one over the natural log of two times the natural, I don't know.\nSpeaker C: Maybe something will think of something, but this is, it may just be that they want to have some very small energies.\nSpeaker G: They want to have some kind of a...\nSpeaker G: I can understand the effect of this, because it's too, to do something like that.\nSpeaker C: Well, since you're taking a natural log, it says that when you get down to essentially zero energy, this is going to be the natural log of one, which is zero.\nSpeaker C: So it'll go down to the natural log being the lowest value for this will be zero.\nSpeaker C: So it restricted to being positive and sort of smoothed it for very small energies.\nSpeaker C: Why they chose 64 and something else, that was probably just experimental.\nSpeaker C: Yeah.\nSpeaker C: And the constant in front of it, I have no idea.\nSpeaker G: What?\nSpeaker G: I will look to try if I move this parameter in the code, what happens?\nSpeaker G: Maybe the thresholds are in basis of this.\nSpeaker C: I agree.\nSpeaker C: They probably have some particular fixed pointer arithmetic that they're using.\nSpeaker E: They're just...\nSpeaker E: They do with hardware.\nSpeaker C: Yeah.\nSpeaker C: I mean, there's probably work with fixed pointer integer or something.\nSpeaker C: I think you're supposed to in this stuff anyway, and maybe that puts it in the right realm somewhere.\nSpeaker C: Yeah.\nSpeaker C: I think given that the level you're doing things in floating point on the computer, I don't think it matters to be my guess.\nSpeaker G: This is more of a thing.\nSpeaker C: Okay, and when did Stefan take off?\nSpeaker G: I think that the stuff I will arrive today, which tomorrow...\nSpeaker C: Well, he was gone this first few days, and then you see her for a couple days before he goes to Salt Lake City.\nSpeaker G: I think that he's in Las Vegas until then, that.\nSpeaker C: Yeah.\nSpeaker C: So he's going to a cast, which is good.\nSpeaker C: I don't know if there are many people here, but I cast, so make sure somebody go.\nSpeaker E: Have people sort of stopped going to a cast?\nSpeaker C: People are less consistent about going to a cast, but I think it's still a reasonable form for students to present things.\nSpeaker C: I think for engineering students of any kind, I think it's if you haven't been there much, it's good to go to, get a feel for things, a range of things, not just speech.\nSpeaker C: But I think for sort of diving the world's speech people, I think that ICSOP and Euro-Speech are much more targeted.\nSpeaker C: And then there are these other meetings like HLT and ASRU.\nSpeaker C: So there's actually plenty of meetings that are really relevant to computational speech processing, of one sort or another.\nSpeaker C: So, I mean, I mostly just ignored it because I was too busy, and didn't get to it.\nSpeaker C: So, I want to talk a little bit about what we're talking about this morning, just briefly.\nSpeaker A: Anything else?\nSpeaker A: Yeah, so I guess some of the progress I've been getting my committee members for the calls.\nSpeaker A: And so far, I'm more again, and being at Mike Jordan, and I asked John O'Halla to be agreed.\nSpeaker A: Cool? Yeah, yeah.\nSpeaker A: So, I just need to ask Malik.\nSpeaker A: And then I talked a little bit about continuing with these dynamic acoustic events.\nSpeaker A: And we're thinking about a way to test the completeness of a set of dynamic events, completeness in the sense that if we pick these X number of acoustic events, do they provide sufficient coverage for the phones that we're trying to recognize, or the words that we're going to try to recognize from Iran?\nSpeaker A: So, Morgan and I were discussing a form of a cheating experiment where we get a chosen set of features or acoustic events.\nSpeaker A: And we train up a hybrid system to do phone recognition on timet. So, the idea is if we get good phone recognition results using the set of acoustic events, then that says that these acoustic events are sufficient to cover a set of phones at least found a timet.\nSpeaker A: So, it would be a measure of, are we on the right track with the choices of our acoustic events.\nSpeaker A: So, that's going on. And also just working on my final project for Jordan's class, which is...\nSpeaker C: Actually, let me hold that thought. Let me back up while we're still on it. The other thing I was suggesting though is that given that you're talking about binary features, maybe the first thing to do is just to count and count co-occurrences and get probabilities for discrete HMM.\nSpeaker C: So, that would be pretty simple because it's just to say if you had 10 events that you were counting each frame would only have a thousand possible values for these 10 bits.\nSpeaker C: And so, you could make a table that would say if you had 39 phone categories, it would be a thousand by 39 and just count the co-occurrences and divide them by the...\nSpeaker C: count the co-occurrences between the event and the phone and divide them by the number of occurrences of the phone.\nSpeaker C: And that would give you the likelihood of the event given the phone. And then just use that in a very simple HMM.\nSpeaker C: And you could do phone recognition then and wouldn't have any of the issues of the training of the net or... I mean, it'd be on the simple side.\nSpeaker C: The example I was giving was if you had onset of voicing and end-of-voicing as being two kinds of events, then if you had those on mark correctly and you counted the co-occurrences, you should get it completely right.\nSpeaker C: So, but you'd get all the other distinctions randomly wrong. I mean, there'd be nothing to tell you that. So, if you just do this by counting, then you should be able to find out in a pretty straightforward way whether you have a sufficient set of events to do the kind of level of classification of phones that you'd like.\nSpeaker C: So, that was the idea. And the other thing that we were discussing was, okay, how do you get your training data?\nSpeaker C: Because the Switchboard Transcription Project was half-thousand people or so working off and on over a couple years.\nSpeaker C: Similar amount of data to what you're talking about with Dimit training. So, it seems to me that the only reasonable starting point is to automatically translate the current timid markings into the markings you want.\nSpeaker C: And it won't have the kind of characteristic that you'd like of catching funny kind of things that maybe aren't there from these automatic markings, but it's...\nSpeaker E: It's probably a good place to start.\nSpeaker C: Yeah, and a short amount of time. Just to, again, just to see if that information is sufficient to determine the phones.\nSpeaker E: And you could even then get an idea about how different it is you could maybe take some subset and go through a few sentences, mark them by hand and see how different it is from the clinical ones.\nSpeaker E: Just to get an idea of rough idea of if it really even makes a difference.\nSpeaker C: You get a little feeling for it that way. Yeah, that's probably right.\nSpeaker C: I guess it would be that this is since Dimit's red speech that this would be less of a big deal if you want to look at spontaneous speech before or after.\nSpeaker C: And the other thing would be say if you have these 10 events you want to see what if you took two events or four events or ten events.\nSpeaker C: And hopefully there should be some point in which having more information doesn't tell you really all that much more about what the phones are.\nSpeaker E: You could define other events as being sequences of these events.\nSpeaker C: You could, but the thing is what he's talking about here is a translation to a per frame feature vector.\nSpeaker C: So there's no sequence in that, I think.\nSpeaker C: Unless you've got your second pass over or something after you've got your...\nSpeaker C: Yeah, but we're just talking about something simple here.\nSpeaker C: Yeah, I'm adding complexity.\nSpeaker C: Yeah, with a very simple statistical structure, could you at least verify that you've chosen features that are sufficient?\nSpeaker C: Okay, and you were saying some of these are going to say something else about your class project?\nSpeaker A: Yeah, so for my class project, I'm tinkering with support vector machines, something that we learned in class, and basically just another method for doing classification.\nSpeaker A: So I'm going to apply that to compare it with the results by King and Taylor who did using recurrent known as they recognized a set of phonological features made of mapping from the MFCC's to phonological features.\nSpeaker A: So I'm going to do a similar thing with support vector machines.\nSpeaker E: So what's the advantage of support vector machines?\nSpeaker A: So support vector machines are good with dealing with less amount of data.\nSpeaker A: So if you give it less data, it still does a reasonable job in learning the patterns.\nSpeaker C: I guess they're sort of succinct.\nSpeaker E: Does there is some kind of a distance metric they use or what do they do for classification?\nSpeaker A: So the simple idea behind the support vector machine is you have this feature space, right? Then it finds the optimal separating play between these two different classes.\nSpeaker A: And so at the end of the day, what it actually does is it picks those examples of the features that are closest to the separating boundary and remembers those and uses them to recreate the boundary for the test set.\nSpeaker A: So given these features are these examples, critical examples, which they call support vectors, then given a new example, if the new example falls away from the boundary in one direction, then it's classified as being part of this particular class.\nSpeaker E: So why save the examples? Why not just save what the boundary itself is?\nSpeaker A: Yeah, that's a good question.\nSpeaker C: That's another way of doing it. So I guess it goes back to nearest neighbor sort of thing.\nSpeaker C: When is nearest neighbor good? Well nearest neighbor good is good if you have lots and lots of examples.\nSpeaker C: But of course if you have lots and lots of examples, then it can take a while to use nearest neighbor's lots of lookups. So a long time ago people talked about things where you would have condensed nearest neighbor where you would pick out some representative examples, which would be sufficient to represent to correctly classify everything that came in.\nSpeaker E: I think support vector stuff goes back to that kind of thing. So rather than doing nearest neighbor where you compare to every single one, you just pick a few critical ones.\nSpeaker C: And the neural net approach or gothse mixtures for that matter are sort of fairly brute force kinds of things where you sort of predefined that there's this big bunch of parameters and then you place them as your best can to define the boundaries.\nSpeaker C: And in fact, as you know, these things do take a lot of parameters. And if you have only a minus amount of data, you have trouble learning them.\nSpeaker C: So I guess the idea to this is that it is reputed to be somewhat better in that regard.\nSpeaker A: It can be a reduced parameterization of the model by just keeping certain selected examples.\nSpeaker C: But I don't know if people have done sort of careful comparisons of this on large tasks or anything. Maybe they have. I don't know.\nSpeaker D: So do you get some kind of number between zero and one at the output?\nSpeaker A: Actually, you don't get a nice number between zero and one. You get either a zero or one.\nSpeaker A: Basically, you get a distance measure at the end of the day. And then that distance measure is translated to a zero or one.\nSpeaker E: But that's looking at it for a binary classification. And you get that for each class. You get a zero or one.\nSpeaker C: But you have the distances to work with. Because actually, Mrs. Cippy stayed people to use support vector machines for speech recognition. And they were using it to make probabilities.\nSpeaker A: Yeah, they had a way to translate the distances into a probability with a simple sigmoidal function.\nSpeaker C: So they used sigmoidal or a softmax type thing. And didn't they like exponential or something and divide by the sum of them.\nSpeaker C: Oh, so it is a sigmoidal.\nSpeaker E: Did they get good results with that?\nSpeaker C: No, they were okay. I mean, I don't think they were earth shattering. But I think that this was a couple years ago.\nSpeaker C: I think people were very critical because it was interesting just to try this. And it was the first time they tried it.\nSpeaker C: So the numbers were not incredibly good. But there was reasonable.\nSpeaker C: I don't remember anymore. I don't remember what the task was. It was broadcast news or something.\nSpeaker A: So I'm not planning on doing speech recognition with it. Just doing detection of phonological features.\nSpeaker A: So for example, this feature set called the sound patterns of English is just a bunch of binary valued features.\nSpeaker A: So I don't know if you can say, is this voicing or is this not voicing this sound or is this.\nSpeaker A: Did you find any more mistakes in their table?\nSpeaker A: I haven't gone through my table. Yesterday I brought Chuck the table and I was like, is the mapping from N to this phonological feature called coronal?\nSpeaker A: Should it be a one? Or should it be a coronal instead of not coronal as it was labeled in the paper?\nSpeaker A: So I haven't hunted down all the mistakes yet.\nSpeaker C: But as I was saying, people do get probabilities of these things. And we were just trying to remember how they do.\nSpeaker C: But people have used it for speech recognition. They have gotten probabilities. So they have some conversion from these distances to probabilities.\nSpeaker C: You have the paper, right, the Mississippi State paper. Yeah, if you're interested.\nSpeaker A: I can show you.\nSpeaker E: So in a thing you're doing, you have a vector of ones and zeros for each phone.\nSpeaker A: Is this the class project?\nSpeaker A: Yeah.\nSpeaker A: Is that what you're right? Right, right. So for every phone there is a vector of ones and zeros.\nSpeaker A: Of course, whether it exhibits a particular phonological feature or not.\nSpeaker E: And so when you do your, what is the task for the class project to come up with the phones or to come up with these vectors to see how close they match the phone?\nSpeaker A: To come up with a mapping from MFCCs or some features that to whether there is an existence of a particular phonological feature.\nSpeaker A: And yeah, basically it's to learn a mapping from the MFCCs to phonological features.\nSpeaker A: Is it the answer to your question? I think so.\nSpeaker E: I guess, I mean, I'm not sure what you get out of your system.\nSpeaker E: Do you get out a vector of these ones and zeros and then try to find the closest matching phoneme to that vector?\nSpeaker A: Oh, no, no. I'm not planning to do any phoneme mapping yet.\nSpeaker A: It's basically, it's really simple, basically, a detection of phonological features.\nSpeaker A: I see.\nSpeaker A: Because the, so King and Taylor did this with recurrent neural nets.\nSpeaker A: And their idea was to first find a mapping from MFCCs to phonological features.\nSpeaker A: And then later on, once you have these phonological features, then map that to folks.\nSpeaker A: So I'm sort of reproducing phase one.\nSpeaker E: So they had one recurrent net for each particular feature.\nSpeaker E: Did they compare that? I mean, what if you just did phone recognition and did the reverse look up?\nSpeaker E: So you recognize a phone, then whichever phone was recognized, you spit out its vector of ones and zeros.\nSpeaker C: The spectrary could do that. That's probably not what you're going to do in this class, Brian.\nSpeaker C: So have you had a chance to do this thing we talked about yet with the search and penalty?\nSpeaker C: No, actually, I was going different, that's a good question too, but I was going to ask about the changes to the data in comparing PLP and no capstrom for the SRI system.\nSpeaker E: Well, what I've been changes to data, I'm not sure.\nSpeaker C: So we talked on the phone about this, that there was still a difference of a few percent, and you told me that there was a difference in how the normalization was done.\nSpeaker C: And I was asking if you were going to redo it for PLP with a normalization done as it had been done for the mock-up stream.\nSpeaker E: Right, no, I haven't had a chance to do that.\nSpeaker E: So what I've been doing is trying to figure out, it just seems to me like there's a, well, it seems like there's a bug because the difference in performance is, it's not gigantic, but it's big enough that it seems wrong.\nSpeaker C: I agree, but I thought that the normalization difference was one of the...\nSpeaker E: Yeah, but I don't know how this is.\nSpeaker E: Yeah, I guess I don't think that the normalization difference is going to count for everything. So what I was working on is just going through and checking the headers of the wave files to see if maybe there was a certain type of compression or something that was done that my script wasn't catching.\nSpeaker E: For some subset of the training data, the features I was computing were junk, which would cause it to perform okay, but the models would be all messed up.\nSpeaker E: So I was going through and just double checking that kind of thing first to see if there was just some kind of obvious bug in the way that I was computing the features.\nSpeaker E: Looking at all the sampling rates, make sure all the sampling rates were what HK, what I was assuming they were...\nSpeaker E: Yeah, that makes sense to check out. So I was doing that first before I did these other things just to make sure there wasn't something.\nSpeaker C: Although really, a couple of 3% difference in order or rate could easily come from some difference in normalization, I would think.\nSpeaker E: Yeah, and I think, I'm trying to remember, but I think I recall that Andreas was saying that he was going to run sort of the reverse experiment, which is to try to emulate the normalization that we did, but with ML Cup Straw features.\nSpeaker E: Sort of, you know, back up from the system that he had. I thought he said he was going to have to look back through my email from him.\nSpeaker C: Yeah, he's probably off at the end of his meeting there.\nSpeaker C: Yeah, but yeah, I just think they should be roughly equivalent. I mean, again, the Cambridge folk found the PLP actually be a little better.\nSpeaker C: I mean, the other thing I wondered about was whether there was something just in the bootstrapping of a system which was based on, but maybe not.\nSpeaker E: So one thing that's a little bit, I was looking, I've been studying and going through the logs for the system that Andreas created.\nSpeaker E: And his way that the SRI system looks like it works is that it reads the way files directly and does all of the capture computation stuff on the fly.\nSpeaker E: Right. And so there's no place where these, where the capture files are stored in, where that I can go look at and compare to the PLP one.\nSpeaker E: So, whereas with our features, he's actually storing the capture.\nSpeaker E: Yes, going to read those in, but it looked like he had to give it, even though the capture was already computed, he has to give it a front end parameter file, which talks about the kind of.\nSpeaker E: Computation that his malfunction thing does. So, I don't know if that probably doesn't mess it up, it probably just ignores it if it determines that it's already in the right format or something, but the two processes that happen are a little different.\nSpeaker E: So, anyway, there's stuff there to start with.\nSpeaker C: Yeah. So, okay, let's go back to what you thought I was asking you.\nSpeaker E: Yeah, no, and I didn't have a chance to do that.\nSpeaker E: Same answer anyway. Yeah, I've been working with Jeremy on his project, and then I've been trying to track down this bug.\nSpeaker E: That's a seed front end features.\nSpeaker E: So, one thing that I did notice yesterday, I was studying the, the, the Rostecode.\nSpeaker E: And it looks like we don't have any way to control the frequency range that we use in our analysis. We basically, it looks to me like we do the FFT, and then we just take all the bins, and we use everything.\nSpeaker E: We don't have any set of parameters where we can say, you know, only process from, you know, 110 hertz to 3750.\nSpeaker C: At least I couldn't see any kind of. Yeah, I don't think it's in there. I think it's in the, the filters.\nSpeaker C: So, the FFT is on everything, but the filters, for instance, ignore the lowest bins and the highest bins.\nSpeaker C: What it does is it copies the filter bank, which is created by integrating over FFT bins.\nSpeaker C: When you get the mail, when you get the mail scale. Yeah, it's a, it, it actually copies the, the second filters over to the first.\nSpeaker C: So, the first filters are always, and you can, you can specify a different number of features, different number of filters, I think.\nSpeaker C: So, you can specify a different number of filters, and whatever you specify, the last ones are going to be ignored.\nSpeaker C: So, that, that's a way that you sort of change what the, what the bandwidth is.\nSpeaker C: You, you can't do it without, I think, changing the number of filters.\nSpeaker E: I saw something about, it looked like it was doing something like that, but I didn't quite understand it.\nSpeaker C: So, maybe, yeah. So, the idea is that the very lowest frequencies, and typically the various highest frequencies are kind of junk.\nSpeaker C: Uh-huh. And so, you just, for continuity, you just approximate them by, by the second to highest and second to lowest.\nSpeaker C: It's just a simple thing we put in.\nSpeaker C: And so, but that's a fixed thing, there's nothing to put in. Yeah, I think that's a fixed thing, but see, see my point, if you had, if you had 10 filters, then you would be throwing away a lot at the two ends.\nSpeaker C: And if you had, if you had 50 filters, you'd be throwing away hardly anything.\nSpeaker C: I don't remember there being an independent way of saying we're just going to make you, just from here to here.\nSpeaker E: I don't know, it's actually been a while since I've looked at it. Yeah, I went through the FICAL code and then looked at, it was calling the RUST Alive and things like that.\nSpeaker E: And I didn't, I couldn't see any place where that kind of thing was done, but I didn't quite understand everything that I saw.\nSpeaker C: Yeah, see, I don't know FICAL at all, but it calls RUST with some options.\nSpeaker C: But I think, I don't know, I guess for some particular database, you might find that you could tune that and tweak that to get that a little better, but I think that in general, it's not that critical.\nSpeaker C: I mean, there's, you can, you can throw away stuff below 100 hertz or so and it's just not going to affect phonetic classification at all.\nSpeaker E: The other thing I was thinking about was, is there a, I was wondering if there's maybe certain settings of the parameters when you compute PLP, which would basically cause it to output Malcapsterum.\nSpeaker E: So that in effect, what I could do is use our code to produce Malcapsterum and compare that directly to...\nSpeaker C: Well, it's not precisely, yeah. I mean, what you can do is, you can definitely change the filter bank from being a trapezoidal integration to a triangular one, which is what the typical Malcapsteral filter bank does.\nSpeaker C: So some people have claimed that they got some better performance doing that, so you certainly can do that easily.\nSpeaker C: But the fundamental difference, I mean, there's other small differences.\nSpeaker E: There's a few groups that happen, sorry.\nSpeaker C: Yeah, but, you know, as opposed to the log and the other case.\nSpeaker C: I mean, the fundamental difference that we've seen any kind of difference from before, which is actually an advantage for PLP, I think is that the, the smoothing at the end is autoregressive instead of being kept from kept from location.\nSpeaker C: So it's a little more noise robust.\nSpeaker C: And that's why when people start getting databases that had a little more noise in it, like broadcast news and so on, that's why Cambridge Switched to PLP, I think.\nSpeaker C: So that's a difference that I don't think we put any way to get around since it was an advantage.\nSpeaker C: But we did, we did hear this comment from people at some point that it, they got some better results with the triangular filters rather than the trapezoidal, so that is an option in Rasta.\nSpeaker C: And you can certainly play with that.\nSpeaker C: But I think you're probably doing the right thing to look for books first.\nSpeaker E: Yeah, just, it just seems like this kind of behavior could be caused by, you know, some of the training data being messed up.\nSpeaker E: Yeah, you're sort of getting most of the way there, but there's a, so I started going through and looking, one of the things that I did notice was that the log likelihoods coming out of the recognizer from the PLP data were much lower, much smaller than for the male kept still stuff.\nSpeaker E: And that the average amount of pruning that was happening was therefore a little bit higher for the PLP features.\nSpeaker E: So since he used the same exact pruning thresholds for both, I was wondering if it could be that we're getting more pruning.\nSpeaker C: Oh, he, he, he used the identical pruning thresholds even though the, the range of the, the likelihood, oh well, that's, that's a pretty good point right there.\nSpeaker C: Yeah.\nSpeaker C: So I would think that you might want to do something like, you know, look at a few points to see where you were starting to get significant search errors.\nSpeaker E: Right. Well, what I was going to do is I was going to take a couple of the utterances that he had run through, then run them through again, but modify the pruning threshold and see if it, you know, affects the score.\nSpeaker C: Yeah.\nSpeaker C: But I mean, you could, if, if that looks promising, you could, you know, run the overall test set with a, with a few different pruning thresholds for both.\nSpeaker C: Yeah. And presumably he's running at some pruning threshold that's, that's, you know, gets very few search errors, but is, is relatively, right?\nSpeaker E: I mean, yeah, generally, these things, you, you turn back pruning really far. So I, I didn't think it would be that big a deal because I was freaking well, you'd have it turned back so far that, you know, but you may be in the wrong range for the PLP features for some reason.\nSpeaker E: And the, the, the runtime of the recognizer on the PLP features is longer, which sort of implies that the networks are busier.\nSpeaker E: You know, there's more things it's considering, which goes along with the fact that the matches aren't as good.\nSpeaker E: So, you know, it could be that we're just pruning too much.\nSpeaker E: So yeah.\nSpeaker C: Yeah, maybe just be different kind of distributions and, and yes, so that's, that's another possible thing.\nSpeaker C: They, they should really shouldn't, there's no particular reason why they would be exactly behave exactly the same.\nSpeaker C: Right.\nSpeaker E: So, so there's lots of little differences. Yeah. Yeah.\nSpeaker C: Trying to track it down. Yeah. I guess this was a little bit off topic. I guess I guess I was, I was thinking in terms of that this is being a core item that once we, once we had it going, we would use for a number of the front end things also.\nSpeaker D: That's as far as my stuff goes. Yeah. Well, I tried this mean subtraction method due to Avandano.\nSpeaker D: I'm taking six seconds of speech. I'm using two second FFT analysis frames, step by a half second. So it's a quarter length step.\nSpeaker D: And I take that frame and four, the four, I take the, sorry, I take the current frame and the four past frames and the four future frames.\nSpeaker D: And that, that adds up to six seconds of speech. And I calculate the spectral mean of the log magnitude spectrum over that.\nSpeaker D: And I use that to normalize the current center frame by mean subtraction. And then I move to the next frame and I do it again.\nSpeaker D: Well, actually I calculate all the means first and then I do the subtraction. And I tried that with HDK, the Aurora setup of HDK training on clean TI digits.\nSpeaker D: And it helped in a phony reverberation case where I just use a simulated impulse response. The error rate went from something like 80, from something like 18% to 4%.\nSpeaker D: And on meeting recorded for my digits, my channel F, it went from 41% error to 8% error.\nSpeaker E: On the real data, not with artificial reverb.\nSpeaker D: Right. And that that was trained on clean speech only, which I'm guessing is the reason why the baseline was so bad.\nSpeaker C: And that's actually a little side point is I think that's the first results that we have of any sort on the far field data for recorded in meetings.\nSpeaker D: Well, I'm actually Adam, around the SRI recognizer on the near field on the far field, also he did one 2CM channel and one PDH channel.\nSpeaker C: Oh, I didn't recall that. What kind of numbers was he getting with that?\nSpeaker D: I'm not sure. I think it was about 5% error for the PCM channel.\nSpeaker C: Five. Yeah. So why were you getting 41?\nSpeaker D: I'm guessing it was the training data. Clean TI digits is pretty pristine training data. And if they trained the SRI system on this TV broadcast type stuff, I think it's a much brighter range of channels.\nSpeaker C: No, but wait a minute. I think he was getting 1% error for the near field.\nSpeaker C: Yeah, I think it was getting around 1% for the close mic. So it's still this kind of ratio. It's just, yeah, it's a lot more training data.\nSpeaker C: So it probably should be something we should try then is to see if at some point is to transform the data and then use it for the SRI system.\nSpeaker C: So you have a system which for one reason or another is relatively poor.\nSpeaker C: Yeah. And you have something like 41% error and then you transform it to 8th by doing this work.\nSpeaker C: So here's this other system which is a lot better. There's still this kind of ratio. It's something like 5% error with the distant mic and 1% close mic.\nSpeaker C: So the question is how close to that one can you get if you transform the data using that system?\nSpeaker D: Right. So I guess the SRI system is trained on a lot of broadcast news or switchboard data.\nSpeaker E: Yeah. So do you know which one it is?\nSpeaker E: It's trained on a lot of different things. It's trained on a lot of switchboard, call home, a bunch of different sources, some digits or some digits training in there.\nSpeaker D: One thing I'm wondering about is what this mean subtraction method will do if it's faced with additive noise.\nSpeaker D: Because I don't know what log magnitude spectral subtraction is going to do to additive noise.\nSpeaker C: Yeah. Well, it's not exactly the right thing. But you were already seeing that because there is added noise here.\nSpeaker D: Yeah, that's true. Yeah. It's a good point.\nSpeaker D: So... Okay. So it's a reasonable to expect. It would be helpful if we used it with the SRI system.\nSpeaker C: Yeah. I mean, as helpful. So that's the question. We're often asked this when we work with a system that isn't sort of industry standard grade.\nSpeaker C: And we see some reduction error using some clever method, then we'll work on a good system.\nSpeaker C: So the other one is pretty good system. I think 1% would aerate on digits strings. It's not stellar.\nSpeaker C: But given that this is real digits is supposed to sort of laboratory.\nSpeaker C: And it wasn't trained on this task.\nSpeaker C: And it wasn't trained on this task. Actually, 1% sort of a reasonable range. People would say, I can imagine getting that.\nSpeaker C: And so the 4% or 5% or something is quite poor. If you're doing a 16-digit credit card number, you basically get it wrong almost all the time.\nSpeaker C: So significant reduction error for that would be great.\nSpeaker C: And then... Yeah. So... Yeah. Cool. Sounds good.\nSpeaker C: Right. I actually have to run. So I don't think I can do the digits. But I guess I'll leave my microphone on.\nSpeaker E: Yeah.\nSpeaker C: Yeah. Yeah. Yeah. Yeah. Yeah. You're actually kind of disinterprets. Yeah? Yeah. Yeah. I know. You're quickly.\nSpeaker C: Sorry. I just have to run for another appointment. Okay. Did I? Yeah, I left it.\nSpeaker C: Okay. Okay. Okay. This is transcript L-110. 902-573-266-166-674208. 5169-477-7950. 5602-2577.\nSpeaker C: 6089-07264-5. 2192-89558. 3250-2934-4612-0885.\nSpeaker A: Transcript L-111-362-8332-207-16654268-9006-39334-754-7960-9903-7432-7542. 493-519-103-7452.\nSpeaker A: 519-788-504-940753-758-441-701-8553253-8.\nSpeaker E: Transcript L-112-07876216-4653-1258-477-727-770. 636-34797-776. 529-327-329.\nSpeaker E: 939-389-141-7604-400-867-458-066-359-777. 7355-3245-08.\nSpeaker D: I'm reading transcript L-113-65697-284-4087-5187-405702-425-037-681-779-217-909-0605-68859.\nSpeaker D: 522-55662-911-7206-7162-023-019-023.\nSpeaker G: Transcript L-115-064-732-271-08758-673-755-119-14.\nSpeaker G: 988-419-3102-410-368-219-628-969-186-565-0-3116-639-8-78-398-9.\nNone: AS6-0810.\n"}, {"meeting_id": "TS3008c", "summary": "This meeting talked about the conceptual design of the remote control product. Industrial Designer gave the first presentation, stating the personal preference for components designs. User Interface delivered the second. It led to a following discussion about basic function types as well as the size of their buttons. Next, Marketing's presentation identified the three most important aspects of remote design and expressed the focus on the trend on the elder market. As such, the group started to discuss how to meet the satisfaction of different age markets with the design. After that, Project Manager steered the meeting to further questions, including technical functions, material, color design, buttons functionality, and docking station design.", "dialogue": "Speaker F: Kay.\nSpeaker E: You're late.\nSpeaker F: Zoom.\nSpeaker B: Okay, all set.\nSpeaker B: Welcome to the conceptual design meeting.\nSpeaker B: The agenda. The opening, I'll again be the secretary in make minutes, take minutes, and it will be three presentations, just like the last meeting. So, who wants to start off?\nSpeaker B: Technical designer again. Okay.\nSpeaker B: Before we begin, I want to say I put minutes of the second meeting in the shared folder, but they're still not quite okay. It's still some technical difficulties, so the first part of the minutes are very hard to read because there are two documents that are layered over each other. Okay.\nSpeaker B: But from now on, I won't use my pen anymore.\nSpeaker B: It's just ordinary keyboard. It's better.\nSpeaker B: Keyboard work. I think it will be more easy for you to read minutes.\nSpeaker C: All right. Okay. When we talk about components design, it's really about material and really the stuff we build the remote controls of.\nSpeaker C: The remote control consists of components and the components of a re-riled control consists of properties and material. We have to choose these wisely and it could affect a kind of growth in buying the remote controls.\nSpeaker C: The components of remote control are of course the case.\nSpeaker C: The properties of the case are it has to be solid, a hard material like hard plastic with software for falling and it feels good in your hand.\nSpeaker C: The buttons has to be solid too and the material is software.\nSpeaker C: I've got an email from the possibilities of real reaction.\nSpeaker C: They are telling me that when we build a remote control of plastic or rubber, the buttons have to be rubber too.\nSpeaker C: Okay. When we use a rubber-doubled curved case, we must use rubber push buttons too. The rubber-doubled curved case is a three-dimensional curve in the design, which is necessary and we want to be trendy.\nSpeaker C: The energy source, I've got a lot of possibilities for that too.\nSpeaker C: The basic battery, which I prefer because of its non-dependent of here.\nSpeaker C: You have to have a kinetic energy also in this one, like in the watches, remote control can lie on the table for a day and then you push a button so you don't have to work with it all the time.\nSpeaker C: Solar shells are also a bit weird for remote controls.\nSpeaker C: Also the case material, I think plastic is the best with rubber because wood or titanium would also be a bit weird.\nSpeaker D: How about taneums? Probably trendy, I think?\nSpeaker E: That's true, I guess.\nSpeaker D: Maybe a little bit expensive.\nSpeaker C: I don't know how they don't tell anything about the cost of titanium.\nSpeaker C: The chip set and the board is all of the shelves.\nSpeaker C: Also the speaker in the remote control when we want to retrieve it.\nSpeaker C: The base station is also of the shelf. All the materials and components are just available in our factory.\nSpeaker C: I've told about the three first points.\nSpeaker C: The simple electronical chip is available with the lead transmitter.\nSpeaker C: It's all of the shelves and even the speaker and the wireless retrievers are all available in our company.\nSpeaker C: Another possibility I looked upon was the LCD displays.\nSpeaker C: Could be something special to our remote control.\nSpeaker C: It's possible, but it only costs a bit more.\nSpeaker C: Maybe it can be within the limits of 25 euros.\nSpeaker B: 12.\nSpeaker C: Actually, I've got an email with some examples.\nSpeaker C: This is where the most trendy is one.\nSpeaker C: You see a coffers which can be...\nSpeaker B: Well, are those the tooth brushes?\nSpeaker C: I don't know.\nSpeaker B: It resembles the design.\nSpeaker C: Maybe we can bring a couple of...\nSpeaker C: A couple of types of...\nSpeaker C: A whole set of different remote controls.\nSpeaker C: Maybe we can bring a whole line with a huge variety of houses.\nSpeaker C: Like maybe radios and television also in the same style.\nSpeaker E: That'll be for the future, I guess.\nSpeaker C: Yes, because we have to bring the logo and all the stuff back into it.\nSpeaker E: Definitely.\nSpeaker D: Thank you, sir.\nNone: I'll look it up.\nSpeaker E: That's okay.\nNone: I'll go to the next slide.\nSpeaker D: I still don't have any information about user requirements.\nSpeaker D: I was thinking about just the basic functions.\nSpeaker B: We decided on that in the last meeting.\nSpeaker D: But then when there are new user requirements.\nSpeaker B: I didn't receive any new requirements.\nSpeaker B: Nothing?\nSpeaker B: No, but we decided to use only basic functions.\nSpeaker D: I have here a couple of basic functions I could think of.\nSpeaker D: I don't know if there are maybe a little bit more.\nSpeaker B: Maybe we can think of that later.\nSpeaker B: These are the ones you already summed up.\nSpeaker D: I pointed them out here.\nSpeaker D: Just to make that a little bit easier.\nSpeaker D: Another function is, of course, we already discussed it on the side.\nSpeaker D: I don't know what costs of it.\nSpeaker D: I was also looking for what you said for I got a name about LCD in front of the remote control.\nSpeaker D: I don't know if that's a good idea or maybe it's a little bit too much for 12.5.\nSpeaker D: If we got already something like a base.\nSpeaker B: Bit redundant also maybe.\nSpeaker B: I don't know what kind of information.\nSpeaker D: I was just thinking about it and I got a problem.\nSpeaker C: Maybe we can bring a tail text to the remote control.\nSpeaker A: They've got to have like a little bit more.\nSpeaker D: Make RC a little bit too big.\nSpeaker D: Exactly.\nSpeaker D: Well, the functions are not more to discuss.\nSpeaker D: It's just the base things we already discussed at the No PCR.\nSpeaker D: That's very easy.\nSpeaker B: But you do mention the next and previous button.\nSpeaker D: Well, that's next channel.\nSpeaker D: I got an email with a remote control with a base.\nSpeaker D: It's just an idea.\nSpeaker D: I think of the button sizes and I'm not sure if they have to be big or just small.\nSpeaker D: I think it depends on the function.\nSpeaker D: Well, I'm the expert for user friendly but not for trendiness.\nSpeaker E: If you save tiny buttons aren't user friendly then we wouldn't implement that.\nSpeaker D: Well, okay.\nSpeaker D: It's a bit of a point.\nSpeaker D: Yeah, I've nothing to do.\nSpeaker B: When we only use base functions we have the possibility to make the buttons larger.\nSpeaker D: It's a little bit larger.\nSpeaker B: I think we already agreed upon the fact that the skip buttons and volume buttons, those two, they have to be large.\nSpeaker B: I mean the two basic buttons, skip channels.\nSpeaker B: I don't know why but I think that's the trend.\nSpeaker B: It puts an extra accent on the simplicity of our remote to make these two most basic functions extra big.\nSpeaker E: I also probably use the most used buttons on the unit.\nSpeaker E: It's from your research.\nSpeaker D: That was all personal preference I didn't have.\nSpeaker D: Anytime left.\nSpeaker E: You don't care.\nSpeaker E: Yeah.\nSpeaker E: Go away.\nSpeaker D: Come on.\nSpeaker F: Yeah, click on it.\nSpeaker E: Oh, great.\nSpeaker E: I've done some research again about trends on the internet.\nSpeaker E: I've done some investigation and I got some information from fashion watchers from Paris and Milan.\nSpeaker E: Some finding the most important thing is fancy look and feel of the most control.\nSpeaker E: We were going to imply that.\nSpeaker E: That's nice.\nSpeaker E: The second important thing is innovative technology in the RC.\nSpeaker E: Our market really likes that and the third point there in this order of importance, the third point is a high ease of use.\nSpeaker E: For the idea, I've put some trends for the markets for elderly people, dark colors, simple recognizable shapes.\nSpeaker E: So we probably won't do that.\nSpeaker E: The younger markets like the themes of this year are surprisingly fruits and vegetables and spongy material.\nSpeaker E: I found this image which is, well, it symbolizes the idea of fruits and vegetables.\nSpeaker E: I don't see the spongy part in it but with a little bit of fancy.\nSpeaker B: Then we have to do something with spongebob.\nSpeaker E: Exactly.\nSpeaker E: I got some ideas.\nSpeaker E: Well, pictures isn't really a good word but some symbols of fruits or vegetables may be.\nSpeaker E: Catch your colors.\nSpeaker E: Fruit is yellow, green, red, whatever.\nSpeaker E: So remote controls in catch colors.\nSpeaker E: No, we don't want dark colors.\nSpeaker E: No, I just put them there for general.\nSpeaker E: For general idea.\nSpeaker E: The darkings, I think, spongey material is very irritating for the remote control itself.\nSpeaker E: But to implement some spongey thing, maybe we can do it in the darking station at the bottom of the darking station or whatever.\nSpeaker E: And we could bring one line with the dark color to...\nSpeaker B: For diversity.\nSpeaker E: Also, a bit for elderly people who are a little bit crazy and wonder.\nSpeaker E: Maybe want a little younger design but still the dark color.\nSpeaker E: I mean, it reaches a different market but it doesn't cost really much effort to bring like a black RC on the market or whatever.\nSpeaker D: How do we use fruit and vegetables in Christ's sake with remote control?\nSpeaker B: I think that our design already resembles some piece of fruit like a pear.\nSpeaker E: There's always empty space of course on remote control.\nSpeaker E: I mean, I think this part of the RC, well...\nSpeaker E: I don't think you have to do it the upper part or whatever.\nSpeaker E: It's not used with buttons, I guess.\nSpeaker E: So you can put some fruit in.\nSpeaker B: It doesn't have to remind you explicitly of a piece of fruit but just the round curves.\nSpeaker B: I think this already resembles something like a pear to me.\nSpeaker E: Yeah, exactly.\nSpeaker E: If you make a little bit of greenish.\nSpeaker B: Do you get the idea? The fruit is kind of round.\nSpeaker E: I mean, we could use one of these for the... What is it?\nSpeaker A: I don't know.\nSpeaker A: Great.\nSpeaker A: Yeah.\nSpeaker D: I don't even need to create an artist or something like that to make it work.\nSpeaker D: Sure.\nSpeaker D: Vegetable or fruit?\nSpeaker D: We have a very big team.\nSpeaker E: We can do multiple things.\nSpeaker E: This is then the pear.\nSpeaker E: I don't know the English word.\nSpeaker E: It's pear, yes.\nSpeaker E: But maybe we don't have bananas.\nSpeaker E: Not easy for remote control, but we can't make all 10 designs.\nSpeaker C: We have to make one design, I think.\nSpeaker B: But I think it's already two or three.\nSpeaker B: What we were up to.\nSpeaker B: Yeah.\nSpeaker B: It doesn't have to resemble what I already said a specific piece of fruit.\nSpeaker B: No, sure.\nSpeaker B: But just like a fruity thing going on.\nSpeaker B: But that's great.\nSpeaker E: I don't know what I was saying.\nSpeaker B: But I do like to catch the colors.\nSpeaker B: Yeah, I do like to do the idea of making a catchy color design.\nSpeaker B: Because I think a dark color would be nice too.\nSpeaker A: Yeah.\nSpeaker A: Picked is a fruit.\nSpeaker A: Maybe it's too much.\nSpeaker E: Yeah.\nSpeaker D: We have to.\nSpeaker D: Not very simple.\nSpeaker D: There have to be the firm colors.\nSpeaker D: Our own color.\nSpeaker B: Okay.\nSpeaker B: We've got our...\nSpeaker C: Yellow.\nSpeaker C: Yellow.\nSpeaker C: A real reaction.\nSpeaker C: She can put a logo on top of it.\nSpeaker B: Yeah, I don't think our company colors are this fashionable.\nSpeaker E: Maybe we can...\nSpeaker E: If we've got a docking station over here, I can draw with this thing, but I'll try.\nSpeaker E: If this is our docking station, we can make our logo over here.\nSpeaker E: It doesn't work.\nSpeaker E: And then...\nSpeaker E: Yeah, in the button then.\nSpeaker C: With the strawberry on top.\nSpeaker E: Yeah, in the bottom of the remote.\nSpeaker E: The button over here.\nSpeaker E: Whatever.\nSpeaker E: I don't know.\nSpeaker E: Okay.\nSpeaker E: On the front, of course.\nSpeaker E: Because else you can find it.\nSpeaker E: So that will my this a little bit.\nSpeaker E: I'll close them down.\nSpeaker E: Go away.\nSpeaker B: Okay, can you open the conceptual design?\nSpeaker B: On the presentation.\nSpeaker B: Yes.\nSpeaker B: See what was on the agenda?\nSpeaker E: Lazy.\nSpeaker B: The agenda.\nSpeaker B: The decisions on the remote.\nSpeaker B: Because maybe we can start with the technical functions, but I don't think...\nSpeaker B: Do we want to use an LCD display, for example?\nSpeaker E: Well, it's nice.\nSpeaker E: Of course.\nSpeaker E: But I don't know what to display on it.\nSpeaker C: Maybe we can make a TV guide on it for the channel you're on.\nSpeaker E: Yeah, but it should be like this big.\nSpeaker E: I don't think we should do it.\nSpeaker C: The TV channel with four programs, you can zap to them with the page down button.\nSpeaker E: Yeah, sure, but it has to show an entire title of a program, or at least a quite large part of it.\nSpeaker E: And then you get a very large LCD.\nSpeaker C: No, when you're mobile phone, you can you can read text also.\nSpeaker C: So why not on your remote?\nSpeaker B: Never.\nSpeaker B: I think it's a bit redundant.\nSpeaker B: And it's also not... I don't even think it looks like sexy or something.\nSpeaker D: What would you display on it?\nSpeaker C: A program information or a guide on TV.\nSpeaker D: A lot of new TVs.\nSpeaker C: Yes, also on the internet.\nSpeaker E: A lot of TVs in each show when you...\nSpeaker E: Which already works in your TV.\nSpeaker B: You're not going to watch your remote control.\nSpeaker C: Yes, but you also want to know what's next.\nSpeaker E: But then we also need to bring out a line of TVs, which we're planning to do, or whatever.\nSpeaker E: Because the TV has to send information back to the RC.\nSpeaker E: And I don't know if that's possible.\nSpeaker C: That's really possible.\nSpeaker E: Yes, yes, of course it's possible.\nSpeaker E: But you've got to implement it in the TVs.\nSpeaker E: And I don't think everyone's going to buy a real reaction TV within a month after the release of our remote control.\nSpeaker B: I really understand you want to make your job more exciting than anything else to be in it.\nSpeaker B: But I really don't think it's a good... because it also doesn't stroke.\nSpeaker B: We wanted... when we talk about the materials, it's a good idea to use these plastic materials with the soft rubber stuff on it.\nSpeaker B: It was our idea to give it a more sturdy look.\nSpeaker B: And you can throw with it.\nSpeaker B: But I don't think LCD display fits in that image.\nSpeaker B: It's more vulnerable.\nSpeaker B: It's not very solid.\nSpeaker E: You can make it.\nSpeaker B: I don't think it's coherent with the design we're after.\nSpeaker E: No.\nSpeaker E: I don't think so.\nSpeaker B: In my opinion, we can vote for it.\nSpeaker B: The LCD display.\nSpeaker B: I don't want to.\nSpeaker B: And he doesn't, so it's up to him.\nSpeaker B: If you...\nSpeaker B: I've read somewhere that I've got some kind of fatal...\nSpeaker B:...right?\nSpeaker B: So I can also say...\nSpeaker B: We can fold your way.\nSpeaker B: Yeah, you can do...\nSpeaker D: So what do you think this is?\nSpeaker B: We're not even sure what information we want to display.\nSpeaker B: No, that's right.\nSpeaker D: And I also have to think about new functions, maybe buttons or something like that, to control that LCD or something?\nSpeaker D: Yeah, I guess.\nSpeaker C: You can use buttons that are already on the remote control.\nSpeaker C: For double function.\nSpeaker D: When I go to the second channel, what does it show me?\nSpeaker C: The title and information about the program.\nSpeaker C: About that program.\nSpeaker C: But what he said was right about the televisions.\nSpeaker C: They have to be customized to that.\nSpeaker B: No, that's not going to work.\nSpeaker C: But maybe in future it will be a giant hit and when you are the first.\nSpeaker C: You have the biggest...\nSpeaker B: Like the bigger universal remotes, the LCD displays.\nSpeaker B: But then it's very functional to indicate what device you are controlling.\nSpeaker B: That's what I've seen.\nSpeaker C: You can put a little LCD display on it with lots of information.\nSpeaker B: But it just doesn't match with our whole basic concept.\nSpeaker C: But when you put a transparent plastic screen on top of it, it isn't vulnerable.\nSpeaker B: No, that's maybe not the most important.\nSpeaker B: Is it fashion?\nSpeaker B: I don't think so.\nSpeaker C: When you put maybe a color LCD screen on it, it's very special and very trendy.\nSpeaker C: I don't know.\nSpeaker C: A remotes control.\nSpeaker D: That's a rocket if it's trendy.\nSpeaker B: What do you have?\nSpeaker B: You have to do it after the trendiness of LCD display.\nSpeaker E: I think it's always pretty trendy to be honest.\nSpeaker E: But I don't know if I'm coming back to the cast again.\nSpeaker E: But I think we've got to build a pretty cheap design to stay within our limits.\nSpeaker E: And I think especially color LCD, which is pretty trendy.\nSpeaker E: I think it will be too expensive.\nSpeaker C: But I've got the email with the possibility.\nSpeaker C: And LCD was a possibility for the control.\nSpeaker C: So why don't we use it?\nSpeaker C: We're going to have it.\nSpeaker E: We're going to have it.\nSpeaker E: Also for monogramme LCD or colored LCD.\nSpeaker D: If you want to be trendy, you have to be colored.\nSpeaker E: Yeah, really.\nSpeaker E: If you have black and white or something or gray, in 2004, you can't put something on the market, which is monogramme.\nSpeaker E: Really?\nSpeaker C: No, but it doesn't say anything about the color.\nSpeaker C: But I also got the possibility to put a scroll button on it.\nSpeaker B: I didn't really know the whole idea of LCD display.\nSpeaker B: I'm sorry.\nSpeaker B: You can't convince me.\nSpeaker B: I don't know.\nSpeaker B: I'll just read you guys.\nSpeaker B: I don't really feel it.\nSpeaker B: We already were...\nSpeaker B: It's too much maybe.\nSpeaker B: We actually have the base station gadget.\nSpeaker B: And it has to be a simple design which...\nSpeaker E: Yes, but we've got a point about a little force.\nSpeaker E: And I think not a whole package of specialty.\nSpeaker B: And really, I don't see how the LCD display is going to add anything on a design level.\nSpeaker B: I think it's...\nSpeaker B: No, wait.\nSpeaker B: Slicker to have no LCD.\nSpeaker B: But at simplicity, you have two big buttons and you can do whatever you want.\nSpeaker B: Yes, but with these two buttons, you don't need an LCD.\nSpeaker B: That remote controls are already on the market.\nSpeaker B: It doesn't fit in our philosophy behind the whole remote.\nSpeaker C: Yes, but when you want to have something special.\nSpeaker B: We already have the docking station which is...\nSpeaker C: Yes, but we have a picture of it from another company.\nSpeaker B: It has to be the fellas.\nSpeaker B: No, but it does...\nSpeaker B: It does.\nSpeaker A: It does.\nSpeaker A: It does.\nSpeaker A: It does.\nSpeaker C: It does.\nSpeaker C: It's already made.\nSpeaker A: The remote controls.\nSpeaker A: We're going to develop our own docking station.\nSpeaker A: Is that something?\nSpeaker E: Yes.\nSpeaker E: The big two are the prototype?\nSpeaker E: Yeah, I don't know.\nSpeaker E: Exactly.\nSpeaker E: I've never seen it in the store.\nSpeaker B: But we really have to cut this off.\nSpeaker B: I guess the idea you really like it, you know, the LCD thing, but I think it's not a good idea.\nSpeaker B: We have already mentioned all arguments.\nSpeaker B: Do you guys think...\nSpeaker B: I think it's too much.\nSpeaker E: I think it's a little too much, yeah.\nSpeaker E: It's overdone.\nSpeaker B: Okay, we...\nSpeaker B: Let's skip the LCD.\nSpeaker B: Okay.\nSpeaker B: I'm sorry, maybe you can do something for your own place or make...\nSpeaker B: No, no, no.\nSpeaker B: Make it happen in your basement or something.\nSpeaker A: I will rule the world with it.\nSpeaker B: Okay, but for the technical part, the material, I think it was a good idea to use the plastic and the rubber.\nSpeaker C: Yes, maybe a bit of cushioning.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker B: Exactly.\nSpeaker B: This is what the...\nSpeaker B: The sponge, yeah.\nSpeaker B: Yeah, but it feels so ready.\nSpeaker B: We're after, you know, to give it the soft touch in your hands and also the...\nSpeaker B: No, like, yeah, this is the airbag kind of thing.\nSpeaker B: Yes, you're just gonna drop it.\nSpeaker B: Throw it at your wheel, brother's head.\nSpeaker E: If you drop it, the airbag comes out.\nSpeaker B: No, no, no, no.\nSpeaker B: Maybe...\nSpeaker C: But then we have to look that it will not be too childish to see.\nSpeaker B: Yeah, yeah.\nSpeaker B: Okay, that's a good point.\nSpeaker B: And that's why I like the dark color bit, you know?\nSpeaker B: Yeah.\nSpeaker B: Because maybe the design...\nSpeaker B: Maybe it's a bit of...\nSpeaker D: But not black, I think.\nSpeaker B: It's a bit 90s, maybe, what we're up to right from to this point.\nSpeaker D: If it's fruit and vegetables, it has to be colorful.\nSpeaker B: Yeah, that's true.\nSpeaker B: But it has to be a little bit solid.\nSpeaker B: It mustn't be too overwhelming when you put it in.\nSpeaker D: Can we combine it or something with yellow and black?\nSpeaker B: Yeah, maybe so.\nSpeaker D: Make it a B?\nSpeaker E: What?\nSpeaker E: B.\nSpeaker E: Oh, a B.\nSpeaker B: No, I don't like to get on that combination.\nSpeaker B: But it is our company colors.\nSpeaker B: Yeah, it's real good colors.\nSpeaker C: Yeah.\nSpeaker C: We have to use yellow.\nSpeaker C: Hmm.\nSpeaker E: Hmm.\nSpeaker B: I don't like yellow.\nSpeaker B: Well, we can ask...\nSpeaker E: But that's not really pretty.\nSpeaker E: Really nicely out there.\nSpeaker E: We can put the logo on our base station.\nSpeaker E: Yeah.\nSpeaker E: And maybe very, very tiny on the remote control itself.\nSpeaker E: But...\nSpeaker B: Okay, but what are other technical things we have to discuss?\nSpeaker C: Front of the...\nSpeaker C: We can have different fronts of the...\nSpeaker C: Should we do that?\nSpeaker B: I don't think you should do that.\nSpeaker B: Maybe just bring it out in different colors, but not off the key.\nSpeaker B: I get that.\nSpeaker B: Fronts afterwards.\nSpeaker B: It's also too much.\nSpeaker B: People don't want to spend more money on their remote control.\nSpeaker D: Yeah, you can...\nSpeaker D: You can design...\nSpeaker D: Use the customer with color you want.\nSpeaker D: Yeah, definitely.\nSpeaker D: Yeah.\nSpeaker E: Just bring more design to your market.\nSpeaker B: For color.\nSpeaker B: Without...\nSpeaker B: Okay.\nSpeaker B: So, are we through the technical part then?\nSpeaker B: Yes.\nSpeaker B: Okay.\nSpeaker B: So we agreed upon...\nSpeaker B: Well, not...\nSpeaker B: Oh, yeah.\nSpeaker C: This is our real young and dynamic style.\nSpeaker B: The materials you mentioned in your personal preferences were quite okay.\nSpeaker B: It's not only the last point.\nSpeaker B: It's not titanium.\nSpeaker B: No titanium is not an out of question.\nSpeaker D: It's just like this titanium.\nSpeaker C: Yes.\nSpeaker B: But when we use both the plastic and soft things and titanium as well.\nSpeaker B: Sure.\nSpeaker C: Makes it in a homogenous design.\nSpeaker C: But then you can throw with it.\nSpeaker C: It will make a huge noise or break other stuff.\nSpeaker C: When you throw with it, you will break other stuff with your remote control.\nSpeaker C: No.\nSpeaker C: Titanium is a bit...\nSpeaker C: No, but you should be harder.\nSpeaker C: But also on the colors.\nSpeaker B: Okay, think of the possibilities to make it completely titanium.\nSpeaker B: What would it be more trendy, more chic?\nSpeaker E: I think that...\nSpeaker D: I think that the titanium nowadays is way more often used than plastic.\nSpeaker C: Yes, but the titanium remote control.\nSpeaker C: When you're watching TV or your hands are a little bit sweaty.\nSpeaker E: On the other hand, if you want to make frozen stuff with it.\nSpeaker B: I really like the idea of the plastic and the big kind of thing.\nSpeaker B: Because then it fits in our philosophy to make it sturdy and simple.\nSpeaker B: Like sports and gaming.\nSpeaker B: When you make titanium, it becomes more like some kind of gadget you actually don't need.\nSpeaker B: And when it's big and plastic, it's like some fun stuff you can always have around.\nSpeaker B: It's always fun to have something big and plastic around.\nSpeaker D: Yes, it's a empty tree player of Nike.\nSpeaker D: So isn't that titanium with a little bit rubber?\nSpeaker C: Yes, it's...\nSpeaker C: But it is plastic.\nSpeaker D: Is it plastic?\nSpeaker D: It's titanium looking.\nSpeaker C: Yes, what we can do that only.\nSpeaker D: Yes, yes, yes.\nSpeaker D: I don't know if you know the empty tree player of Nike.\nSpeaker D: Okay, that's very...\nSpeaker D: It's rubber.\nSpeaker C: We can make this a style too.\nSpeaker C: Maybe this is difficult.\nSpeaker E: Because that's different material.\nSpeaker E: And then you're going to have two material lines.\nSpeaker C: So we can make it from the same kind of plastic.\nSpeaker E: If it's just a color, which you change, then I guess it's nice to have one of these.\nSpeaker B: I do like the idea of maybe a titanium type of body and then with plastic coloration.\nSpeaker B: You don't like the soft stuff, but I don't know if it's possible.\nSpeaker C: I don't have the information.\nSpeaker C: I didn't cut it.\nSpeaker B: But you can make the plastic that titanium look.\nSpeaker B: Yes, true.\nSpeaker B: But make it just like shine.\nSpeaker B: Yeah, true.\nSpeaker D: Maybe that's a good idea.\nSpeaker D: But if you want to lay it longer, then two weeks from the end, you can maybe...\nSpeaker B: Maybe we should.\nSpeaker B: I don't know if we should talk about our months.\nSpeaker D: I don't know what time?\nSpeaker C: 40 minutes.\nSpeaker C: In a lot of other product categories, like even in bags industry, they began with typical leather bags, but then they became stylish with all sort of colors.\nSpeaker C: Kind of fun of fronts.\nSpeaker C: Like we can use on the telephone.\nSpeaker C: Like Eastpac began a revolution with all these kind of bags and colors.\nSpeaker C: And different colors.\nSpeaker C: Yes, and styles.\nSpeaker C: They have also kind of roses on it.\nSpeaker B: Oh, yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah, it's a possibility.\nSpeaker B: But let's think about the base.\nSpeaker C: Then we can always use the same design for greater resemblance, but with new colors, new prints, on this.\nSpeaker B: Yeah.\nSpeaker B: But our basic idea, I mean, you're probably going to have like two type of materials, like the plastic enclosure and then the pets that surround it and in light.\nSpeaker B: We have to incorporate the lights too.\nSpeaker B: Are we going to give it a two tone color look like the plastic mold is in one color and the caution pets around it are in another color.\nSpeaker B: Is that the idea?\nSpeaker B: Is that a good idea?\nSpeaker E: How do you mean the rubber?\nSpeaker B: How many colors are we going to have?\nSpeaker B: We're only five minutes left by the way.\nSpeaker B: Yes.\nSpeaker B: How many colors are we going to give it?\nSpeaker B: Like two tone color?\nSpeaker B: Yeah.\nSpeaker B: There are three components.\nSpeaker C: Three components type.\nSpeaker C: You have the buttons.\nSpeaker C: The case.\nSpeaker B: I think maybe the case of rubber should be in one color and then the rubber of the buttons and the cushions as well should be in another color.\nSpeaker D: Or you just make one color maybe with a kind of like a big wave or something like.\nSpeaker D: Okay, but not more than another color.\nSpeaker B: Not more than two colors.\nSpeaker D: No, definitely not.\nSpeaker E: Maybe we should talk about it.\nSpeaker C: Or when you use the buttons as black, you can use two colors as well.\nSpeaker C: Okay.\nSpeaker B: Well, do we have to think of some other important things?\nSpeaker B: Oh, yeah.\nSpeaker B: The functionalities or the devices?\nSpeaker D: I was thinking about the, do we still want the joystick idea?\nSpeaker E: No.\nSpeaker E: No, I think that's too vulnerable.\nSpeaker B: I think this is okay.\nSpeaker B: So we have basic, we have the numbers, we have the power button, we have a data text.\nSpeaker B: And maybe one to access a menu or something.\nSpeaker E: Yeah, but that's how I was thinking that's got to be on the television.\nSpeaker B: Yeah, but I think I really need a menu button.\nSpeaker B: It's just the only button.\nSpeaker B: What kind of menu?\nSpeaker B: Yeah, I...\nSpeaker D: Is it a different from every television?\nSpeaker B: No, I think most TVs have an...\nSpeaker B: Yeah, it's a different...\nSpeaker B: A menu nowadays to access the screen settings.\nSpeaker E: I think it's okay to add a menu button for...\nSpeaker E: But that covers all the...\nSpeaker B: All the other settings, it covers everything then.\nSpeaker C: But then you have to put up and down and left and right.\nSpeaker E: You can use the two, eight, four and six.\nSpeaker B: And you can also use the normal skip buttons for that.\nSpeaker B: In that way, we have only the numbers, the power button, skip and volume.\nSpeaker B: And then a remote.\nSpeaker B: And then a text.\nSpeaker B: And a menu.\nSpeaker B: And then a menu.\nSpeaker B: That's it.\nSpeaker B: So we need...\nSpeaker B: Yeah.\nSpeaker A: Okay.\nSpeaker A: Another style of some stuff.\nSpeaker B: About the design of the docking station, something important about it.\nSpeaker B: We should remind us of the remote itself, I guess.\nSpeaker B: Yeah, definitely.\nSpeaker E: In one corner.\nSpeaker E: Are we gonna do something with the spongy thing there?\nSpeaker B: I think the spongy thing already comes in the cushions, bats and things on the side.\nSpeaker B: We will make it spongy.\nSpeaker B: And the fruity thing is just the shape should be fruity.\nSpeaker B: I think this is kind of fruity.\nSpeaker B: You know, just round shapes with...\nSpeaker E: Yeah, it's kind of fruity.\nSpeaker E: Catchy colors.\nSpeaker B: Yeah, but we really have to think...\nSpeaker B: I think colors is very important because it has to be fleshy.\nSpeaker B: But it doesn't have to be annoying that when you have...\nSpeaker B: You know, some things just over the top.\nSpeaker B: And when you have the table for more than two weeks, it just gets annoying because it's so big and fleshy.\nSpeaker B: It has to be some level of subtlety.\nSpeaker B: But we still have to think of how we managed to get to that.\nSpeaker B: Okay.\nSpeaker B: I guess so.\nSpeaker B: But I think also we just show we have to do something with color.\nSpeaker B: But also I think we have to keep the dark color thing in mind.\nSpeaker B: I think that adds to the too much color maybe.\nSpeaker B: Too much color.\nSpeaker D: When you got it in the living room, it's much maybe...\nSpeaker D: But aren't it designed experts?\nSpeaker B: It has to be.\nSpeaker B: Okay.\nSpeaker B: Well, I think the meeting will be over within a minute.\nSpeaker B: So we will wrap up.\nSpeaker B: Or is there anything we'd like to discuss?\nSpeaker E: I guess not.\nSpeaker B: Do you think that?\nSpeaker B: No.\nSpeaker B: Okay.\nSpeaker B: Well, you will read the minutes in the...\nSpeaker B: You can find them in there.\nSpeaker B: Sure.\nSpeaker B: Probably.\nSpeaker B: Yeah.\nSpeaker B: Oh no, for sure.\nSpeaker B: Because I will now dive them out.\nSpeaker C: What are we going to do now?\nSpeaker E: We'll see in your email.\nSpeaker D: I don't know.\nSpeaker D: I don't know.\nSpeaker D: I hope so.\nSpeaker D: And the other thing is that you don't have kind of prototype or something like that.\nSpeaker D: You see kind of prototype you can...\nSpeaker C: I will make one more in the next 20 minutes.\nSpeaker C: Yeah.\nSpeaker E: Construct one.\nSpeaker E: Yeah.\nSpeaker E: With toilet paper or...\nSpeaker F: Okay.\nSpeaker B: Alright.\nSpeaker E: Shall we get back to work?\nSpeaker B: Yeah.\nSpeaker B: Great.\nSpeaker B: I was waiting for you.\nSpeaker B: I was going to...\nSpeaker C: Back to the pen.\nSpeaker C: You lazy.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: Yeah.\nNone: Oh.\nNone: Oh yeah.\nNone: What would you do?\nNone: I think it's a smart time.\nNone: Oh, okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nNone: Okay.\nNone: We have a lot of commissing.\nNone: I've got half the time.\nNone: Back to you.\nNone: Yes.\nNone: I'm going to do a bokeh.\nNone: Yeah.\nNone: Oh.\nNone: Okay.\nNone: Have fun.\nNone: Yeah.\nNone: Okay.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Shall we?\nNone: Oh yeah.\nNone: Oh yeah.\nNone: That's what I'm going to do.\nNone: Oh, come on.\nNone: Back up.\nNone: No.\nNone: Come on on.\nNone: No.\nNone: I'm going to make a javatar.\nNone: Yeah.\nNone: Oh.\nNone: Oh.\nNone: Oh.\nNone: I'm going to do this.\nNone: Yeah.\nNone: Okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nNone: Okay.\n"}, {"meeting_id": "TS3005d", "summary": "This is the fourth meeting. Project Manager began a detailed design meeting with Industrial Designer and User Interface's overall design presentation. Industrial Designer and User Interface introduced the remote control was green with a light-blue screen of LCD, scroll and microphone function and the RR logo. Marketing put up a questionnaire so group members discussed and evaluated remote design. The final average score was two. Then the group calculated production cost. After calculation, the production cost was eighteen fifty and surpassed the company's budget. So the design was changed and functions such as scroll and microphone were lost to satisfy the budget. The group evaluated the project and gave some suggestions about the whole process.", "dialogue": "Speaker A: Good.\nSpeaker A: Me.\nNone: Me.\nSpeaker A: Me.\nNone: Oh.\nSpeaker F: So, well, welcome everyone.\nSpeaker F: As you may have noticed, I created separate folders because it was, they've had a little busy in our shared project documents folder.\nSpeaker F: I don't know if everybody put their own documents in the right folder, which is from our detailed design meeting.\nSpeaker F: That's new one.\nSpeaker F: We didn't make any yet.\nSpeaker C: We should say that one.\nSpeaker B: Oh, we move.\nSpeaker B: We just do that.\nSpeaker C: Yeah, safe in the folder.\nSpeaker F: Oh, no, this is just one big document, so you can leave that whatever it is.\nSpeaker F: Oh.\nSpeaker F: And we have an evaluation.\nSpeaker F: Okay.\nSpeaker C: Agenda.\nSpeaker F: Well, not many documents this time.\nSpeaker F: Oh, yes.\nSpeaker F: I have it open myself, I guess.\nSpeaker F: Well, the detailed design meeting, we're finally getting somewhere, hopefully.\nSpeaker F: What are we going to do?\nSpeaker F: I've opened it already.\nSpeaker F: I'm still going to take some minutes.\nSpeaker F: And if I'm right, you two are going to give a prototype presentation.\nSpeaker F: No, we could, aren't you?\nSpeaker F: And did you do something with evaluation criteria?\nSpeaker F: Yeah.\nSpeaker F: Good.\nSpeaker F: We have a correct agenda.\nSpeaker F: And then we have to look at something which is less nice, the finance aspect, where we can afford what we have designed.\nSpeaker F: And if we can, we can commence the final part, which is the production or project evaluation, how did we work together and what are the results and how happy are we with those.\nSpeaker F: Okay.\nSpeaker F: Well, finance, uh, later.\nSpeaker F: Now I have to give a word to you too.\nSpeaker A: Okay.\nSpeaker C: Get up, start up.\nSpeaker C: Hey, it is.\nSpeaker B: We made a prototype.\nSpeaker B: We first start with the overall, uh, this is about the total remote control.\nSpeaker B: We made it green.\nSpeaker C: Just an example color.\nSpeaker C: So, uh, there's one of the colors we would like to see in our prototype.\nSpeaker B: And the screen light blue.\nSpeaker B: So, uh, this is the scroll button and the microphone on the, on the side here on there.\nSpeaker B: And the R and R logo just says R and R now.\nSpeaker C: Okay.\nSpeaker B: Any questions so far?\nSpeaker C: Big microphone.\nSpeaker C: Big, yeah, just, uh, just an idea about how the, the, the, the, the later's going to be.\nSpeaker C: Uh, well, it's an idea in that.\nSpeaker B: Oh, you, perhaps you should make it a bit big so people know it's there and uh.\nSpeaker A: Do not forget it.\nSpeaker C: To function it, it doesn't really have to be small, uh, big of course.\nSpeaker C: Yeah, okay.\nSpeaker C: The microphone could be just the minor, uh, whole, uh, on the left, uh, button.\nSpeaker C: Okay.\nSpeaker C: Um, we also made some, uh, ideas about how, uh, the options menu would work.\nSpeaker C: Uh, using the scroll button on the sides.\nSpeaker C: Uh, uh, uh, good.\nSpeaker B: You push the scroll button.\nSpeaker B: Yeah, you push the scroll button.\nSpeaker B: And it's claps out.\nSpeaker C: It's a drop down menu or a full out menu comes out and, uh, you get, uh, get the options, uh, becoming available.\nSpeaker C: For example, uh, TV settings, uh, remote settings, et cetera.\nSpeaker C: So, uh, you can scroll down to the scroll button.\nSpeaker C: Uh, as you can see, uh, it's here.\nSpeaker C: Just push it in.\nSpeaker C: Uh, the menu comes out like this.\nSpeaker C: And, uh, it all becomes visible.\nSpeaker B: Um, you could also touch it so that it comes out and use the, the, the scroll thing as, uh, what you fingers.\nSpeaker B: Indeed.\nSpeaker B: Yes.\nSpeaker C: Okay.\nSpeaker C: Um, it's also, uh, nice to see that, um, we made a small, uh, menu.\nSpeaker C: Uh, the options menu becoming available when pressing the scroll button.\nSpeaker C: Any opportunity to use the data text, which is used, uh, which is still be used.\nSpeaker C: And we think that it's, uh, very handy to put it, uh, not, uh, under the options menu, but in, uh, you know, in and apart, uh, in a separate button.\nSpeaker B: In a separate button, yeah.\nSpeaker B: That we should use the data text sign.\nSpeaker C: A sign in.\nSpeaker C: Just like, okay.\nSpeaker C: Indeed.\nNone: Okay.\nSpeaker C: We can, uh, modify that layer.\nSpeaker C: Okay.\nSpeaker C: Should I like to make any comments about next time?\nSpeaker B: Uh, well, this is the total interface, uh, of the LCD screen.\nSpeaker B: Uh, the numbers, which is pretty, we put an, an extra button in, we can erase it, but it's the button where you can switch channels.\nSpeaker B: I mean, just when you are on one and you go to two, you can, you go to five, you can go back to one with that button.\nSpeaker B: Yeah.\nSpeaker B: That one.\nSpeaker B: It has an area.\nSpeaker B: And, uh, we put that in.\nSpeaker B: Thought it would be handy here.\nSpeaker B: Uh, this, uh, one number, two numbers, but, uh, below that, the page and the sound and, uh, in the middle, uh, the mute.\nSpeaker B: Uh, battery indicator.\nSpeaker B: It's quite large.\nSpeaker B: And this is the, uh, the on off off, uh, standby, uh, or at least it should look like it.\nSpeaker B: And the options, uh, tail text.\nSpeaker C: You can see very clearly now that the, uh, tail text options menu isn't, uh, taking much, uh, uh, taking much part of the screen.\nSpeaker C: So it's very, uh, when you, uh, when you use it, it doesn't, uh, become irritating to see.\nSpeaker C: Because if you put it on the top, you always get to see the options menu.\nSpeaker C: Because people regularly, uh, read from left to right bound.\nSpeaker C: So.\nSpeaker B: Yeah.\nSpeaker B: Well, it's about it, I think.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nNone: Thank you.\nSpeaker B: I will put deck on the.\nSpeaker B: And I did miss when I was talking the RR stands for.\nSpeaker B: This is a logo of the.\nSpeaker B: Okay.\nSpeaker B: It's the.\nSpeaker B: Okay.\nSpeaker B: My nose only RR.\nSpeaker C: Full screen.\nSpeaker F: I would have recognized it if you would have right color.\nSpeaker A: Sorry.\nSpeaker A: Okay.\nSpeaker D: Full screen.\nSpeaker D: Okay.\nSpeaker E: Okay.\nSpeaker E: My task was this time.\nSpeaker E: To put up a questionnaire.\nSpeaker E: By which we can evaluate the design.\nSpeaker E: Of the remote control.\nSpeaker E: By the questions we come out requirements from the of the users.\nSpeaker E: My name, my job.\nSpeaker E: Okay.\nSpeaker E: I'm not sure if you have any methods questionnaire with seven point skill from one to seven from true to false.\nSpeaker E: Like question.\nSpeaker E: Is remote big enough?\nSpeaker E: We can say it's true or false.\nSpeaker E: By steps.\nSpeaker E: One means absolutely not true.\nSpeaker E: Seven means true.\nSpeaker E: The three important things of.\nSpeaker E: You used to point.\nSpeaker E: The point of this year is.\nSpeaker E: Are is the remote control fancy enough?\nSpeaker E: Is it in a fade if enough?\nSpeaker E: And is it easy enough to use?\nSpeaker E: And then the evaluation itself.\nSpeaker C: Right.\nSpeaker E: Okay.\nSpeaker E: First question.\nSpeaker E: Is the design fancy enough?\nSpeaker E: Project measure.\nSpeaker E: What do you think?\nSpeaker F: Well, it looks fancy.\nSpeaker F: Actually with green color.\nSpeaker F: But does it.\nSpeaker F: And the curves which we decided.\nSpeaker F: Yeah.\nSpeaker F: Only what happened to the single curve.\nSpeaker F: We should look.\nSpeaker C: No, the single curve idea was.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: You should make a side where you said.\nSpeaker C: A side way few.\nSpeaker C: I guess.\nSpeaker B: Oh, not that bad.\nSpeaker B: Not that bad.\nSpeaker A: Okay.\nSpeaker F: Smart word.\nSpeaker B: Can I draw here?\nSpeaker B: Supposing.\nNone: Oh.\nSpeaker B: Yeah.\nSpeaker B: It's going to work.\nSpeaker B: Something like this from the side.\nSpeaker B: But with a bit of a curve here, right?\nSpeaker B: Yeah.\nSpeaker C: That's a single curve.\nSpeaker B: Yeah.\nSpeaker B: So if you flip it like this.\nSpeaker B: Okay.\nSpeaker B: There's no theory.\nSpeaker C: It's also very handy if you make a side curve.\nSpeaker C: Way to make the road control very thin at the bottom.\nSpeaker C: At the bottom.\nSpeaker C: At the bottom.\nSpeaker C: I make it rather thick on the top because on the top it has the screen which takes in some space.\nSpeaker C: And the batteries can be located over there.\nSpeaker C: Yeah.\nSpeaker B: So.\nSpeaker B: Can I just make the back of this part a bit bigger so that it plays a good job?\nSpeaker B: Yeah.\nSpeaker E: It's going to be a little bit heavy at the top.\nSpeaker C: Let's have now.\nSpeaker C: Yeah.\nSpeaker B: That's a good problem.\nSpeaker E: With two batteries.\nSpeaker E: The whole print plate at the top.\nSpeaker E: And if you were holding it.\nSpeaker E: I think the battery should be in here because it's just nothing.\nSpeaker B: Yeah.\nSpeaker C: Okay.\nSpeaker F: Entity.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker E: We have to rate.\nSpeaker E: Yeah.\nSpeaker E: We have to rate.\nSpeaker E: Is this a one?\nSpeaker E: False.\nSpeaker E: This is seven.\nSpeaker E: So fancy enough means does it comes to the younger people and the elderly people?\nSpeaker B: I think it's pretty fancy.\nSpeaker B: I think it's pretty fancy.\nSpeaker B: If you don't make it green then the other people will.\nSpeaker E: Yeah.\nSpeaker E: You have to green.\nSpeaker E: I think it's better than that much now.\nSpeaker E: It's already designed.\nSpeaker B: I think it does.\nSpeaker F: Yeah.\nSpeaker F: And it's designed.\nSpeaker F: Oh, I think that's actually because of the microphone and also the screen.\nSpeaker B: Yeah.\nSpeaker B: I don't know whether older people will use this.\nSpeaker F: Well, so.\nSpeaker B: I would make it two or something.\nSpeaker D: The two?\nSpeaker B: Yeah.\nSpeaker D: That's a fit.\nSpeaker D: That's true.\nSpeaker C: It's one.\nSpeaker B: All right.\nSpeaker B: It's a very fancy.\nSpeaker B: It's a one.\nSpeaker F: A little bit strange.\nSpeaker F: We have to adjust that ourselves.\nSpeaker E: Yeah.\nSpeaker E: I should use that one, but it doesn't.\nSpeaker E: But it's a one.\nSpeaker B: Okay.\nSpeaker E: No, it's two.\nSpeaker E: True is a one.\nSpeaker E: Very true is a very true.\nSpeaker E: Isn't that true?\nSpeaker B: They think it's very true.\nSpeaker B: It's very true because we designed it to be very fancy.\nSpeaker C: Yeah.\nSpeaker C: It's very fancy.\nSpeaker C: I think.\nSpeaker C: Have you ever seen a remote control like this?\nSpeaker F: No.\nSpeaker F: Okay.\nSpeaker F: So it's fancy.\nSpeaker F: Yeah.\nNone: Okay.\nSpeaker F: One, two.\nSpeaker F: That's one of that.\nSpeaker F: Okay.\nSpeaker E: Let's give it a two.\nSpeaker E: Is it innovative?\nNone: I think it is.\nSpeaker B: Because it has an else, these gram, a microphone.\nSpeaker B: And a search for all this rubber.\nSpeaker C: Yeah.\nSpeaker C: The scroll.\nSpeaker D: I think it's a one.\nSpeaker D: I think it's a one.\nSpeaker D: It's a one.\nSpeaker E: Yeah.\nSpeaker E: True.\nNone: Also.\nSpeaker E: The buttons are very easy to find.\nSpeaker E: That was a big requirement of the old people.\nSpeaker B: Yeah, because they're right on your screen.\nSpeaker B: So you can use the, the, the, the right to your screen.\nSpeaker B: So I don't know where you.\nSpeaker B: With the ones and all the buttons.\nSpeaker E: Easy to find.\nSpeaker E: Not only this buttons, all the buttons.\nSpeaker B: Yeah.\nSpeaker B: I think they are.\nSpeaker B: The options are a little bit harder.\nSpeaker B: But if you take the hard, okay.\nSpeaker E: Sure.\nSpeaker C: I think it's, it's easier than the regular, uh, remote control.\nSpeaker C: Yeah.\nSpeaker C: I think it's easy.\nSpeaker E: I think the options buttons are not the easiest way to, to.\nSpeaker B: No, they're not, but they're, they are easy to find.\nSpeaker D: True.\nSpeaker C: They are a lot easier to find than, uh, than on the regular remote controls where you have to, uh, find out what which sign or icon means on everybody.\nSpeaker C: Yeah.\nSpeaker C: That's true.\nSpeaker C: So you have to, you have to use the, the manual.\nSpeaker C: The manual to understand.\nSpeaker C: You can add next.\nSpeaker F: So which number are we going?\nSpeaker C: I think it's, uh, it's a two.\nSpeaker C: At least.\nNone: Yeah.\nSpeaker F: Two.\nSpeaker E: Yeah.\nSpeaker E: Two.\nSpeaker E: Yeah.\nSpeaker E: It's not perfect, but I think it's a three.\nSpeaker C: Three.\nSpeaker C: And why is that?\nSpeaker E: Because I, I don't think it, maybe it's easy to use.\nSpeaker E: It has to be easy to find right away.\nSpeaker E: I, I think if you have the button at the right, I don't think you can find the option button.\nSpeaker B: But you don't have to use the button.\nSpeaker B: You can touch it.\nSpeaker B: You can touch it.\nSpeaker B: Yeah.\nSpeaker C: It's at both, both the options.\nSpeaker C: Yeah.\nSpeaker E: Okay.\nSpeaker E: But you have, then you have your written option on here, the data text button, right?\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: Then, okay.\nSpeaker B: Then I think also to you.\nSpeaker C: You can touch option.\nSpeaker C: I do.\nSpeaker C: Okay.\nSpeaker B: I do.\nSpeaker C: The idea.\nSpeaker C: The next question.\nNone: The next question.\nSpeaker F: Oh my God.\nSpeaker F: Otherwise we have two results in one question.\nSpeaker E: It's different.\nSpeaker E: It's easy to use as well for younger elderly people.\nSpeaker B: For young people, I think it's easy to use.\nSpeaker E: Young means 16 to 40 years at elderly from 40 to their death.\nSpeaker C: I think it's the most useful remote control ever to be manufactured.\nSpeaker C: Very enthusiastic.\nSpeaker F: Also, if you're a 6 year old.\nSpeaker C: Yeah, because it has the regular controls, like as you can see on the screen now.\nSpeaker C: And you don't have all the other options always on your screen.\nSpeaker C: So it's, it's really easy to use.\nSpeaker E: You want these options to, if you're a 6 years old.\nSpeaker B: You're holding one of those things in your head.\nSpeaker B: They don't want the extra options.\nSpeaker E: No, but we're going to make this for all kind of people.\nSpeaker F: So it would be easy for them to use the speech recognition.\nSpeaker B: I think it is.\nSpeaker B: If they read the menu.\nSpeaker B: Because they might.\nSpeaker C: Yeah, perhaps that is one of those.\nSpeaker C: It's not difficult.\nSpeaker B: You say record.\nSpeaker B: To do the speech recognition.\nSpeaker B: Then you should question and the answer.\nSpeaker B: And that's everything it does to the speech recognition.\nSpeaker F: Yeah, well, maybe that would make it even more easy to use.\nSpeaker B: I think it would be difficult.\nSpeaker C: Yeah, I think it is.\nSpeaker E: So, because all the people who can't make it to.\nSpeaker E: Yeah, I think it's difficult to do.\nSpeaker E: Yeah.\nSpeaker E: I chose three.\nSpeaker E: Three.\nSpeaker F: I would also say three.\nSpeaker F: Three, three, two, two.\nSpeaker F: Two.\nSpeaker F: No.\nSpeaker F: No, what are we going to do?\nSpeaker F: Two and all.\nSpeaker F: Okay, yeah, three, I see.\nSpeaker F: Give me more.\nSpeaker F: Another question.\nSpeaker E: Remote overwhelmed with buttons.\nSpeaker E: No.\nSpeaker E: No.\nSpeaker E: No, that's, that's it.\nSpeaker B: I mean, that's definitely one.\nSpeaker B: That's definitely one.\nSpeaker B: There's definitely an hour.\nSpeaker B: I mean, a seven is it.\nSpeaker B: It is.\nSpeaker B: Yeah, that's it.\nSpeaker F: Yeah, but I think you should make it one scale with one.\nSpeaker E: Yeah, I think it isn't.\nSpeaker F: There's a way to become something like calculate any.\nSpeaker F: Isn't overwhelmed.\nSpeaker C: Yeah.\nSpeaker C: So, yeah, any.\nSpeaker F: True.\nSpeaker F: Okay.\nSpeaker F: All right.\nSpeaker F: Designed for that.\nSpeaker E: Through all this colors that different.\nSpeaker E: Yeah.\nSpeaker B: Meet different target groups.\nSpeaker B: We make them in different colors.\nSpeaker C: Yeah, there's option.\nSpeaker F: Yeah.\nSpeaker F: I thought we had about single colors, but you can also make a wood color, not just one single color, but a wood.\nSpeaker B: Yeah, it's not.\nSpeaker B: Like, it looks like wood.\nSpeaker B: Also with rubber.\nSpeaker B: I think you can.\nSpeaker F: But it looks like wood.\nSpeaker B: It isn't.\nSpeaker B: Yeah, okay.\nSpeaker F: It feels like rubber.\nSpeaker F: You can make a print on rubber, can't you?\nSpeaker F: Yeah.\nSpeaker B: Yeah.\nSpeaker B: But then when you square two, it's just come up.\nSpeaker E: That's the one.\nSpeaker C: Okay.\nSpeaker E: It is harder to.\nSpeaker E: Yeah.\nSpeaker E: Money questions.\nSpeaker F: I, uh, okay.\nSpeaker F: Well, we have time.\nSpeaker F: We're getting paid.\nSpeaker C: We're getting paid.\nSpeaker E: You're all used to this.\nSpeaker E: That's, that's a one.\nSpeaker E: That's rubber.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: It's not the most spongy thing.\nSpeaker B: I think it's a three.\nSpeaker B: Yeah.\nSpeaker C: It's a three because you want to make it, uh, rather flexible, but not too flexible because it has an L.C.E. screen.\nSpeaker A: Yeah.\nSpeaker E: How about across hard to lose?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker E: You can, you can lose it if you're six years old.\nSpeaker E: You first, I'll see the thing you didn't, didn't adjust, uh, set the microphone and then you lose it.\nSpeaker E: Then you have lost it.\nSpeaker B: You can lose it, right?\nSpeaker B: It isn't hard to lose.\nSpeaker E: It is a dar, no.\nSpeaker E: I think, I think this is a two.\nSpeaker B: Personally.\nSpeaker B: Two.\nSpeaker B: That it's hard to lose?\nSpeaker B: No.\nSpeaker B: Yeah.\nSpeaker B: It isn't hard to lose.\nSpeaker B: It's a six.\nSpeaker D: You think.\nSpeaker D: It's hard to lose.\nSpeaker B: So it's a two.\nSpeaker B: Yeah.\nSpeaker B: You can lose it.\nSpeaker A: So I don't know.\nSpeaker B: You can make it a three.\nSpeaker B: It doesn't.\nSpeaker A: If you're six years old.\nSpeaker A: Yeah, but a hard.\nSpeaker F: A hard to lose is good.\nSpeaker F: So it's, this question should be hard to lose.\nSpeaker F: It's difficult to lose.\nSpeaker C: Yeah.\nSpeaker C: It's hard to lose.\nSpeaker B: Yeah.\nSpeaker B: So then it is almost true.\nSpeaker B: Yeah.\nSpeaker B: I think also.\nSpeaker C: Okay.\nSpeaker C: Mostly because of the option to.\nSpeaker C: Whoa.\nSpeaker A: Yeah.\nSpeaker F: It's okay.\nSpeaker F: It happened above also.\nSpeaker F: Maybe when you scroll away and back it will be normal.\nSpeaker F: Yeah.\nSpeaker F: Oh.\nSpeaker C: Well, I actually, I remember.\nSpeaker C: If you put the cursor on the table.\nSpeaker E: Okay.\nSpeaker E: The most control mainly be sold to younger people.\nSpeaker E: True.\nSpeaker B: I think it will.\nSpeaker B: Yeah.\nSpeaker E: True.\nSpeaker E: Very true.\nSpeaker E: Yeah.\nSpeaker E: I don't think very true because the colors.\nSpeaker E: We have the colors.\nSpeaker E: We have the buttons.\nSpeaker E: Aren't that much?\nSpeaker E: Materials.\nSpeaker E: Not material.\nSpeaker E: Isn't that.\nSpeaker E: Okay.\nSpeaker E: So I think, I think it's a three.\nSpeaker F: Well, I think it's a lower number.\nSpeaker F: So better because we designed it for young people especially.\nSpeaker D: Yeah.\nSpeaker D: But I.\nSpeaker D: What do you think?\nSpeaker D: I think.\nSpeaker C: I think it's two.\nSpeaker C: I think two, two, two, two, two.\nSpeaker C: That's very good.\nSpeaker A: You have features?\nSpeaker F: To satisfy younger people.\nSpeaker F: Younger people.\nSpeaker F: No.\nSpeaker E: Perhaps not.\nSpeaker E: Because younger people are critical and they wanted a lot of features on this thing.\nSpeaker E: Well, not exactly.\nSpeaker C: But if it had some games on it or something like that.\nSpeaker C: But it's for remote control.\nSpeaker C: I think it would satisfy those things.\nSpeaker A: Yeah.\nSpeaker A: The speech, possibility of color.\nSpeaker C: Yeah, the scroll.\nSpeaker C: Scroll options.\nSpeaker E: I think they'll be quite met with their expectations.\nSpeaker F: Yeah, but those are more offensive functions.\nSpeaker F: Not really many features or something.\nSpeaker F: I mean, those are features.\nSpeaker F: Yeah, three features.\nSpeaker E: Basically, LCNIs.\nSpeaker E: LCNIs.\nSpeaker E: They'll keep the touch screen as a feature.\nSpeaker E: Yeah.\nSpeaker E: The microphone is a feature.\nSpeaker B: The microphone is a feature.\nSpeaker B: And that you can change the channels feature and that you can change the volume as a feature.\nSpeaker B: And that you can change the options of the remote.\nSpeaker B: Yeah, I think.\nSpeaker B: I don't know.\nSpeaker E: I think.\nSpeaker C: Easy volume output.\nSpeaker C: The most obvious.\nSpeaker C: The most obvious.\nSpeaker C: Younger people usually use remote control to put it on the volume.\nSpeaker C: Yeah.\nSpeaker C: Turn out.\nSpeaker E: I think it's one.\nSpeaker E: Enough features.\nSpeaker E: Yeah, I think because you have channel features, you have the screen features, audio features.\nSpeaker E: You have all buttons on it.\nSpeaker E: Which is like microphone, extra LCD screen, extra scroll thing, extra.\nSpeaker F: Okay, you think one?\nSpeaker F: What do you think?\nSpeaker B: I think two or three.\nSpeaker F: Yeah.\nSpeaker F: Two.\nSpeaker F: I'd say three.\nSpeaker F: Two.\nSpeaker A: Make it two.\nSpeaker A: One, two, three.\nSpeaker A: It's another two.\nSpeaker A: One, two.\nSpeaker E: You can see the remote control is the R&R.\nSpeaker B: Yeah, there's R&R in front.\nSpeaker B: It's not the color.\nSpeaker B: So maybe you should make two, but it has R&R.\nSpeaker E: Oh, yeah.\nSpeaker E: I did have.\nSpeaker E: Now you have the black one.\nSpeaker E: Yeah.\nSpeaker E: And we'll probably make also yellow one.\nSpeaker E: Yeah.\nSpeaker B: Yeah.\nSpeaker F: Maybe two.\nSpeaker E: Okay, true.\nSpeaker E: One, one, one.\nSpeaker E: It's the color.\nSpeaker E: And the menu is a fail.\nSpeaker E: That is a fail.\nSpeaker E: It's easy to use.\nSpeaker B: I think it is, but I don't know what you think.\nSpeaker C: I think it's easier to use than a regular remote control because when you push on the options menu, you get the various options entirely explained, entirely explained.\nSpeaker C: Instead of just finding out what an icon means on a button.\nSpeaker B: Yeah.\nSpeaker B: I think the navigation is easier.\nSpeaker B: Yeah, you can have again.\nSpeaker B: You have to push the arrows and with the normal to the remote.\nSpeaker D: I think that would be quite hard to learn.\nSpeaker C: You're not satisfied.\nSpeaker C: Okay.\nSpeaker C: Let's start over again.\nSpeaker C: Yeah.\nSpeaker E: That's very good.\nSpeaker E: Let's go for menu.\nSpeaker E: I think it would be a, yeah, too.\nSpeaker E: Not lower.\nSpeaker F: I tell you.\nSpeaker F: We only have to.\nSpeaker F: It's pretty good.\nSpeaker D: I think so.\nSpeaker B: Yeah.\nSpeaker B: So, okay.\nSpeaker D: We have one, three, a one that I have to cut open.\nSpeaker D: I'll have to cut.\nSpeaker D: Oh, go.\nSpeaker D: Two, two, two, two.\nSpeaker D: Oh, you have to.\nSpeaker B: So it's pretty good.\nSpeaker B: The average is two.\nSpeaker B: Yeah.\nSpeaker E: That is quite good.\nSpeaker B: Yeah.\nSpeaker B: It might be.\nSpeaker D: I think so.\nSpeaker B: We can be happy.\nSpeaker D: Safe.\nSpeaker D: Okay.\nSpeaker D: What is it?\nSpeaker B: It's like a bird.\nSpeaker B: It's a fly.\nSpeaker B: A fly.\nSpeaker B: A bird of fly.\nSpeaker B: A bird of fly.\nNone: Yeah.\nSpeaker B: That's it.\nNone: Okay.\nSpeaker A: That was your evaluation.\nSpeaker F: Yes.\nSpeaker F: Show.\nSpeaker F: Okay.\nSpeaker F: We don't have to calculate anything.\nSpeaker F: No, it's two.\nSpeaker E: Okay.\nSpeaker E: Good.\nSpeaker F: That's true.\nSpeaker F: Let's see.\nSpeaker F: We didn't ask to shave.\nSpeaker F: Yeah.\nSpeaker E: Yeah.\nSpeaker E: I saved it.\nSpeaker F: Okay.\nSpeaker F: Well, the next part might be a little bit difficult to do in a group, but I am willing to try it.\nSpeaker F: Because we are going to look at the finance and I have a nice Excel sheet to do that.\nSpeaker B: Redesign.\nSpeaker B: No.\nSpeaker F: I'm not sure I've ever put it for the look on that.\nSpeaker F: And we're going to calculate the production costs and if they are under or at 12, you know, 50, we're good.\nSpeaker F: Yeah.\nSpeaker F: And if we're not, we're going to redesign.\nSpeaker B: We're going to erase the two versions.\nSpeaker F: Very quick.\nSpeaker F: I think yes.\nSpeaker F: I don't know if I.\nSpeaker B: Do you have the cost, sir?\nSpeaker F: Put the Excel sheet in the.\nSpeaker F: Let's hope.\nSpeaker B: We're going to read an 8 o'clock.\nSpeaker F: I think it's still in my own documents folder.\nSpeaker B: We're still going to be read an 8 o'clock.\nSpeaker B: I doubt it.\nSpeaker B: Perhaps we've got features that don't exist in the Excel sheet.\nSpeaker E: Yeah.\nSpeaker D: Maybe the microphone.\nSpeaker D: That was in my information.\nSpeaker D: Yeah.\nSpeaker C: It wasn't too much.\nSpeaker C: Now what maybe I got.\nSpeaker C: I grabbed the LCD screen.\nSpeaker F: Scrap it.\nSpeaker F: Okay.\nSpeaker F: Well, this is it.\nSpeaker F: Maybe I could ask one of you to fill it in so that I can also take minutes.\nSpeaker F: And maybe the other can use that thing to count numerous functions.\nSpeaker F: But you shoot a unit like right.\nSpeaker F: We have to count some things and we have to think about some things.\nSpeaker F: Count it.\nSpeaker C: You've got Excel to count.\nSpeaker F: Yeah.\nSpeaker F: No, count number of functions.\nSpeaker F: Oh, okay.\nSpeaker C: I draw a different screen shot.\nSpeaker C: I'll do that because I draw the screen.\nSpeaker B: We've got a battery.\nSpeaker B: One or two batteries.\nSpeaker B: And I want battery.\nSpeaker B: But two small batteries.\nSpeaker F: Yeah, it's more about the energy source.\nSpeaker F: Do you use a hand dynamo of battery?\nSpeaker B: A solar cell.\nSpeaker B: Solar cell.\nSpeaker B: No, no, no, no, no.\nSpeaker B: One battery.\nSpeaker B: Right?\nSpeaker B: No, hand dynamo.\nSpeaker C: Yeah.\nSpeaker C: Hand.\nSpeaker B: We have simple chip advanced chip, right?\nSpeaker B: No, I have simple.\nSpeaker E: I have advanced chip.\nSpeaker E: Well, but we also have sample speaker.\nSpeaker E: Yeah.\nSpeaker B: So this one and this one.\nSpeaker B: We're already on mine.\nSpeaker B: Single.\nSpeaker B: What?\nSpeaker B: Are we?\nSpeaker B: Are we?\nSpeaker E: We have double curved.\nSpeaker E: Single.\nSpeaker C: Nice single curved.\nSpeaker C: Single curved as you drew three dimensional.\nSpeaker C: But it isn't three dimensional.\nSpeaker B: It's not going to work, people.\nSpeaker E: This one is curved like this, right?\nSpeaker E: It's curved like this.\nSpeaker E: No, no, no, single curved like this.\nSpeaker C: That's the only curve you made, not curved like that.\nSpeaker C: That's, oh, but we have curves like this.\nSpeaker E: And the two curves.\nSpeaker E: 13.\nSpeaker E: Oh, I understand.\nSpeaker B: We have the scroll wheel, right?\nSpeaker B: Yeah.\nSpeaker B: Is he integrated?\nSpeaker B: No.\nSpeaker C: Push.\nSpeaker C: Yeah, we don't have.\nSpeaker C: We got an integrated scroll wheel and push button.\nSpeaker C: Because when you push it, we want this pull.\nSpeaker C: Oh, yeah, right.\nSpeaker B: It's not, it's not, no.\nSpeaker B: Else you can play.\nSpeaker B: Okay.\nSpeaker B: You're 15, too bad.\nSpeaker B: Okay.\nSpeaker B: Oh, but the special color, yeah.\nSpeaker B: The special form, right?\nSpeaker C: But now button supplements.\nSpeaker C: We don't got the button supplements.\nSpeaker B: Oh, we don't have any buttons.\nSpeaker B: So.\nSpeaker B: 80 and a half.\nSpeaker E: Yeah, we have to lower it with six points.\nSpeaker B: Okay.\nSpeaker B: 15 and...\nSpeaker B: 12 and a half.\nSpeaker B: All right.\nSpeaker B: We could lose the curve.\nSpeaker E: We could use...\nSpeaker E: We could lose the scroll wheel.\nSpeaker C: We could make just a regular scroll wheel.\nSpeaker B: But you can push it.\nSpeaker B: So you can see that.\nSpeaker C: Yeah, you can push it.\nSpeaker C: You have to tap the options window button here.\nSpeaker C: And then scroll down with the button.\nSpeaker C: Yeah.\nSpeaker B: All right.\nSpeaker B: Yeah, I think that will be...\nSpeaker B: So normal scroll wheel.\nSpeaker B: Normal scroll wheel.\nSpeaker B: And I think we should...\nSpeaker E: I think we should...\nSpeaker E: You can just grab the sample speaker.\nSpeaker E: It's four units.\nSpeaker B: Yeah, but what if you would...\nSpeaker B: It is a new feature.\nSpeaker C: Okay, so we don't exactly need the same, but...\nSpeaker C: We don't need a curve.\nSpeaker B: No.\nSpeaker B: It's possible to lose that.\nSpeaker C: The curve that's really...\nSpeaker B: Square.\nSpeaker B: No, then it won't stand up from the table.\nSpeaker B: That is what you...\nSpeaker B: That means with...\nSpeaker C: Yeah, that's meant with what's curve.\nSpeaker C: The curve is in a dimension.\nSpeaker C: Oh, okay.\nSpeaker C: If you make it a flat one, there's no curve.\nSpeaker C: You got no curve.\nSpeaker C: You lose this one?\nSpeaker E: Yeah, that only is one.\nSpeaker E: Yeah, we can see it a bit.\nSpeaker B: No, one.\nNone: 60.\nSpeaker B: Okay, Andy.\nSpeaker E: So we don't...\nSpeaker E: Yeah, we also have to...\nSpeaker C: Is it possible to make...\nSpeaker B: Could we do it on a regular chip on print or something?\nSpeaker B: Otherwise we don't have a LCD screen.\nSpeaker E: No.\nSpeaker B: You just can't do that.\nSpeaker B: What did you change?\nSpeaker F: You changed the...\nSpeaker F: You changed the scroll wheel.\nSpeaker F: Yeah, and the single curve to uncurl.\nSpeaker F: Oh, but it's just one.\nSpeaker F: Yeah, so that doesn't...\nSpeaker E: Maybe you should...\nSpeaker E: You...\nSpeaker E: Scrap a sample speaker.\nSpeaker C: Yeah.\nSpeaker C: The sample speaker, too.\nSpeaker C: It's four points.\nSpeaker C: You have four points.\nSpeaker E: Yeah, but it is a...\nSpeaker E: And then you can keep the curve.\nSpeaker B: It is a new feature.\nSpeaker B: It is a new feature.\nSpeaker B: It's a new feature.\nSpeaker B: When you lose, what else do you want to scrap?\nSpeaker E: I don't know.\nSpeaker E: We have to scrap four points.\nSpeaker B: Yeah.\nSpeaker B: That's difficult.\nSpeaker F: Or maybe I'm going to hand them over.\nSpeaker F: Yeah.\nSpeaker E: You can take it with wood instead of rubber.\nSpeaker C: No, that's not...\nSpeaker C: When you made a remote control of wood...\nSpeaker A: Yeah.\nSpeaker C: Don't make a remote control.\nSpeaker E: It all takes one point less.\nSpeaker F: Yeah, but a wooden remote control only helps for...\nSpeaker B: Oh, can I ask something?\nSpeaker B: All people, we...\nSpeaker B: What is special color?\nSpeaker B: Is that the wood...\nSpeaker B: I think it is.\nSpeaker B: This we have to have that one, too.\nSpeaker F: Yeah, but it's only a hop.\nSpeaker F: But I think the only option is to drop the sample speaker.\nSpeaker C: To knock the sample speaker in.\nSpeaker C: And sample sensor.\nSpeaker B: Then we still have too much.\nSpeaker E: Yeah, okay.\nSpeaker C: Three, four, three.\nSpeaker C: Of course, but...\nSpeaker C: Why 12 and a half?\nSpeaker C: That's making 13 or 14.\nSpeaker C: We need 14.\nSpeaker B: Three.\nSpeaker B: We need point three.\nSpeaker E: There's a scroll view.\nSpeaker B: It's a color.\nSpeaker E: Don't make a wood.\nSpeaker E: Yeah, but a wood.\nSpeaker E: We can make a brown, dark brown, not a wood.\nSpeaker B: Yeah.\nSpeaker B: But it's a special color, is it?\nSpeaker C: Oh, my gosh.\nSpeaker C: Oh, my gosh.\nSpeaker B: It's also green or...\nSpeaker E: That's just normal color, fruit color, normal color.\nSpeaker B: Yeah, but it's a special color than just rubber color.\nSpeaker B: You have to add something to the rubber to make it green.\nSpeaker B: Yeah.\nSpeaker B: You don't say here, green rubber.\nSpeaker A: They don't sell green rubber.\nSpeaker E: I don't think you can ever move to a 12 and a half.\nSpeaker B: Yeah, you can.\nSpeaker B: You have to lose.\nSpeaker E: Well, then we have to scrap LCD display.\nSpeaker E: We have to scrap...\nSpeaker E: No, no, no.\nSpeaker E: Then we have to scrap everything.\nSpeaker E: Because how many colors we're going to make?\nSpeaker E: Five.\nSpeaker C: If we lose a scroll wheel...\nSpeaker C: And make it totally dependent on the touch screen.\nSpeaker C: Yeah.\nSpeaker C: Then it's possible to make.\nSpeaker C: Then you can add two colors.\nSpeaker C: And then you can add the colors.\nSpeaker C: Yeah, two colors with it.\nSpeaker C: Special.\nSpeaker C: Okay, if you lose...\nSpeaker C: If you lose...\nSpeaker C: That's such a great idea.\nSpeaker B: You do this one.\nSpeaker A: And you get a level of.5.\nSpeaker C: And you make...\nSpeaker C: And then you can make a special single curve, for example.\nSpeaker E: But the colors...\nSpeaker E: The colors like she told...\nSpeaker E: Is that all the colors we add?\nNone: Or...\nSpeaker C: What do you mean?\nSpeaker C: Special colors.\nSpeaker C: How all the colors you want.\nSpeaker E: We have yellow, red, black.\nSpeaker C: When you use more than one color, it's a special color.\nSpeaker C: Okay.\nSpeaker B: I think when you use the color that's not originally the rubber, then you use special color.\nSpeaker B: Because you have to add...\nSpeaker E: Yeah, but the rubber is all original black.\nSpeaker B: So you always lose the special.\nSpeaker B: You could make it always black.\nSpeaker E: Yeah, but we're going to make it yellow, red.\nSpeaker E: And then you add two special colors on top of the one.\nSpeaker C: We also want to make another color.\nSpeaker B: Yeah, we have to make this like four or five or something.\nSpeaker E: Because they have more colors than only black.\nSpeaker E: Yeah, that isn't it.\nSpeaker E: And then I think...\nSpeaker E: I don't think they made me this special.\nSpeaker B: Is it per remotes?\nSpeaker F: I think you pay half per remotes.\nSpeaker F: Yeah, that's right.\nSpeaker B: And each remotes...\nSpeaker B: One color per remotes.\nSpeaker B: Yeah, and indeed...\nSpeaker C: You don't need four of those for those special colors in one remote.\nSpeaker C: Yeah.\nSpeaker E: We have two points spare.\nSpeaker F: So the battery we have advanced chip on print.\nSpeaker B: So it would be curved.\nSpeaker B: Single curves.\nSpeaker B: Because of...\nSpeaker F: Yeah, well you can at least make it curved again.\nSpeaker F: Single curves.\nSpeaker B: You can make a nice remote.\nSpeaker B: Because that was very important.\nSpeaker F: Too bad for the speaker.\nSpeaker F: So it's curved, so we dropped the speech recognition.\nSpeaker F: Should we change that?\nSpeaker B: That's one of not.\nSpeaker F: We dropped the scroll wheel.\nSpeaker F: Could you copy it?\nSpeaker B: And make the rest of the same.\nSpeaker E: The scroll wheel is dropped.\nSpeaker B: The entire...\nSpeaker B: Yeah.\nSpeaker B: Actually, I'm copying the page.\nSpeaker B: So...\nSpeaker B: Oh.\nSpeaker B: Oh.\nSpeaker B: I made the entire...\nSpeaker B: I do.\nSpeaker B: I do.\nSpeaker C: I do.\nSpeaker C: I do.\nSpeaker A: No, no, no, no.\nSpeaker A: Okay.\nSpeaker C: Are you there?\nSpeaker C: I do.\nSpeaker B: Perhaps you can say this one.\nSpeaker B: And I'll copy your...\nSpeaker B: I just copy page.\nSpeaker B: Left off.\nSpeaker F: No, but you...\nNone: Yeah.\nSpeaker B: All right.\nSpeaker B: Something went wrong.\nNone: Tap.\nSpeaker F: Okay, but this new remote we can have for?\nSpeaker B: Let's forget.\nSpeaker C: It should have worked.\nSpeaker C: Okay.\nSpeaker C: So it has this list as start.\nSpeaker C: No.\nSpeaker C: One degree this list.\nSpeaker C: I just received it.\nSpeaker A: Okay.\nSpeaker C: Yeah.\nSpeaker F: It worked so hard at the final department.\nSpeaker C: Okay.\nSpeaker C: Well, though...\nSpeaker C: I suppose this is...\nSpeaker C: Okay.\nSpeaker C: So we lose the scroll wheel.\nSpeaker C: Yeah.\nSpeaker B: The microphone.\nSpeaker F: Yeah.\nSpeaker F: That's it.\nSpeaker B: And the microphone.\nSpeaker B: And we change something, I guess.\nSpeaker B: We...\nSpeaker B: I don't know.\nNone: Okay.\nSpeaker F: Twelve, you have fifty.\nSpeaker F: And that you try to make a new design or what?\nSpeaker B: Yeah, I tried to copy that one, but it didn't work.\nSpeaker B: It didn't work?\nSpeaker B: So we could fix it like this.\nSpeaker F: Hmm.\nSpeaker F: You could select it all.\nSpeaker B: It's longer.\nSpeaker B: Then you can erase.\nSpeaker C: Oh, you can only erase?\nSpeaker C: Erase.\nSpeaker B: Erase.\nSpeaker B: Earlier when we selected it.\nSpeaker B: Erase anything.\nSpeaker B: Okay.\nSpeaker F: Can't you just say copy, blink, new page.\nSpeaker F: Paste.\nSpeaker F: Erase.\nNone: Erase.\nSpeaker F: Select none.\nSpeaker F: Hold this.\nSpeaker C: Just tap down somewhere.\nSpeaker C: Just tap somewhere.\nSpeaker C: Just tap somewhere.\nSpeaker C: You decided.\nSpeaker C: Right.\nSpeaker F: Okay.\nSpeaker F: And now you can erase.\nSpeaker B: I don't think I can.\nSpeaker B: You can try.\nSpeaker B: Or it should be possible.\nSpeaker B: Oh, yeah.\nSpeaker B: No.\nSpeaker F: Oh, oh.\nSpeaker F: Well, you can draw over it with white.\nSpeaker F: Yeah.\nSpeaker B: We tried it earlier.\nSpeaker B: It's very much work.\nSpeaker F: Yeah.\nSpeaker F: Sorry.\nSpeaker F: Well, but that's also useful for the evaluation because I think we have a prototype now which we can afford and we only need to draw a little bit to get a good design.\nSpeaker F: Doodle and I think we should then move on to the production evaluation because of the time.\nSpeaker B: And erase the mic.\nSpeaker B: Yeah.\nSpeaker F: Goodbye, Mike.\nSpeaker F: I need is no mic.\nSpeaker C: Oh, I erased half of the line.\nSpeaker F: This.\nSpeaker F: Pushing on.\nSpeaker F: Move back to here.\nNone: Too bad.\nNone: Oh.\nSpeaker B: Like this.\nSpeaker B: Still looks like.\nSpeaker F: Okay.\nSpeaker F: Oh, thank you.\nSpeaker F: Looks like iPod.\nSpeaker F: No, no.\nSpeaker F: No.\nSpeaker F: I can erase that.\nSpeaker B: Yeah.\nSpeaker C: Oh, so now we align.\nSpeaker D: Oh, so now we align.\nSpeaker D: Difference between lines and text and the pen.\nSpeaker F: Oh.\nSpeaker A: All right.\nSpeaker F: All right.\nSpeaker F: I'll erase this.\nSpeaker F: No.\nSpeaker F: Traits.\nSpeaker F: Okay.\nSpeaker D: We'll just leave it at this.\nSpeaker D: And quickly face.\nSpeaker F: And then we are going to the project or product evaluation.\nSpeaker F: We just did.\nSpeaker F: Oh, I think I can sit for that.\nSpeaker F: Almost my final slide.\nSpeaker F: What did you think about the process?\nSpeaker F: How satisfied are we?\nSpeaker C: That lines are sometimes are short.\nSpeaker C: Yeah.\nSpeaker B: That's stressful.\nSpeaker B: You think that's what my presentation is already?\nSpeaker E: I think it would be better if we worked a little bit together.\nSpeaker E: Yeah.\nSpeaker E: Now we worked through each other and you could.\nSpeaker B: And you had information I didn't have.\nSpeaker E: Yeah.\nSpeaker E: You had information I also had.\nSpeaker E: Oh.\nSpeaker E: Some things I had in my presentation they already told.\nSpeaker F: And for me it sometimes was a surprise who was going to present what.\nSpeaker E: So yeah, I don't think that is the best way to work for such a project.\nSpeaker F: You would say communicate during our individual.\nSpeaker E: Yeah.\nSpeaker E: Or maybe a session of five minutes together or something and then work separate.\nSpeaker C: Yeah, but why not work here together?\nSpeaker C: For example, why should we be separated from each other in those different rooms?\nSpeaker F: Not probably to simulate the whole working process.\nSpeaker F: You can work together too.\nSpeaker F: You can't have a meeting.\nSpeaker E: Yeah.\nSpeaker E: Several weeks.\nSpeaker E: Then you can work together by mail or by, I don't know, chats something.\nSpeaker E: Now we are completely separated from each other.\nSpeaker E: I don't think that was the best way.\nSpeaker C: But the technology was fantastic.\nSpeaker C: Yeah, the technology was so good.\nSpeaker B: I don't really like that board.\nSpeaker B: It doesn't really work.\nSpeaker B: Work now.\nSpeaker E: Sometimes it would be good.\nSpeaker B: It does work, but sometimes it does do right.\nSpeaker C: Perhaps it is easier to get one of those digital pens or so and to and lay next to that keyboard over there.\nSpeaker D: So you can draw and see it on the screen.\nSpeaker D: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: So you don't think the smart board is really useful?\nSpeaker B: It's useful, but it doesn't really work all the time.\nSpeaker C: So when you put the pen on the screen, the line is being drawn at two or three centimeters below.\nSpeaker F: Yeah, so it may be a bit natural.\nSpeaker F: Yeah.\nSpeaker B: All right.\nSpeaker B: Yeah, you can point to where you want the line to be.\nSpeaker C: The project, because of the deadlines, you didn't have time to make a very qualitative presentation.\nSpeaker C: So you used different PowerPoint presentations in which you put your material on the...\nSpeaker A: Oh.\nSpeaker F: That wasn't me.\nSpeaker F: That wasn't me.\nSpeaker F: So the means we discussed the smart board and what about this digital pen?\nSpeaker B: I didn't use it at all.\nSpeaker B: I used it.\nSpeaker B: You can use it.\nSpeaker F: I used it as a normal pen and only you used it to...\nSpeaker B: I used it to...\nSpeaker B: On the computer.\nSpeaker B: Yeah.\nSpeaker C: It did work.\nSpeaker C: I used it to, but...\nSpeaker C: Hello.\nSpeaker B: I don't think why you want to use it.\nSpeaker B: It does work.\nSpeaker B: To make some designs is very easy.\nSpeaker E: It's very easy for it to decide something and then load it to your computer.\nSpeaker B: Yeah.\nSpeaker C: And then you can show it to everyone.\nSpeaker C: Yeah.\nSpeaker B: That's right.\nSpeaker B: It doesn't really work normally.\nSpeaker C: It doesn't work normally.\nSpeaker C: Yeah, it's too big.\nSpeaker C: It's too fat.\nSpeaker C: Fat dog members.\nSpeaker F: Okay.\nSpeaker F: And what about the teamwork?\nSpeaker F: The teamwork was okay.\nSpeaker B: I think it was great.\nSpeaker D: Yeah.\nSpeaker D: I think so too.\nSpeaker D: Only thing that we worked through post-iture.\nSpeaker D: Yeah.\nSpeaker D: Something that was all the problem.\nSpeaker E: But it wasn't the assignment.\nSpeaker E: Yeah.\nSpeaker F: Yeah.\nSpeaker F: But for the more...\nSpeaker F: Okay.\nSpeaker F: And maybe I should walk out of the room when you discuss at this point.\nSpeaker F: The leader.\nSpeaker B: I thought it was good.\nSpeaker D: Yeah.\nSpeaker D: No problem.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker D: Okay.\nSpeaker F: And creativity.\nSpeaker F: Well, when we look at this, I'd say.\nSpeaker F: It's been creative.\nSpeaker C: Yeah.\nSpeaker C: The room for...\nSpeaker C: It was the idea to be creative, so...\nSpeaker A: Yeah.\nSpeaker C: You got some standard ideas in your head.\nSpeaker C: Yeah.\nSpeaker C: And this is what came out.\nSpeaker B: You get stuff from the...\nSpeaker B: Yeah.\nSpeaker B:...\nNone:...\nSpeaker C:...\nSpeaker C: Information was sometimes...\nSpeaker C:...\nSpeaker C:... a little bit too late.\nSpeaker C: Yeah.\nSpeaker E: Too late.\nSpeaker E: It took a lot of time before you got your...\nSpeaker B: Yeah.\nSpeaker E: Okay.\nSpeaker E: I played, I think, seven times only three or something.\nSpeaker E: Well, I didn't know...\nSpeaker B: Oh, did you...\nSpeaker B: It's on there.\nSpeaker B: It's on there.\nSpeaker B: I didn't find that.\nSpeaker B: I'm so...\nSpeaker F: I'm so interested in...\nSpeaker F:...\nNone:...\nSpeaker F:...\nNone:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F:...\nNone:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker E:...\nSpeaker F:...\nSpeaker F:...\nSpeaker F: satisfied but it's better when we get more information.\nSpeaker C: More information about the costs.\nSpeaker C: First of all I didn't think that we were able to make an LCD screen.\nSpeaker C: It was possible.\nSpeaker B: Actually you could make an LCD screen but no mic or you could make mic.\nSpeaker F: What we ordered was to improve this whole process.\nSpeaker D: I think that's about it.\nSpeaker B: Let's have you laptop.\nSpeaker C: They were just fine.\nSpeaker E: The network was okay.\nSpeaker E: Everything was also available there.\nSpeaker F: More time might have improved the quality of work.\nSpeaker E: It's now half past four, half past three.\nSpeaker B: It's just the half hours between the work alone.\nSpeaker B: Then it pops up, pop up screen.\nSpeaker B: Maybe.\nSpeaker F: More time during the work phases.\nSpeaker F: I just got my warning for the last five minutes.\nSpeaker F: I'll move on to my last slide.\nSpeaker F: We managed to do that quickly.\nSpeaker F: We had to do it because we have to have design and that is within the budget.\nSpeaker F: We evaluated.\nSpeaker F: We evaluated the project.\nSpeaker F: Everybody is very happy at least I am.\nSpeaker F: I have to write the final report now.\nSpeaker F: Thank you for your cooperation.\nSpeaker F: Very nice day.\nSpeaker B: Thank you.\nSpeaker B: Do we get another email?\nSpeaker F: You are fine.\nSpeaker F: Maybe you can try to make a screenshot of this.\nSpeaker F: You can save it.\nSpeaker B: It is not a picture.\nSpeaker C: You can save it.\nSpeaker F: Please put it in the project folder.\nSpeaker D: Do you find it as a JPEG?\nSpeaker B: No.\nSpeaker B: I have a close meeting.\nSpeaker F: I hope to see you soon.\nSpeaker F: In the five minutes.\nSpeaker C: Happy celebration.\nSpeaker C: Happy celebration.\nSpeaker C: It is not too big.\nSpeaker D: This one?\nSpeaker B: This one?\nSpeaker D: This one is too big.\nSpeaker B: You can only save it in my document.\nSpeaker B: Oh my god.\nSpeaker B: It is very expensive.\nSpeaker B: Three.\nSpeaker B: I have a question.\nSpeaker D: Why can't we stay here?\nSpeaker D: I will take it to maybe up.\nNone: Consider that.\nNone: What might be a good trick I've never got.\nNone: You don't need to prisoners.\n"}, {"meeting_id": "ES2016b", "summary": "First, the project manager briefly reviewed the last meeting. According to market research, Marketing suggested adopting modern appearance, reducing buttons, increasing positioning and voice recognition functions. From the perspective of technical function design, User Interface emphasized that the user should be the centre, reduce the buttons, focus on modifying the shape of the remote control, and do not waste the budget to design the universal remote controller. Industrial Designer had come up with the option of cheap and sustainable batteries, infrared lights and voice recognition technology. Industrial Designer also offered mouse-like scrolling and clicking functions, as well as specific remote control shapes.", "dialogue": "Speaker E: Okay.\nSpeaker E: Thanks for coming to this meeting.\nSpeaker E: How are we doing on our remote?\nSpeaker E: We have some ideas and some ideas for what people want.\nSpeaker E: Well, yeah.\nSpeaker A: I was asked to do it anyway.\nSpeaker A: I think.\nSpeaker E: All right.\nSpeaker E: Let's just go over the agenda real quick for this meeting.\nSpeaker E: See if I can't get things moving along here a little bit better.\nSpeaker E: I'll go over what we went over last time, which shouldn't take long.\nSpeaker E: And I believe each of you have a presentation.\nSpeaker E: I've sent a couple more requirements for our remote, but what they want.\nSpeaker E: Then we can come to a conclusion on what we want the remote to do and how it's going to do it, hopefully.\nSpeaker E: And then we'll have the closing, which we'll have 40 minutes for.\nSpeaker E: Let's see.\nSpeaker E: The last meeting, we went over who was responsible for what?\nSpeaker E: I'm responsible for leading the meetings, keeping the notes and coming up with the final presentation.\nSpeaker E: Corinne is our marketing expert.\nSpeaker E: She's going to figure out what the consumer wants.\nSpeaker E: Ryan is our user interface designer.\nSpeaker E: And Manuel is the industrial designer.\nSpeaker E: So you're going to come up with the ideas, Ryan, and you're going to pick them apart.\nSpeaker E: We decided our remote.\nSpeaker E: We wanted to be a universal remote that everyone would want.\nSpeaker E: We want to be modern.\nSpeaker E: It's fun, different.\nSpeaker E: It needs to be sturdy, easy to find.\nSpeaker E: So we're going to have that locator function.\nSpeaker E: And we want to be different.\nSpeaker E: And then we went over a couple of different ideas, ball-shaped phone, the keyboard shape.\nSpeaker E: We decided that it should probably be one-handed, something we can use with one-hand.\nSpeaker E: And that was our last meeting.\nSpeaker E: So why don't each of you have a presentation?\nSpeaker E: I'll hand it off to you.\nSpeaker E: Does anyone want to go first so we can maybe see what the people want?\nSpeaker B: What was it?\nSpeaker D: What was it?\nSpeaker B: How do I get it?\nSpeaker E: Slice to go to the next one.\nSpeaker B: This is my report, which is going to be based pretty much on a survey that I've sent.\nSpeaker B: Sorry, I actually need to see something else on my screen.\nSpeaker B: Hit F8 again.\nSpeaker E: Do you want to be on both screens?\nSpeaker B: No, I want something else.\nSpeaker E: Is that possible?\nSpeaker E: I don't know.\nSpeaker E: I'm not sure you can maybe minimize that screen and then have it in both up at the same time.\nSpeaker B: First of all, the method that I used was by doing some marketing research on some interviews that were conducted and then some internet research.\nSpeaker B: I was sent a report that was, I think, there were 100 remote users that they interviewed.\nSpeaker B: I will show you some of the results from that, which I think will be helpful.\nSpeaker B: Here are some of the findings.\nSpeaker B: They said that the users dislike the look and feel of their current remote controls.\nSpeaker B: 75% of the users find their remote controls to be ugly, which is a fairly significant number, I would say.\nSpeaker B: 80% of the users would be willing to spend more money if they could get a remote that would look fancy.\nSpeaker B: I think that earlier we were on to something when we were talking about having a modern, cool look.\nSpeaker B: They say the current remote controls do not match well the operating behavior of the user.\nSpeaker B: 75% of the users said they zap a lot.\nSpeaker B: If anyone could clarify what that means.\nSpeaker B: Does that just mean like changing the channel?\nSpeaker B: 50% of the users say that they're only using 10% of the buttons on the remote control.\nSpeaker B: There was something else.\nSpeaker B: They kept track of the frequency per hour and using certain buttons.\nSpeaker B: Some of them, it looks like, barely needs to be included at all.\nSpeaker B: Of course, channel selection is used the most frequently.\nSpeaker B: Then teletext was the next volume and then power.\nSpeaker B: Then audio settings and screen settings and channel settings were practically never used.\nSpeaker B: I think we could definitely eliminate or somehow combine a lot of the functions into one button.\nSpeaker B: The biggest user frustrations, as we said 50% of people find that their remotes are lost somewhere.\nSpeaker B: I think a tracking device of some sort would be a good idea.\nSpeaker B: They said 34% said it takes too much time to learn how to use a new remote.\nSpeaker B: 26% said that the controls are bad for RSI.\nSpeaker A: What is it?\nSpeaker B: Just a capacitive strain injury.\nSpeaker B: So base, as far as speech recognition goes, the younger group looks like they're all for it from the 15-25 age group.\nSpeaker B: Over 90% said they would pay more.\nSpeaker B: It went down incrementally.\nSpeaker B: The older they get, it looks like the less willing they are to pay.\nSpeaker B: Maybe we could discuss this and think, decide if we think it's worth investing in this, at least if we're targeting the younger groups.\nSpeaker B: In conclusion, some things that I drew from this are that I think we were correct.\nSpeaker B: We definitely need to focus on a new modern appearance, since so many people seem to be concerned about the ugliness of the remote control.\nSpeaker B: A multifunctional remote could be a good thing to explore, so you only have one rather than five different remotes sitting all over your room.\nSpeaker B: We need to simplify the remote and reduce the number of buttons, get rid of the ones that don't seem to be serving much purpose.\nSpeaker B: And then lastly, I thought that maybe we could discuss the idea of speech recognition.\nSpeaker B: And that's it.\nSpeaker E: Okay, very nice. Now we actually have some ideas of what people want, what we should focus on.\nSpeaker B: Let's make a look at that real quick.\nSpeaker A: Having just listened, I'll draw on some of the things as well.\nSpeaker A: I'm just going to approach the technical functions design.\nSpeaker A: The method I used was to explore the technical functions of remote control.\nSpeaker A: I had some things sent to me not very much to look at similar devices, defined things from them, and then personal preferences that I would suggest.\nSpeaker A: We discussed the universal one.\nSpeaker A: It's just been brought up again then, but I think a universal remote control is actually quite a difficult object to design and possibly within maybe the budget that we do it because you need to know all the specifications of all the electronic companies.\nSpeaker A: I'm not sure if you ever come across a universal remote control yourself, but you have to do a nightmare to use, you have to set them, reset them to everything, and that would only add buttons, whereas I think there is to take away buttons.\nSpeaker A: I think it better maybe to concentrate on maybe it's a universal one for TVs, or maybe just one that you could design and then different people manufacturers could use it to set their specifications.\nSpeaker A: If the aim is to get something that's unique in design.\nSpeaker A: Okay, here's just two pictures of remote controls.\nSpeaker A: There's just simple TV remote controls, but one is user-centered, that is the one on the left.\nSpeaker A: You need to straight away see there's less buttons, and the other one is some engineered centered where that's more specified for the sort of elaborate piece of equipment that's trying to control.\nSpeaker A: I'd ask which appeals more to the product that we want, and on what defines the standard market research and stuff, we're probably looking at something that should be user-centered, fewer buttons, simpler to use, and if 10% is hidden away, if 10% is what's used, maybe the other 50%, the buttons that I use very rarely, like programming, they could be hidden, maybe some remote controls that you might have come across, and maybe a little flip thing where they're hidden away, and the main buttons, the ones you come across.\nSpeaker A: And finally, I'm sorry, covered that, our product I think should be user interface orientated.\nSpeaker A: Like I said, a concentrate on TV control, universal mode, might be too complex.\nSpeaker A: And as what the major finders and market research has said, it's the image and the appearance that people dislike, so that we should concentrate on something that would set a trend, and then change the fashion of a remote controls.\nSpeaker D: And that's it.\nSpeaker E: What was your last conclusion on that one? Focus on the image of it.\nSpeaker A: Good, good.\nNone: Okay.\nSpeaker D: I'll talk to you about the working design.\nSpeaker D: The interior is basically what dictates the design, both the interface and the outer appearance, because this is all the stuff that needs to go in there.\nSpeaker D: Unfortunately, the people who are supposed to do this little presentation for me, obviously, were too drunk to actually accomplish it.\nSpeaker D: I'm going to do lots of the stuff on the board.\nSpeaker D: This is the basic premise of a remote control. The basic function is to send messages to another system. Okay, so much is clear.\nSpeaker D: And energy source feeds an integrated circuit chip that can compose messages.\nSpeaker D: Often in the form of infrared bits, this is the most mostly used. There's also some sound systems, but infrared is the better, the more used system parts are cheaper as well.\nSpeaker D: A user interface controls the chip, and accordingly, the messages. This is where my people screwed up, basically. So I'm going to explain that on the board, other.\nSpeaker D: What we have is different components that obviously need to go in there, we'll start with an energy source.\nSpeaker D: It's usually a battery, right, since it's not feasible to add a cable to that.\nSpeaker D: This energy source, of course, is connected to the user interface itself, which can be buttons, whatever, which in fact controls.\nSpeaker D: Controls a chip.\nSpeaker D: This is the user interface, where we have the chip.\nSpeaker D: The way this goes normally is that this chip then controls an infrared lamp that sends out the signal.\nSpeaker D: Of course, the signal differs accordingly, depending on what the chip tells the infrared lamp, and of course that's controlled.\nSpeaker D: The chip itself is controlled by the user interface.\nSpeaker D: The way you normally do it is that you add a little device such as a lamp to the whole thing as well, so that you know that it's working.\nSpeaker D: Basically, you press something, you get a response, which is also comparatively important on one of those devices.\nSpeaker D: Now, what we're talking about here are what I think should be discussed are these two components, mostly the energy source for one thing can be altered.\nSpeaker D: What we probably cannot alter is, of course, the infrared, the sending device, basically the infrared lamp.\nSpeaker D: We cannot change the chip, which controls the infrared lamp, right, these two are components that we have to use, and these are dictated by the whole function of the whole thing.\nSpeaker D: The lamp can be put on to the device.\nSpeaker D: It doesn't have to be there.\nSpeaker D: This can be discussed as well.\nSpeaker D: The user interface, that's something we can also discuss.\nSpeaker D: As we've heard, speech recognition is the hype, obviously, in the moment.\nSpeaker D: Speech recognition interface, we don't know that, or if we just do the usual button thing, or we have a touchpad or something like that, that's something we can discuss.\nSpeaker D: Of course, the energy source, batteries, solar cells, who knows.\nSpeaker D: Of course, it's always a question whether these components are, in fact, available cheap enough, developed enough, but that's, I suppose, rather up to marketing, not to you.\nSpeaker E: To the industry of design.\nSpeaker E: The more complex we make it, of course, the more expensive it's going to be.\nSpeaker E: People have said that they would, well, younger generations of people said that they would pay more for a speech recognition remote.\nSpeaker E: So, possibly, it might be worth the investment.\nSpeaker A: I think speech recognition was one of those things where it has to be really good for them to work.\nSpeaker A: Sometimes you find yourself saying things over and over and over and do something you phone.\nSpeaker A: I agree.\nSpeaker A: Different dialects, I suppose.\nSpeaker E: Myself, I find when you, when there's something that's speech recognition, like you call on the phone and you try to change your telephone or power or something, sometimes they have a speech recognition on the other hand and you say one for this and you find your, like you said, saying the same thing over and over and over.\nSpeaker E: I find myself, especially if I'm in a crowd of people, looking really silly.\nSpeaker A: I don't know if, what do you want to keep saying stuff?\nSpeaker A: If you're watching stuff, if you're watching something, what do you want to do?\nSpeaker E: Volume up, volume down, change the channel, you know, channel up, channel down.\nSpeaker E: I don't know.\nSpeaker B: Another thing about these figures is 91% of the youngest age group said they'd do it, but probably a lot of them, that's actually their parents money.\nSpeaker B: Like, I don't know if they would actually go out and purchase this themselves, a 15 year old, you know.\nSpeaker A: As well, it'd be the gimmick factor for the younger people.\nSpeaker E: It's a gimmick factor that they like at first.\nSpeaker A: Gets old here.\nSpeaker E: Okay.\nSpeaker E: Let's see here.\nSpeaker E: Yeah, I guess so.\nSpeaker E: Now, I was sent a couple of things to modify our new requirements.\nSpeaker E: The remote's only going to be for the television, which is good because we already decided your research showed that not only is the universal remote more complicated, it's more costly, and your research showed that, you know, most of the people don't even use it.\nSpeaker E: You said 50% of the people only use 10% of the buttons, so if we remove the universal remote, then that solves that problem.\nSpeaker E: No teletext.\nSpeaker E: So we don't have to worry about that.\nSpeaker E: But we do have to use the company wants us to incorporate the corporate color and our slogan, which is we put the fashion in electronics.\nSpeaker E: And our corporate colors are gray and yellow.\nSpeaker E: We could probably get away with black too.\nSpeaker E: Those are the three new requirements that I was told we need to use.\nSpeaker E: From all three of your presentations, I think that we were on the right track a lot in our last meeting.\nSpeaker E: We want something that looks good. We want something that's simple. We want something that you can find easily.\nSpeaker E: And the speech recognition, I guess, is kind of give or take. It's going to cost more.\nSpeaker E: The younger people say that they like it, but it's probably, I would say, probably not worth the investment at this point in time.\nSpeaker E: So maybe we should just do away with speech recognition.\nSpeaker E: And that way we can focus on our form.\nSpeaker A: I did have a thought about the tracking thing is that it came with maybe a holder.\nSpeaker A: A whole store, if you want to call it, you should put it back in your mode.\nSpeaker A: If you don't put it back in, you press something like that.\nSpeaker A: It just sends out a bit to find where it is.\nSpeaker E: That would be good if we were going with our ball.\nSpeaker E: Or with any form that would be good. That could be the charger for, you know, we could use rechargeable batteries in the remote.\nSpeaker E: And that would be a solar or, you know, however you want to go about it, the holder could also be the charging unit with the locator button.\nSpeaker E: And if it were the ball, you'd no longer have to have a flat space on it.\nSpeaker E: We still have the how to hold on to it.\nSpeaker D: Yeah, you put it on the couch table while you're watching.\nNone: It's going to roll away.\nSpeaker E: So I guess after the meeting, we'll have some questionnaires and some summaries for what's going on.\nSpeaker E: Then we'll take lunch.\nSpeaker E: We can come back and work on our individual work.\nSpeaker E: I'll do the minutes.\nSpeaker E: Let's see.\nSpeaker E: It looks like you're already on your way for working with the components.\nSpeaker E: What chips we need and what, you know, how to power it and whatnot.\nSpeaker E: User interface concept. We want it to be something simple.\nSpeaker E: Minimal number of buttons.\nSpeaker E: I guess our main thing that we should focus on until then is probably deciding on a certain look.\nSpeaker D: Yeah.\nSpeaker E: I think we should be able to work on that.\nSpeaker A: I think we should be able to work on that.\nSpeaker A: You have time to work on that.\nSpeaker A: You have enough time between our meetings to get everything done that you need to get done.\nSpeaker A: I think it still has a bit of variation.\nSpeaker A: On a grad time group, but maybe not as silly as a boy.\nSpeaker E: Should we maybe shoot for a specific target group?\nSpeaker E: That way, if we were shooting for young guys, then it's a certain look to the remote or girls or older people.\nSpeaker E: You think that would help us find a specific form that we would want to pick out?\nSpeaker A: I don't know if that might come up from the train watching.\nSpeaker A: Do you find anything for now?\nSpeaker B: Yeah, not yet, but maybe by the next meeting we'll have something to do on that.\nSpeaker E: We know that the remote is going to have to be, we'll just say we've got a color scheme for it.\nSpeaker E: Yellow, it's nice and bright with the buttons being gray or black and are slogan pasted somewhere on it.\nSpeaker E: The bottom.\nSpeaker E: Maybe we could sketch.\nSpeaker E: Basically, all of our typical remote are just a rectangular sort.\nSpeaker E: It's more, of course, this will look like a bone.\nSpeaker E: Go ahead and erase this.\nSpeaker E: I hope everyone memorized that.\nSpeaker A: I mean, Matt, to look at the shape and trend of things like mobile phones and shapes that they've been going on.\nSpeaker A: They've gone from big, big block things, which is what the remote control is.\nSpeaker E: Everybody's got a mobile phone, except for me now.\nSpeaker E: But they are all mobile phones.\nSpeaker E: They've taken a turn to where they're really small, which may defeat our purpose for being able to locate our remote all the time.\nSpeaker E: But then again, you don't want the first mobile phone that was this big and you had a handbag to carry it around in.\nSpeaker E: So we've got basically what the remote looks like now is what we're stuck with.\nSpeaker E: Maybe we could go with a square or something.\nSpeaker E: You know, with minimal number of buttons.\nSpeaker E: I guess you've got one through nine typing in your channels.\nSpeaker E: We've got volume up and down, channel up and down.\nSpeaker E: Power.\nSpeaker E: Power.\nSpeaker E: Usually at the top.\nSpeaker E: A mute.\nSpeaker E: That's pretty much all you need, I think.\nSpeaker E: A menu button, maybe.\nSpeaker E: So, you know, if we want the remote to do other things like, I guess, the TV to be able to change the tint and the color and all those kinds of things that are built into TVs, we just have that under one standard menu button where you go and press the menu button, scroll up and down to select it.\nSpeaker D: Talking of which, maybe a scrolling function is not too bad.\nSpeaker D: Okay.\nSpeaker D: So, talk about maybe look at that from the side there, maybe, which is technically the easiest option would probably be like a scrolling wheel like this.\nSpeaker E: Like a wheel on your mouse.\nSpeaker D: Yeah, sort of like that.\nSpeaker D: You can even, if we're coming from mice, we can add a click function where you, in order to verify the information, you just press it down.\nSpeaker D: It's a good idea.\nSpeaker D: Also, when it comes to the ergonomics of the whole thing, if you want to make it square for the looks of it, then maybe make it more comfortable to hold the whole thing.\nSpeaker D: So, I'm going to go bulge down here, which maybe from an engineering point, this could be holding the batteries and so on.\nSpeaker D: I know you know what I'm getting at here.\nSpeaker D: Yeah, it's good.\nSpeaker D: Look at it from the side.\nSpeaker D: You like that.\nSpeaker D: Then you hold it in your hand like this and maybe you bring the buttons nearer to the, or the more important buttons nearer to the thumb side.\nSpeaker A: Yeah, right.\nSpeaker A: Maybe the buttons could like decrease the bigger ones.\nSpeaker A: You know, that is actually bigger in size than the ones that are less important, smaller here.\nSpeaker D: You can also have it maybe talking about mobile phones again, writing open.\nNone: Yeah.\nSpeaker D: With the lesser used functions, this part slides into that part.\nSpeaker D: Exactly.\nSpeaker A: Just been sketching something, just like on a sort of side view.\nSpeaker A: Something that, you know, sort of fit in the hand better.\nSpeaker A: So you know, you just sort of have your hand.\nSpeaker D: Oh, it's just thinking this, this of course causes a problem for left or right handed people actually.\nSpeaker A: Yeah, it's also true.\nSpeaker A: That, yeah, instantly, so it's always going through the right hand first.\nSpeaker A: Right.\nSpeaker A: I mean, it's the older.\nSpeaker A: Yeah.\nSpeaker E: Is everyone who in here is right handed, left handed?\nSpeaker C: Right.\nSpeaker E: You guys all right handed?\nSpeaker E: Yeah.\nSpeaker E: So I'm left handed, but I, so I can say that most things are designed for right handed people.\nSpeaker D: That's right.\nSpeaker E: Like the written language.\nSpeaker E: Or English.\nSpeaker B: Maybe you could buy like a special left handed version of the remote.\nSpeaker E: Maybe.\nSpeaker E: Oh, special order.\nSpeaker E: I would say, I mean, it should be probably designed for a right handed person.\nSpeaker E: But that's good.\nSpeaker E: That's, I think we've got kind of a good plan there.\nSpeaker E: At least for what buttons we're going to use.\nSpeaker E: So we've got, I like to scroll, scroll action and the.\nSpeaker A: I mean, I just, just thinking maybe.\nSpeaker A: If it was a.\nSpeaker A: Circular.\nSpeaker A: Um.\nSpeaker A: This sort of, that sort of, it's sort of an early idea we had, but the way that was more like that.\nSpeaker A: Whereas this is just a one handed thing that you sort of almost wrapped around the thumb.\nSpeaker A: So you'd have your hand into the thumb there.\nSpeaker A: You'd have all the buttons sort of around here.\nSpeaker A: And that I think could work.\nSpeaker A: Yeah.\nSpeaker A: On any hand.\nSpeaker A: Kind of like holding a, if you just had it like right around there, left, all right.\nSpeaker D: Minimizes its size wise.\nSpeaker A: Minimizes its size.\nSpeaker A: You know, really quite small.\nSpeaker D: And then there's maybe.\nSpeaker D: Another point of making it a two and one kind of thing.\nSpeaker D: If we have a smaller, a smaller device that actually fits into the big one.\nSpeaker D: You have full functions in here and just a couple of functions in there.\nSpeaker D: Yeah.\nSpeaker D: Like your zap, zapping device.\nSpeaker D: It's just channels up down volume.\nSpeaker D: And on off.\nSpeaker D: That's it.\nSpeaker D: Yeah, maybe.\nSpeaker E: So two remotes and one.\nSpeaker E: But then would be easier to lose one for the other.\nSpeaker E: That's right.\nSpeaker D: The big one has to be sufficiently big and perhaps heavy so you can't move it around.\nSpeaker D: Make it a piece of furniture.\nSpeaker D: It actually is your coffee table.\nSpeaker D: Yes.\nSpeaker D: Yes, there you go.\nSpeaker D: State you.\nSpeaker E: Okay.\nSpeaker E: Well, we need to finish up here.\nSpeaker E: So for the next.\nSpeaker E: So for the next before the next meeting.\nSpeaker E: Shall we work on your watching trends to see what kind of new information we've got going on.\nSpeaker E: Manual, you're going to work on the components for what we're going to use.\nSpeaker E: Let's go with our rounded.\nSpeaker E: Kind of fit in your hand.\nSpeaker E: Let's explore the possibility of having the two piece.\nSpeaker E: So one for the quick zapping and possibly one with just that rounded part that fits into your hand.\nSpeaker E: And Ryan, you work on how well they'll work with the user.\nSpeaker E: So both of those concepts.\nSpeaker E: Okay, I think we made some good progress here.\nSpeaker E: We'll see you in, I don't know, I guess we'll take lunch in about half an hour.\nSpeaker E: Or I guess we'll be taking lunch shortly and then about half an hour after that we'll probably have another meeting.\nSpeaker C: All right.\nNone: Okay.\n"}, {"meeting_id": "TS3004b", "summary": "This meeting was about the functional design of the remote control. Firstly, Marketing gave a presentation on functional requirements. Group decided to focus on the fancy and fashionable look, usability, and different colors. Next, User Interface gave a presentation on the technical function design. Also, the group discussed this topic, and they decided to design the menu buttons of the remote similar to the mobile phone. Then, Industrial Designer gave a presentation on the working design. Group mates discussed deciding on the use of LED light on the buttons to indicate the transmitting of the Morse code when pressing the button. They also decided to use a more intelligent chip than the standard one when the circuit was closed, it would produce the pattern. For the age group, they would target the age group below forty since it was a young market.", "dialogue": "Speaker D: I just forgot their names.\nSpeaker D: So you're sorry, you just forgot them all.\nSpeaker D: Okay.\nSpeaker D: So do you know them or?\nSpeaker E: The names.\nNone: Yeah.\nSpeaker E: For for master.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker D: I know.\nSpeaker D: But do you know your name?\nSpeaker E: Oh, dumb one.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker D: And what's your name?\nSpeaker D: Paul.\nSpeaker D: W-I-E-S.\nSpeaker D: Z.\nSpeaker D: Z.\nSpeaker D: E-R.\nSpeaker C: Okay.\nSpeaker C: What's your name?\nSpeaker C: Martine.\nSpeaker C: Yeah, but just your name.\nSpeaker C: What?\nSpeaker C: Just your name.\nSpeaker C: Uh, I'll be, uh, baby.\nSpeaker C: E-F-A.\nSpeaker C: Okay.\nSpeaker C: I was a little short on time.\nSpeaker C: Yeah, me too.\nSpeaker D: Yeah.\nSpeaker D: Same.\nSpeaker D: No, no, no, no, just for first my...\nSpeaker D: Oh, sorry.\nSpeaker B: Oh, let's see what you put in here.\nSpeaker D: So, let's have a look.\nSpeaker D: We have 40 minutes on it.\nSpeaker D: I don't know.\nSpeaker D: I can't...\nNone: Okay.\nSpeaker D: Perfect.\nSpeaker D: So, we have...\nSpeaker D: Oh, we'll see.\nSpeaker D: So, so we have 40 minutes for this, uh, for this second meeting.\nSpeaker D: And we have to make, uh, sure that we're going...\nSpeaker D: that we are...\nSpeaker D: that we are...\nSpeaker D: that we know where they're going to make, uh, what the product is going to, like, look like.\nSpeaker D: Uh, first I have to note of the last meeting, so I show them to you.\nSpeaker D: No, sorry.\nSpeaker D: I'll just escape this one.\nSpeaker D: How do I escape this?\nNone: What?\nSpeaker D: How do I escape this?\nSpeaker D: This is the presentation.\nSpeaker D: Uh, oh, let's see.\nSpeaker D: And show, sorry.\nSpeaker D: Okay, so...\nSpeaker D: have a look at this one.\nSpeaker D: Okay, so the points we had last meeting was, um, should be a universal remote control.\nSpeaker D: Now that I just got an email from the...\nSpeaker D: from the personal coach, and it should be a TV remote control only.\nSpeaker D: So, have you changed that part?\nSpeaker D: Okay.\nSpeaker D: Um, so, yeah, it still has to be, uh, a remote control for kids and all of that.\nSpeaker D: It's still the same.\nSpeaker D: Um...\nSpeaker D: Now these points, we have to...\nSpeaker D: look at, you all know them.\nSpeaker D: But, uh, there's another point, um, uh, the main, uh, people of interest of this company are 40 plus people, so they're old and not younger people.\nSpeaker D: You have to look at that as well.\nSpeaker D: Especially old people, maybe bigger buttons or something like that.\nSpeaker D: Yeah, okay.\nNone: So...\nSpeaker D: So, yeah, that's it.\nSpeaker D: So, you can do your presentation for...\nSpeaker E: Which one?\nSpeaker E: Okay.\nSpeaker D: Oh, it doesn't matter.\nSpeaker D: Okay.\nSpeaker B: Uh, thanks for all the time.\nNone: Okay.\nSpeaker E: Well, my name is Jens Doman, but we're in group.\nSpeaker E: I will start it.\nNone: Wait.\nSpeaker E: Um, I've...\nSpeaker E: I used a marketing report on the site.\nSpeaker E: Uh, I think you've read it too.\nSpeaker E: Uh, and, uh, I'm very more, I served as in the other side.\nSpeaker D: I didn't read it, so it's not for me.\nSpeaker D: I didn't get it.\nSpeaker D: Oh, okay, I was the only one who got it.\nSpeaker E: Okay, it was, uh, um...\nSpeaker E: A report about an experiment with a lot of users, and they had a lot of findings in their report with statistical, uh, uh, I think, uh, with statistical proof.\nSpeaker E: So, I, um, I had three pages with findings, and a lot of, a lot of findings, so we can use this to create our own remote control.\nSpeaker E: Uh, 75% of the users find, uh, most remote controls ugly.\nSpeaker E: Yeah, I think, uh, that's a lot, so we have to make a beautiful remote control.\nSpeaker E: Uh, 80% of the users will spend more money when, uh, a remote control will, uh, look fancy.\nSpeaker E: I think this fits, uh, at the, uh, what, what Michil said about older people.\nSpeaker E: Older people will, uh, spend more money, uh, for, uh, something, uh, what's good, because younger people are more critical, uh, about, uh, where they spend their money, money at.\nSpeaker E: Uh, 75% of the users said they zap a lot.\nSpeaker E: Well, okay. That's, uh, normal, I think.\nSpeaker E: Uh, we have to make good zap buttons, but this one of our requirements...\nSpeaker E: The last point is quite interesting.\nSpeaker E: Yeah, 50% of users say they only use 10% of the buttons.\nSpeaker E: Uh, but, uh, my channel already said it.\nSpeaker E: Yeah.\nSpeaker E: And, uh, maybe our, uh, full open system is, is a good one, but I don't think it's, uh...\nSpeaker D: Yeah, we should have the 10% on the, on the top.\nSpeaker D: Then you're...\nSpeaker D: Yeah, the 10% on the top.\nSpeaker E: Yeah.\nSpeaker E: That's a good one.\nSpeaker E: Uh, uh, page two, remote controls are often lost somewhere in the room.\nSpeaker E: That's exactly what we said about, uh, maybe a home station for, uh, for it, uh, to, uh, recharge the batteries or something.\nSpeaker E: Uh, I thought more, maybe we could make, uh, a clap system, so when you...\nSpeaker E: Clap your hands, it will beep or something.\nSpeaker E: Uh, you must find it, uh, quickly.\nSpeaker B: Maybe just a button on the home station, so...\nSpeaker B: Remote control beeps when you click that button on the home station.\nSpeaker E: Okay, yeah.\nSpeaker E: And we can combine that.\nSpeaker E: Uh, it takes too much time to learn how to use a new remote control.\nSpeaker E: Uh, I think we must, uh, take a look at this.\nSpeaker E: It's only, uh, 34... of, uh, 34% but it's, uh, a tough one.\nSpeaker E: Because if we make a whole new product or own style, we...\nSpeaker E: Uh, this is our difficult, uh, uh, difficulty, I think.\nSpeaker E: Uh, next, remote controls are bad for Air Z.\nSpeaker E: Yeah, but only if they zap a lot and they watch over five hours TV or something.\nSpeaker E: I don't... we haven't, uh, we mustn't look too much at the last point.\nSpeaker E: Okay, last page.\nSpeaker E: Uh, the last, uh, experiment, uh, was about gadgets, like, uh, speech recognition.\nSpeaker E: We didn't, uh, think about that already.\nSpeaker E: And, uh, an LCD on the remote control.\nSpeaker E: We already thought about that.\nSpeaker E: Uh, uh, they finally had a conclusion that younger people, uh, uh, under an age of 45 are more interested in new features.\nSpeaker E: And, but they are more critical.\nSpeaker E: And older people want to spend more money.\nSpeaker E: But, uh, they, uh, they don't want to, uh, have a lot of new features because they're in their old, uh, thinking way.\nSpeaker E: And they want to keep the old, uh, things, the old things.\nSpeaker E: I mean, I think.\nSpeaker D: So we have a new, uh, a new...\nSpeaker E: But you already said that the company was about, uh, 40 plus, uh, clients.\nSpeaker D: Yeah.\nSpeaker D: So we just can skip the LCD on the remote control.\nSpeaker E: I think we can speak, uh, we can skip speech recognition, uh, directly because it's not reachable for 25 euros.\nSpeaker E: Um, then I have my personal, uh, preference.\nSpeaker E: Okay, that's, uh, not very good because I talked about telephishing, the D-place area and PCR, uh, the question about, but it's already out of the question.\nSpeaker E: Um, my point is, well, uh, if we may, uh, make, uh, remote, uh, control for only telephishing, I think it's hard to, uh, sell it for 25 euros, but it's the exercise.\nSpeaker D: And because it's too expensive.\nSpeaker D: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Only, only for telephishing, uh, on the market you can now buy for 20 euros, so remote control for everything.\nSpeaker E: And we only make it for telephishing, so we might be, uh, must make it, make it, uh, very special.\nSpeaker E: Yeah, but it's a good usability.\nSpeaker B: So you can use it.\nSpeaker E: Okay, I talked about the home station.\nSpeaker E: Uh, it must be simple because, uh, the, the elderly people needs to use it.\nSpeaker E: And I found a motto and we put the fashion in electronics.\nSpeaker E: And that's, uh, the motto we are, uh, referenced to, uh, for our, uh, our style.\nSpeaker E: We, we have to make, uh, a new product.\nSpeaker E: We have to be, um, yeah.\nSpeaker E: One of a kind.\nSpeaker E: So it has to look, uh, you know, the companies about our, uh, their own fashion, their own\nSpeaker D: style. But all people are not looking for that.\nSpeaker B: Oh, I think not really.\nSpeaker B: Yeah.\nSpeaker C: I think the main thing is usability.\nSpeaker C: That's when we can, uh, make it a special product.\nSpeaker C: Keep it simple.\nSpeaker B: We also have to stand out because there are already, like you said, so many controls out that support a lot of stuff.\nSpeaker B: But we have to make sure that we're better in usability and stand out by just the looks of it.\nSpeaker B: So make it just a different color or a different shape.\nSpeaker B: So, okay.\nSpeaker E: Okay.\nSpeaker E: This was my presentation.\nSpeaker E: Okay.\nSpeaker E: I don't know.\nSpeaker D: So Paul, you can do the next one.\nSpeaker D: Yeah.\nSpeaker D: It's on the, on the, uh, net, net, uh, thing.\nSpeaker D: Okay.\nSpeaker B: Okay.\nSpeaker B: Technical functions design.\nSpeaker B: Okay.\nSpeaker B: Well, so we have, uh, broad audience.\nSpeaker B: That isn't true anymore, but, um, we have elderly people, so we need to keep it simple.\nSpeaker B: Uh, the way I want to keep it simple is to use the send, uh, standards.\nSpeaker B: So, uh, standardize the methods, like, uh, on all, uh, remote controls.\nSpeaker B: Not too full, like, uh, you already said, only 10% is being used.\nSpeaker D: So what do you think of only the numbers and the on off button?\nSpeaker B: Yeah, I have some, uh, the next page, but yes, uh, basic functions like numbers and that's sort of stuff.\nSpeaker B: Um, so the option is that we put on there should be easy to use and cause you have more room and for elderly people, big buttons, uh, an icon on it or text on it, so it's very clear what that button does.\nSpeaker B: So it doesn't take much time to figure out how to use it.\nSpeaker B: Um, a way to find out, uh, what people use.\nSpeaker B: So maybe just use, uh, questionnaires on the internet or just ask some random people, elderly people what they use, what they want on, uh, remote control to find out that there was already in a one done, well, uh, functions I could think of, uh, volume, channels, the basic, I call it the numpath, just one, two, three, et cetera.\nSpeaker B: Uh, text service options, um, basic on off and I found in, uh, could I think of favorites?\nSpeaker B: I always look up the same, um, pages on text and I always have to click the number.\nSpeaker B: Okay.\nSpeaker B: But if you could make a new option that you just have to press one button and you get on your...\nSpeaker C: Yeah, but then you have to remember what favorite is what channel, so you might as well remember the number.\nSpeaker C: Or not.\nSpeaker D: No, maybe it's too complicated, I'm not sure.\nSpeaker B: Well, it's a good idea.\nSpeaker B: It was just a thought, so I'm, I would find it handy, I think, when you just press one button and you get on six, six, six.\nSpeaker C: Yeah, okay, but, um, how can you remember what, uh, channel, uh, connects to what favorite\nSpeaker B: button. Uh, but I was reading on the page, uh, a writing field just sends commands, basic commands to, uh, the television.\nSpeaker B: So switch to channel six.\nSpeaker B: Yeah.\nSpeaker B: So a button six says six, and if you make favorites, it can say six, six, six in a row.\nSpeaker B: Just numbers.\nSpeaker C: Yeah, okay, but for a user to, to remember, if I press that button, it goes to that channel.\nSpeaker B: Well, if you set a favorite.\nSpeaker C: Yeah, but isn't it hard to remember?\nSpeaker C: Like favorite one and...\nSpeaker D: Uh, maybe, for me, if I use my telephone, I never use those buttons to, to, to caution.\nSpeaker D: No, no, neither do I.\nSpeaker B: So, and...\nSpeaker B: Okay, now, maybe not.\nSpeaker D: If I don't do it, maybe old people...\nSpeaker B: Nah, maybe old people.\nSpeaker B: Yeah, I know.\nSpeaker B: They don't like new features, so maybe not.\nSpeaker B: Um, well, play pause.\nSpeaker B: I don't know if it's usable.\nSpeaker B: It's not, it's still not, it's not anymore.\nSpeaker B: Not anymore for TV.\nSpeaker B: Right, on and off.\nSpeaker B: I don't know, did I miss any other buttons?\nSpeaker B: Basic buttons?\nSpeaker B: I couldn't think of any other, especially not for TV.\nSpeaker D: Just on the front as well.\nSpeaker D: No, that's the only thing you need.\nSpeaker C: The next and previous, previous I know, but next channel?\nSpeaker B: Just a channel, um, what I mean is that...\nSpeaker B: Like a web browser?\nSpeaker C: Oh, okay.\nSpeaker C: Just a channel.\nSpeaker C: Um, there's also a button to, uh, go to the channel you've been before.\nSpeaker C: Like a web browser back button.\nSpeaker B: Um, I don't have...\nSpeaker D: I know what it is, but I think it's ultra difficult for me.\nSpeaker B: Yeah, I don't think you use that.\nSpeaker B: Only when you want to go to, yeah, just use, uh, when you want to switch between channels all the time.\nSpeaker D: Yeah, you want to, uh, do it like if you have a channel above 10, normally you can press 1, 0, or...\nSpeaker B: Yeah, I think, um, especially for older people, use standard buttons, so you have that, uh...\nSpeaker D: The 10 plus, or...\nSpeaker B: Yeah, the 10 plus button.\nSpeaker B: Just, uh, 1-1 pressing, or quickly after each other.\nSpeaker D: Okay.\nSpeaker B: Just to keep it simple, and standard features.\nSpeaker C: Yeah, wouldn't it be a problem to, uh, because you have to be fast enough.\nSpeaker C: Maybe the elderly people.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker B: But I think that's in the TV as well.\nSpeaker B: That's how the TV handles it.\nSpeaker B: But you can have a button that says, um...\nSpeaker B: Yeah, it's not, it's not the...\nSpeaker B: It's not the...\nSpeaker B: So you have, like, uh...\nSpeaker B: Yeah, but just...\nSpeaker B: It's a very, uh, to press those two buttons.\nSpeaker E: And what Paul says, it's not remote control.\nSpeaker E: Uh, that, um, that makes the TV, um, do 1-2, it's the TV.\nSpeaker E: It depends.\nSpeaker E: That it must be 2 seconds or 3 seconds or 1 seconds that you must type 1.\nSpeaker B: So if you have a universal TV controller, you need that 1 button that has 2, uh, stripes.\nSpeaker B: Yeah, so you have a period of, I don't know, 5 seconds to press those buttons.\nSpeaker B: And that, not for elderly people to look 1, 2, press and aim and...\nSpeaker D: But do we still need the 2 level remote control?\nSpeaker D: Because if you only have that...\nSpeaker D: It's only for television now.\nSpeaker B: So I don't think so.\nSpeaker B: No.\nSpeaker B: Uh, I just thought of another one.\nSpeaker B: Most things in modern TVs are also on the menu.\nSpeaker B: Yeah.\nSpeaker B: So you also need a menu button.\nSpeaker B: And then, uh, navigation.\nSpeaker B: Uh, but we can integrate that with...\nSpeaker B: We can integrate that with volume and previous and next.\nSpeaker B: So we have...\nSpeaker C: I think you really have to divide between functions you often use.\nSpeaker C: Like, um, uh, maybe, uh, switch channels and, uh, volume and get that kind of things.\nSpeaker C: And, uh, the menu button.\nSpeaker C: Because you almost never use menu button.\nSpeaker C: Yeah, okay.\nSpeaker B: Uh...\nSpeaker D: Maybe it's still a good idea.\nSpeaker D: I'm not sure.\nSpeaker D: You also have to use a mute button to...\nSpeaker D: Yeah, mute button.\nSpeaker B: Yes.\nSpeaker B: Maybe.\nSpeaker D: I don't know where you have to put it.\nSpeaker B: Yeah, well, that's...\nSpeaker B: I think that's the latest step.\nSpeaker B: You just now have to figure out what to put on.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: And later we can do the design.\nSpeaker B: Okay.\nSpeaker B: Uh, now my personal preference is using the standards basic.\nSpeaker B: Um, I think that we should stand out, uh, unique...\nSpeaker B: being unique with the design.\nSpeaker B: So we have to, I don't know, uh, make a different shape than you as well.\nSpeaker B: So when you are in the shop and you see our TV controller hanging, that it stands out not just in a row and you see all the same, uh, remote controls.\nSpeaker D: Yeah.\nSpeaker D: There's have to be...\nSpeaker D: There's have to be, uh, with different colors, or...\nSpeaker D: I don't know, different colors.\nSpeaker E: Um, maybe we can give it out in different colors.\nSpeaker E: You can choose...\nSpeaker E: It's like, uh, I make something.\nSpeaker E: It's a naked...\nSpeaker E: Yeah, well, why not?\nSpeaker E: Or, uh...\nSpeaker E: That's fancy.\nSpeaker D: That's, uh, fancy.\nSpeaker D: It looks cheap as well because it's a small thing.\nSpeaker D: It's only twenty-five years.\nSpeaker D: It looks very cheap if you make it.\nSpeaker B: Well, maybe we can look at, um, mobile phones.\nSpeaker B: Oh, you just...\nSpeaker D: I don't know what happened.\nSpeaker E: Another front.\nSpeaker E: Who?\nSpeaker D: Yeah, we have to look at mobile phones.\nSpeaker B: That's right.\nSpeaker B: The design very well.\nSpeaker B: And, well, basically, hard the same.\nSpeaker B: Just a bit smaller.\nSpeaker B: You think you can't make it if you control it too small because then you will over-load Always lose it.\nSpeaker B: Yeah.\nSpeaker B: But, uh, well, I think that's a good example.\nSpeaker B: Okay.\nSpeaker D: Okay, yes, you're really up your bed because of all that.\nSpeaker D: Okay.\nSpeaker D: You finished, okay.\nSpeaker D: Yeah, first.\nSpeaker B: I think we discussed everything.\nSpeaker B: Okay.\nNone: Well, the working design.\nNone: The model.\nSpeaker C: Okay.\nSpeaker C: Well, the working design.\nSpeaker C: The method I use is search the web.\nSpeaker C: Just a web page provided.\nSpeaker C: Um, basically, I'm not very technical, uh, educated, but I could figure it out.\nSpeaker C: Um, basically what happens is you press a button.\nSpeaker C: It's closed by pressing the button, uh, like a switch.\nSpeaker C: And by closing that certain circuit, the chip knows what button is pressed.\nSpeaker C: So, like, you press the one, that circuit is closed and, uh, then the chip produces a pattern, like Morse code, to, uh, and sends that to the, uh, LED.\nSpeaker C: That's the, uh, light emitting diode, I think.\nSpeaker C: Um, and the light emitting diode is, uh, producing infrared light.\nSpeaker C: That's, uh, invisible to the human eye and, um, transmit that to the TV.\nSpeaker C: However, there's also, uh, a visible, uh, diode.\nSpeaker C: That's, uh, blinking red, if you look in it.\nSpeaker C: And that's, uh, another diode, I believe, because infrared is not visible.\nSpeaker C: So, that's, uh, two different things.\nSpeaker D: We also have to have a LED.\nSpeaker C: Yeah, I think so.\nSpeaker B: I'm not sure if it's the same thing that you do to indicate that it's active.\nSpeaker D: Yeah.\nSpeaker D: Or just a green one, because if you use it, it's green on the red.\nSpeaker E: It's real green.\nSpeaker E: I think it's, in the case of it's active.\nSpeaker E: Yeah, when you press it.\nSpeaker E: It's just, uh, the...\nSpeaker D: Redstone shows up like something's wrong and green is like it's okay.\nSpeaker D: You press it.\nSpeaker C: Yeah, I don't know.\nSpeaker C: Maybe.\nSpeaker C: Oh, almost, almost, uh, remote controls is red.\nSpeaker C: Yeah.\nSpeaker D: Yeah.\nSpeaker B: Okay, we make it.\nSpeaker B: So maybe, well, we don't have to make it red.\nSpeaker B: Maybe integrate it into design as well.\nSpeaker D: Yeah, the whole, the whole, uh, remote control becomes, uh, green, if you press the button.\nSpeaker D: That's cool.\nSpeaker D: Yeah.\nSpeaker C: Oh, maybe.\nSpeaker C: I think the batteries will be, uh, now we have little recharge.\nSpeaker C: Yeah, okay.\nSpeaker D: If we can still make that.\nSpeaker B: Yeah, but it doesn't have to be red.\nSpeaker B: It's just to indicate something.\nSpeaker B: Yeah.\nSpeaker C: That is working.\nSpeaker C: It's not, not the batteries are low.\nSpeaker B: It's not very important.\nSpeaker C: Green or red or whatever.\nSpeaker A: Yeah.\nSpeaker C: Just to indicate it's working.\nSpeaker C: So really.\nSpeaker C: And, uh, the receptor in the TV senses the pattern.\nSpeaker C: So we have to understand what patterns are used to, you know, to make it universal so that it can be used with all the TVs.\nSpeaker C: We have to really understand what patterns are used so we can, uh, otherwise it won't\nNone: work. So we have to, uh, make buttons for that as well.\nSpeaker B: To make.\nSpeaker C: No, uh, the chip, um, uh, is producing the pattern.\nSpeaker C: So we have to make a chip that produces patterns to, you know, to the other working.\nSpeaker E: An automatic research function for each television.\nSpeaker C: Yeah, or I don't know exactly how it works.\nSpeaker C: It wasn't explained there.\nSpeaker B: Well, I, I use, uh, universal, uh, remote control and, uh, before end you have a list of all the TVs you have, etc.\nSpeaker B: And you have to put in a number.\nSpeaker B: Yeah.\nSpeaker B: So it works on your TV.\nSpeaker E: Okay.\nSpeaker E: Oh, I have a modern one.\nSpeaker E: And the modern one.\nSpeaker E: You, uh, you type, uh, search and the lead became to blink, blink, blink.\nSpeaker E: And, uh, uh, after a short time the television turns off.\nSpeaker E: And then you know, oh, it's the right one.\nSpeaker E: You can stop it.\nSpeaker E: And then it's okay.\nSpeaker E: So you don't have to search for your television or your code.\nSpeaker E: Okay.\nSpeaker E: It's, uh, the, the pattern for itself.\nSpeaker E: Okay.\nSpeaker C: Yeah.\nSpeaker C: And we use that.\nSpeaker C: And it's a bit more intelligent chip than the standard one because you have to, yeah.\nSpeaker D: I just say, can you just say it again because I was just looking at this.\nSpeaker C: Okay.\nSpeaker C: Sure.\nSpeaker C: Well, uh, you have this chip that's, uh, when the circuit is closed produces the pattern.\nSpeaker C: But, um, basically for, uh, brands of TV, these patterns are different.\nSpeaker C: So like when you press a one on one TV, it goes to one.\nSpeaker C: And on another TV, it won't work basically.\nSpeaker C: So, uh, basically what you have to do is, uh, get these patterns right.\nSpeaker C: Yeah.\nSpeaker C: And, uh, by, uh, that can be done by, uh, just, uh, changing the signal.\nSpeaker C: Yeah, changing this pattern all the time.\nSpeaker C: And, uh, what he's doing when he's configuring this pattern, this chip, uh, is trying to switch off the television.\nSpeaker C: And when it's, uh, switched off, you can push your button.\nSpeaker C: It's working now.\nSpeaker C: So, and then he saves that setting.\nSpeaker C: Okay.\nSpeaker C: And then, um, he's working.\nSpeaker E: Yeah, that's the right option.\nSpeaker C: Okay.\nSpeaker C: Well, uh, the components.\nSpeaker C: Yeah, that's a bit technical and I hadn't, I was a little short on time.\nSpeaker C: Um, but I think I understand it.\nSpeaker C: Um, the energy source is, uh, the battery, basically.\nSpeaker C: Um, that's connecting to all the components because it has to be fed with energy.\nSpeaker C: Okay.\nSpeaker C: Uh, the sub component is, uh, I, I think it is, uh, the button pressed.\nSpeaker C: So basically when you press a button, a switch gets closed.\nSpeaker C: Um, that's connecting to a chip.\nSpeaker C: So the chip knows what button you pressed.\nSpeaker C: And the chip sends out, uh, the pattern to the infrared bulb.\nSpeaker C: I didn't put the description by this one.\nSpeaker C: This is a normal bulb.\nSpeaker C: So the normal flashing light.\nSpeaker C: Okay.\nSpeaker C: So that's the lead.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: And this is the lead too, but this one is producing infrared light.\nSpeaker C: It's invisible and this one is producing normal light.\nSpeaker D: So we can make the normal one also a normal light.\nSpeaker D: No, the lead light is a normal one.\nSpeaker D: Yeah.\nSpeaker D: Flash up your.\nSpeaker E: Uh, if you, if you use the buttons, uh, both of them works.\nSpeaker E: But one you can't see and one indicates that you use a button.\nSpeaker E: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Because, uh, when you button, press a button and it doesn't work, it can, basically, if the battery is low, it won't work.\nSpeaker C: Yeah.\nSpeaker D: So you have to indicate that it's, maybe you have to, uh, put, uh, a small light around the button you push.\nSpeaker D: So you see green if you push that button.\nSpeaker E: I think it's been necessary power.\nSpeaker E: You use that.\nSpeaker D: Maybe if you, if you do that, you know that you're, uh, sending a signal.\nSpeaker D: Okay.\nSpeaker D: And it's, you also know which button.\nSpeaker E: It's gonna have to be on the wall, uh, remote control.\nSpeaker E: Have to be let's.\nSpeaker E: Yeah, that's right.\nSpeaker D: Everywhere.\nSpeaker D: And, uh, it doesn't make any difference for the energy you use.\nSpeaker D: You'll still want left.\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker B: It will look different.\nSpeaker B: And I think we need to find something else.\nSpeaker B: That looks different.\nSpeaker B: Yeah.\nSpeaker B: Cause otherwise we will just make another standard and our motto also.\nSpeaker E: It's, uh, let's, uh, beneath the buttons.\nSpeaker B: Yeah.\nSpeaker B: Around the buttons or in the buttons even.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker C: Yeah, but like when you push it.\nSpeaker A: Yeah.\nSpeaker C: Yeah.\nSpeaker C: You have your finger over the button.\nSpeaker C: So you can see.\nSpeaker C: Must be around it then.\nSpeaker C: Yeah.\nSpeaker C: Or, or maybe on top of the green light splashing.\nSpeaker E: Not, not, not here.\nSpeaker E: That's.\nSpeaker E: Yeah.\nSpeaker A: There.\nNone: Yeah.\nSpeaker D: Maybe, uh, the same as a telephone or a mobile phone or.\nSpeaker B: Yeah.\nSpeaker B: We're thinking about that.\nSpeaker B: A mobile phone in the dark lights up.\nSpeaker B: Everything lights up.\nSpeaker B: That's a good idea.\nSpeaker D: It only takes a little energy.\nSpeaker B: Yeah.\nSpeaker B: A mobile phone can do it.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Maybe that's a good idea for old people as well.\nSpeaker B: Yeah.\nSpeaker B: If you were in the dark, you can't see the rumors.\nSpeaker D: It's only few let's only four or something for let.\nSpeaker B: No, I don't know.\nSpeaker B: But if we use a battery station, which I think we'll use.\nSpeaker B: Yeah.\nSpeaker C: Yeah.\nSpeaker C: That's a good idea.\nSpeaker D: Yeah.\nSpeaker D: We'll have enough power to everything.\nSpeaker D: Everyone agrees to that or.\nSpeaker D: Yeah.\nNone: Yep.\nSpeaker E: Maybe we'll post it on the home station.\nSpeaker E: A button to, to call your remote control.\nSpeaker E: Okay.\nSpeaker D: Yeah.\nSpeaker D: But you have to make a sound device in it.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker B: Okay.\nSpeaker C: Yeah.\nSpeaker C: Maybe it's that that's a little too hard to make.\nSpeaker C: Especially for that kind of money because it has to be.\nSpeaker C: The production cost is 12 and a half.\nSpeaker E: 12 and a half.\nSpeaker E: Okay.\nSpeaker E: But we only have to make it for television and.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: We must have something special.\nSpeaker B: We also have remote controls with a lot of options.\nSpeaker B: But we lose about 90% of those options.\nSpeaker B: So I think you can.\nSpeaker E: You have to give customers some extra.\nSpeaker E: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I think we will save money with that.\nSpeaker D: Okay.\nSpeaker D: You're almost finished.\nSpeaker C: Yeah.\nSpeaker C: The personal preference I didn't fill it out because it was short on time.\nSpeaker C: But basically what we could do too is have a Bluetooth integration.\nSpeaker D: Yeah.\nSpeaker D: I just wanted to talk some more.\nSpeaker D: So maybe.\nSpeaker D: Yeah.\nNone: Okay.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: So.\nSpeaker D: This.\nSpeaker D: So.\nSpeaker D: Sorry.\nSpeaker D: You have some new project requirements as if they look at what they are.\nSpeaker D: Yeah.\nSpeaker D: Teletext has become outdated since the popularity of Internet.\nSpeaker D: So a teletext option.\nSpeaker D: Maybe we have to skip that one.\nSpeaker D: I'm not sure.\nSpeaker C: I think a lot of people use teletext.\nSpeaker D: It's on your television.\nSpeaker D: It's only one button.\nSpeaker D: Yeah.\nSpeaker D: We definitely should use it.\nSpeaker D: Yeah.\nSpeaker D: What I told you the remote controls should only be used for a television.\nSpeaker D: So that's maybe easier.\nSpeaker D: And the 40 plus people already told you.\nNone: Oh, no.\nSpeaker D: Sorry.\nSpeaker D: Oh, this is a problem.\nSpeaker D: Oh, sorry about that.\nSpeaker D: The new product should reach a new market with customers are the younger and the 40.\nSpeaker D: So it should be flashy or interesting.\nSpeaker C: Sorry about that.\nSpeaker D: I just didn't read it.\nSpeaker D: So did you make some.\nSpeaker D: Change.\nSpeaker C: Yeah, that kind of changes the whole situation.\nSpeaker B: Yeah, then we have to make some nice features.\nSpeaker B: I think the thing in the dark is a good way to make more like mobile phone.\nSpeaker B: More modern.\nSpeaker B: No.\nSpeaker B: That's important.\nSpeaker E: I think LCD doesn't work for 12 and a half euros.\nSpeaker E: I think.\nSpeaker B: I don't think it's very usable in remote control, especially when you only have TV functions on.\nSpeaker B: No.\nSpeaker C: I don't think you'll use it often because you can see on the television what channel you\nSpeaker E: are. That's channel one.\nSpeaker E: OK.\nSpeaker B: No, I think that's not usable now.\nSpeaker B: So but what other features can we put in?\nSpeaker C: Yeah.\nSpeaker C: I think you have to make it a bit flashy and popular.\nSpeaker C: But then the usability is not that required because like in the mobile phones usability is not that good.\nSpeaker E: I think some of you had something to read about speech recognition about you said one and the television turns on one.\nSpeaker E: Is that reachable?\nSpeaker E: Maybe?\nSpeaker E: That's very fancy.\nSpeaker E: That's cool.\nSpeaker E: That's 12.\nSpeaker C: Yeah.\nSpeaker C: OK.\nSpeaker C: Only the numbers.\nSpeaker E: Only numbers.\nSpeaker E: Only the numbers, one to 20 or something.\nSpeaker D: That should be cool.\nSpeaker D: Yeah.\nSpeaker D: Maybe we have to integrate that as well.\nSpeaker D: If it's possible.\nSpeaker D: If it's possible.\nSpeaker D: I don't know.\nSpeaker D: I don't know.\nSpeaker D: I think it's very expensive actually.\nSpeaker D: Why should it?\nSpeaker D: Yeah.\nSpeaker D: You only have a microphone in there.\nSpeaker D: Yeah.\nSpeaker C: It has to work.\nSpeaker C: And does it have to work only in English or in Dutch too?\nSpeaker E: Ah, maybe.\nSpeaker E: Ah, that's a problem.\nSpeaker C: And, and.\nSpeaker C: Oh, in English.\nSpeaker B: I don't know how it's worth it.\nSpeaker B: Yeah, it's probably my job to figure that out.\nSpeaker B: But you can use cream from wherever.\nSpeaker B: No, one.\nSpeaker B: It's an additional one.\nSpeaker E: I didn't.\nSpeaker E: We have to think about that.\nSpeaker E: But do we do it?\nSpeaker E: No, I don't know.\nSpeaker B: If we do it.\nSpeaker B: Is it very usable?\nSpeaker B: That's what I'm looking at.\nSpeaker C: And do you have to speak in like anything or?\nSpeaker D: Maybe it's too hard to realize that.\nSpeaker D: We have a short time to put it on the market.\nSpeaker D: OK.\nSpeaker D: You have to do a lot of testing before we can do that.\nSpeaker C: Yeah, thanks.\nSpeaker D: We can make the remote control very flashy.\nSpeaker E: Indeed, the languages are a difficult thing.\nSpeaker E: Because we, we, we, we, we, we, we, we.\nSpeaker B: And it's useful enough to take the effort to design something with speech recognition.\nSpeaker C: And also, if you have a good speech recognition, you can just throw the.\nSpeaker E: Yeah, someone says, give me one Coke and the TV turns on to one.\nSpeaker E: It's not.\nSpeaker B: Yeah, OK.\nSpeaker B: That's the right one.\nSpeaker B: But then you have to say TV channel one or something.\nSpeaker A: Not just one.\nSpeaker A: But TV one.\nSpeaker E: That becomes your, because that's very hard for your speech recognition.\nSpeaker B: But I don't think it's just anxious enough.\nSpeaker D: So we have to make some decisions so you can see on the.\nSpeaker D: OK, no speech recognition.\nSpeaker D: So we have to know what we're going to put on.\nSpeaker D: We do make a light under the, under the numbers and everything.\nSpeaker D: If you press something, it lights up for a few seconds.\nSpeaker D: So you can see what other buttons there are.\nSpeaker D: OK.\nSpeaker D: Yeah, that's good.\nSpeaker D: OK, we just take that one.\nSpeaker D: And what else we have.\nSpeaker B: Also the one in the dark.\nSpeaker B: So, lights up when it's dark.\nSpeaker D: Yeah, but it's, oh, I thought that would be the same.\nSpeaker D: If you push something.\nSpeaker E: Yeah, it, it doesn't work all the time.\nSpeaker E: Yes, that's right.\nSpeaker E: It's working only when you use it.\nSpeaker E: So you can switch it on or something.\nSpeaker D: No, maybe we use one button.\nSpeaker D: It must turn on for 20 seconds.\nSpeaker D: And then it must turn on.\nSpeaker A: Yeah, that's right.\nSpeaker A: It's the same as the telephone.\nSpeaker A: Yeah, all right.\nSpeaker A: Yeah.\nSpeaker D: And do we use, what's it called?\nSpeaker D: Like iMac that you can look through it or just normal.\nSpeaker B: Maybe it's good idea.\nSpeaker B: Like we discussed, like iPod.\nSpeaker B: Different colors.\nSpeaker B: Maybe use even different fonts.\nSpeaker B: Yeah, like phones.\nSpeaker D: Different colors, OK.\nSpeaker D: So, just make it.\nSpeaker B: Yeah, it's cool.\nSpeaker B: It has an option, maybe.\nSpeaker D: And so the buttons we have, this is normal.\nSpeaker D: We put in the simple buttons on the top and probably the more complicated buttons.\nSpeaker B: Yeah, we have a lot of different buttons.\nSpeaker B: We don't really have any.\nSpeaker D: Yeah, maybe you have the, we use teletext.\nSpeaker D: You can press a green and a red button to go to the next one or to go to the free button.\nSpeaker B: Well, I think the buttons are very easy.\nSpeaker B: We just stand the buttons.\nSpeaker D: Maybe you can put the options of teletext on the second level of the remote control because\nSpeaker B: you almost never use it. Teletext options are only for a button or something.\nSpeaker C: Yeah, but younger people, I think more often use the internet and elderly people often use teletext.\nSpeaker E: Yeah, that's why I use teletext every day, I think.\nSpeaker D: OK, yeah.\nSpeaker D: For me too.\nSpeaker D: So, we just keep it one level then?\nSpeaker B: Yeah, yeah, one level.\nSpeaker B: I think, because we don't have that many buttons, I think the design is most important.\nSpeaker B: You can integrate the buttons in the design very much because there are the simple buttons.\nSpeaker B: If you put one above it, it's clear.\nSpeaker B: OK.\nSpeaker D: So, that's it for, we're going to have a lunch.\nSpeaker D: Lunch break.\nSpeaker D: How long is lunch break?\nSpeaker D: So, yeah, you know what you have to do with it.\nSpeaker D: This is it.\nSpeaker D: You get your meal and everything.\nSpeaker E: Transwetching.\nSpeaker E: OK.\nSpeaker E: Yeah.\nSpeaker E: That's fine.\nSpeaker E: Yo.\nSpeaker D: Yo.\nSpeaker D: So, we finished for this time.\nSpeaker A: OK.\nSpeaker E: Bye.\nSpeaker C: Yeah.\nSpeaker D: Yeah.\nNone: OK.\n"}, {"meeting_id": "TS3008a", "summary": "This is a kick-off meeting for a new remote control project for television sets. Project Manager started it with a review of project objectives and then the team got acquainted with each other and tools. Through discussion, the team decided to design a remote control with grouped buttons, useful with DVD players and carrying some special features to be confirmed.", "dialogue": "None: ISoE recipeors Short battery and BOOM a bit at this point Thank you.\nNone: Thank you.\nSpeaker A: Then you have to place your laptop exactly on the marked spot.\nNone: All right.\nNone: Thank you for tuning.\nSpeaker A: It's important to place your laptop exactly on the marked spot over here.\nSpeaker D: Okay.\nSpeaker C: It's okay.\nSpeaker C: You used your mouth.\nSpeaker C: What?\nSpeaker C: No mouth needed.\nSpeaker B: I've got a touch pad.\nSpeaker B: Do you know how I can wake it up?\nSpeaker C: With touch pad?\nSpeaker B: No.\nSpeaker B: My laptop.\nSpeaker C: Slap it?\nSpeaker B: You with your brilliant ideas.\nSpeaker B: I don't know if I can touch the power button.\nSpeaker B: Do you know how I can wake it up?\nSpeaker A: No.\nSpeaker A: Try the power button.\nSpeaker B: Come on, move it.\nSpeaker B: Now wake up, bitch.\nSpeaker C: F5.\nSpeaker C: F5, you've lost my screen.\nSpeaker B: Yes, so did I.\nSpeaker B: I closed it.\nSpeaker B: That wasn't very smart, I guess.\nNone: Come on.\nSpeaker D: Get back to me.\nSpeaker A: I closed it.\nSpeaker B: You've got your name?\nSpeaker B: My name is name.\nSpeaker B: I did not know if nothing happens.\nNone: Yes.\nSpeaker D: Hope if we're working.\nNone: All right.\nSpeaker C: Close your laptop.\nSpeaker A: Everybody's ready?\nSpeaker A: Great.\nSpeaker A: Thanks.\nSpeaker A: Welcome to the kickoff meeting.\nSpeaker A: I forgot to put my name over here.\nSpeaker A: It's Martin.\nSpeaker A: So you all know.\nSpeaker A: This is the agenda for today.\nSpeaker A: The opening is what I'm doing right now.\nSpeaker A: We're going to do some acquaintance things.\nSpeaker A: Here are some examples of the tool training project plan discussion and closing.\nSpeaker A: We have 25 minutes.\nSpeaker A: The project aim is to design a new remote control.\nSpeaker A: Some of the objectives are that it has to be original, trendy, and user-friendly.\nSpeaker A: So now we all know what our goal is.\nSpeaker A: I forget the whole acquaintance part, but we all know each other.\nSpeaker A: We all know each other's names.\nSpeaker C: Yes.\nSpeaker C: Contact.\nSpeaker A: Contact.\nSpeaker A: I think we've already been to that part.\nSpeaker A: Okay. It consists of three levels of design.\nSpeaker A: We begin with a functional design.\nSpeaker A: Then we go to the conceptual design and the detailed design.\nSpeaker A: Every level of design consists of some individual work.\nSpeaker A: And we close it with a meeting.\nSpeaker A: You all receive an email with an example of our explanation of what a particular level of design means to the different functions.\nSpeaker A: You probably read that already.\nSpeaker A: I'll tell you about that.\nSpeaker A: Okay. First we're going to try some different things with the tools we have over here.\nSpeaker A: So you get acquainted with these meeting tools.\nSpeaker A: We have the smart boards.\nSpeaker A: Those two boards.\nSpeaker A: This is the presentation board, which I'm using right now.\nSpeaker A: There's a document folder called the share document folder.\nSpeaker A: You can upload your documents to that folder and then you can open them over here.\nSpeaker A: So you can do your PowerPoint presentations on this screen.\nSpeaker A: We also have the white board.\nSpeaker A: Can we skip the white board on our left?\nSpeaker A: No, no.\nSpeaker C: I saw a screen of files.\nSpeaker A: Oh, well, probably.\nSpeaker A: I don't know if the software is on the left.\nSpeaker A: It's mainly your thing for in the meeting.\nSpeaker A: I don't know if it's important.\nSpeaker A: This is an explanation of the smart boards.\nSpeaker A: There's a tool bar over here.\nSpeaker A: It's quite simple.\nSpeaker A: You have the pen function, razor function.\nSpeaker A: It's like a very simple paint application.\nSpeaker A: We use the same file during the whole day.\nSpeaker A: You can make new sheets by pressing on the blank button.\nSpeaker A: It works like this.\nSpeaker A: The pen is selected.\nSpeaker C: Oh, no, no.\nSpeaker A: It's not working like a pen yet.\nSpeaker C: Oh.\nSpeaker A: Doing some stuff now.\nSpeaker A: You can use a pen.\nSpeaker A: You can use an eraser.\nSpeaker A: You can make new blanks.\nSpeaker A: You can change the line width and the color of the pen by pressing on forward.\nSpeaker A: Select pen format.\nSpeaker A: Select current color or line width.\nSpeaker A: It's quite easy.\nSpeaker A: Now you're all acquainted with different tools.\nSpeaker A: We're going to try out the electronic whiteboard.\nSpeaker A: Every participant should draw its favorite animal and some of its favorite characteristics.\nSpeaker A: On blank sheet with different colors with different pen widths.\nSpeaker A: I'll start off then.\nSpeaker A: I'll use this same sheet.\nSpeaker A: I think different colors.\nSpeaker A: I'm going to draw a perone.\nSpeaker C: I'll see you.\nSpeaker A: I'll show you some different color right now.\nSpeaker A: Looks like a fish.\nSpeaker C: Oh.\nSpeaker A: This is just useless drawings.\nSpeaker A: They're not supposed to be green or whatever color it is.\nSpeaker A: I have some of its favorite characteristics.\nSpeaker A: I like its razor sharp teeth.\nSpeaker A: I think it's the big forehead and the small actual face.\nSpeaker A: And I like its overall aggressive look.\nSpeaker A: And that's what I like about piranha.\nSpeaker A: I think that's kind of what the intention should be.\nSpeaker A: Well, it wants to be next.\nSpeaker B: Nobody, I guess.\nSpeaker C: Alg\nSpeaker A: Yeah.\nSpeaker C: Okay. Then pen again.\nSpeaker A: Yeah.\nSpeaker C: Okay.\nSpeaker A: And so in the format menu you can choose the different\nSpeaker C: words and the format. Purple.\nSpeaker C: I don't know what my favorite animal is, but the easiest animal I can think of is a bird.\nSpeaker B: You know I thought of that actually.\nSpeaker B: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It's a little bit light.\nSpeaker C: Another color maybe.\nSpeaker C: A red one.\nSpeaker C: A small one.\nSpeaker C: Two.\nSpeaker C: Three.\nSpeaker A: That's another one.\nSpeaker A: I'll tell you something about your favorite characteristics of these particular words.\nSpeaker C: It's simplicity.\nSpeaker C: Yeah.\nSpeaker C: It's the most simple animal I know.\nSpeaker C: There's a free and maybe in the sky or something like that.\nSpeaker C: Maybe a little bit.\nSpeaker C: Yeah.\nSpeaker C: I don't know.\nSpeaker C: Okay.\nSpeaker C: More birds?\nSpeaker C: No.\nSpeaker A: We'll get your point.\nSpeaker A: Okay.\nSpeaker B: What's it be next?\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Whatever.\nNone: I'll go next.\nSpeaker B: Thanks.\nSpeaker B: I haven't got a favorite animal to sell.\nSpeaker B: Bictionary.\nSpeaker B: What should I draw?\nSpeaker B: I'll go.\nSpeaker B: Thank you.\nSpeaker B: I'll draw a penguin.\nNone: Yeah.\nSpeaker B: I'll draw a penguin.\nSpeaker B: Whatever.\nSpeaker B: I can draw.\nSpeaker B: So you can start to laugh already.\nSpeaker B: Whatever.\nNone: Something like that.\nSpeaker B: Come on.\nSpeaker B: Orange.\nSpeaker B: Whatever.\nSpeaker B: It's better than your bird.\nSpeaker B: Whatever.\nSpeaker B: It's blue.\nSpeaker B: Whatever.\nSpeaker B: I like its ugliness.\nSpeaker B: Yeah.\nSpeaker B: Whatever.\nSpeaker B: The way it walks or whatever.\nSpeaker B: Okay.\nSpeaker B: Your turn.\nSpeaker B: Okay.\nSpeaker D: I'm drawing.\nSpeaker D: I'm going to draw a cat.\nSpeaker D: I don't know why, but cat is a very smart animal.\nSpeaker D: And you can have them at home.\nSpeaker D: Which is not a case with penguins.\nSpeaker A: You can have a piranha at home.\nSpeaker B: Or a lion.\nSpeaker B: I'm in a bird.\nSpeaker C: I'm messing with my bird.\nSpeaker D: It's not very good.\nSpeaker D: Very good.\nSpeaker C: I can see the cat.\nSpeaker A: I don't think Darwin would agree with that.\nSpeaker A: It's not a cat.\nSpeaker D: He's crying.\nSpeaker D: He's crying.\nSpeaker D: He's crying.\nSpeaker A: What do you like about it?\nSpeaker D: Most cats are small.\nSpeaker D: You can handle that.\nSpeaker A: Okay.\nSpeaker A: I wouldn't call it training.\nNone: Okay.\nSpeaker A: This is something about the project finance.\nSpeaker A: The sending price of our remote control is going to be 25 euros.\nSpeaker A: And our profit aim is 15 million euros.\nSpeaker A: We're very ambitious on this one.\nSpeaker A: The market range is international, so it's going to be so worldwide.\nSpeaker A: And the production costs should be a maximum of 12 euro 50 per remote control.\nSpeaker A: So that's clear.\nSpeaker A: Yeah?\nSpeaker A: All right.\nSpeaker A: Okay.\nSpeaker A: We're now going to discuss some stuff.\nSpeaker A: We're going to brainstorm about what kind of remote control is going to be.\nSpeaker A: Tell me about your experiences with remote controls.\nSpeaker A: Do you have good experiences with remote controls?\nSpeaker A: Do they annoy you sometimes?\nSpeaker A: Are they difficult to understand?\nSpeaker A: Maybe they don't interact with different kind of equipment very well?\nSpeaker B: I don't think the four of us got problems with remote controls, but if you see elderly people, all these buttons, and then they buy new TV because their previous one was stolen or whatever.\nSpeaker B: And then it's totally different from out control with different functions on different places.\nSpeaker B: And half of the functions are removed or whatever.\nSpeaker B: So I think what we need is a clear remote control with grouped buttons.\nSpeaker B: All the buttons which apply to the text functions in one area or whatever, not like the button to enter text on top of the remote control and the button to minimize it to this or whatever.\nSpeaker B: Other functions are totally somewhere else.\nSpeaker B: I think we should group them the same for the following buttons.\nSpeaker A: Is it going to be a remote control?\nSpeaker A: That's what it can be used for different kind of equipment like your TV and your stereo.\nSpeaker B: I was thinking since TV is mostly used together with a VCR or DVD player or recorder and not with stereo.\nSpeaker B: I think it should be good to include functions for VCRs and DVD players.\nSpeaker A: The DVD players and home cinema sets often double S stereo high-five sets probably.\nSpeaker A: That's what my experience is.\nSpeaker B: You know high-five set is not often used as I know of in combination with television.\nSpeaker B: But we're going to...\nSpeaker C: It's only for television.\nSpeaker A: Yes.\nSpeaker C: It is only for television.\nSpeaker C: Yes, it is only for television.\nSpeaker C: So we probably don't have to have to have to have functions for DVD players.\nSpeaker B: We're going to bring some about that.\nSpeaker B: If we think it's useful, we do it.\nSpeaker A: Where did you find that?\nSpeaker C: In the email.\nSpeaker C: That's right. It's a television route control.\nSpeaker B: I was thinking since in the video...\nSpeaker A: We have a most dedicated remote control support other functions as well.\nSpeaker A: We have to think about that.\nSpeaker A: But we've got to put some...\nSpeaker A: So user friendliness is a priority.\nSpeaker A: Yeah, else no one's going to buy it.\nSpeaker A: I guess.\nSpeaker A: I need experts.\nSpeaker A: Maybe there's some aspect of which we can excel by making it very useful.\nSpeaker A: Then you're the usability man.\nSpeaker A: So there's going to be a very important task for you then.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Well other ideas, how can we make it trendy or something?\nSpeaker A: Just shape and look at it.\nSpeaker D: Go with fashion.\nSpeaker A: Maybe a can opener underneath it.\nSpeaker A: I don't know.\nSpeaker A: Or something special, like an MPD player inside of it.\nSpeaker A: Well then the production cost is going to be time.\nSpeaker C: Yeah, way too high.\nSpeaker C: I think we have to keep it simple to get a whole market.\nSpeaker C: It's international.\nSpeaker A: Maybe with different type of fronts or what must be something.\nSpeaker A: That's an idea of course.\nSpeaker A: In particular, which makes it...\nSpeaker A: We can skip back to the goals probably.\nSpeaker A: Original, trendly and user friendly.\nSpeaker A: We already talked about user friendliness.\nSpeaker A: Well, something trendy and original.\nSpeaker A: That goes hand in hand I guess.\nSpeaker A: Yeah.\nSpeaker A: When something is original, tends to be trendy.\nSpeaker A: You should make a combination of that.\nSpeaker A: Okay.\nSpeaker A: So the technical part of the process is something you're going to look after for.\nSpeaker A: So you have to think about what kind of equipment you want to manage with it.\nSpeaker A: Well, that's an important part for you then.\nSpeaker A: With regards to the user friendly part of it.\nSpeaker A: Well, you should look out for what makes it trendy.\nSpeaker A: Well, you know, like some special future or something.\nSpeaker A: Did it get some gadgetness or something?\nSpeaker A: Yes.\nSpeaker A: What the market wishes.\nSpeaker A: Okay.\nSpeaker A: Well, closing.\nSpeaker A: Next meeting starts in 30 minutes.\nSpeaker A: Individual acting, the industrial designer will...\nSpeaker A: The working design, of course.\nSpeaker A: Already said that.\nSpeaker A: The user interface designer...\nSpeaker A: Is it user interface?\nSpeaker A: Yeah.\nSpeaker A: Technical functions design.\nSpeaker A: And the management expert...\nSpeaker A: The marketing expert user requirements, especially well.\nSpeaker A: That's what was already clear to us.\nSpeaker A: Specific instructions will be sent to you by your personal coach.\nSpeaker A: I don't know how much time we have left.\nSpeaker A: Not many, I guess.\nSpeaker A: We started at 12.\nSpeaker A: You just got a message.\nSpeaker B: Oh.\nSpeaker B: And what does it say?\nSpeaker B: And it's sad.\nSpeaker B: Oh.\nSpeaker B: Five minutes.\nSpeaker B: So we got four and a half.\nSpeaker A: Okay.\nSpeaker C: Well, I didn't get a message.\nSpeaker B: No.\nNone: He's the...\nSpeaker B: Oh.\nSpeaker B: The team leader.\nSpeaker B: Team leader.\nSpeaker A: I had a schedule then.\nSpeaker A: He is the whatever.\nSpeaker A: We close it.\nSpeaker A: I'm going to make some minutes or take some minutes.\nSpeaker A: And it's clear you can put stuff in the project documents, presentations.\nSpeaker B: Project documents is shown on the whiteboard.\nSpeaker A: Yeah.\nSpeaker A: We're all familiar with Microsoft PowerPoint.\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker A: We're going to write some stuff down and then we're ready.\nSpeaker B: Okay.\nSpeaker A: Well, we can leave already, I guess.\nSpeaker A: Should we...\nSpeaker A: Or is it important that we leave it to exact the...\nSpeaker D: No.\nSpeaker B: No.\nSpeaker B: Okay.\nSpeaker B: I don't think so.\nSpeaker B: Yes.\nSpeaker B: Yes.\nSpeaker A: Go ahead.\nSpeaker A: We'll see each other in 30 minutes.\nSpeaker B: Yeah.\nSpeaker B: Good luck.\nSpeaker A: Yeah.\nSpeaker B: Good luck.\nSpeaker B: I won't need it.\nSpeaker B: I won't need it.\nNone: I'll be right back.\nNone: I'll be right back.\nNone: I'll be right back.\nNone: I'll be right back.\nNone: Okay.\nNone: Okay.\nNone: Good luck.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: Well, I'm going to make some more minutes.\n"}, {"meeting_id": "IS1000b", "summary": "The whole meeting was focused on the target group and the functionality of the new remote control product. After Project Manager briefly reaffirmed the meeting procedure, Marketing, Industrial Designer, and User Interface each gave a presentation about trend-watching, working design, technical function respectively. Then, Project Manager started a group discussion about the demand for teletext and video, for which thick key design was accepted as a feasible solution. Then, the group discussed the target group based on the predetermined budget, and accepted universality as an important feature.", "dialogue": "None: Yes, this one?\nNone: This light.\nNone: Here are my slides.\nNone: For participants.\nNone: Yeah, exactly.\nSpeaker D: Yeah.\nNone: I'm going to come on.\nNone: You're doing something?\nNone: Oh, good.\nNone: I had the camera in this case.\nNone: You should have it easier to write.\nNone: I'll put it in.\nNone: I'll put it in.\nNone: Okay, I didn't get it.\nNone: It's good.\nNone: You have me.\nNone: I just saved it.\nNone: It's for the radius.\nNone: I need it.\nNone: You have to save my camera.\nNone: I'll just open it.\nNone: You can switch it.\nNone: Yes.\nNone: I'll just open it.\nNone: I'll start off.\nNone: Okay, you.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Is it on the runway?\nNone: Yes.\nSpeaker B: Okay.\nSpeaker B: So, I see everybody's here.\nSpeaker B: Okay.\nSpeaker C: What's the agenda for this meeting?\nSpeaker B: I will present the agenda with slides to you.\nSpeaker B: As you can see here.\nSpeaker B: Perfect.\nSpeaker B: So, first, just to mention, I will take notes of this meeting, and I'll try to work them out and give them to you.\nSpeaker B: So, I'll make notes of the previous meeting, and I was about to send them to you, but then I had to go to this meeting, so you will get them too.\nSpeaker A: You are the secretary.\nSpeaker B: Yes.\nSpeaker B: Indeed.\nSpeaker B: I will have worked out some presentations about the task given to you in the previous meeting.\nSpeaker B: In a minute, we will start with them.\nSpeaker B: We will see which order.\nSpeaker B: We will handle them off. Then I will bring in some new requirements.\nSpeaker B: I got from the account manager. I tried to work them out. They were quite abstract, and we can have maybe have come from discussion about it.\nSpeaker B: About the functions. Well, in this meeting, we should really try to reach the decision about the target group and the functionality of the...\nSpeaker A: You mean the social target group?\nSpeaker B: Yes, I mean, well...\nSpeaker B: Yes. Who are we going to get?\nSpeaker B: The customers.\nSpeaker B: The customers.\nSpeaker B: Indeed. Yes.\nSpeaker B: I think that's important.\nSpeaker C: Great question.\nSpeaker B: So, and then we will close this meeting, and after this meeting, we will have a lunch.\nSpeaker B: Good.\nSpeaker B: Maybe...\nSpeaker B: Why Anna, do you have a press tape?\nSpeaker B: No, I don't.\nSpeaker B: You don't have a press tape?\nSpeaker B: I wasn't.\nSpeaker C: No?\nSpeaker B: You are not able to...\nSpeaker B: I can talk about it, but I have no slides.\nSpeaker B: Yes. Maybe you can just talk about it, or maybe you can use the whiteboard if necessary.\nSpeaker C: Well, I have just been presented with some research we have done in a small focus group, so 100 people just asked them about their remote control usage habits and what they wanted in remote control.\nSpeaker C: It's probably got email as to you.\nSpeaker C: I have just got a webpage with some data on it.\nSpeaker C: Basically, it's saying that users generally dislike the look and feel of their remote controls.\nSpeaker C: 75% of users find most remote controls ugly.\nSpeaker C: 80% of users would spend more money when a remote control would look fancy.\nSpeaker C: Current remote controls do not match well the operating behaviour of the user.\nSpeaker C: 75% of users said they zap a lot, so they use their remote control quite frequently while they're watching television.\nSpeaker C: 50% of users say that they only use 10% of the buttons.\nSpeaker C: So they've got a remote control with a lot of functionality, but really most of the time they only use a small part of that.\nSpeaker B: Do you have this information on the webpage?\nSpeaker B: I have a webpage, yes.\nSpeaker B: Maybe you can send an email to me later.\nSpeaker B: Yep, sure.\nSpeaker C: About this.\nSpeaker C: Basically, there's a breakdown of how much they use the different functions that are remote control.\nSpeaker C: Power and volume selection are only used a few times within this per hour.\nSpeaker C: Channel selection is used 168 times.\nSpeaker C: And then there's things like channel settings, audio settings, which are only used very infrequently.\nSpeaker C: Kelly Texas used 14 times in the hour, so it is used but not nearly as much as the channel selection is used.\nSpeaker C: An interesting thing that this report has brought up is that 50% of users report that the remote control gets lost a lot of the time in the room.\nSpeaker C: Yeah, yeah, I have to.\nSpeaker C: So my location in the remote control would be very useful to a lot of users.\nSpeaker C: 34% said it takes too long to learn to use the remote control that once it's easier to use straight away more intuitive, perhaps.\nSpeaker A: Easy to learn or do you say?\nSpeaker C: 34% said it took too much time to learn to use new one.\nSpeaker C: And 26% said remote controls are bad but RSI. I don't know how we just got about one comment in that.\nSpeaker C: What did you do?\nSpeaker C: For RSI, are you prepared to destroy an injury?\nSpeaker C: Okay.\nSpeaker B: I think that our doctor says that it's the opinion of the user.\nSpeaker C: Yeah, that's what the report says.\nSpeaker C: And then it's got a demographic breakdown.\nSpeaker A: You cannot put this webpage online on the...\nSpeaker C: I should be able to actually. If I'm out of here now.\nSpeaker D: You can disconnect it then.\nSpeaker D: You can maybe just...\nSpeaker D: Oh, yeah.\nSpeaker A: Okay. It's a webpage on C.\nSpeaker A: Yeah.\nSpeaker A: It's a file. Okay.\nSpeaker A: Otherwise you...\nSpeaker A: Yeah.\nSpeaker A: You can connect this one.\nSpeaker A: You can connect this one.\nSpeaker D: Yeah.\nSpeaker D: What?\nSpeaker D: What do you have?\nSpeaker D: It is.\nNone: Okay.\nSpeaker C: Well.\nSpeaker A: So these are important numbers that Matthew and I need to take into account for...\nSpeaker A: Oh, they took a background with us.\nSpeaker C: It's probably easier if you put on yours than the amount to you.\nSpeaker C: This is a web link.\nSpeaker D: Okay.\nSpeaker A: Yeah.\nSpeaker A: These numbers have to be taken into account for both user interface and functional design.\nSpeaker C: One thing it goes on to talk about, which is interesting, is the...\nSpeaker A: Because if there are many numbers and we need to select to constrain our design based on what is more important.\nSpeaker C: Yeah.\nSpeaker C: One thing that's interesting is talking about speech recognition in a row of control.\nSpeaker A: Speech recognition in?\nSpeaker C: And who would pay more for that and whether people would find it useful?\nSpeaker C: Do you have no thought on that?\nSpeaker C: Yes.\nSpeaker C: I'll just get this up.\nSpeaker A: So they would not need any button?\nSpeaker A: Or potentially, yeah.\nSpeaker A: Or based on speech.\nSpeaker A: Okay.\nSpeaker A: I think even for...\nSpeaker C: Interesting idea.\nSpeaker C: Yeah, I think that would not work so well.\nSpeaker C: You wouldn't have both options.\nSpeaker B: Okay.\nSpeaker B: Well, it would be a solution for when your remote control is lost.\nSpeaker B: I mean, when it has speech recognition, then...\nSpeaker B: Yeah.\nSpeaker B: Then it doesn't matter where it is.\nSpeaker B: Well, it should be in range.\nSpeaker B: Or maybe it can respond and produce sound, so say where it is.\nSpeaker B: Yeah.\nSpeaker B: But these are all quite fancy futures.\nSpeaker B: I'm not sure whether...\nSpeaker B: Go ahead.\nSpeaker B: We can make this for 12 euro, 15 cents.\nSpeaker A: No, you can't.\nSpeaker A: And we don't know where the state of the earth of speech recognition is.\nSpeaker A: Maybe, you know...\nSpeaker D: Oh.\nSpeaker D: Well, it depends.\nSpeaker D: There is a very small vocabulary where you want to do the operations.\nSpeaker D: Like, you want to say, on, off, one, two...\nSpeaker D: But it's quite noisy.\nSpeaker A: You're doing the TV.\nSpeaker A: Yeah.\nSpeaker D: Yeah, it's not going to be so easy, but...\nSpeaker D: Usually, it's going to be more of an isolated...\nSpeaker D: Do you have some more?\nSpeaker B: Okay.\nSpeaker D: But I don't know.\nSpeaker D: It's 25 cents.\nSpeaker A: So, yeah, just to summarize maybe the...\nSpeaker C: This is now talking about who would pay for speech recognition in a remote control, who would pay more for it.\nSpeaker C: 90% of the 15 to 25 year old market said that they would pay more.\nSpeaker C: It goes down from there, 76% for 25 to 35, 35 to 45, 22% for 45 to 55, and then 8% for 55 to 65.\nSpeaker C: Okay.\nSpeaker C: Big glare.\nSpeaker C: Okay.\nSpeaker C: It really depends where we're going to be targeting this product, which we'll be talking about later, I think.\nSpeaker B: Yes, we will talk about this later.\nSpeaker C: Did you get the email?\nSpeaker C: Yeah, that one.\nSpeaker C: Just follow that link.\nSpeaker A: It'll be in a different window.\nSpeaker C: Okay.\nSpeaker C: Left.\nSpeaker A: Okay, perfect.\nSpeaker C: So, that's the biggest I was just talking about there, with the different demographics.\nSpeaker C: Another thing that's talking about there is the LCD screen, but there's no figures apparently on that.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Maybe, can you give your presentation now?\nSpeaker B: Okay.\nSpeaker A: I stay.\nSpeaker D: No, you can move.\nSpeaker B: You can move.\nSpeaker A: I can move as far as...\nSpeaker A: Maybe I take your chair.\nSpeaker B: Yes, you can take my chair.\nSpeaker B: It's a tenor solution.\nSpeaker D: How much is that?\nSpeaker D: It's actually good.\nSpeaker D: Yeah.\nSpeaker A: So, I think, as everybody knows, I'm the industrial designer.\nSpeaker A: And in this presentation, this quick presentation, is going to focus on the working design of the product.\nSpeaker A: And I'd like first to give a quick, simple introduction.\nSpeaker A: How does it work so that everybody knows, even if you don't have a very technical background, or is it, I think, in the project is important.\nSpeaker A: So, basically, the basic function of the country is to send messages to another system that is fixed.\nSpeaker A: And so an energy source fits an integrated circuit, the chip, that can compose messages, usually through infrared bits.\nSpeaker A: And the user interface controls the chip, and accordingly the messages.\nSpeaker A: So, my method for designing the work design, first, the main point is that I would wish to make a really functional product.\nSpeaker A: I would prefer to have very functional capabilities, whether than fancy stuff that, in fact, is not used and doesn't work.\nSpeaker A: So, for that, yeah, it's important to take into account the user requirements from the marketing expert, Anna.\nSpeaker A: And we should agree on what are the technical functions for this remote control, and I'll show you the working design.\nSpeaker A: So, basically, here is a very large view of what we want, on a button.\nSpeaker A: It can be simple, but it's important.\nSpeaker A: And also enabling the two broad channels, as well as other buttons that come after.\nSpeaker A: So, the components I quickly draw here is that in this part you have the remote control, the sender, and the other part, the receiver.\nSpeaker A: So, that's my method is, well, my aim would be to design the, and choose the chips and the infrared components to build the remote control.\nSpeaker A: So, of course, we need energy sources and other receiver.\nSpeaker A: This is a very quick design.\nSpeaker A: You stop me or interrupt me if you don't agree on it on that.\nSpeaker A: So, what I found, after a lot of work, actually, I draw for you this schema that can be maybe too technical for you, but it's really important for me.\nSpeaker A: You do it a long time ago.\nSpeaker A: Oh, well, I won't.\nSpeaker A: And, see, so I won't go into details about that, but these are my preferences to use the kind of components.\nSpeaker B: And why do you want this kind of component?\nSpeaker B: Are they cheap or are they reliable?\nSpeaker B: What were your...\nSpeaker A: So, the main components you see here are the cheapest I found and you have always a compromise with reliability and it's expensive, but these ones are also really reliable.\nSpeaker A: So, yeah, that's it for the working design.\nSpeaker A: I hope you get clear view on what a remote control is in terms of technical components, but maybe, yeah.\nSpeaker B: It's more clear now, I think.\nSpeaker D: But is it... can you just buy it on the market and plug it in or you want to...\nSpeaker A: No, no, no, no, this is a preference, but we can always change.\nSpeaker B: What I was thinking about, the schema about the sender and the receiver, can you get back to it?\nSpeaker B: Yes, the receiver is, of course, already in the television and we are not able to change it.\nSpeaker B: Of course, yeah.\nSpeaker B: We must adapt to the receiver.\nSpeaker A: I suppose there is a standard way of communicating to the television.\nSpeaker A: We need to adapt to that protocol that already exists, but what we can do is adapting the chips inside to the best chips and infrared bowls.\nSpeaker B: Yes.\nSpeaker A: OK.\nSpeaker A: Thank you.\nSpeaker D: It's just you have to change the frequencies.\nSpeaker A: The frequencies? Yeah.\nSpeaker A: Of course, in the chip you have...\nSpeaker D: But you should be careful people are something becoming a problem like a guy has recently designed a remote with a switch off any other TVs.\nSpeaker D: Yeah. So basically it goes through all the things, so maybe you should think of...\nSpeaker D: Of course, yeah, we should take that into account.\nSpeaker C: Just take the TV in the next apartment through the lab.\nSpeaker C: So you can just tell us...\nSpeaker D: Just watch the seat and then switch off everyone's TV and you can just walk away.\nSpeaker D: You don't have to be near the TV at all.\nSpeaker B: I like that idea.\nSpeaker B: I think Miles will consider this thing.\nSpeaker B: Maybe we can go to your presentation.\nSpeaker B: Yes, I assume we were finished.\nSpeaker B: Yes.\nSpeaker B: OK.\nSpeaker D: So I can take my...\nSpeaker D: No, no.\nSpeaker D: OK.\nSpeaker D: So...\nSpeaker D: Well, I...\nSpeaker D: I can take mine. It's OK.\nSpeaker D: Well, I...\nSpeaker D: So...\nSpeaker D: OK.\nSpeaker B: So...\nSpeaker B: Sorry.\nSpeaker B: I know where this...\nSpeaker D: It's on a desktop. It's a technical...\nSpeaker D: Oh, OK.\nSpeaker D: That's all.\nSpeaker D: That's all. So... I'm going to talk a little bit about the technical function.\nSpeaker D: So what...\nSpeaker D: Actually, it's about what is the user going to do?\nSpeaker D: I think Miles has presented what is going inside.\nSpeaker D: So what's the user is going to see from the outside and how he's going to use it?\nSpeaker D: So...\nSpeaker D: Well, the approach is that...\nSpeaker D: Basically, the idea is to send a message to the TV set.\nSpeaker D: As Miles has pointed, and it will be decoded by the TV.\nSpeaker D: And usually, we... it is easier to have keys or buttons with which people can press and then...\nSpeaker D: Changing a button will basically change the message which is being sent to the TV.\nSpeaker D: And...\nSpeaker D: And basically, it sends an infinite signal and decoded by the receiver.\nSpeaker D: So... as... Anna has said that these people are interested in things which are...\nSpeaker D: You don't need to press the keys.\nSpeaker D: People can have a speech recognition, but this is a question which we have to see later.\nSpeaker D: But in the present scenario is that you have certain keys and you press it like your phone.\nSpeaker D: And it sends a message to the TV.\nSpeaker D: Yeah. And...\nSpeaker D: So...\nSpeaker D: Generally, I don't have some figures, sorry, but...\nSpeaker D: So, there are two kinds of remote if you popularly in the household.\nSpeaker D: Actually, so you have a standard TV remote where you have just an on-off button and play...\nSpeaker D: Volume change and...\nSpeaker D: Keys for the number and more than one digit option.\nSpeaker D: And if you see, for example, right now...\nSpeaker D: Even the one... more than one digit option is for two digit... which is like 99.\nSpeaker D: But tomorrow you might have 150 channels, you know, to browse or 200 channels to browse.\nSpeaker D: Who knows? But... then there is a...\nSpeaker D: This is a standard one without any fancy thing, you know, like...\nSpeaker D: It doesn't have to let text option. It can... without any...\nSpeaker D: It's a very simple thing... which you can browse.\nSpeaker D: And then you have a... what's the... video remote... which is like...\nSpeaker D: Usually it has almost all the keys over there, but then it has other options like...\nSpeaker D: Stop... and then you play the movie or...\nSpeaker D: Or fast forward the movie or something like that.\nSpeaker D: So... it has those... so these are the standard commonly found remote controls in the...\nSpeaker D: market. And then... which is generally used by the people.\nSpeaker D: And then... well, personal preferences, I would...\nSpeaker D: Basically think of having a kind of aim for the next generation thing where we could have both...\nSpeaker D: The... TV and the remote... video remote control, because...\nSpeaker D: Some of the keys in the videos remote control and the TV...\nSpeaker D: They could be integrated together so that we could...\nSpeaker D: Aim for the... like in the coming future...\nSpeaker D: The type of... applications.\nSpeaker C: Okay. How would that work? So I've got...\nSpeaker C: Say maybe a VCR and a TU which is separate.\nSpeaker C: So on my one at home, I've got a VCR remote which then changes the channel on the VCR...\nSpeaker C: Yeah. And then the journey on the TV.\nSpeaker C: Yeah. So is there gonna be like a switch on the remote that says...\nSpeaker C: Use a TV or use a VCR or does it know which one you want to use?\nSpeaker D: Actually, you could think of having...\nSpeaker D: You can have a key which could tell... it could go to the video thing.\nSpeaker D: But... you still can't... in that case, when you use that, the function should be able to take up the VCR option and you could play it all.\nSpeaker D: You can also think about having like...\nSpeaker D: In a few days, you will be in a few coming years.\nSpeaker D: You might even have a system where you have a separate sitting setup box...\nSpeaker D: And you have something like...\nSpeaker D: You don't... you suppose you are not able to watch some program and actually...\nSpeaker D: It downloading all the time for you and you can just...\nSpeaker D: You know, when you come back, you can just switch on that thing and watch the program.\nSpeaker D: In that case, you want to browse fast or browse slow.\nSpeaker D: You want to have those kind of functionalities.\nSpeaker D: This is a kind of next generation thing.\nSpeaker D: But it is going to come in a couple of years.\nSpeaker B: I think it's already there. I mean, the blog is recorded.\nSpeaker B: I've seen them in the show.\nSpeaker D: Yeah. So it's going to record your things and you...\nSpeaker D: You need basically the functionalities, what you need in both the...\nSpeaker D: Video as well as in the standard TV thing.\nSpeaker C: But I don't think we are trying to make a universal remote here.\nSpeaker C: No, we are not making a universal remote.\nSpeaker D: We are just looking at giving a scenario.\nSpeaker D: I have a TV and tomorrow I am going to have a setup box which is going to sit there and...\nSpeaker D: It's going to do that for me.\nSpeaker B: We need to decide on how far we go to in this...\nSpeaker B: I mean, can go pretty far, I think, with functions and possible future...\nSpeaker D: Yeah. So...\nSpeaker B: But it's good to keep in mind.\nSpeaker D: Okay. So that ends my presentation.\nSpeaker D: So we can always discuss about it.\nSpeaker D: For example, the video market actually...\nSpeaker D: This demand, video over demand or what we call it as...\nSpeaker D: It's presently booming up actually.\nSpeaker D: So people are providing things like...\nSpeaker D: Movies you can select actually.\nSpeaker D: So you want to watch a movie and your provider gives a list of movies.\nSpeaker D: And then you select those lists.\nSpeaker D: And it basically you go out, it downloads a movie, it gives for you.\nSpeaker D: And then when you come, you want to watch it on your TV.\nSpeaker D: This is going to be...\nSpeaker A: Or even you don't need to download it streamed online.\nSpeaker A: Yeah, it can be streamed online for you.\nSpeaker D: And you can say what time I want to watch.\nSpeaker D: Yeah.\nSpeaker B: So I have received some points of thinking over of my account manager.\nSpeaker B: And I would like to share them with you.\nSpeaker B: First thing is teletext is a well-known feature of televisions.\nSpeaker B: But it's getting used less and less.\nSpeaker B: That's especially because of the internet of course.\nSpeaker B: So we should think about it.\nSpeaker B: Do we include it?\nSpeaker B: And do we give it a prominent place on the remote module control itself?\nSpeaker B: As Anna indicated, it's not used very much, but it is still used.\nSpeaker B: So do we stay to the television only?\nSpeaker B: The television as we all know it with broadcasting signals and you can't go back.\nSpeaker B: Or do we go further as Matthew indicated by supporting recording devices?\nSpeaker C: So David is in BCS?\nSpeaker B: Indeed.\nSpeaker B: And the hardest recorders.\nSpeaker B: Furthermore, we need to really interest younger customers.\nSpeaker B: And then with younger customers, I mean people below the age of 40.\nSpeaker B: And how current customers are mainly 40 plus.\nSpeaker B: So it is a market, but they will grow older and they always need to have the future with younger people.\nSpeaker B: Therefore, younger people like friendly designs.\nSpeaker B: So we should make our city as friendly as possible, but it should also be a reliable image.\nSpeaker B: So when it looks too spacey or too fancy, people will think, well does it work at all?\nSpeaker D: Well, you can follow the ideas how you want to keep the keys.\nSpeaker D: You know, if you take it, you have like 0, 1, 2, 3, like keys separately.\nSpeaker D: But suppose if you take the present end of mobile phones, there are like big thick keys.\nSpeaker D: You press on the top, it takes one number, you press on the bottom, you take another number.\nSpeaker D: So the space covered, so you don't see two separate keys there actually.\nSpeaker D: So it is like, maybe you can draw it on the board.\nSpeaker C: I think taking the idea of getting inspiration from mobile phones is interesting.\nSpeaker C: Especially if we're going after a younger market.\nSpeaker A: That's already used to that.\nSpeaker C: The new and the funky things.\nSpeaker C: Yeah, there's lots of very pretty mobile phones, not too many pretty remote controls.\nSpeaker B: And they're skilled by using it.\nSpeaker D: So for example, for example, you have a key like 1, 2, 3 like this.\nSpeaker D: And you can have keys like this form like that.\nSpeaker D: How much longer have we got for the meeting by the way?\nSpeaker C: Because you have to have a demographic at all, that's a very important issue.\nSpeaker D: So you can have keys like, which are like, so, sorry.\nSpeaker D: So you basically don't change the original order of them.\nSpeaker D: But then the keys are more spacious, they don't look, so there is a very slight thing.\nSpeaker D: So if you press on the top, it takes the 1, it takes the 3, 4, sorry, 4 here, 5 or 6.\nSpeaker D: So the keys can be, it looks, you know, not very much cluttered.\nSpeaker D: But it looks nice.\nSpeaker D: You don't have too many keys, but you can have a lot of options.\nSpeaker D: I think now the idea is clear.\nSpeaker B: We should now try to decide on our direct group.\nSpeaker C: Yeah, which I think is quite tricky.\nSpeaker C: Basically we're trying to get people to buy remote control.\nSpeaker C: Wouldn't they already have a remote control with their television when they buy one?\nSpeaker B: Of course they have already one.\nSpeaker B: So our remote control has to be better.\nSpeaker C: But it's not going to have more functionality because it's only a low market, it's a cheap end or a control. We can't beat them on functionality.\nSpeaker C: We might be able to beat them on the look of it, the design of it.\nSpeaker C: But that's not a big seller.\nSpeaker C: If they're not going to just go and buy a new remote control, just because it looks pretty.\nSpeaker C: They have to actually need it as well.\nSpeaker C: So I'm not sure how we can get people to buy this thing.\nSpeaker B: Well, I think many people said in your research the appearance of the RC is important when they're buying one.\nSpeaker C: But why are they buying one in the first place?\nSpeaker C: Indeed.\nSpeaker B: So that will be about functionality.\nSpeaker C: But if people are buying new remote control for functionality, they'll buy a universe remote.\nSpeaker C: I've got friends who have got so many things they need a universe remote, otherwise they're using five different modes of their older things.\nSpeaker C: In that case they wouldn't buy our product because it doesn't give them what they need in terms of functionality.\nSpeaker B: So you think we should go for a more universal type of thing?\nSpeaker C: Well, we can't with the price range.\nSpeaker C: We're not building a universe remote, we're not building a high-end product.\nSpeaker A: But we will think about what we're buying.\nSpeaker A: Yeah, we have 12.5 euros per RC.\nSpeaker A: And I think with this, now you know the chips are very cheap.\nSpeaker A: And we can include it in our control some new features.\nSpeaker A: But yeah.\nSpeaker C: If we're getting into universal remote territory, we're getting to LCD screens and things like that, which would drive the cost up a lot.\nSpeaker B: I don't know whether that's necessary.\nSpeaker B: I don't think the university is necessary.\nSpeaker B: I think this could be a market because universal remote controls tend to be quite expensive.\nSpeaker B: And quite complicated to use.\nSpeaker B: So we can try to go in between and offer a product which is not as expensive and not as complicated.\nSpeaker B: No, it's flexible maybe.\nSpeaker B: But still people have the idea that this is more functional than a normal universal RC because it has more...\nSpeaker B: It's some kind of universal.\nSpeaker C: But if we're going for the say 15 to 25 age group, then not many that would actually own TVs to use the remote control on.\nSpeaker B: Yes, we're targeting I think more on the 20 to 40 group.\nSpeaker B: So people would just have or already have a job and have the money, but may not want to spend that much money on a universal.\nSpeaker C: I don't know really what the price range rate control is.\nSpeaker C: Are we going to be at the very bottom of the price range or are we kind of middle to bottom?\nSpeaker B: I don't know.\nSpeaker B: Well, I think when we think it over I think we are trying to offer the kind of universal control for less money.\nSpeaker B: Yeah.\nSpeaker B: So do you agree?\nSpeaker D: Well, it's fine with me.\nSpeaker D: Like the price as long as it is not too expensive.\nSpeaker A: Yeah, because we have to take into account that we're going to sell around 4 million.\nSpeaker A: So when we speak about these numbers, the price of a cheap is very cheap.\nSpeaker A: So I'm okay for designing less, a kind of universal RC.\nSpeaker B: You think it's possible for the 12, your 50?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So then we decide on going to this more universal kind of control.\nSpeaker D: Oh, yeah.\nSpeaker D: That's what we need and basically that's needed right now.\nSpeaker C: Yeah, that's needed.\nSpeaker C: Basically you can look through the standard.\nSpeaker C: And you've found together market really needed that.\nSpeaker C: So I guess what I'd like for my university remote is maybe choosing between three devices, being able to switch between them, maybe stereo, VCR and TV.\nSpeaker C: Exactly.\nSpeaker C: And just be able to use them all from the same remote, but not at the same time.\nSpeaker D: You can also browse through all the standards, you know.\nSpeaker D: There are the limited standards for all of them and you can just browse through them.\nSpeaker B: So given we are going for this universal type, maybe it's good when you try to find out and you definitely need it and you will try to get more specific user interface consent.\nSpeaker B: Okay.\nSpeaker B: And maybe you can look on what trends are in this type of market.\nSpeaker B: Well, yeah.\nSpeaker B: So anyone has a point to bring in or shall we?\nSpeaker D: Oh, I don't have anything right now.\nSpeaker D: That's right.\nSpeaker D: We'll go and we'll be sure we'll come up with something good for you.\nSpeaker D: Yes, we can have plans now.\nSpeaker B: So, yeah, so we'll meet in the next meeting.\nSpeaker B: We'll often learn to do a few things.\nSpeaker B: We have 30 minutes of work and then we'll have the next meeting.\nSpeaker B: But you will inform fire at the computer.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Perfect.\nSpeaker A: We'll do it.\nNone: Okay.\n"}, {"meeting_id": "TS3012b", "summary": "In this meeting, the team first gave presentations about their ideas most about remote control's functional design based on research and knowledge they had already known. And then Project Manager delivered new requirements for the look, functions, and some others about the remote control. Later on the team brainstormed about many functional designs\u00e2\u0080\u0094where to put buttons and how much size they should be, the necessity of beep and flash, the design of covers, and the choice of LCD screen and rechargeable batteries\u00e2\u0080\u0094of the remote control based on new requirements, researches, current technology and production price. Finally they decided about the remote's button design and had an optimistic attitude towards rechargeable things but there still needed more market research when they gave a sketch of the new product.", "dialogue": "Speaker B: I'll wait for the marketing director, actually.\nSpeaker B: Anyways, I knew Disney's me.\nNone: Nothing.\nSpeaker B: So you show me a way to make sure he's later, or later, whatever.\nSpeaker B: I'm going to start to have, no, don't have much time anyway.\nSpeaker B: Sorry, okay.\nSpeaker B: A little bit of a problem.\nSpeaker B: We're about to start, so I'll see.\nSpeaker B: Okay, welcome again today, functional design phase.\nSpeaker B: I'm going to take you over the minutes of last meeting.\nSpeaker B: Okay, that was just to get to know each other, have a little thoughts on what your vision is on this project.\nSpeaker B: I put the minutes on the, I made on the project chair, so if you want to review them, they're there.\nSpeaker B: I'll do this after every meeting.\nSpeaker B: So if you have some information, you want to take back, you can find it there.\nSpeaker B: Anyways, today's presentation is for everyone of you.\nSpeaker B: After I got some new project requirements from the project board, so I'm going to go over those later.\nSpeaker B: But I want to start with the stuff you did first, so we can see what everybody came up with.\nSpeaker B: And after that, we can have the new requirements and shares and thoughts.\nSpeaker B: We want to go first.\nSpeaker D: Yeah, share no problem.\nSpeaker D: Take it.\nSpeaker D: There was a little problem with my computer, so not the wool press station.\nSpeaker B: Anyway, let's see what you have.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker B: I still have it open.\nSpeaker D: I want to open the mice.\nSpeaker B: Oh, no.\nSpeaker B: I should close it on your own notebook, I guess.\nSpeaker D: Oh, no.\nNone: Okay.\nSpeaker D: It's not true.\nSpeaker D: Yes.\nSpeaker D: The functional requirements, it's very important for the user.\nSpeaker D: It wants to, yeah.\nSpeaker D: The method we used, it's not a slide because it went wrong.\nSpeaker D: But the method we used, we tested it with 100.\nSpeaker D: 100 men.\nSpeaker D: And we asked them to what the remote feel like and what's important.\nSpeaker B: Is it people or men?\nSpeaker B: People, sorry.\nSpeaker B: People.\nSpeaker B: Okay.\nSpeaker B: Both women were the only men.\nSpeaker D: Okay.\nSpeaker D: The findings, 75% of the use is fine.\nSpeaker D: Most remotes are ugly.\nSpeaker D: That's pretty shocking.\nSpeaker C: Yeah, we have to do something about that.\nSpeaker D: Yeah, they want to spend money for a better system, for better remote control.\nSpeaker D: So we can do a little nice things with it.\nSpeaker D: And the use, yeah, the use, a lot of it doesn't say that.\nSpeaker D: So that's the most important thing.\nSpeaker D: Oh, yeah.\nSpeaker D: Not all is it on my PowerPoint presentation.\nSpeaker D: The relevant buttons are the power, the channel selection and the volume selection.\nSpeaker D: It's the most basic buttons the user wants to use.\nSpeaker D: And less important teletext.\nSpeaker D: They use it, but it's not very important on a scale of 0 to 10.\nSpeaker D: They 6.5.\nSpeaker D: Okay.\nSpeaker D: But not important is the channel selection.\nSpeaker D: That's a little weird.\nSpeaker D: Which channel?\nSpeaker D: No, no, no, no.\nSpeaker D: That's very important.\nSpeaker D: But not important is audio settings, display settings.\nSpeaker D: And we could hide those in a menu or something.\nSpeaker D: And new preferences, and beep to find your control.\nSpeaker D: That was open the test.\nSpeaker D: The most people find it irritating when they cannot find their remote control.\nSpeaker D: So I think it's beep to sound it and you can find it.\nSpeaker D: And another thing they want was speech recognition.\nSpeaker D: So they can say what they want.\nSpeaker D: Let's go to channel one.\nSpeaker D: That's a kind of thing.\nSpeaker D: Anyone that may be on LCD screen to look at what's on every channel.\nSpeaker B: What they want.\nSpeaker B: Preview on the remote control.\nSpeaker B: Yeah.\nSpeaker B: Preview on the channel.\nSpeaker C: Okay.\nSpeaker C: Is that manageable?\nSpeaker C: Because it sounds pretty expensive.\nSpeaker B: It sounds too easy.\nSpeaker B: It's possible.\nSpeaker B: I think it's best to continue.\nSpeaker D: Yeah.\nSpeaker D: My personal preferences.\nSpeaker D: It's better for my favorite channel so I can zip to my quick to my favorite channel.\nSpeaker D: So the remote must see or see what my preferences are for which channel.\nSpeaker B: I can tell if you have a step to the channel that you're on most.\nSpeaker B: You want it to be programmed, for example, or you want it to recognize your favorite channel.\nSpeaker B: You spend 20 million each day on that channel so it recognizes your favorite channel.\nSpeaker D: Yeah.\nSpeaker D: That's what my personal preference is.\nSpeaker B: Okay.\nSpeaker B: So it does recognize itself.\nSpeaker B: You don't have to say it.\nSpeaker D: Okay.\nSpeaker D: Maybe it's easier to say it.\nNone: Okay.\nSpeaker A: I don't know.\nSpeaker D: It's manageable.\nSpeaker D: I see.\nSpeaker D: We will see.\nSpeaker D: It's a little bit the end of it.\nSpeaker D: Okay.\nSpeaker D: A little bit.\nSpeaker D: I lost the computer request.\nSpeaker B: No problem.\nSpeaker B: It's okay.\nSpeaker B: Should I go?\nSpeaker B: Yeah.\nNone: Go ahead.\nNone: Yeah.\nSpeaker C: So some technical functions.\nSpeaker C: Basically, I have some issues which you discussed earlier.\nSpeaker C: Let's just start with the method.\nSpeaker C: It sounds really easy.\nSpeaker C: What does the user do?\nSpeaker C: What does the remote control do?\nSpeaker C: But there are quite some issues.\nSpeaker C: So the things I'm going to concentrate on are the user aspect because the technical aspect, that's pretty much covered.\nSpeaker C: We can do that.\nSpeaker C: What goes wrong with the user gets the remote control.\nSpeaker C: Where is the remote control?\nSpeaker C: We've all had it once.\nSpeaker C: I want to watch some television.\nSpeaker C: Where's the remote control?\nSpeaker C: That was one of your ideas which you posted in the network folder.\nSpeaker C: It's a very good idea.\nSpeaker C: These are just the issues.\nSpeaker C: I come to some personal experience findings, possible solutions later.\nSpeaker C: Search is for the button.\nSpeaker C: There are many buttons on a remote control which are not clear.\nSpeaker C: So either we lose those or we try to make it a little bit more clear.\nSpeaker C: Also, symbols tend to fade after a while.\nSpeaker C: There's nothing more annoying than faded symbols because you don't know which channel is this button.\nSpeaker C: So possibly we could find something for that.\nSpeaker C: Yes, covered that.\nSpeaker C: Oh yes, user presses the button.\nSpeaker C: Usually when you have a lot of buttons, buttons are small.\nSpeaker C: So you press once, remote control goes completely or something like that.\nSpeaker C: So we have to pay attention not to put too many buttons on the remote control.\nSpeaker C: And possibly also the size, so more important buttons.\nSpeaker C: This is bigger.\nSpeaker C: So this is basically what I had in mind in that.\nSpeaker C: Fade-proof symbols, locator, sound, so clear.\nSpeaker C: We should stick to existing symbols but maybe we could do a little investigation to see whether some symbols need to be replaced by others.\nSpeaker C: This I pretty much covered.\nSpeaker C: So what we want to go to is not this one.\nSpeaker C: But more or less buttons, easy, some bigger buttons.\nSpeaker C: So that's basically where I had in mind.\nSpeaker C: This is not the final design.\nSpeaker C: This is just the general idea of how I'd like to see a general idea.\nSpeaker C: So that was it.\nSpeaker B: That was it.\nSpeaker B: Okay, that was good.\nSpeaker B: So we agree on the part that we need to get something on the remote to find it somewhere.\nSpeaker B: It gives you a really good idea.\nSpeaker B: The durability of the thing.\nSpeaker B: Other aspects we'll just see what he came up with and what's possible for that budget.\nSpeaker A: Yeah.\nSpeaker E: Okay.\nSpeaker E: Okay, that's fine.\nSpeaker E: Okay, now, working on with me.\nSpeaker E: Okay.\nSpeaker E: Let's start.\nSpeaker E: Let's just...\nSpeaker E: Okay, the method.\nSpeaker E: There are a few questions that need to be answered.\nSpeaker E: You already talked about it a little bit.\nSpeaker E: Which buttons are I wanted?\nSpeaker E: Is our remote control universal or should it be programmable?\nSpeaker E: If it should be programmable, then we need something like a mode that you can switch it.\nSpeaker E: Because then the buttons have to send out a different signal than they would normally do.\nSpeaker E: And how big is the remote control going to be?\nSpeaker E: I'll tell you why that's important to me.\nSpeaker E: There are a lot of technical parts in remote control.\nSpeaker E: So that's why I also would like to say go a little bit easy on the designs.\nSpeaker E: You're talking about beeps and about video screens, but the material inside and the technical aspects are quite complex already.\nSpeaker E: So keep in mind that everything that you think of, it has to be built.\nSpeaker E: So that's not as easy as it might look like.\nSpeaker E: Material study, I'm working on that for the costs.\nSpeaker E: I have to check out how far I can go with that.\nSpeaker E: Normally a circuit board is made of fiberglass and wires are made of copper.\nSpeaker E: That is how it is done and all remote controls work that way.\nSpeaker E: I think we can just go on with that.\nSpeaker E: Then I've read more integration of materials, it means less cost for the production.\nSpeaker E: The more we can make at once in one piece, that is cheaper.\nSpeaker B: You may integrate them all into the circuit board.\nSpeaker E: Exactly.\nSpeaker E: So if we make a circuit board with the connections already on it, then that's cheaper.\nSpeaker E: So we have to make something that's not too difficult in the design again.\nSpeaker E: This is what it looks like if you press one button.\nSpeaker E: So this is not the entire thing.\nSpeaker E: You have the power coming in, then you have like a switch.\nSpeaker E: If you press it, then some electrical charge goes into the processor.\nSpeaker E: That thinks of a Morse code.\nSpeaker E: That's how you should see it.\nSpeaker E: A Morse code goes to the amplifier, then the signal ascends to two light bulbs.\nSpeaker E: You have infrared and an, I would say, a light and indication light that you know that it's functioning.\nSpeaker E: Here again, that's my story about the different modes.\nSpeaker E: If you want to make remote universal, then the processor has to make up a different Morse code when some button is pressed.\nSpeaker E: That makes it much more complex, so we really need to have a look at what we want.\nSpeaker E: I don't have any personal preferences so far, except for the materials to be used light at your light.\nSpeaker B: Okay.\nSpeaker B: That was it.\nSpeaker E: I'm going to go back to my thing.\nSpeaker B: Okay, back to this screen.\nSpeaker B: So I got some new information on the project specifications or changing a little.\nSpeaker B: Like you said, TEDx is not very popular anymore because the internet, people don't use TEDx anymore or heartily, so we can either, well, I don't think we should remove the button because there are always people who are using it, but I don't think it should be very, it should be one of the big buttons, for example.\nSpeaker B: Just put it somewhere or under a second option or whatever.\nSpeaker B: It's not important anymore.\nSpeaker B: We're targeting young people now because this is a new product and with this new product we want to appeal to younger people, which are younger people we define under 40.\nSpeaker B: So I think it's also good with the fashion and everything.\nSpeaker D: And they want to pay for it.\nSpeaker B: They want to pay for it.\nSpeaker B: People are willing to spend money actually to buy a remote that they like.\nSpeaker B: They like the way it looks, the way it functions, so they're actually going to spend money on it.\nSpeaker D: With more technical specifications.\nSpeaker B: Yeah, and of course, we can go with that anyway.\nSpeaker B: And one thing that should be important is that the corporate color and slogan are recognizable, which is apparently black and yellow, but I'm not sure if we, I think we should keep the logo in mind because with colors, you can have a lot of fashionable colors and everything on it, which suits everybody's stay.\nSpeaker B: So with that concept, I started thinking, so why not just steal Nokia's idea and just make changeable covers for your, I mean, those cost hardly anything, I think, and people could even spend extra money on buying a cover.\nSpeaker B: So I have an entirely remote control that they like to see, or we could, for example, we could make a different, a basic design and sell the cover separately.\nSpeaker B: For example, it's just a little marketing idea that could be applied.\nSpeaker B: So we can, into people suit really everyone.\nSpeaker B: So you don't have to, I think you don't have to make, Ontario remote controls make a basic one and manufacture the cover separately.\nSpeaker B: So that was, that was my idea on what we could do to appeal this product to everyone.\nSpeaker B: So just, I'm not sure if you came up with anything.\nSpeaker B: In the meantime, after making a presentation, let's go back to share.\nSpeaker E: No, I think this is a good idea.\nSpeaker E: Okay.\nSpeaker A: But that is it manageable.\nSpeaker D: Yeah.\nSpeaker D: With, with an LCD screen, you can.\nSpeaker B: Oh, yeah, I think we should lose the LCD screen.\nSpeaker B: I think, for example, it's huge.\nSpeaker B: I think the LCD is huge.\nSpeaker B: It consumes batteries like hell.\nSpeaker B: I think it takes over a lot of power.\nSpeaker B: It costs too much to have a lot.\nSpeaker B: Okay.\nSpeaker B: What we could do, what could be possible is maybe not an LCD screen with, with a preview, but I'm not sure if it's even possible.\nSpeaker B: For example, a little TV guide.\nSpeaker B: Like you have a, just a text only, not a color, just a little text thing so you can use a remote as a TV guide.\nSpeaker B: I'm not sure it's even possible.\nSpeaker B: After checking that out, maybe you could.\nSpeaker B: Yeah, find a little compromise in that, but, yeah, I think the, yeah, the beep is very simple.\nSpeaker B: Think to implement, just make a button on your TV and just hit the button in beep somewhere.\nSpeaker B: Yeah.\nSpeaker B: I think it's easy to implement.\nSpeaker E: We want to hit, we want a button on the television.\nSpeaker E: I think, yeah.\nSpeaker E: I mean, where else should you put it?\nSpeaker E: Okay.\nSpeaker E: But that rules out universal remote control.\nSpeaker B: Yeah, but how are you going to use that if you're, I mean, if you're running a remote, if you're, I mean, if your remote control is lost, how are you going to press, where are you going to press the button?\nSpeaker B: Exactly.\nSpeaker C: Maybe just a slap on sticker with a button, which sends out a.\nSpeaker B: A slap on sticker.\nSpeaker B: Oh, you mean like a separate thing you can attach to your TV?\nSpeaker A: Yeah, that could be possible.\nSpeaker B: A little box.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker E: I'd like to know now if we want the universal remote to fall in the house because that's.\nSpeaker E: I think it's universal.\nSpeaker B: Everything I'm going to do.\nSpeaker B: I think we should go for universal because apparently we're a separate company making separate remote controls to sell to a lot of people.\nSpeaker B: I think universal remote controls to be possible.\nSpeaker B: Okay.\nSpeaker D: Every one wants to buy it.\nSpeaker D: Yeah.\nSpeaker D: I think we're targeting everyone.\nNone: Yeah.\nSpeaker E: It's fine with me, but then I know what's okay.\nSpeaker B: Universal is good.\nSpeaker B: Speech recognition.\nSpeaker B: I think it's very hard because we're selling across multiple countries.\nSpeaker B: So I think implementing speech recognition is such a small.\nSpeaker B: One.\nSpeaker B: And then I think it's very hard to do.\nSpeaker D: When you say one to it.\nSpeaker B: See, I see Arabian people speaking one to whatever.\nSpeaker C: Besides that, the technology isn't really super yet.\nSpeaker C: So that is a problem.\nSpeaker B: It's not a mature technology.\nSpeaker B: I think it takes a lot of memory and everything in the remote control.\nSpeaker C: It's a good idea, but it's just not.\nSpeaker C: I don't think the market's right for that yet.\nSpeaker B: I don't think it should be implemented in the remote control yet.\nSpeaker B: Yeah.\nSpeaker B: What else do we have?\nSpeaker B: Well, you should look into the materials that are a little durable.\nSpeaker B: So the symbols won't fade, maybe a little harder plastic.\nSpeaker B: Maybe you don't have to do all the buttons, especially the ones that are the popular buttons.\nSpeaker B: So those always fade first.\nSpeaker B: All right.\nSpeaker B: I think that's the most important things that we have now.\nSpeaker B: Let's see.\nSpeaker E: If we make remote control with changeable covers, then we can also make a cover for every language area.\nSpeaker E: That's especially for all the people that they can read it in their own language.\nSpeaker B: We're not targeting older people.\nSpeaker B: Do you remember that?\nSpeaker B: Everything.\nSpeaker B: We target is under 40.\nSpeaker B: You assume that they read correctly.\nSpeaker B: I think the most important thing about young people is that they're really sensitive to trends that are passing through the world.\nSpeaker C: But should the exchangeable covers include the buttons themselves?\nSpeaker B: No, for now, I think it's just something you put over them.\nSpeaker B: That's not even a bad idea.\nSpeaker B: I mean, for example, if you're into a durability issue, for example, if your buttons are faded, if you make a durable remote there are faded, you can just buy a new cover.\nSpeaker B: I'm not sure it's hard to make.\nSpeaker B: But I have no bad idea.\nSpeaker E: The buttons are like an ochit telephone on one sleeve, so you don't have to change your whole cover.\nSpeaker E: If you can make something that you can only remove the sleeve.\nSpeaker E: It works the same as an ochit telephone.\nSpeaker B: Yeah, I know.\nSpeaker B: It's just one piece of rubber.\nSpeaker B: Yeah, that's what I mean.\nSpeaker B: Something you have to look into.\nSpeaker B: Either change both the buttons and the cover or just the cover.\nSpeaker B: I'm not sure which is easy.\nSpeaker B: Anyway, yeah.\nSpeaker B: This is what we're going to do after just lunch break and have more individual work after that.\nSpeaker B: So let's see where we can see the...\nSpeaker B: So you have to come up with the components concepts.\nSpeaker B: Yeah, you want some...\nSpeaker B: You just don't trend watching because even if we're going to do those covers and everything, what people really want, that's what we need to know in this phase because that's going to be the essential final design we're going to come up with.\nSpeaker B: Okay.\nSpeaker B: Yeah, I think usually in the phase it's pretty obvious.\nSpeaker B: I mean, it should be very intuitive.\nSpeaker B: Yeah, it should speak for itself.\nSpeaker B: For example, I bought a remote control last week with an Intv8.\nSpeaker B: It's like old buttons and you have no idea what it does at first.\nSpeaker B: And I'm not a total TV nuke anyway.\nSpeaker B: So I think it should be...\nSpeaker B: Have less or very little buttons and maybe a second level menu for the advanced things.\nSpeaker B: Or maybe just stick them under a menu.\nSpeaker B: Like you said, the sound options and the surround or whatever.\nSpeaker B: The more complicated, just stick them under one menu and give it a...\nSpeaker B: Just put it in a software piece, a menu and you can select everything you want to set on your TV.\nSpeaker C: Is it technically possible to send a signal to a television and then pops up a menu because we're working with different types of television?\nSpeaker C: So we're going to work with that.\nSpeaker E: That is true.\nSpeaker E: I don't think so.\nSpeaker E: No, that's true.\nSpeaker E: I don't think so.\nSpeaker E: Because the television needs to respond to a signal.\nSpeaker C: Yeah, that's true.\nSpeaker E: That's not how it is.\nSpeaker E: Exactly.\nSpeaker C: That's how possible.\nSpeaker C: So basically we...\nSpeaker B: I'm not sure if it's impossible, but there's a chance it's not.\nSpeaker C: Or we could use a double-sided for less-used functions.\nSpeaker B: A double-sided remote control?\nSpeaker B: Yeah.\nSpeaker C: I think that's useful.\nSpeaker C: With the cover, it's basically an idea to overcome these issues.\nSpeaker B: Yeah, but then you're going to have a lot of wasted buttons.\nSpeaker B: For example, you have a Sony TV, the half of the buttons won't function if you have a...\nSpeaker B: Or a Sony.\nSpeaker C: Now, that one's a portfolio of TV.\nSpeaker C: Basic functions, but functions which are not frequently used.\nSpeaker C: I don't think we should.\nSpeaker C: Yeah.\nSpeaker C: Universal remote control.\nSpeaker C: We're going to have to have most buttons on it.\nSpeaker C: A notch.\nSpeaker B: Yeah.\nSpeaker B: You have to make it a little decision between the part if you want to get your universal remote control.\nSpeaker B: That it should do what people usually do with our TVs.\nSpeaker B: The very complicated settings, they can usually do that with either the old remote control if they really hate that thing.\nSpeaker B: But you cannot take into consideration all the different brands of TVs.\nSpeaker B: I think there is a standard, for example, between...\nSpeaker B: Because usually the menu button is usable between different brands, especially the big ones, the big brands.\nSpeaker B: I have a universal remote control and I can use the menu button.\nSpeaker B: I'm not sure if you can put some reason to that if it's possible.\nSpeaker B: I think just the navigation is very basic.\nSpeaker B: It's usually...\nSpeaker E: But I think there must be a way to invoke the more complicated functions of the television.\nSpeaker E: For instance, if your old remote control is broken and you buy hours, then you should be able to get everything out of the television that's in it.\nSpeaker E: Yeah, that's true.\nSpeaker B: Otherwise, you lose functions by buying all of them.\nSpeaker B: I would just reach into that if we can open the menu.\nSpeaker B: I think it's possible.\nSpeaker B: Okay.\nSpeaker B: It's a way how to.\nSpeaker B: Yeah.\nSpeaker B: Then they can do everything, I suppose, because usually the TVs have that inside their little pieces of software.\nSpeaker B: So, it's okay.\nSpeaker E: I think that our remote control should not look like any other.\nSpeaker E: For instance, Sony makes all the remote controls exactly the same for all the devices.\nSpeaker E: And you always grab the wrong one because the formatting of the buttons is exactly the same, only the labels are different.\nSpeaker B: It should be a little distinct from everything else.\nSpeaker B: It's either...\nSpeaker B: Yeah, absolutely.\nSpeaker B: Maybe the shape can be a little different.\nSpeaker D: Maybe it's a little more curved or whatever.\nSpeaker D: Yeah, division.\nSpeaker D: And with different colors.\nSpeaker C: I'll put some on paper and present them next time.\nSpeaker C: Yeah.\nSpeaker B: The idea is very nice.\nSpeaker B: So, I'm on work.\nSpeaker B: Okay.\nSpeaker B: I'm not sure what...\nSpeaker B: Because we had 40 minutes, I'm not sure how much time we have left for the meeting anyway.\nSpeaker B: I heard it beep.\nSpeaker B: Yeah, but it wasn't me.\nSpeaker B: It was him closing something.\nSpeaker B: So, anyway, we could do some work now on the design, for example.\nSpeaker B: What do you want?\nSpeaker B: Not a trilogut thing.\nSpeaker C: Yeah, but we like some curves or...\nSpeaker B: Yeah, maybe something like this.\nSpeaker B: With a little smooth inside.\nSpeaker B: So, you have the transmitter here, for example.\nSpeaker B: Let's see that you...\nSpeaker B: What we'll be having.\nSpeaker B: I think that the button should be...\nSpeaker B: The channel button should be...\nSpeaker B: And it'll be on the round.\nSpeaker B: Let's see.\nSpeaker B: One, two...\nSpeaker B: God damn it.\nSpeaker C: Oh, we get the general idea here.\nSpeaker B: Yeah, okay.\nSpeaker B: Another one here.\nSpeaker B: Let's see what...\nSpeaker B: I think those are always on top of the controls.\nSpeaker B: So, they should be here.\nSpeaker C: And since you're holding it like this, I suggest you put the channel up, channel down.\nSpeaker B: With it.\nSpeaker B: Yeah, just take it.\nSpeaker C: So, you have the up channel.\nSpeaker C: Channel.\nSpeaker C: The volume.\nSpeaker C: Do we want that horizontal or vertical?\nSpeaker B: You take triangles or...\nSpeaker D: Well, that's fine.\nSpeaker B: This is basically what people are accustomed to.\nSpeaker B: Maybe we should make them bigger or whatever.\nSpeaker B: Yeah, but it just is.\nSpeaker B: General idea.\nSpeaker C: Maybe the menu button can go in the middle.\nSpeaker B: In the middle.\nSpeaker B: Usually there.\nSpeaker E: Perhaps you should also make something like a flash on it, if it's lost.\nSpeaker E: For people that are deaf, they won't do it at the beep.\nSpeaker B: Yeah, but for example, if it's lost in your armchair, we'll see a flash.\nSpeaker B: You won't be able to find it.\nSpeaker B: And flash takes up a lot of batteries again.\nSpeaker E: That's true, but it only has to do so when you press a button that is lost.\nSpeaker E: Yeah.\nSpeaker C: We could make a combination that it goes beep and that's...\nSpeaker B: That's an other idea.\nSpeaker B: Yeah, I thought for deaf people, for example.\nSpeaker B: We could do that.\nSpeaker B: Just a little bit.\nSpeaker C: So, we have the basic channels we've got here.\nSpeaker C: The power button.\nSpeaker C: Yeah, that's true.\nSpeaker C: Somewhere over here.\nSpeaker B: It should be central.\nSpeaker B: You shouldn't press it by accident, but it shouldn't be stacked away somewhere.\nSpeaker C: I usually press it on top.\nSpeaker C: At least that's what I'm accustomed to.\nNone: Yeah.\nNone: I have no idea.\nSpeaker E: I'm not sure if it's possible.\nSpeaker E: What would you like to use?\nSpeaker B: Yeah, I thought maybe we should move the button's down.\nSpeaker B: And you're a reading from...\nSpeaker D: You always read from the top to the bottom.\nSpeaker D: The bottom will help the bottom.\nSpeaker B: Yeah, that's true.\nSpeaker B: I think the power button should be on top.\nSpeaker B: Yeah.\nSpeaker B: First thing you do is turn it on.\nSpeaker B: So, a power button on top.\nSpeaker C: Okay, mute button.\nSpeaker C: Is that somewhere here?\nSpeaker B: I think it should be at the bottom somewhere.\nSpeaker C: Is that used often?\nSpeaker C: Sorry?\nSpeaker C: The mute button.\nSpeaker C: The picture you don't have.\nSpeaker C: Oh.\nSpeaker C: I'm pretty much accustomed to it right over here.\nSpeaker C: At least in general.\nSpeaker B: I don't think it's important.\nSpeaker B: I think it should be.\nSpeaker D: Or if the volume selection...\nSpeaker B: No, because people are accustomed to it.\nSpeaker D: It's not at top, but around the volume selection.\nSpeaker D: I don't know where I'm.\nSpeaker B: Exactly.\nSpeaker B: Take this out.\nSpeaker B: See.\nSpeaker B: I'm going to take triangles anyway.\nSpeaker B: I'm drawing triangles, but...\nSpeaker C: That's pretty much what people are accustomed to.\nSpeaker B: Yeah, it's maybe a bigger, lesser than thing.\nSpeaker B: So, anyway, I think this should be the channels.\nSpeaker B: What?\nSpeaker B: Well, I'm accustomed to the channels being...\nSpeaker B: Here, okay.\nSpeaker B: That's better.\nSpeaker B: Okay, should we channel up and down?\nSpeaker E: So, we also look if it's possible to make a rechargeable remote.\nSpeaker E: That you don't have to buy no batteries.\nSpeaker B: So, that is one part.\nSpeaker B: It's a good thing to recharge it.\nSpeaker B: Maybe what could be possible is one with rechargeable batteries, for example.\nSpeaker B: You have to just put pet-night batteries that are rechargeable.\nSpeaker B: It's very annoying.\nSpeaker B: I'm not sure if it's expensive, but just take a look at it.\nSpeaker B: It might be a very good idea, because it's useful to have it rechargeable.\nSpeaker B: But at the same time, if you don't want to recharge it, and you want to watch TV now, you want to be able to put different batteries in it to use it now, and not in half an hour.\nSpeaker E: Okay.\nSpeaker E: Now, you could make a device, but I'm not sure how that's possible with the cost, that you can put in normal batteries rechargeable.\nSpeaker B: I think I have a nice idea.\nSpeaker E: The remote also can act as a rechargeer.\nSpeaker E: So, then you can choose.\nSpeaker E: You have every decision.\nSpeaker E: Know what I mean?\nSpeaker B: Not exactly.\nSpeaker E: You can put in normal pen lights, rechargeable pen lights, but they can also be recharge with the remote, with a wire.\nSpeaker B: I think it's a pretty good idea to have a, like, sort of, maybe a base station that you put on the TV, it could be flat, you could insert your remote into it.\nSpeaker E: But I think that it will cost a lot.\nSpeaker E: Normal wire would be better.\nSpeaker E: Like a PDA, a handheld.\nSpeaker E: You can just put it in the electricity and it charges itself.\nSpeaker B: We were talking about the fact that we wanted to insert either a beep or a flash, or a little separate signaler.\nSpeaker B: So, you could put that on the TV, for example.\nSpeaker B: It could be very flat.\nSpeaker B: It could be very small.\nSpeaker B: It's a very small, yeah, I'm drawing it big now, but.\nSpeaker B: So, you can put your remote on flat, for example.\nSpeaker B: And at the backside of your remote, just a little hole.\nSpeaker B: For example, you just put it on and it recharges, for example.\nSpeaker B: I think it's very good.\nSpeaker C: But again, isn't that too expensive?\nSpeaker C: Because that means that we have to implement that.\nSpeaker B: That's what you buy yourself.\nSpeaker E: I wanted to try to find out that.\nSpeaker E: I'm not sure if there's information about it.\nSpeaker C: It's just an idea.\nSpeaker C: If it's possible.\nSpeaker C: Do people actually want that to pay extra for it?\nSpeaker B: They want to pay for it.\nSpeaker B: They want to pay for it.\nSpeaker B: They want to charge for it.\nSpeaker B: I'm not sure.\nSpeaker B: You should find out if it's rechargeable.\nSpeaker D: There was not an ask.\nSpeaker D: Yes.\nSpeaker E: These are comfort issues.\nSpeaker E: So, I think people will pay, they want to pay for comfort.\nSpeaker B: They want to pay for comfort, which is just so we could either make a separate station or just send a signal to the remote control to either beep or flash to find it.\nSpeaker B: We have to either that or make it integrated with a docking station.\nSpeaker E: I think this is a brilliant project.\nSpeaker E: I would buy it myself.\nSpeaker E: I think it would be good.\nSpeaker B: I like to be part of it.\nSpeaker B: I like the coffee.\nSpeaker B: Can we save this?\nSpeaker B: Yeah, we can save this.\nSpeaker B: We even save the end.\nSpeaker B: Everything.\nSpeaker B: Okay.\nSpeaker B: So, what do we have here?\nSpeaker B: I think you have just a lot of work on if it's possible for the cost.\nSpeaker E: I hope if I have information about it.\nSpeaker B: Maybe you could find out what people are willing to pay.\nSpeaker B: We are going for 24 euro sales price.\nSpeaker B: Just try to find out what they are willing to pay for because if they are willing to pay more, we could lose a little profit and maybe attract more customers and just have to see what that looks like.\nSpeaker E: I would like to make a decision model.\nSpeaker E: What it costs and what kind of materials.\nSpeaker E: Choose what we want to do.\nSpeaker E: We have some financial information.\nSpeaker B: Exactly.\nSpeaker B: That would be nice.\nSpeaker C: Could you post some other essentials of what people want so that I can work with some buttons?\nSpeaker C: Where to put it?\nSpeaker C: You seem to have information on that.\nSpeaker C: I would like to see some of it.\nSpeaker E: Was it not possible to send emails around the office?\nSpeaker E: No, it wasn't.\nSpeaker B: It was possible not allowed.\nSpeaker B: So, that's not why I'm not sure that you're allowed to share.\nSpeaker A: Very popular.\nSpeaker B: So, I don't care.\nSpeaker D: I don't have any information yet.\nSpeaker D: I have the...\nSpeaker C: I have your PowerPoint presentation.\nSpeaker D: The homepage of our Internet.\nSpeaker B: I would include the new one.\nSpeaker B: Where would we...\nSpeaker D: Here's my marketing report.\nSpeaker D: Maybe you can look at that.\nSpeaker C: Where would we want the teletext button?\nSpeaker C: We decided that it's not that important.\nSpeaker B: Do we put it somewhere over here?\nSpeaker D: It just increases a little.\nSpeaker C: This is something for the next meeting.\nSpeaker B: It draws up some designs of the phosphory.\nSpeaker B: I have not a figure.\nSpeaker B: Keep in mind that the important button should be on top.\nSpeaker C: We have decided more or less the basic structure.\nSpeaker B: I can put the other buttons in.\nSpeaker B: Just play a little with this, shift a little up or down, and we'll see what looks best.\nSpeaker E: What did you want to say?\nSpeaker B: Post your designs from time to time on the product share.\nSpeaker D: What I already said is that the remote controls are always lost.\nSpeaker D: But it's also for people, they want to learn it fast.\nSpeaker B: So we don't want very little buttons.\nSpeaker B: Just the buttons use a lot.\nSpeaker D: It's easy to learn.\nSpeaker C: It should cover all the functions.\nSpeaker B: Possibly just an idea that pops.\nSpeaker B: We had a function of what people do.\nSpeaker B: People change channels, people change volume, and they change channels.\nSpeaker B: They turn TV off and on.\nSpeaker B: That's what you do.\nSpeaker B: I'm not sure who presented that again.\nSpeaker B: Those are the basic functions that people use it for.\nSpeaker B: So those should be very well represented.\nSpeaker C: We could go a step further because some TVs have the possibility to adjust brightness.\nSpeaker C: That kind of menus.\nSpeaker B: That's true.\nSpeaker B: That's why we stick under the menu button.\nSpeaker B: Usually in every TV that's configured under the menu.\nSpeaker E: That's why we make this.\nSpeaker E: Because it needs to be configured in a television.\nSpeaker E: If it isn't, then we cannot reach it.\nSpeaker E: We need to adjust.\nSpeaker E: I think most modern TVs have it in their menu.\nSpeaker E: True.\nSpeaker E: Isn't there a possibility to do research in that?\nSpeaker E: We know that for sure.\nSpeaker E: If you rule out functions, then that gets known.\nSpeaker E: Then people are not going to buy it.\nSpeaker E: Then the consumer bond or something says you cannot do this and that with it.\nSpeaker E: If that's commercial.\nSpeaker E: We'll see what we can come up with.\nSpeaker D: Another thing I want to say is that we are looking at the market for the age younger than 40.\nSpeaker D: On my report, I didn't show it in my presentation because of my computer press.\nSpeaker D: They want to pay for an LCD screen and a speech recognition.\nSpeaker E: I want to pay for...\nSpeaker E: Did I really say that like that?\nSpeaker E: Yeah.\nSpeaker D: Those two things.\nSpeaker E: Did I realize how much that costs?\nSpeaker C: That's how much I'm doing.\nSpeaker C: We're not going to be able to sell it for 25.\nSpeaker E: But the LCD screen.\nSpeaker E: If they own it, then in features, we cannot do it for that price.\nSpeaker B: Okay.\nSpeaker B: So we're not focusing on this.\nSpeaker B: All the interesting features, not really the LCD.\nSpeaker B: Would you pay for speech recognition or control?\nSpeaker E: Okay.\nSpeaker D: So we can...\nSpeaker D: Speech recognition is...\nSpeaker D: We can look at the possibilities for an LCD and...\nSpeaker D: Yeah, just look at the possibilities.\nSpeaker B: I don't know how it's about...\nSpeaker B: Apparently it's what people want.\nSpeaker B: How much it's supposed to be the most...\nSpeaker B: It's called a serious remote.\nSpeaker B: Maybe it's not even that expensive.\nSpeaker B: Or find a compromise, maybe just black and white.\nSpeaker B: Or if it works some extra information on it.\nSpeaker E: I doubt it.\nSpeaker E: But I really need finance information.\nSpeaker E: Me too.\nSpeaker B: We all do.\nSpeaker B: I think it's something we should put into consideration.\nSpeaker B: Apparently it's what people want.\nSpeaker B: We should see what it costs if it's possible.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nNone: Yeah.\nSpeaker B: We should do a little thing about the design because it looks boring, really, to me.\nSpeaker B: Even if you put a different cover on it, it looks different.\nSpeaker E: It looks taller, maybe.\nSpeaker E: And the design.\nSpeaker E: It's different.\nSpeaker B: This is like phillipsa.\nSpeaker B: I have no clue.\nSpeaker B: I just do something where it would fit into your hand easily.\nSpeaker C: Something like this to make it kind of futuristic.\nSpeaker E: Oh, I realize if we make it small, then it needs to be a little thicker.\nSpeaker E: I think it's a very old electronica in it.\nSpeaker E: Sorry.\nSpeaker E: If we make it smaller, that's why, then we need to make it a little thicker.\nSpeaker E: Because I have to put all the electrons in it.\nSpeaker B: Yeah, we have to keep in mind it shouldn't be too heavy.\nSpeaker B: Maybe we can stick it in there, I think.\nSpeaker B: Even if in the worst case, you can even, you give me the pen back?\nSpeaker B: Yeah, sure.\nSpeaker B: So let's say people would want an LCD thing.\nSpeaker B: Let's take the basic design again.\nSpeaker B: Oh, crap.\nSpeaker B: What it came up with.\nSpeaker B: Let's make it a little bigger now.\nSpeaker B: Let's make it here.\nSpeaker B: Anyway, we could, let's say we have an LCD screen.\nSpeaker B: People want an LCD screen.\nSpeaker B: So then we should probably put it here.\nSpeaker B: It doesn't have to be really big, but it just has to be there.\nSpeaker B: Things are good place.\nSpeaker B: People do it.\nSpeaker D: If you're reading from top to bottom, I think it's better to put it at the top.\nSpeaker B: It's not the most important function.\nSpeaker B: It's just an extra thing.\nSpeaker B: You press the buttons on top, but your finger is on top.\nSpeaker D: If you are going to put an LCD on it, I think it's very important to use it because it's...\nSpeaker B: How can you use an LCD screen?\nSpeaker D: The function of it.\nSpeaker D: So you can use it maximum because it's caused a lot.\nSpeaker C: Now it's pretty much tucked away in your hand.\nSpeaker B: If you press a button, you can see it right.\nSpeaker B: I'm trying to imagine myself what it would look like.\nSpeaker E: I personally would prefer it on the top.\nSpeaker E: You would prefer it on the top.\nSpeaker B: So we have three people saying it should be on top.\nSpeaker B: Then you would have to stack away your buttons somewhere else.\nSpeaker B: Anyways.\nSpeaker D: It's expensive to build it.\nSpeaker D: I'm still not going to stick it to the maximum.\nSpeaker B: Well, the LCD thing.\nSpeaker B: If it's possible, if it's not too expensive, we should include it because it's cool.\nSpeaker E: We should just try to make out if it's possible.\nSpeaker E: If it is possible, we should really do it.\nSpeaker E: We need that information.\nSpeaker E: Also, keep in mind again, the LCD screen is very flat, but it needs transistors, resistors, I don't know what more, and that needs space.\nSpeaker E: So I'll have to look if that's possible.\nSpeaker D: Yeah, I know.\nSpeaker C: So basically, we have to decide now, is what goes on top.\nSpeaker C: Do we put the volume control and the channel control here?\nSpeaker C: Would we like to press A?\nSpeaker B: We could have something that has it on the bottom, maybe in a circle, for example, like in this example.\nSpeaker B: I think this is not good, but we could make circular buttons, for example, for up and down.\nSpeaker C: Do you agree that we keep this at the center because it's basically the most important?\nSpeaker B: I think the channel button should be in the center.\nSpeaker B: The channel button should be stick together, for example, here in this section.\nSpeaker C: Well, that would make them quite small.\nSpeaker B: We're not sure about the size, anyway, it's just a general design to make it as big as we want.\nSpeaker C: For example, we could take a big remote control, probably not something we could keep in mind.\nSpeaker E: Let's see.\nSpeaker E: If we have a space issue, we can also lose one battery.\nSpeaker E: It is rechargeable, so we can stick with one instead of two minutes.\nSpeaker B: If we even use an LCD screen, we need two.\nSpeaker B: Definitely.\nSpeaker E: If we need two batteries and an LCD screen, that means enough space.\nSpeaker B: We have to see what the size is.\nSpeaker B: LCD doesn't have to be very big.\nSpeaker E: No, but the things behind it, any space.\nSpeaker D: Finish meeting now.\nSpeaker C: So, would you like this or would we like the...\nSpeaker B: I have to decide with what people want.\nSpeaker B: Either this one, this could be the luxurious one.\nSpeaker B: So, we have channel buttons here.\nSpeaker B: I'm just drawing something.\nSpeaker B: I think this wouldn't look cool.\nSpeaker B: You could have, for example, small buttons here for some extra functions.\nSpeaker B: So, if your basic functions here, which, like we decided before, programs up and down,\nNone: we can do that. And what about the speech recognition?\nSpeaker B: You could just...\nSpeaker D: I'm not sure if we could do that, but you could put a microphone in here.\nSpeaker D: Or an LCD or a speech.\nSpeaker B: I think LCD is better now.\nSpeaker B: Just do research into both.\nSpeaker B: What it costs and what it takes for space.\nSpeaker E: But let's cut the meeting for now.\nSpeaker E: One more thing, I'd like to say.\nSpeaker E: Let's give this a name and not like some serial number that no one understands.\nSpeaker E: Let's go.\nSpeaker B: Shall we try to think about a name?\nSpeaker B: Let's say a name.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: Anyways, the minutes will be on the web shortly.\nSpeaker B: There we go.\nSpeaker C: Just go and finish up and I'll see you at lunch.\nNone: Yeah.\nSpeaker E: Good luck.\nSpeaker E: Yeah.\nSpeaker D: Same.\nSpeaker E: Do we have any other...\nSpeaker B: Okay.\n"}, {"meeting_id": "Bro026", "summary": "The meeting occurred close to the finalization of the model for Aurora. The team had put together a model and was tweaking it to see what achieves best performance. The professor suggested that it was time to make some decisions, have several constants, and then test other theories on remaining features. The neural net, for instance, was yet to come together. Still, the model was second in ranking for the task, behind the leader by only one point. The evaluation criteria for the tasks was still not completely clear, but the team seemed to be doing well.", "dialogue": "Speaker C: Okay, so we had a meeting with Heneck, in which snow and stuff on the\nSpeaker D: What was the update? What was the update?\nSpeaker D: Yeah.\nNone: So there is nothing.\nNone: All the new features.\nSpeaker D: All right, suppression.\nSpeaker D: The CVS.\nSpeaker F: Is the CVS mechanism working well?\nSpeaker F: Are people epidemiogi-gramming code via that?\nSpeaker B: I don't think...\nSpeaker B: I don't think they use it?\nSpeaker B: Yeah, I don't think anybody up there is working on it right now.\nSpeaker C: I think it's more likely that what it means is that when Sunil is up there, he will grab it.\nSpeaker B: Yeah.\nSpeaker B: So right now nobody is working on it.\nSpeaker C: Yeah, they're working on different tasks.\nSpeaker C: But what will happen is he'll go back up there and Prattibah will come back from these coasts.\nSpeaker C: And I guess actually after your speech for a little bit, he'll go up there too.\nSpeaker C: So actually everybody who is working on it will be up there for at least a little while.\nSpeaker C: So they'll remotely access it.\nSpeaker F: So has anybody tried remotely accessing the CVS using SSH?\nSpeaker D: I don't know if I read that.\nSpeaker B: I can actually do it today.\nSpeaker B: I can just log into...\nSpeaker B: Have you tried it yet?\nSpeaker B: No, I didn't.\nSpeaker D: I tried it.\nSpeaker D: Yeah.\nSpeaker D: It worked really well.\nSpeaker D: Okay.\nSpeaker D: So right now it's a mechanism with SSH.\nSpeaker D: I didn't set up...\nSpeaker D: You can also set up a CVS server on a new core.\nSpeaker F: Yeah, right.\nSpeaker F: And that's used in CVS password.\nSpeaker F: So when you came in from Belgian, Belgian using SSH, was it asking you for your own password into ICSI?\nSpeaker F: So if you can only do that if you have an account to ICSI.\nSpeaker F: Because there isn't a way to set up anonymous CVS.\nSpeaker D: Yeah, this way you have to set up a CVS server or put them...\nSpeaker D: Yeah, you can access it.\nSpeaker D: Okay, so the anonymous mechanism...\nSpeaker D: You can access it.\nSpeaker F: Because a lot of the open-source stuff works with anonymous CVS.\nSpeaker F: And I'm just wondering...\nSpeaker F: I mean, for our transcripts, we may want to do that.\nSpeaker C: Yeah.\nSpeaker C: Yeah, for this stuff, I don't think we're quite up to that.\nSpeaker C: I mean, we're still so much in development.\nSpeaker F: We want to have just the insiders.\nSpeaker F: Of course.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nNone: Cool.\nNone: Yeah.\nSpeaker C: So...\nSpeaker F: What's name?\nSpeaker C: Well, I mean, I think maybe the thing to be...\nSpeaker C: I'm sure you've just been working on details of that since the meeting, right?\nSpeaker C: I mean, so...\nSpeaker D: Since the meeting?\nSpeaker C: Yeah, that was Tuesday.\nSpeaker C: Training a new PD and a new feature name.\nSpeaker E: Okay.\nSpeaker D: So this should be right?\nSpeaker C: But I guess maybe the thing...\nSpeaker C: Since you guys weren't at that meeting, might be just to...\nSpeaker C: sort of recap the conclusions of the meeting.\nSpeaker C: You're talking about meeting with Heena.\nSpeaker C: Yeah, because that was sort of...\nSpeaker C: We'd sort of been working up to that that he would come here this week.\nSpeaker C: And we would sort of...\nSpeaker C: He's going out of town like now.\nSpeaker C: And I'm going out of town a couple of weeks.\nSpeaker C: And time is marching, sort of...\nSpeaker C: Given all the many wonderful things we could be working on, what will we actually focus on?\nSpeaker C: And what do we freeze?\nSpeaker C: And, you know, what do we...\nSpeaker C: So, I mean, this software that these guys created was certainly a key part.\nSpeaker C: So then there's something central.\nSpeaker C: And there are at least a bunch of different versions going off in ways that differ trivially.\nSpeaker C: And...\nSpeaker C: That's nice.\nSpeaker C: And then within that, I guess the idea was to freeze a certain set of options for now.\nSpeaker C: To run it a particular way and decide on what things are going to be experimented with, as opposed to just experimenting with everything.\nSpeaker C: So, a certain set of things constant.\nSpeaker C: So, maybe describe roughly what we are keeping constant for now.\nSpeaker D: Yeah, well, so we've been working like six weeks in the nice compensation.\nSpeaker D: We endeavor to...\nSpeaker F: So, you're going to use which of the two techniques?\nSpeaker D: So, finally, it's a winner-fragile on FSD-based, and it uses two steps, smoothing the transfer function.\nSpeaker D: And the first step, that's on the kind which is the direction.\nSpeaker D: And after these steps, there is a further smoothing and a frequency which use, and it's lighting and there are different techniques.\nSpeaker F: So, this is on the...\nSpeaker F: Before any mel scaling.\nSpeaker F: Yeah, that's it.\nSpeaker C: So, this smoothing is done on the estimate of what you're going to subtract, or on the thing that has already had something subtracted.\nSpeaker D: It's on the transfer function.\nSpeaker C: Oh, it's on the transfer function for the winner filter.\nNone: Yeah, okay.\nSpeaker D: Yeah, so basically we tried different configuration within this idea.\nSpeaker D: We tried to add down these some mel bands, having spectroscopy, instead of phenophiltering.\nSpeaker D: But finally, we ended up with this configuration that works quite well.\nSpeaker D: So, we're going to fix this for the monitor work of the other aspects.\nSpeaker D: The world system, so...\nSpeaker C: Actually, let me...\nSpeaker C: Dave isn't here to talk about it, but let me just interject.\nSpeaker C: This module, in principle, I mean you would know whether it's true, in fact, is somewhat independent from the rest of it, because you recent the size speech, right?\nSpeaker C: So, well, you know, I guess you don't recent the size speech, but you could.\nSpeaker D: Well, we do, but we don't recent the size.\nSpeaker D: In the program, we don't recent the size, and then realize once again, we just used to clean the evidence.\nSpeaker C: But you have a recent the size thing that you, that's an option.\nSpeaker C: Yeah, I guess my point is that in some of the work he's doing in reverberation, one of the things that we're finding is that for an artificial situation, we can just deal with reverberation, and his techniques work really well.\nSpeaker C: But for the real situation, problem is that you don't just have reverberation, you have reverberation in noise, and if you don't include that in the model, it doesn't work very well.\nSpeaker C: So, in fact, it might be a very nice thing to do, to just take the noise, remove a part of it, and put that in front of what he's looking at, and generate new files or whatever, and then do the reverberation part.\nSpeaker C: So, it's...\nSpeaker C: Anyway.\nSpeaker F: The Dave hasn't tried that yet.\nSpeaker C: No, no, he's...\nSpeaker F: I guess he's busy with...\nSpeaker F: Prelims, right?\nSpeaker C: Yeah, so...\nSpeaker C: But, you know, that'll...\nSpeaker C: It's clear that we are not...\nSpeaker C: With the real case that we're looking at, we can't just look at reverberation and isolation because the interaction between that noise is considerable.\nSpeaker C: And that's...\nSpeaker C: I mean, in the past, we've looked at...\nSpeaker C: And this is hard enough.\nSpeaker C: The interaction between channel effects and...\nSpeaker C: And the additive noise is so convolutional effects and additive effects.\nSpeaker C: And that's hard enough.\nSpeaker C: I mean, I don't think we really...\nSpeaker C: I mean, we're trying to deal with that.\nSpeaker C: In a sense, that's what we're trying to deal with in this Aurora task.\nSpeaker C: And we have the...\nSpeaker C: LDA stuff that, in principle, is doing something about convolutional effects.\nSpeaker C: And we have the noise suppression that's doing something about noise.\nSpeaker C: Even that's hard enough in the online normalization as well.\nSpeaker C: There's all these interactions between these two, and they work so hard and...\nSpeaker C: And struggling everything around.\nSpeaker C: But now, when you throw in the reverberation, it's even worse, because not only do you have these effects, but you also have some long-time effects.\nSpeaker C: And so Dave has something which...\nSpeaker C: Is doing some nice things under some conditions with long-time effects, but when there's noise there too, it's pretty hard.\nSpeaker C: So we have to start.\nSpeaker C: Since any...\nSpeaker C: Almost any real situation is going to have...\nSpeaker C: Where you have the microphone distant, it's going to have both things.\nSpeaker C: We actually have to think about both at the same time.\nSpeaker C: So, there's this noise suppression thing, which has sort of worked out, and then maybe just continued telling what else is in the...\nSpeaker C: The former half.\nSpeaker D: What are parts of the system?\nSpeaker D: The blocks that were present before...\nSpeaker D: Would you have put it back in?\nSpeaker C: So that's again, that's the Wiener filtering followed by...\nSpeaker C: That's done at the FFT level.\nSpeaker D: Then...\nSpeaker D: Yeah, then the mail, sit or bank...\nSpeaker D: And the lug of the ratio.\nSpeaker C: Well, the filtering is done...\nSpeaker C: In frequency domain?\nSpeaker C: Yeah.\nSpeaker C: Then the mail, then the lug, and then the LDA filter.\nSpeaker C: And then...\nSpeaker C: Downsample...\nSpeaker C: DCT...\nSpeaker C: Online...\nSpeaker D: Online...\nSpeaker D: The same thing.\nSpeaker D: And then compute that.\nSpeaker C: And replace...\nSpeaker C: Then you run natural.\nSpeaker C: And then in parallel with...\nSpeaker C: In neural net and then following neural net,\nSpeaker D: and finally friend dropping, which... Would be a neural network, so you first need to sign up for these, and the input of this neural network would be somewhere between lug and mail band.\nSpeaker E: So...\nSpeaker C: So that's sort of...\nSpeaker C: Most of the stuff is...\nSpeaker C: Yeah, it's operating parallel with this other stuff.\nSpeaker C: Yeah.\nSpeaker C: So the things that we...\nSpeaker C: I guess we sort of...\nSpeaker C: There's some neat ideas for VADs.\nSpeaker C: So, I mean, I think there's sort of like...\nSpeaker C: There's a bunch of tuning things to improve stuff.\nSpeaker C: There's questions about raceplaces where there's an exponent.\nSpeaker C: If it's the right exponent, or ways that we're estimating noise, that we can prove estimating noise, and there's going to be a host of those.\nSpeaker C: And, certainly, it seemed like the things...\nSpeaker C: The main things that we were brought up that are going to need to get worked on seriously are...\nSpeaker C: A significantly better VAD...\nSpeaker C: Putting the neural net on...\nSpeaker C: Which, you know, we haven't been doing anything with the neural net at the end there.\nSpeaker C: And the...\nSpeaker C: Opening up the second front.\nSpeaker F: The other happens to sound.\nSpeaker C: Yeah, yeah, I mean, because we have...\nSpeaker C: We have half the data rate that they allow.\nSpeaker C: And so the initial thing, which came from the meeting that we had down south, was that we'll initially just put in a mouse spectrum.\nSpeaker C: That's the second one.\nSpeaker C: It's, you know, TEPZ.\nSpeaker C: There's a question about exactly how we do it.\nSpeaker C: We probably will go to something better later.\nSpeaker C: But, initial thing is that...\nSpeaker C: Capster and Spectre behave differently, so.\nSpeaker C: I think Tony Robinson used to do...\nSpeaker C: I was saying this before, I think he used to do...\nSpeaker C: Mel, Spectre and Mel Capster used them as alternate features.\nSpeaker C: Put them together.\nSpeaker F: So, if you took the system, the way it is now, the way it's... you're going to freeze it.\nSpeaker F: And ran it on the last evaluation.\nSpeaker F: Where would it be?\nSpeaker F: In terms of ranking.\nSpeaker F: So, again...\nSpeaker C: Well, you know, you haven't tested it actually on the German and Danish have you.\nSpeaker C: No, you didn't.\nSpeaker F: Yeah.\nSpeaker F: So, on the ones that you did test it on, it would have been second.\nSpeaker C: Yeah.\nSpeaker C: What did...\nSpeaker C: I mean, but when you're saying second, you're comparing to the numbers that the...\nSpeaker C: Yeah, the best system before got on...\nSpeaker C: We also without German and Danish.\nSpeaker B: And the ranking actually didn't change after the German and Danish.\nSpeaker B: So, yeah.\nSpeaker C: Well, ranking didn't before, but I'm just asking where this is to...\nSpeaker C: Yeah.\nSpeaker C: Where there was without the German and Danish, right?\nSpeaker F: Where were we actually on the last test?\nSpeaker C: Oh, we were also essentially second, although there were...\nSpeaker C: I mean, we had a couple systems and they had a couple systems.\nSpeaker C: And so, I guess by that, we were third, but I mean, there were two systems that were pretty close.\nSpeaker C: They came from the same place.\nSpeaker C: So, institutionally, we were second.\nSpeaker C: So, the second that you're saying now is...\nSpeaker C: Same-wide.\nSpeaker C: Oh, no, I think it's also institutionally.\nSpeaker C: Still, it's next-wide.\nSpeaker C: I mean, I think both of their systems probably...\nSpeaker C: We're between their two systems.\nSpeaker C: Oh, are we?\nSpeaker B: Yeah.\nSpeaker B: So, their first system is 54 points, something, and we are 53 points, something.\nSpeaker B: And their second system is all the 53 points, something.\nSpeaker B: Yeah, 1%.\nSpeaker C: Yeah, so basically, they're all pretty close.\nSpeaker C: And in some sense, we're all doing fairly similar things.\nSpeaker C: I mean, good argue about the LDA and so forth, but I think, yeah, a lot of ways are doing very similar things.\nSpeaker F: So, how do they fill up all these bits?\nSpeaker F: I mean, for...\nSpeaker C: Why are we using half?\nSpeaker C: Yeah.\nSpeaker C: Well, so you could...\nSpeaker F: How are they using more than half, I guess, maybe?\nSpeaker C: Yeah, so I think, guys, closer to it, me, so correctly, from wrong.\nSpeaker C: But I think that what's going on is that in both cases, some kind of normalization is done to deal with convolutional effects.\nSpeaker C: They have some capital modification, right?\nSpeaker C: In our case, we have a couple things.\nSpeaker C: We have the online normalization, and we have the LDA roster.\nSpeaker C: And they seem to complement each other enough and be different enough that they both seem to help us.\nSpeaker C: But then, anyway, they're both doing the same sort of thing, but there's one difference.\nSpeaker C: The LDA roster throws away high modulation frequencies, and they're not doing that.\nSpeaker F: So...\nSpeaker C: So that if you throw away high modulation frequencies, then you can downsample.\nSpeaker F: I see.\nSpeaker F: I see.\nSpeaker F: So...\nSpeaker F: So what if you didn't...\nSpeaker F: So do you explicitly downsample them?\nSpeaker F: Do we explicitly downsample them?\nSpeaker F: Yeah.\nSpeaker F: And what if we didn't do that?\nSpeaker C: Would we get worse performance?\nSpeaker F: I think it doesn't affect it, doesn't it?\nSpeaker C: Yeah.\nSpeaker C: I see.\nSpeaker C: Yeah.\nSpeaker C: So I think the thing is, since we're not evidently throwing away useful information, let's try to put it into useful information.\nSpeaker C: Yeah.\nSpeaker C: And so, you know, we've found in a lot of ways for quite a while that having a second stream helps a lot.\nSpeaker C: So that's put in.\nSpeaker C: And you know, I may even end up with Mel Spectrum, even though I'm saying I think we can do much better just because it's simple.\nSpeaker C: And, you know, in the long run having something everybody will look at and say, oh yeah, I understand, is very helpful.\nSpeaker F: So you're thinking to put the Mel Spectrum in before any of the noise removal stuff?\nSpeaker F: Well, that's a question.\nSpeaker C: I mean, we were talking about that.\nSpeaker C: It looks like it'd be straightforward to remove the noise.\nSpeaker F: And...\nSpeaker F: Because that happens before the Mel conversion, right?\nSpeaker C: Yeah.\nSpeaker C: So, I mean, to do it after the Mel conversion, after the noise removal, after the Mel conversion.\nSpeaker C: And it's even a question in my mind, anyhow, whether you should take the log or not.\nSpeaker C: I sort of think you should, but...\nSpeaker E: I don't know.\nSpeaker D: What about normal ideas?\nSpeaker C: Right.\nSpeaker C: Well, but normalizing Spectrum instead of Capster?\nSpeaker C: Yeah.\nSpeaker C: Yeah, probably.\nSpeaker C: Some kind would be good, you know, I would think.\nSpeaker B: Well, it actually makes it dependent on the overall energy of the frame.\nSpeaker B: If you do or don't, don't analyze.\nSpeaker B: If you don't, don't analyze.\nSpeaker B: And if you don't, don't analyze.\nSpeaker C: Right.\nSpeaker C: Yes, I mean, one would think that you'd want to normalize.\nSpeaker C: But my thought is, particularly if take the log, try it.\nSpeaker C: And then if normalization helps, then you have something to compare against and say, okay, this much effect.\nSpeaker C: I mean, you don't want to change six things and then see what happens.\nSpeaker C: You want to change them one at a time.\nSpeaker C: So, adding this other stream in, that's simple in some way.\nSpeaker C: And then, saying, particularly because we found the past, there's all these different results you get with slight modifications of how you do normalization.\nSpeaker C: Normalization is a really tricky sensitive thing and you learn a lot.\nSpeaker C: So, I would think you would want to have some baselines that says, okay, we don't normalize this is what we get.\nSpeaker C: And we do this normalization and we do that normalization.\nSpeaker C: But the other question is, so I think ultimately one depends on normalization, I agree.\nSpeaker F: So, this second stream, will it add latency to the system?\nSpeaker C: No, it's in parallel.\nSpeaker C: We're not talking about computation time here.\nSpeaker C: We're actually pretty far out.\nSpeaker C: So, it's just in terms of what data it's depending on.\nSpeaker C: It's depending on the same data as the other.\nSpeaker C: Same.\nSpeaker A: So, with this new stream, would you train up the VAD on both features somehow?\nSpeaker B: No, I guess the VAD has its own set of features.\nSpeaker B: Okay.\nSpeaker B: Which could be one of these streams or it can be something derived from these streams.\nSpeaker D: Okay.\nSpeaker D: And there is also the idea of using Crap's maybe.\nSpeaker B: Yeah, it's also.\nSpeaker D: Well, right, you bought for an empty show.\nNone: It's a good idea.\nSpeaker A: With that fit on the handset or...\nSpeaker A: Oh.\nSpeaker A: Okay.\nSpeaker B: Let me test.\nSpeaker B: Yeah.\nSpeaker B: Let me test to fit the delays and all this stuff.\nSpeaker C: Well, this is delays in the storage, yeah.\nSpeaker C: But I don't think the storage is so big for that.\nSpeaker C: I think the biggest thing we've run into for storage is neural net.\nSpeaker C: You, right.\nSpeaker C: And so, I guess the issue there is, are we using neural net-based traps?\nSpeaker C: And how big are they?\nSpeaker C: That'll be an issue.\nSpeaker A: Maybe they could be little ones.\nSpeaker A: Right.\nSpeaker A: Because she also does the correlation-based traps without the neural net just looking at the coordinates.\nSpeaker C: Maybe for VAD, they would be okay.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It's true.\nSpeaker C: Or simple neural net, right?\nSpeaker C: I mean, the thing is, if you're doing correlation, you're just doing a simple dot product with some weights, which you happen to learn from the data.\nSpeaker C: And so, putting it on linearity on it is not that big a deal.\nSpeaker C: It certainly doesn't take much space.\nSpeaker C: Right.\nSpeaker C: So, the question is, how complex a function you need to have an added layer or something, in which case, potentially, you know, it could be big.\nSpeaker C: So, what's next?\nNone: Maybe.\nSpeaker C: Not like this.\nSpeaker F: So, the meeting with Henik that you guys just had was to decide exactly what you were going to freeze in this system.\nSpeaker F: Is that, or was there, are you talking about what the new stuff for?\nSpeaker C: What the freeze and then what to do after we froze?\nNone: Yeah.\nSpeaker C: And, like I said, I think that the basic directions are, I mean, there's lots of little things, such as improvement noise estimator, but the bigger things are adding on the neural net and the second stream and then improving the VAD.\nSpeaker B: So, I have actually, after the meeting, added the second stream to the data and maybe I'll start with the feature net in that case.\nSpeaker B: It's like you're looking at the VAD, right?\nSpeaker D: Yeah.\nSpeaker B: For the VAD?\nSpeaker B: No.\nNone: But the network on the VAD and one.\nSpeaker B: So, you already have it?\nSpeaker B: Okay.\nSpeaker B: So, just to take the features from the final.\nSpeaker B: Okay.\nSpeaker D: But, yeah, I think that's your issue.\nSpeaker F: What about the new part of the evaluation, the Wall Street Journal part?\nSpeaker C: Right.\nSpeaker C: Have you ever, very good question.\nSpeaker C: Have you ever worked with the Mississippi State Software?\nSpeaker E: No.\nSpeaker C: Oh.\nSpeaker C: Well, you may be called up a hand to help.\nSpeaker C: I think that kind of all the work in this stuff here has been in the small vocabulary.\nSpeaker F: So, how was the interaction supposed to happen?\nSpeaker F: I remember last time we talked about this, it was sort of up in the air, whether they were going to be taking people's features and then running them or they were going to give the system out.\nSpeaker F: So, they're going to just deliver a system base.\nSpeaker F: Do we already have it?\nSpeaker B: Yeah, I guess it's almost ready.\nSpeaker B: So, that's what, so they have released a document describing the system.\nSpeaker C: Maybe you could point it at Chuck, because we'll have to grab this over CVS or something.\nSpeaker B: No, it's just downloadable from their website.\nSpeaker C: Because one of the things that might be helpful if you've got time and all of this is, is if these guys are really focusing on improving all the digit stuff, maybe you got the front end from them, maybe you could do the runs for the, and you know, iron out hassles that you have to tweak Joe about, or whatever, because you're more experienced who's going to urge vocabulary stuff.\nSpeaker B: So, I'll point you to the website and the mail corresponding.\nSpeaker F: And it's not ready yet, this is the thing.\nSpeaker B: I think they are still tuning something on that.\nSpeaker B: So, they're like varying different parameters like the insertion penalty and all those stuff and then seeing what's the performance.\nSpeaker F: So, what is going to be parameters that are frozen, nobody can change?\nSpeaker B: I guess there is time during which people can make suggestions.\nSpeaker F: Oh, but everybody's going to have to use the same value.\nSpeaker F: After that.\nSpeaker B: So, this suggestion, this, this, this, Peter, during which people can make suggestions is to know whether it is actually biased towards any set of features or.\nSpeaker C: Yeah, so, certainly the thing that I would want to know about is whether we get really hurt on insertion penalty, language model scaling, sorts of things.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: In which case, Hary or Henik will need to, you know, push the case or about this.\nSpeaker F: And we may be able to revisit this idea about, you know, somehow modifying our features to.\nSpeaker C: Yes, in this case, that's right.\nSpeaker C: Yeah.\nSpeaker C: That's right.\nSpeaker C: Some of that may be a last minute rush thing because if our features are changing.\nSpeaker C: Yeah.\nSpeaker C: But, yeah, the other thing is that even though it's a month away, it's starting to seem to me now like November 15th is right around the corner.\nSpeaker C: And if they haven't decided things like this, what the parameters are going to be for this.\nSpeaker C: When deciding is not just somebody deciding, I mean, in fact, there should be some understanding behind the deciding, which means some experiments and so forth.\nSpeaker C: It seems pretty tight to me.\nSpeaker F: So, what's the significance of November 15th?\nSpeaker C: That's when the evaluation is.\nSpeaker C: Yeah.\nSpeaker C: So, yeah.\nSpeaker C: So, yeah.\nSpeaker C: So I have to, but, you know, they may even decide in the end to push it off.\nSpeaker C: It wouldn't, you know, entirely surprise me.\nSpeaker C: But due to other reasons, like some people are going away, I'm hoping it's not pushed off for a long while.\nSpeaker C: That would be, but it's an awkward position.\nSpeaker C: But anyway.\nSpeaker C: Okay.\nSpeaker C: Great.\nSpeaker C: Yeah, I think that'll be helpful.\nSpeaker C: There's not anybody, OGI currently, who's working with this.\nSpeaker F: Is this part of the evaluation just a small part of how important is this to the overall?\nSpeaker C: I think it depends how badly you do.\nSpeaker C: I mean, I think that it is...\nSpeaker C: This is one of those things that will be debated afterwards.\nSpeaker C: Yeah.\nSpeaker C: Well, I mean, it's conceptually, my impression, again, you guys correct me if I'm wrong, my impression is that they want it as a double check that you haven't come across, you haven't invented features, which are actually going to do badly for a significantly different task, particularly one with larger vocabulary.\nSpeaker C: But it's not the main emphasis.\nSpeaker C: I mean, the truth is most of the applications they're looking at are pretty small vocabulary.\nSpeaker C: So it's a double check.\nSpeaker C: So they'll probably assign it some sort of low weight.\nSpeaker F: It seems to me that if it's a double check, they should give you a 1 or a 0.\nSpeaker F: You passed the threshold, you didn't pass the threshold, and they shouldn't even...\nSpeaker F: Yeah, you're not going to score it.\nSpeaker C: But I mean, we'll see what they come up with.\nSpeaker C: But in the current thing, for instance, where you have this well-matched, moderately matched, and this highly mismatched, the emphasis is somewhat on the well-matched, but it's only a marginal, right?\nSpeaker C: It's a 40, 35, 25 or something like that.\nSpeaker C: So you still, if you were way, way off on the highly mismatched, it would have a big effect.\nSpeaker C: And it wouldn't surprise me if they did something like that with this.\nSpeaker C: So again, if you get...\nSpeaker C: If it doesn't help you much for noisy versions of this large vocabulary data, then it may not hurt you that much.\nSpeaker C: But if you don't... if it doesn't help you much at all, or put another way, if it helps some people a lot more than it helps other people, there are strategies too.\nSpeaker F: So is this...\nSpeaker F: Goodter was putting a bunch of Wall Street Journal data on our disks.\nSpeaker F: So that's the data that we'll be running on.\nSpeaker C: Yeah.\nSpeaker C: I see.\nSpeaker C: So if the data just not the recognized...\nSpeaker C: Okay.\nSpeaker F: So this test may take quite a while to run, then.\nSpeaker F: A judging by the amount of data that he was putting.\nSpeaker C: Well, it's training and test, right?\nSpeaker F: I guess I'm not sure.\nSpeaker C: No, I mean, if it's like the other things, there's data for training the HMMs and data for testing it.\nSpeaker C: So I wouldn't... so it's...\nSpeaker F: Okay.\nSpeaker C: So training, the recognizer.\nSpeaker C: But I think it's trained on clean.\nSpeaker C: Is it trained on clean?\nSpeaker C: Wall Street?\nSpeaker D: Yeah.\nSpeaker D: Probably no.\nSpeaker D: It's training range between 10 and 20 dB.\nSpeaker D: I think it's testing between 5 and 15 dB.\nSpeaker D: Yeah.\nSpeaker C: Okay.\nSpeaker B: So it's like a medium mismatch condition, sort of.\nSpeaker E: I see.\nSpeaker D: So the noise is...\nSpeaker D: There is a range of different noises.\nSpeaker D: So...\nSpeaker D: The jar select can run on the media.\nSpeaker D: Okay.\nSpeaker D: And to the files.\nSpeaker D: There are noises that are different.\nSpeaker D: And the noise is used on the 8 digits.\nSpeaker E: Yeah.\nNone: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I mean, I wouldn't imagine that the amount of testing data was that huge.\nSpeaker C: I probably put training...\nSpeaker C: Almost certainly, put training data there, too.\nSpeaker C: Maybe not.\nSpeaker C: So, that's that.\nSpeaker F: Okay.\nSpeaker F: One last question on that.\nSpeaker F: Did they estimate that they would have that system available for download?\nSpeaker B: I guess one...\nSpeaker B: Some preliminary version is already there.\nSpeaker F: Oh, so there's something you can download to just learn.\nSpeaker B: Yeah.\nSpeaker B: But they're actually paralleling doing some modifications also, I think.\nSpeaker B: Okay.\nSpeaker B: So I guess the final system will be frozen by middle of like one more week, maybe.\nSpeaker C: Oh, that's pretty simple.\nSpeaker B: Yeah, that's just one more.\nSpeaker A: Is this their SVM recognizer?\nSpeaker B: No, it's just a straight-forward achievement.\nSpeaker C: You know, they have a lot of options.\nSpeaker C: And they recognize your SVM is one of the things they've done with it.\nSpeaker C: But it's not their more standard thing.\nSpeaker C: For most part, it's Gaussian mixed using.\nSpeaker A: Okay.\nSpeaker B: Just the HTML.\nSpeaker B: Gaussian mixed with momentum.\nSpeaker A: Gaussian mixed with momentum.\nSpeaker C: Yeah, the SVM thing was an HTML also.\nSpeaker C: Yeah, this is a hybrid.\nSpeaker F: Yeah.\nSpeaker F: So just so that I understand they're providing scripts and everything so that basically you push a button and it does training and then it does test and everything.\nSpeaker F: Is that the idea?\nSpeaker B: I think, yeah, I guess something like...\nSpeaker B: It's like...\nSpeaker B: As painless as possible is work.\nSpeaker B: Do they provide all the scripts and everything and then...\nSpeaker B: Yeah, there's books to put your features out.\nSpeaker C: Yeah, in fact, I mean, if you look into it a little bit, it might be reasonable, you know, Joe, right?\nSpeaker C: Yeah, just to sort of ask him about the issue of different features having different kinds of scaling, characteristics and so on.\nSpeaker C: So that, you know, possibly having entirely different optimal values for the usual twittles, factors and what's the plan about that?\nSpeaker B: Should we like add check all through to the mailing list?\nSpeaker B: Maybe better, I mean, in that case, if he's going to...\nSpeaker B: Yeah.\nSpeaker F: So there's a mailing list for this?\nSpeaker F: Yeah, that'd be great.\nSpeaker B: Yeah, I guess maybe Harry or Heneke, one of them has to send a mail to Joe.\nSpeaker B: Or maybe if you...\nSpeaker B: I could send it.\nSpeaker B: Yeah, I know, maybe.\nSpeaker B: I know, really.\nSpeaker B: Yeah, so that's just fine.\nSpeaker B: So...\nSpeaker C: Yeah, and just maybe...\nSpeaker C: You have Harry's...\nSpeaker C: Have Harry's?\nSpeaker C: Yeah, so maybe just CC Harry and say that you've just been asked to handle the larger vocabulary part here.\nSpeaker C: And, you know...\nSpeaker F: Would it be better if I asked Harry to ask Joe?\nSpeaker C: Why don't you just ask Joe but CC Harry and then in the notes say Harry, hopefully this is okay with you.\nSpeaker C: And then if Joe feels like he needs a confirmation, Harry can answer it.\nSpeaker C: That way you can get started asking Joe quickly while he's maybe still putting in nails and screws and do it.\nSpeaker B: And there is an archive of all the mails that has been...\nSpeaker B: That has gone between these people, among these people.\nSpeaker B: So, just you can see all those mails in the ISIP website.\nSpeaker B: Okay.\nSpeaker B: This is the website.\nSpeaker B: Is that a password?\nSpeaker B: Yeah, it's password-productive.\nSpeaker B: So, like...\nSpeaker B: It's...\nSpeaker B: Like...\nSpeaker C: What do you think about how long would be most useful for you to go up to OJ?\nSpeaker D: I don't know.\nSpeaker D: We can.\nSpeaker D: For September we can sit up.\nSpeaker D: Work schedule.\nSpeaker D: At some point, maybe we better...\nSpeaker C: Oh, so you're imagining more that you would come back here first for a while and then go up there.\nSpeaker C: I mean, it's up to you guys, are you?\nSpeaker C: Well, anyway, you don't have to decide the second.\nSpeaker C: But think about it, about what you think would be the best way to work it all supported either way.\nSpeaker C: Okay.\nSpeaker C: Anything to tell us?\nSpeaker A: Well, I've been reading some literature about clustering of data.\nSpeaker A: Just...\nSpeaker A: I guess, let me put it in context.\nSpeaker A: Okay.\nSpeaker A: So, we're talking about discovering intermediate categories to classify.\nSpeaker A: And I was looking at some of the work that San Gita was doing on these traps things.\nSpeaker A: So, she has... she has temporal patterns for a certain set of phonemes from Timmit, right?\nSpeaker A: The most common phonemes.\nSpeaker A: Each one of them has a nice pattern over time, one second window, and it has these patterns.\nSpeaker A: So, she has a trap for each one of the phonemes times 15 for each of the 15 critical bands.\nSpeaker A: And she does this agglomerative hierarchical clustering, which basically is a clustering algorithm that starts with many, many, many different points, many different clusters corresponding to the number of data patterns that you have in the data.\nSpeaker A: And then you have this distance metric, which measures how closely related they are.\nSpeaker A: And you start by merging the patterns that are most closely related.\nSpeaker A: You created three.\nSpeaker F: Yeah, yeah, yeah, a dendrogram tree.\nSpeaker F: You can think of values anywhere along that tree to fix your set of clusters.\nSpeaker A: Right.\nSpeaker A: Usually, it's when the similarity measures don't go down as much.\nSpeaker A: And so, so you stop at that point.\nSpeaker A: And what she found was that there were five broad categories corresponding to things like fricatives and vocalic and stops and one for silence and another one for schwa sounds.\nSpeaker A: And I was thinking about ways to generalize this because it's not a completely automatic way of clustering because beforehand you have these traps, you're saying that these frames correspond to this particular phonemes.\nSpeaker A: And that's constraining your clustering to the set of phonemes that you already have.\nSpeaker A: Whereas, maybe we want to just take a look at arbitrary windows in time, a varying length, and cluster those.\nSpeaker A: And I'm thinking if we do that, then we would probably, at some point in the clustering algorithm, find that we've clustered things like, okay, this is a transition, this is a relatively stable point.\nSpeaker A: Now, hoping to find other things of similarity and maybe use these things as the intermediate categories that later classify.\nSpeaker C: Are you looking at this in their own bands?\nSpeaker A: Right.\nSpeaker A: I guess what you're going to be using. Yeah, I've been exactly figured out the exact details for that.\nSpeaker A: But the representation of the data that I was thinking of was using critical band energies over different lengths of time.\nSpeaker C: Yeah, I mean, it seems somehow that needs to, there's a couple things that I wonder about with this. So one is, again, looking at the same representation.\nSpeaker C: I mean, if you're going for this sort of thing where you have little detectors that are looking at narrow bands, then what you're going to be looking for should be some category that you can find with the narrow bands.\nSpeaker C: That seems to be kind of fundamental to it.\nSpeaker C: And then the other thing is that I wonder about with it and don't take this in the wrong way like I know what I'm doing or anything.\nSpeaker C: But I mean, just wondering, really, the sort of standard answer about this sort of thing is that if you're trying to find the right system in some sense, whether you're trying to categorize or parameters.\nSpeaker C: And your goal is discrimination. Then having choices based on discrimination as opposed to unsupervised nearness of things is actually better.\nSpeaker C: And I don't know if that means since you're dealing with issues for busness, you know, maybe this isn't right. But it'd be something I'd be concerned about because, for instance, you can imagine.\nSpeaker C: If you remember from your quarrels, John O'Halla saying that Bon Pa differed, not really because of voicing, but because of aspiration.\nSpeaker C: I mean, as far as what's really there in the acoustics.\nSpeaker C: So if you looked, if you're doing some coarse clustering, you probably would put those two sounds together. And yet I would guess that many of your recognition errors were coming from screwing up on this distinction.\nSpeaker C: So in fact, it's a little hard because recognizers to first order sort of work. And the reason we're doing the things we're doing is because they don't work as well as we'd like.\nSpeaker C: And since they sort of work, it means that they are already doing, if you go and take any recognizer that's already out there, and you say, how well is it distinguishing between schwa's and stops?\nSpeaker C: Boy, I bet they're all doing nearly perfectly on this. So these big categories that differ in huge, obvious ways, we already know how to do.\nSpeaker C: So what are we bringing to the party? I mean, in fact, what we want to do is have something that, particularly in the presence of noise, is better at distinguishing between categories that are actually close to one another. And hence would probably be cluster together.\nSpeaker C: So that's the hard thing. I mean, I understand that it's this other constraint that you're considering is that you want to have categories that would be straightforward for, say, a human being to mark if you had manual annotation.\nSpeaker C: And something that you really think you can pick up. But I think it's also essential that you want to look at what are the confusions that you are making. And how can you come up with categories that can clarify these confusions?\nSpeaker C: So I mean, the standard sort of way of doing that is take a look at the algorithms, you're looking at them, throw in some discriminative aspect to it. This is more like, how does LDA differ from PCA?\nSpeaker C: I mean, the same sort of thing, they're both authorizing, but, you know, and this is a little harder because you're not just trying to find parameters, you're actually trying to find the categories themselves.\nSpeaker C: A little more like brain surgery, I think, on yourself.\nSpeaker C: So, anyway, that's my thought.\nSpeaker C: Okay.\nSpeaker C: You've been thinking about this for a long time, actually, I mean, well, actually you stopped thinking about it for a long time, but you used to think about it a lot.\nSpeaker C: You were thinking about it more now. Yeah. Yeah. It's categories.\nSpeaker F: I guess, I don't, I don't, it's not clear to me how to reconcile, you know, what you're saying, which I think is right, with the way I've been looking at it, it's, it's all not very clear to me.\nSpeaker F: But it seems to me that the desire, the desirable feature to have is something that is bottom up.\nSpeaker F: Now, however we do that, and so, I guess what I don't understand is how to do that and still be discriminative because to be discriminative, you have to have categories, and the only categories that we know up to use are, sort of, these human, human-significant categories that are significant to humans, like phonemes, things like that.\nSpeaker C: But that's sort of what you want to avoid. Well, here's a, here's a, here's a generic and possibly useless thought, which is, what do you really, I mean, in a sense, the only systems that make sense are ones that have something from top down in them.\nSpeaker C: Right, because if even the smallest organism is trying to learn to do anything, if it doesn't have any kind of reward for, or penalty for doing anything, then it's just going to behave randomly.\nSpeaker C: So whether you're talking about something being learned through evolution or being learned through experience, it's got to have something come down to it that gives its reward, or at least some sort of reinforcement.\nSpeaker F: So the question is how far down, and stop at words, but we don't, right, we go all the way down to phonemes.\nSpeaker C: Right, but I think it maybe in some ways, part of the difficulty is trying to deal with these phonemes.\nSpeaker C: And it's almost like you want categories, if our metric of goodness, or correction, if our metric of badness is word error rate, then maybe we should be looking at words.\nSpeaker C: I mean, for very nice reasons, we've looked, well, it's syllables, and they have a lot of good properties.\nSpeaker C: But if you go all the way to words, I mean, that's really, I mean, in many applications, you want to go further, you want to go to concepts or something.\nSpeaker C: Yeah, I have concepts, actions, the sort of thing.\nSpeaker C: Words aren't bad, yeah.\nSpeaker F: So the common, right, the common wisdom is you can't do words because there's too many of them, right? So you have to have some smaller set that you can use.\nSpeaker F: And so everybody goes to phonemes. But the problem is that we build models of words in terms of phonemes, and these models are really cartoonish, right?\nSpeaker F: So when you look at conversational speech, for example, you don't see the phonemes that you have in your word models.\nSpeaker C: But we're not trying for models of words here. See, so here's maybe where, if the issue is that we're trying to come up with some sort of intermediate categories, which will then be useful for later stuff, then maybe it doesn't matter that we can't have enough.\nSpeaker C: I mean, what you want to do is build up these categories that are best for word recognition. And somehow if that's built into the loop of what the categories, I mean, we do this every day in this very gross way of running over a thousand experiments because we have asked computers picking the thing that has the best word error, right?\nSpeaker C: In some way, I mean, we derive that all the time. In some ways, that's not a bad thing to do because it tells you, in fact, how your adjustments at the very low level affect the final goal.\nSpeaker C: So maybe there's a way to even put that in a much more automatic way, where you take something about the error at the level of the word or some other, it could be so, but it's some large unit.\nSpeaker C: And, yeah, you may not have word models, you may have phone models, whatever, but you just sort of don't worry about that. And just somehow feed it back through.\nSpeaker F: So that's what I call a useless comments because I'm not really telling you how to do it. But I mean, it's, it's, it's, you know, I think the important part is there is that, you know, if you want to be discriminative, you have to have, you know, categories.\nSpeaker F: And I think this, the important categories are the words. Yeah.\nSpeaker F: Not the phones, maybe. And so, right, if you can put the words in to the loop, somehow for determining goodness of your sets of clusters.\nSpeaker C: Now, that being said, I think that, that if you have something that is, once you start dealing with spontaneous speech, all the things you're saying are really true.\nSpeaker C: If you have read speech that's been manually annotated like Tim at, yeah, then, you know, the phones are going to be right, actually, yeah, for the most part.\nSpeaker C: So, it doesn't really hurt them to, to do that, to put in discrimination at that level. If you go to spontaneous speech, then it's, it's trickier. And, and the phones are, you know, it's going to be based on bad pronunciation models that you have.\nSpeaker F: And it won't allow for the overlapping phenomena. So, it's almost like there's this mechanism that we have that, you know, when, when we're hearing read speech and all the phones are there, you know, we deal with that.\nSpeaker F: But, when we go to conversational and all of a sudden, not all the phone names are there, it doesn't really matter that much to us as humans, because we have some kind of mechanism that allows for these word models, whatever those models are to be.\nSpeaker F: And it doesn't really hurt them.\nSpeaker F: I'm not sure how, how to build that in.\nSpeaker C: Yeah, I mean, I guess the other thing is, is to think a little bit, when you, when you start looking at these kind of results, I think it usually is, is pretty intuitive. But, start looking at what are the kinds of confusions that you do make, you know, between words if you want or, or even between phones in, in, in red speech, say, when there is noise.\nSpeaker C: You know, so, is it more across place or more across manor or is it court, you know, is it, I mean, I know one thing that happens is that you, you, you lose low energy phones.\nSpeaker C: I mean, if it's added noise, then low energy phones sometimes don't get heard. And if that, if that is, if it's, if that turns it into another word or, or different, you know, another pair of words or something, then it's more likely to happen.\nSpeaker C: But, I don't know, I would, I would guess that you, I don't know, anyway, that's.\nSpeaker F: I think part of the difficulty is that a lot of the robustness that we have is probably coming from a much higher level, you know, we understand the context of the situational work, having conversation.\nSpeaker F: And so, there's noise in there, you know, our brain fills in.\nSpeaker A: Well, that's what, what should be the prediction?\nSpeaker C: Oh, sure, that's really big. But, I mean, even if you do diagnostic rhyme tests, kind of things, you know, where there really isn't any information like that, people are still better in noise than they, than they are in, then the machines are.\nSpeaker C: So, I mean, that's, right, we can't, we can't get it at all without any language models, language models are there and important. But, but, if we're not working on that, then we should work on something else and improve it, but, especially if it looks like the potential is there.\nSpeaker C: Should we just do this? Yeah. So, that's right here.\nSpeaker C: Okay, transcript, L-338.\nSpeaker C: 86556-481-5134-2727-1998-2524-8858-0302-7580-0105-992-945-8519-69249-692-497249-339-07342-89.\nSpeaker C: 734330 20417517\nSpeaker D: Let's click L-389 8892 640700 1759 8489 284 851 551 671 09 5065 288 895 282 793 078 707 079 36 2614 9813 8997 4467 0405\nSpeaker F: Transcript L-333 6370 690 60028327 777607174 7381141066 957431990 440680058 081543724 527481348\nSpeaker A: Transcript L-336 760149538 179649399 890002238 1195436059 0477569867 303895392 861464648 638406072019\nSpeaker B: Transcript L-334 1374091804 4794665647 1248047514 8221735705 9106363594 3960392838 0691690 238458830 Preserved\nSpeaker C: Exactly\nNone: 1000\n"}, {"meeting_id": "Bro027", "summary": "The team was coming close to finalizing the model for Aurora. They were still trying to make some improvements to improve their score. The team began with a discussion about how reverberation could be better accounted for. The professor thought that experimenting with different filters could help achieve a better signal-to-noise ratio. The new system for estimating silence probabilities that the team had added to the model was very effective, but it created a 220ms latency in the VAD. The team was not sure what kinds of constraints would be placed on latency eventually. They wanted to play it safe. The models, over all, were performing well, though the team intended to keep improving them. The team was also trying to figure out how to deal with different kinds of background noise. The meeting ended with a discussion of logistical issues.", "dialogue": "Speaker D: A, A.\nSpeaker D: It's a string.\nSpeaker D: Test.\nSpeaker D: Test.\nSpeaker E: Test.\nSpeaker E: Test.\nSpeaker E: Test.\nSpeaker E: So, let's see.\nSpeaker E: Yeah, Barry's not here and Dave's not here.\nSpeaker E: I can say about this quickly to get through it.\nSpeaker E: David, I submitted this.\nSpeaker E: This is for you.\nSpeaker G: Yes, I do.\nSpeaker E: Yeah, it's interesting.\nSpeaker E: I mean, basically, we're dealing with reverberation.\nSpeaker E: When we deal with pure reverberation technique, it's using works really, really well.\nSpeaker E: When they have reverberation in here, we'll measure the signalized ratio.\nSpeaker E: It's about 9 dB.\nSpeaker E: So, fair amount of...\nSpeaker F: I mean, for that actual recording.\nSpeaker E: Yeah.\nSpeaker E: And actually, it brought up a question, which may be relevant to the Aurora stuff, too.\nSpeaker E: I know that when you figured out the filters that we're using for the MellScale, there were some experimentation that went on at OGI.\nSpeaker E: But one of the differences that we found between the two systems that we were using, the Aurora HTK system, baseline system, and the system that we were...\nSpeaker E: The other system we were using, the SRI system, was that the SRI system had maybe 100 hertz high pass.\nSpeaker E: And the Aurora HTK was like 24.\nSpeaker D: 64.\nSpeaker E: Yeah. If you're using the baseline...\nSpeaker E: Is that the band center?\nSpeaker D: No, the edge.\nSpeaker E: The edge is really 64.\nSpeaker E: Some reason...\nSpeaker A: So, the center would be somewhere around like 100 and...\nSpeaker A: 100 and 100 and maybe it's like 100 hertz.\nSpeaker E: But do you know, for instance, how far down there would be at 20 hertz?\nSpeaker E: What, how much rejection would there be at 20 hertz?\nSpeaker E: At 20 hertz.\nSpeaker E: Yeah, any idea what the curve looks like?\nSpeaker A: 20 hertz frequency...\nSpeaker A: Oh, it's zero at 20 hertz, right?\nSpeaker A: The filter?\nSpeaker B: Actually, the left edge of the first filter is 64.\nSpeaker A: So anything less than 64 is zero.\nSpeaker E: It's actually set to zero.\nSpeaker E: Yeah. You can filter that.\nSpeaker E: Oh, from the...\nSpeaker B: This is the filter bank.\nSpeaker B: Oh, so you're going to be doing this?\nSpeaker A: Yeah, it's zero.\nSpeaker A: Yeah, so it's a weight on the power spectrum.\nSpeaker A: Triangular weighting.\nSpeaker E: Right.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: So that's still different than Dave thought, I think.\nSpeaker E: But still, it's possible that we're getting in some more noise.\nSpeaker E: So I wonder, was there...\nSpeaker E: There was experimentation with, say, throwing away that filter or something?\nSpeaker E: And...\nSpeaker A: Throwing away the first?\nSpeaker A: Yeah.\nSpeaker A: Okay, we've tried including the full band, right, from zero to four k.\nSpeaker A: And that's always worse than using 64 hertz.\nSpeaker E: Right.\nSpeaker E: But the question is whether 64 hertz is too low.\nSpeaker A: Yeah, I mean, make it 100 or something?\nSpeaker A: Yeah.\nSpeaker A: I think I've tried 100 and it was...\nSpeaker A: More or less the same as likely was.\nSpeaker A: On the same speech that car.\nNone: Aurora.\nSpeaker E: It was on the speech that car.\nSpeaker A: Yeah.\nSpeaker A: So I tried 100 to 4 k.\nSpeaker A: Yeah.\nSpeaker E: So it was...\nSpeaker E: And on the...\nSpeaker E: The address also.\nSpeaker D: No, no, no.\nSpeaker A: I think I just tried it on speech that car.\nSpeaker E: Maybe something to look at sometime because what he was looking at was performance in this room.\nSpeaker E: Would that be more like...\nSpeaker E: Well, you think that'd be more like speech that car, I guess, in terms of the noise.\nSpeaker E: The speech that car is more sort of roughly stationary, a lot of it.\nSpeaker E: Yeah.\nSpeaker E: And TI digits maybe is not so much.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Okay, well maybe it's not a big deal.\nSpeaker E: But anyway, that was just something we wondered about.\nSpeaker E: But certainly a lot of the noise is below 100 hertz.\nSpeaker E: The signalized ratio looks fair and mouth better if you have passed filter it from this room.\nSpeaker E: But it's still pretty noisy.\nSpeaker E: Even 100 hertz up, it's still fairly noisy.\nSpeaker E: The signalized ratio is actually still pretty bad.\nSpeaker E: So, I mean, the main...\nSpeaker E: So that's something.\nSpeaker F: The far field.\nSpeaker F: Yeah, that's the far field.\nSpeaker E: Yeah, the near field would be pretty good.\nSpeaker F: So what is...\nSpeaker F: What's causing that?\nSpeaker E: Well, we got a video projector in here.\nSpeaker E: And which we keep on during every session we record.\nSpeaker E: We were aware of, but we thought it wasn't a bad thing.\nSpeaker E: That's a nice noise source.\nSpeaker E: And there's also the air conditioning, which is pretty low frequency kind of thing.\nSpeaker E: So those are major components, I think, for the stationary kind of stuff.\nSpeaker E: But I guess I've maybe said this last week too, but it really became apparent to us that we need to take a count of noise.\nSpeaker E: So I think when he gets done with his pre-limps study, I think one of the next things we want to do is to take this noise processing stuff and synthesize some speech from it.\nSpeaker E: What are his pre-limps?\nSpeaker E: I think in about a little less than two weeks.\nSpeaker C: Oh, wow.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: So.\nSpeaker D: It might even be sooner.\nSpeaker D: I see this is 16th, 17th.\nSpeaker D: Yeah, I don't know if it's before.\nSpeaker F: It might even be a week.\nSpeaker F: So a week we can have.\nSpeaker E: I guess they were going to do it sometime during the semester.\nSpeaker E: They seem to be...\nSpeaker E: Well, the semester actually is starting out.\nSpeaker E: Is it already?\nSpeaker E: Yeah, the semester is late August.\nSpeaker E: They start here.\nSpeaker E: So they do it right at the beginning of the semester.\nSpeaker E: Yeah.\nSpeaker E: So, yeah, I mean, that was sort of one... I mean, the overall results seemed to be first place in the case of either artificial reverberation or a modest sized training set either way.\nSpeaker E: It helped a lot.\nSpeaker E: But if you had a really big training set, a recognizer system that was capable of taking advantage of a really large training set.\nSpeaker E: So one thing with the HTK is that it has...\nSpeaker E: As we're using... the configuration we're using is being gone by the terms of Aurora.\nSpeaker E: We have all those parameters just set as they are.\nSpeaker E: So even if we had 100 times as much data, we wouldn't go out to, you know, 10 or 100 times as many Gaussian or anything.\nSpeaker E: So it's kind of hard to take advantage of big chunks of data.\nSpeaker E: Whereas the other one does sort of expand as you have more training data.\nSpeaker E: So that's what's about the matric rate actually.\nSpeaker E: And so that one really benefited from the larger set.\nSpeaker E: And it was also a diverse set with different noises and so forth.\nSpeaker E: So that seemed to be...\nSpeaker E: So if you have that better recognizer, that can build up more parameters.\nSpeaker E: And if you have the natural room, which in this case has a pretty bad signal noise ratio, then in that case...\nSpeaker E: The right thing to do is just to use speaker adaptation.\nSpeaker E: And not bother with the psychostica processing.\nSpeaker E: But I think that that would not be true if we did some explicit noise processing as well as the kind of additional kind of things we were doing.\nSpeaker E: So that's sort of what we found.\nSpeaker F: I started working on the city state recognizer.\nSpeaker F: Oh, OK.\nSpeaker F: So I got a touch with Joe and from your email and things like that.\nSpeaker F: They added me to the list.\nSpeaker F: Oh, good question.\nSpeaker F: The mailing list.\nSpeaker F: He gave me all the pointers and everything that I needed.\nSpeaker F: So I downloaded the... There were two things that they had to download.\nSpeaker F: One was the software.\nSpeaker F: And another was a sample.\nSpeaker F: So I downloaded the software.\nSpeaker F: And piled all of that.\nSpeaker F: Oh, great.\nSpeaker F: And I grabbed the sample stuff that I haven't...\nSpeaker A: That sample was released only yesterday or the day before, right?\nSpeaker F: Well, I haven't grabbed that one yet.\nSpeaker F: So there was another short sample.\nSpeaker F: So I haven't grabbed the latest one that he just...\nSpeaker F: Oh, OK.\nSpeaker F: But the software is going to be fine and everything.\nSpeaker E: Is there any word yet about the issues about adjustments for different future sets?\nSpeaker F: No.\nSpeaker F: You asked me to write to him and I think I forgot to ask him about that.\nSpeaker F: I don't remember yet.\nSpeaker F: I'll check that.\nSpeaker F: Yeah.\nNone: Yeah.\nSpeaker E: So that turned out to be an important issue for us.\nSpeaker E: Yeah.\nSpeaker F: Because they have...\nSpeaker F: The old send is the best.\nSpeaker A: Because they have already frozen those insertion penalties and all those stuff is what I feel.\nSpeaker A: Because they have this document explaining the recognizer and they have this tables with various language model weights insertion penalties.\nSpeaker F: OK.\nSpeaker A: I haven't seen that one yet.\nSpeaker A: It's there on that.\nSpeaker A: And on that, I mean, they have run some experiments using various insertion penalties and all those things.\nSpeaker A: Yeah.\nSpeaker E: I think they picked the values for what test set?\nSpeaker A: So the one that they have reported is NIST evaluation Wall Street Journal.\nSpeaker E: But that has nothing to do with testing.\nSpeaker A: So they are actually trying to fix those values using the clean training part of the Wall Street Journal, which is...\nSpeaker A: I mean, the Aurora.\nSpeaker A: Aurora has a clean subset.\nSpeaker A: I mean, they're going to train it and then they're going to run some evaluations.\nSpeaker E: So they're setting it based on that.\nSpeaker E: OK. So now we may come back to the situation where we may be looking for a modification of the features to account for the fact that we can't modify these parameters.\nSpeaker E: But it's still worth, I think, just chatting with Joe about the issue.\nSpeaker F: Do you think that something I should just send to him or do you think I should send it to this?\nSpeaker F: I'm mailing this.\nSpeaker E: Well, it's not a secret. We're certainly willing to talk about it with everybody.\nSpeaker E: But I think that it's probably best to start talking with him just to...\nSpeaker E: Yeah, it's a dialogue between two of you about what, you know, what does he think about this and what could be done about it.\nSpeaker E: If you get 10 people involved in it, there'll be a lot of perspectives based on you.\nSpeaker E: But I think it all should come up eventually. But if there's any way to move in a way that would be more open to different kinds of features.\nSpeaker E: But if there isn't, it's just kind of shut down.\nSpeaker E: And that's also, it's probably not worthwhile bringing it into a larger form where political issues will come in.\nSpeaker E: OK.\nSpeaker A: So this is now... It's compiled on a Solaris.\nSpeaker A: Yeah. Because there was some mail saying that it's not stable for Linux and all those.\nSpeaker F: Yeah, that was a particular version.\nSpeaker F: Soos. Yeah. Soos. Yeah.\nSpeaker D: Yeah.\nSpeaker D: OK.\nSpeaker D: So it should be OK.\nSpeaker F: It could be compiled by actually no errors.\nSpeaker E: There's a slightly off topic, but I noticed just glancing at the Hopkins Workshop website that...\nSpeaker E: I don't know, we'll see how much they accomplished. One of the things that they were trying to do.\nSpeaker E: And the graphical models thing was to put together a toolkit for doing arbitrary graphical models for speech recognition.\nSpeaker E: So Jeff, two Jeffs.\nSpeaker F: Who's the second Jeff?\nSpeaker E: Oh, do you know Jeff Swag?\nNone: Yeah.\nSpeaker E: Oh, he was here for a couple of years and he got his PhD.\nSpeaker E: And he's been an IBM last couple of years.\nSpeaker E: OK.\nSpeaker E: So he did his PhD on dynamic basenets for speech recognition.\nSpeaker E: And he had some continuity built into the model, presumably to handle some inertia in the production system.\nSpeaker E: So...\nSpeaker B: I've been playing with first the VAD.\nSpeaker B: So it's exactly the same approach, but the features that the VAD neural network use are MFCC after noise compensation.\nSpeaker D: I think it results. What was it used before?\nSpeaker B: It was just PLP.\nSpeaker A: Yeah, it was actually...\nSpeaker A: No, it was just the noisy features, I guess.\nSpeaker B: Yeah, you see, not compensated features.\nSpeaker B: This is what we get after...\nSpeaker B: So actually, the features are noise compensated and there is also the LDA filter.\nSpeaker B: And then it's a pretty small neural network which use 9 frames of 6 features from C0 to C5 plus the first derivatives.\nSpeaker B: And it has 100 hidden units.\nSpeaker F: Is that 9 frames centered around the current frame?\nSpeaker E: So I'm sorry, there's how many inputs?\nSpeaker E: So it's 12 times 9 inputs and 100 hidden items.\nSpeaker B: Two outputs.\nSpeaker E: OK, next we have 11,000 parameters which actually shouldn't be a problem.\nSpeaker F: So what is different between this and...\nSpeaker B: So the previous system, it's based on the system that has a 53.66% improvement.\nSpeaker B: It's the same system, the only thing that changes the estimation of the silence probabilities.\nSpeaker B: Which now is based on cleaned features.\nSpeaker B: A lot better.\nSpeaker B: So it's not bad, but the problem is still that the latency is too large.\nSpeaker B: What's the latency?\nSpeaker B: The latency of the VAD is 220 milliseconds.\nSpeaker B: And the VAD is used for online normalization.\nSpeaker B: And it's used before the delta computation.\nSpeaker B: So if you add this components, it goes to 170.\nSpeaker E: I'm confused, you start off with 220 and you're under the output 170.\nSpeaker B: With 270.\nSpeaker B: If you add the delta computation, which is done afterwards.\nSpeaker E: So it's 220.\nSpeaker E: Is this at least 20 milliseconds frames?\nSpeaker E: Is that why?\nSpeaker E: Is it after-down?\nSpeaker B: The 220 is 100 milliseconds for the...\nSpeaker B: No, it's 40 milliseconds for the cleaning of the speech.\nSpeaker B: Then there is the neural network, which uses 9 frames.\nSpeaker B: So it adds 40 milliseconds.\nSpeaker B: After that, you have the filtering of the silence probabilities.\nSpeaker B: Which is a median filter.\nSpeaker B: It creates 100 milliseconds delay.\nSpeaker A: Plus there is a delta input.\nSpeaker B: And there is a delta input, which is...\nSpeaker E: 1.2 milliseconds for smoothing.\nSpeaker E: So it's...\nSpeaker D: 40 plus...\nSpeaker D: And then 40 is 40 plus...\nSpeaker D: 40 plus 100.\nSpeaker A: So it's 200 actually.\nSpeaker B: Yeah, there are 20 that comes from...\nSpeaker B: There is 10 that comes from the LDA filters also.\nSpeaker B: So it's 210, yeah.\nSpeaker A: If you're using...\nSpeaker A: But if you're using 3 frames...\nSpeaker A: If you're using 3 frames, it is 30 here for delta.\nSpeaker B: I think it's 5 frames.\nSpeaker A: So 5 frames are just 20.\nSpeaker A: Okay, so so 200 in time.\nSpeaker E: 40 for the cleaning of the speech.\nSpeaker E: 40 for the NN.\nSpeaker E: 100 for the smoothing.\nSpeaker B: 24 delta.\nSpeaker A: I think 24 delta.\nSpeaker A: I mean that's the input of the net.\nSpeaker D: And delta inputs in that.\nSpeaker A: Yeah, so it's like...\nSpeaker A: 5, 6, 7 plus delta.\nSpeaker A: 9 frames of.\nSpeaker E: And then 10 milliseconds for...\nSpeaker A: This is an LDA filter.\nSpeaker E: 10 milliseconds for LDA filter.\nSpeaker E: And another 10 milliseconds you said for the frame.\nSpeaker B: For the frame, I guess I computed 220 here.\nSpeaker B: Yes, it's for the...\nSpeaker E: Okay, and it's delta beside that.\nSpeaker B: So this is the features that are used by the network.\nSpeaker B: And then afterwards you have to compute the delta on the main feature stream, which is...\nSpeaker B: Delta and double delta, which is 15 milliseconds.\nSpeaker E: Yeah, now I mean after the noise part, the 40, the other 180...\nSpeaker E: Well, I mean...\nSpeaker E: Wait a minute, some of this is in parallel, isn't it?\nSpeaker E: I mean the LDA...\nSpeaker E: Well, the LDA is part of the VAD, right?\nSpeaker B: The VAD use LDA filter at features also.\nSpeaker B: Oh, it does.\nSpeaker E: Ah.\nSpeaker E: So in that case there isn't too much in parallel.\nSpeaker B: No, there is...\nSpeaker B: Just don't sampling, up sampling.\nSpeaker E: So the delta at the end is how much?\nSpeaker D: It's 50.\nSpeaker D: 50.\nSpeaker B: All right, so...\nSpeaker B: Well, we could probably put the delta before online normalization.\nSpeaker B: Should not make a big difference because...\nSpeaker F: We use this smaller window for the delta.\nSpeaker F: I guess there's a lot of things.\nSpeaker E: Yeah, if you put the delta before the...\nSpeaker E: Yeah, because then I click on parallel.\nSpeaker B: Yeah, because the time constant of the online normalization is pretty long compared to the delta window.\nSpeaker E: Okay, so...\nSpeaker E: And you ought to be able to pull off 20 milliseconds from somewhere else to get under 200, right?\nSpeaker E: I mean, it's 200 milliseconds for smoothing, it's sort of an arbitrary amount, it could be 80.\nSpeaker F: Yeah, probably.\nSpeaker F: So what's the baseline to be under?\nSpeaker E: Well, we don't know.\nSpeaker E: They're still arguing, but...\nSpeaker E: I mean, if it's 250, then we can keep the delta where it is if we shaved off 20.\nSpeaker E: If it's 200, if we shaved off 20, we could meet it by moving the delta back.\nSpeaker F: So, have you know that what you have is too much if they're still the same?\nSpeaker E: Oh, we don't, but it's just...\nSpeaker E: I mean, the main thing is that since we got burned last time, but not worrying about it very much, we're just staying conscious of it.\nSpeaker E: Oh, okay.\nSpeaker E: And so, I mean, if a week before we have to be done, someone says, well, you have to have 50 milliseconds less than you have now,\nSpeaker F: it would be pretty magic around here, so. That's still best.\nSpeaker F: That's a pretty big win, and it doesn't seem like you're in terms of your delay or...\nSpeaker F: That.\nSpeaker E: He added a bit on, I guess, because before we were able to have the noise stuff and the LDA being parallel, and now he's requiring it to be done first.\nSpeaker B: Well, I think the main thing, maybe it's the cleaning of the speech, which takes 40 milliseconds or so.\nSpeaker B: Right.\nSpeaker E: Well, so you say, let's say 10 minutes, I'll take seconds for the LDA.\nSpeaker E: LDA is pretty short, right now.\nSpeaker E: Well, 10, yeah.\nSpeaker E: And then four...\nSpeaker A: Yeah, the LDA, we don't know, is it very crucial for the features, right?\nSpeaker B: No, I just...\nSpeaker B: Yeah.\nSpeaker B: This is the first try, I mean, maybe it's not very useful.\nSpeaker E: But I think you have, I mean, you have 20 for Delta computation, which you now is already doing twice, right?\nSpeaker E: Were you doing that before?\nNone: Hmm.\nSpeaker B: Well, in the proposal, the input of the VAD network were just three frames.\nSpeaker A: Yeah, just a static.\nSpeaker B: No, that's the features.\nSpeaker E: So what you have now is 40 for the noise, 20 for the Delta and 10 for the LDA.\nSpeaker E: That's 70 milliseconds of stuff, which was formerly in parallel, right?\nSpeaker E: So I think, you know, that's the difference, as far as the timing.\nSpeaker E: Yeah.\nSpeaker E: Right.\nSpeaker E: And you can experiment with cutting various pieces of these back a bit.\nSpeaker E: But, I mean, we're not...\nSpeaker E: We're not in terrible shape.\nSpeaker F: Yeah, that's what it seems like.\nSpeaker E: Yeah.\nSpeaker E: It's not like it's heading up to 100 milliseconds or something.\nSpeaker F: Where is this 57.02?\nSpeaker F: And in comparison to the last evaluation?\nSpeaker F: Well, it's, I think,\nSpeaker E: it's better than anything in any bodyguide. I was alright.\nSpeaker B: The best was 54.\nSpeaker B: Yeah, 5.\nSpeaker B: And our system was 49, but with the neural network.\nSpeaker B: With the neural net.\nSpeaker E: Yeah.\nSpeaker A: So this is like the first proposal.\nSpeaker A: The proposal wanted was 44, actually.\nSpeaker A: Yeah.\nSpeaker E: Yeah.\nSpeaker E: And we still have a neural net in.\nSpeaker E: So it's, you know, it's, you know, so it's, we're doing better.\nSpeaker E: I mean, we're getting really good.\nSpeaker E: Better recognition.\nSpeaker E: I mean, I'm sure other people work in this, I'm not sitting still either.\nSpeaker E: But, yeah.\nSpeaker E: But, I mean, important thing is that we learn how to do this better.\nSpeaker E: So.\nNone: Yeah.\nSpeaker E: So our...\nSpeaker E: Yeah, you can see the kind of numbers that we're having, say, in speech deck car, which is a hard task.\nSpeaker E: It's really, you know, it's a sort of...\nSpeaker E: sort of reasonable numbers.\nSpeaker E: It's starting to be.\nSpeaker B: Yeah, even for a well-matched case, it's 60% ever-right reduction.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker D: Okay.\nSpeaker G: Yeah.\nSpeaker B: So actually, this is in between what we had with the previous VAD and what Sunil did with an ideal VAD, which gave 62% improvement.\nSpeaker A: Yeah, it's almost...\nSpeaker A: It's almost an average.\nSpeaker A: It's a little around here.\nSpeaker A: Yeah.\nSpeaker G: What is that?\nSpeaker B: So if you use, like, an ideal VAD for dropping the frames...\nSpeaker B: All of the best we can get.\nSpeaker B: The best that we can get, that means that we estimate the second probability on the clean version of the utterances.\nSpeaker B: Then you can go up to 62% ever-right reduction, probably.\nSpeaker G: Yeah.\nSpeaker F: So that would be even...\nSpeaker F: That would change this number down here to 62.\nSpeaker B: Yeah.\nSpeaker B: Yeah, so you had a good, very good VAD that works as well as VAD working on clean speech.\nSpeaker B: Yeah.\nSpeaker F: Then you would go...\nSpeaker F: So that's sort of the best you could hope for.\nSpeaker E: So 53 is what you were getting with the old VAD.\nSpeaker E: Yeah.\nSpeaker E: And 62 with the...\nSpeaker E: You know, quote unquote cheating VAD and 57 is what you got with the real VAD.\nSpeaker E: That's great.\nNone: Yeah.\nSpeaker B: The next thing is I started to play...\nSpeaker B: But I don't want to worry too much about the delay now.\nSpeaker B: Maybe it's better to wait for the decision.\nSpeaker B: Yeah.\nSpeaker B: The committee.\nSpeaker B: But I started to play with the...\nSpeaker B: TAN-DEM, your own network.\nSpeaker B: I just did the configuration that's very similar to VAD.\nSpeaker B: I just did the configuration that's very similar to what we did for the February proposal.\nSpeaker B: And...\nSpeaker B: So there is a first feature stream that use straight MFCC features.\nSpeaker B: Well, these features actually.\nSpeaker B: And the other stream is the output of a neural network using as input also these cleaned MFCC.\nSpeaker F: I don't know what is going into the 10 and that.\nSpeaker B: So there is just this video stream, the 15 MFCC plus delta and the bell delta.\nSpeaker B: So it makes 45 features that are used as input to the HTK.\nSpeaker B: And then there are more inputs that comes from the TAN-DEM MFCC.\nSpeaker B: Oh, okay.\nSpeaker E: Yeah, he likes using both.\nSpeaker E: So then it has one part that's discriminating with one part that's not.\nSpeaker B: So yeah, right now it seems that I just tested on speech that current while the experiment are running on the IDGs.\nSpeaker B: Well, it improves on the well matched and the mismatched conditions, but it gets worse on the highly mismatched.\nSpeaker B: Compared to these numbers?\nSpeaker B: Compared to these numbers, yeah.\nSpeaker B: Like on the well matched and medium mismatch, the gain is around 5% relative, but it goes down a lot more like 15% on the HM case.\nSpeaker E: You're just using the full 90 features.\nSpeaker E: You have 90 features?\nSpeaker B: From the networks, it's 28.\nSpeaker E: And from the other side, it's 45.\nSpeaker E: It's 45, yeah.\nSpeaker E: It's the 73 features.\nSpeaker E: You're just feeding them like that.\nSpeaker E: There isn't any KLT or anything like that.\nSpeaker B: There is a KLT after the neural network.\nSpeaker F: That's how you get added 28.\nSpeaker F: Yeah.\nSpeaker B: I don't know.\nSpeaker B: It's because it's what we did for the first proposal.\nSpeaker B: We tested it.\nSpeaker B: So we're trying to go down.\nSpeaker B: 27.\nSpeaker B: Yeah.\nSpeaker B: So I wanted to do something very similar to the proposal as a first try.\nSpeaker B: But we have to, for sure, we have to go down because the limit is now 60 features.\nSpeaker B: We have to find a way to decrease the number of features.\nSpeaker F: So it seems funny that I quite understand everything, but that adding features, I guess if you're keeping the backend fixed, maybe that's it.\nSpeaker F: Because it seems like just adding information shouldn't get worse results.\nSpeaker F: But I guess if you're keeping the number of Gaussian fixed in the recognized system.\nSpeaker E: Well, yeah.\nSpeaker E: But I mean, just in general, adding information, suppose the information you added was a really terrible feature and all that brought in was noise.\nNone: Right?\nSpeaker E: So, or suppose it wasn't completely terrible, but it was completely equivalent to another one feature that you had, except it was noisier.\nSpeaker E: In that case, you wouldn't necessarily expect it to be better at all.\nSpeaker F: Oh, yeah.\nSpeaker F: I wasn't necessarily saying it should be better.\nSpeaker F: But it's worse.\nSpeaker E: On the highly mismatched condition, so highly mismatched condition means that in fact your training is a bad estimate of your test.\nSpeaker E: So having a greater number of features, if they aren't maybe the right features that you use, certainly can easily make things worse.\nSpeaker E: I mean, right, if you have lots and lots of data, and you have your training as representing a group of your tests, then getting more source of information should just help.\nSpeaker E: But it doesn't necessarily work that way.\nSpeaker E: So I wonder, well, what's your thought about what to do next with it?\nSpeaker B: I don't know, I'm surprised because I expected the neural net to help more when there is more mismatch as it was the case for the...\nSpeaker A: So was the training set the same as the February proposal?\nSpeaker A: Or the DLG?\nSpeaker B: Yeah, it's the same training set, so it's the same with the DLG.\nSpeaker B: Now, it's added.\nSpeaker E: Well, we might have to experiment with better training sets.\nSpeaker E: But the other thing is, I mean, before you found that was the best configuration, but you might have to retest those things now that we have different, the rest of it is different, right?\nSpeaker E: So, for instance, what's the effect of just putting the neural net on without the other path?\nSpeaker E: You know what the straight features do that gives you this.\nSpeaker E: You know what it does in combination, you know, necessarily, you know what.\nSpeaker F: What if you did what it makes sense to do the KLT on the full set of combined features?\nSpeaker F: Instead of just on the...\nSpeaker B: Yeah, I guess the reason I did it this way is that in February we tested different things like that.\nSpeaker B: So, I think two KLT, I think just a KLT for a network or having a global KLT.\nSpeaker B: So, you try the global KLT?\nSpeaker B: Yeah, and the difference is between these configurations were not huge, but it was marginally better with this configuration.\nSpeaker E: But, yeah, that's obviously another thing to try since things are different.\nSpeaker E: And I guess if the...\nSpeaker E: These are all...\nSpeaker E: So, all of these 73 features are going into the HMM.\nSpeaker E: Yeah.\nSpeaker E: And are there any delta's being computed of them?\nSpeaker B: Of the straight features, yeah.\nSpeaker E: So, not of the...\nSpeaker B: The tandem features are...\nSpeaker B: I know.\nSpeaker B:...used that there are...\nSpeaker B: So, yeah, maybe we can add some context from these features.\nSpeaker B: So, it's good.\nSpeaker B: And, dating, this last work.\nSpeaker E: Yeah, but the other thing I was thinking was...\nSpeaker E: No, I lost track of what I was thinking.\nSpeaker F: What is the...\nSpeaker F: You said there's a limit of 16 features or something.\nSpeaker F: What's the relation between that and the...\nSpeaker F: I know, I should say.\nSpeaker F: 4800 bits per second.\nSpeaker B: No relation.\nSpeaker B: Not the relation.\nSpeaker B: The 4800 bits is for transmission of some features.\nSpeaker B: And, generally, it allows you to transmit like 15...\nSpeaker B:...capsetrums.\nSpeaker E: The issue was that this is supposed to be a standard that's then going to be fed to somebody's recognize or somewhere, which might be...\nSpeaker E: It might be concerned how many parameters are used and so forth.\nSpeaker E: And so, they felt they wanted to settle a limit.\nSpeaker E: So, they chose 60.\nSpeaker E: Some people wanted to use hundreds of parameters and...\nSpeaker E:...that bothered some other people.\nSpeaker E: So, it just chose that.\nSpeaker E: I think it's kind of arbitrary too.\nSpeaker E: But that's kind of what's chosen.\nSpeaker E: I remember what it was going to say.\nSpeaker E: What I was going to say is that maybe...\nSpeaker E:...maybe with the noise removal, these things are now more correlated.\nSpeaker E: So, you have two sets of things that are kind of uncorrelated within themselves.\nSpeaker E: But they're pretty correlated with one another.\nSpeaker E: And they're being fed into these variants only, Gaussian and so forth.\nSpeaker E: And so, maybe it would be better idea now than it was before to have one kLT over everything.\nSpeaker E: To decore a little.\nSpeaker B: Yeah, I see.\nSpeaker E: Maybe.\nSpeaker A: What are the SNLs in the training set limit?\nSpeaker B: It's ranging from zero to clean.\nSpeaker B: Yeah, from zero to clean.\nSpeaker E: Yeah.\nSpeaker E: So, we found this macrophone data and so forth that we were using for these other experiments to be pretty good.\nSpeaker E: So, that's after you expose other alternatives that might be another way to start looking, is just improving the training set.\nSpeaker E: I mean, we were getting lots better recognition using that.\nSpeaker E: Of course, you do have the problem that we're not able to increase the number of Gaussian's or anything to match anything.\nSpeaker E: So, we're only improving the training of our feature set.\nSpeaker E: But that's still probably something.\nSpeaker F: So, you're saying at the macrophone data, the training of the neural net?\nSpeaker E: Yeah, that's the only place that we can train.\nSpeaker E: We can't train the other stuff with anything other than the standard amount.\nSpeaker F: What was the train on again?\nSpeaker B: The one that you?\nSpeaker B: It's timid with noise.\nSpeaker B: So, yeah, it's rather small.\nSpeaker D: How big is the net, by the way?\nSpeaker B: It's 500 unit units.\nSpeaker E: And again, you did experiments back then where you made it bigger.\nSpeaker E: That was sort of the threshold point.\nSpeaker E: Much less than that, it was worse.\nSpeaker E: Much more than that.\nSpeaker E: It wasn't much better.\nSpeaker A: So, is it the performance degradation in the high mismatch is something to do with the cleaning up that is done on the timid after adding noise?\nSpeaker A: All the noises are from the TI digits.\nSpeaker A: Well, it's like the high mismatch of the speech that car after cleaning up, maybe having more noise than the training set of timid after cleaning after you do the noise cleaning.\nSpeaker A: Earlier, you never had any compensation.\nSpeaker A: You just trained it straight away.\nSpeaker A: So, you had like all these different conditions of SNS.\nSpeaker A: Actually, in the training set of neural net, after cleaning up, you have now a different set of SNS for the training of the neural net.\nSpeaker A: And is it something to do with the mismatch that's created after the cleaning up, like the high mismatch?\nSpeaker B: You mean the more noisy utterances on the speech that car might be a lot more noisy than that?\nSpeaker A: I mean, SNR after the noise compensation of the speech deck.\nSpeaker E: So, the training that is being trained with noise compensated.\nSpeaker E: Yeah, it's tough, which makes sense.\nSpeaker E: But you're saying, yeah, the noisy or ones are still going to be even after our noise compensation.\nSpeaker E: You're still going to be pretty noisy.\nSpeaker A: So, now, the after-noise compensation, the neural net is seeing a different set of SNS than that was originally there in the training set of timid.\nSpeaker A: So, the net saw all the SNR conditions.\nSpeaker A: Now, after cleaning up, it's a different set of SNR.\nSpeaker A: That SNR may not be covering the whole set of SNS that you're getting in the speech deck.\nSpeaker E: Right, but the speech deck car that you're seeing is also reduced in noise.\nSpeaker A: Yeah, it is, but there could be some issues of...\nSpeaker B: Well, if the initial range of SNR is different, the problem was already there before.\nSpeaker E: Yeah, because...\nSpeaker E: Yeah, I mean, it depends on whether you believe that the noise compensation is equally reducing the noise, the test set and the training set.\nSpeaker E: I mean, you're saying there's a mismatch in noise that wasn't there before, but if they were both the same before, then if they were both reduced equally, then there would not be a mismatch.\nSpeaker E: So, I mean, this may be...\nSpeaker E: Haven't forbid this...\nSpeaker E: Most compensation process may be imperfect, but...\nSpeaker E: It's maybe stringing something differently.\nSpeaker A: No, that could be seen from the TI digits testing condition, because the noises are from the TI digits, right?\nSpeaker A: Noise.\nSpeaker A: Yeah, so...\nSpeaker A: So, cleaning up the TI digits and if the performance goes down in the TI digits mismatch, high mismatch like...\nSpeaker A: Cleaning training.\nSpeaker A:...or a clean training or a zero-db test.\nSpeaker A: Yeah, so we'll see.\nSpeaker A: Yeah, then it's something to do.\nSpeaker E: I mean, one of the things about... I mean, the macrophone data, I think, was recorded over many different telephones.\nSpeaker E: And so, there's lots of different kinds of acoustic conditions.\nSpeaker E: It's not artificially added, no, I swear, anything.\nSpeaker E: So, it's not the same. I don't think there's anybody recording a record from a car, but I think it's varied enough that if...\nSpeaker E: If doing this adjustments, and playing around with it doesn't make it better, it seems like the most obvious thing to do is to improve the training set.\nSpeaker E: I mean, the condition... it gave us an enormous amount of improvement in what we were doing with meeting recorded digits, even though...\nSpeaker E: There again, these macrophone digits were very, very different from what we were going on here.\nSpeaker E: We weren't talking over telephone here, but it was just, I think, just having a nice variation in acoustic conditions was just a good thing.\nSpeaker G: Yeah.\nSpeaker B: Yeah, actually, to... when I observed on the HM cases that the number of deletions dramatically increases.\nSpeaker B: It doubles... when I had the neural network doubles, the number of deletions.\nSpeaker B: So, you don't know how to interpret that, but... I'm here either.\nSpeaker F: And... and... and...\nSpeaker F: I don't understand the same insertion substitution.\nSpeaker B: Maybe they're a little bit lower.\nSpeaker B: They are a little bit better for me, but...\nSpeaker E: Did they increase the number of deletions, even for the cases that got better?\nSpeaker E: No, it doesn't. So, it's only the highly mismatched.\nSpeaker E: And remind me again, the highly mismatched means that...\nSpeaker B: It's clean training, well, close microphone training.\nSpeaker B: Close microphone training, distant microphone, I speed, I think.\nSpeaker B: Well, the most noisy cases of the distant microphone from testing.\nSpeaker E: Right.\nSpeaker E: So, maybe the noise subtraction is... subtracting half speech.\nSpeaker B: But... I mean, but we thought the neural network is... well, it's a better one.\nSpeaker B: It's just when we add the neural networks.\nSpeaker B: Yeah, right.\nSpeaker E: The feature of the same insertion.\nSpeaker F: Well, that says that, you know, the models and the recognizer are really paying attention to the neural net features.\nSpeaker E: Yeah. But, yeah, actually, the timet noises are sort of a range of noises, and they're not so much the stationary driving kind of noises, right?\nSpeaker E: It's pretty different, isn't it?\nSpeaker B: There is a car noise, so there are just four noises.\nSpeaker B: The car, I think, Babel, Subway, right, and Street, or Airport or something.\nSpeaker B: Train station.\nSpeaker B: So, it's mostly while car is stationary, Babel is stationary background, plus some voices, some speech over it, and the other two are rather stationary also.\nSpeaker E: Well, I think that if you run it, actually, maybe you remember this, when you, in the old experiments, when you ran with the neural net only, and didn't have the side path with the pure features as well, did it make things better to have the neural net?\nSpeaker E: Was it about the same...\nSpeaker E: It was a little bit worse.\nSpeaker B: Then, just the features.\nSpeaker E: So, until you put the second path in with the pure features, the neural net wasn't helping at all.\nSpeaker B: What's interesting?\nSpeaker B: It was helping if the features were bad, just playing the B's around the MCC's.\nSpeaker B: But, as soon as we added the linearization, they were doing similar enough things.\nSpeaker E: Well, I still think it would be interesting to see what would happen if you just had the neural net without the side thing.\nSpeaker E: And the thing I have in mind is, maybe you'll see that the results are not just a little bit worse, maybe that there are a lot worse.\nSpeaker E: But if, on the other hand, it's somewhere in between what you're seeing now and what you'd have with just the pure features, then maybe there is some problem of a combination of these things, or correlation between them somehow.\nSpeaker E: If it really is the net, is hurting you at the moment, then I think the issue is to focus on improving the net.\nSpeaker E: So what's the overall, I mean, you haven't done all the experiments, but you said it was somewhat better, say, 5% better for the first two conditions and 15% worse for the other one.\nSpeaker E: But of course that one's rated lower, so I wonder what the net effect is.\nSpeaker B: I think it was 1 or 2%. That's not that bad, but it was like 2% relative worse once we did that curve.\nSpeaker B: I have to check that to a live.\nSpeaker A: Well, overall it will be still better, even if it is 15% worse, because the 15% worse is given like 25.\nSpeaker G: 0.25 weight.\nSpeaker E: Right, so the worst it could be if the other is exactly as it is 4% and in fact, since the others are somewhat better.\nSpeaker A: So either you get cancel load or you'll get almost the same?\nSpeaker B: Yeah, it was slightly worse.\nSpeaker E: Yeah, it should be pretty close to cancel that.\nSpeaker F: You know, I've been wondering about something. In a lot of the Hub 5 systems recently have been using LDA.\nSpeaker F: And they run LDA on the features right before they train models.\nSpeaker F: So there's the LDAs right there before the HMMs.\nSpeaker F: So you guys are using LDA, but it seems like it's pretty far back in the process.\nSpeaker A: This LDA is different from the LDA that you are talking about. The LDA that you are saying is like you take a block of features like 9 frames or something and then do an LDA on it and then reduce the dimensionality to something like 24 or something like that.\nSpeaker F: And then feed it to HMM.\nSpeaker A: Yeah, so this is like a 2-dimensional time.\nSpeaker A: So this is a 2-dimensional time.\nSpeaker A: And the LDA that we are applying is only in time, not in frequency, across frequency. So it's like more like a filtering in time rather than doing it.\nSpeaker F: So what about, I mean, I don't know if this is a good idea or not, but what if you put, ran the other kind of LDA on your features right before they go into the HMM?\nSpeaker B: Actually, I think, well, what we do with the HMM is something like that, except that it's not linear.\nSpeaker B: But it's like a nonlinear or discriminant.\nSpeaker F: Yeah, so sort of like the tandem stuff is kind of like it's nonlinear, I'll be.\nSpeaker B: But the other features that you have, the non-hand ones, while in the proposal they were transformed using PCA, but it might be that LDA.\nSpeaker E: The argument is kind of, and it's not like we really know, but the argument anyway is that we always have the problem.\nSpeaker E: So, the other thing is that LDA, they're good. They're good because you learn to distinguish between these categories that you want to be good at distinguishing between.\nSpeaker E: And PCA doesn't do that. PCA, or PCA, throws away pieces that are maybe not going to be helpful just because they're small.\nSpeaker E: But the problem is, training sets are perfect and testing sets are different. So you face the potential problem with discriminative stuff, B-L-D-A, or neural nets, that you are training to discriminate between categories in one space, but what you're really going to be getting is something else.\nSpeaker E: And so, Stefan's idea was, let's feed both this discriminatively trained thing and something that's not.\nSpeaker E: So, you have good set of features that everybody's worked really hard to make, and then you discriminatively train it, but you also take the path that doesn't have that, and putting those in together.\nSpeaker E: And that seems, so it's kind of like the combination of what Dan has been calling a feature combination versus posterior combination or something.\nSpeaker E: You have the posterior combination, but then you get the features from that and use them as feature combination with these other things.\nSpeaker E: And that seemed, at least on the last one, as he was saying, when he only did discriminative stuff, it actually didn't help at all in this particular case.\nSpeaker E: There was enough of a difference, I guess, between the testing and training. But by having them both there, the fact is, some of the time, the discriminative stuff is going to help you.\nSpeaker E: And some of the time, it's going to hurt you by combining two information sources.\nSpeaker F: So, you wouldn't necessarily then want to do LDA on the non-tandom features because that you're doing something to them.\nSpeaker E: I think that's counter to that idea. Now, again, we're just trying these different things. We don't really know what's going to work best.\nSpeaker E: But if that's the hypothesis, at least to be counter to that hypothesis, to do that.\nSpeaker E: And in principle, you would think that the neural net would do better at the discriminant part than LDA.\nSpeaker F: Well, maybe not. Yeah, exactly.\nSpeaker F: I mean, we were getting ready to do the tandem stuff for the Hubbive system. And Andreas and I talked about it.\nSpeaker F: And the idea, the thought was, well, yeah, the neural net should be better. But we should at least have a number to show that we did try the LDA in place of the neural net.\nSpeaker F: So we can show a clear path. You have a valid, then you have LDA, then you have the neural net.\nSpeaker F: You can see theoretically. So, I was just wondering, I think that's a good idea.\nSpeaker D: Did you do that?\nSpeaker F: No, that's what we're going to do next, as soon as I finish this.\nSpeaker F: Yeah. Yeah. No, well, that's a good idea.\nSpeaker E: You just want to show. I mean, it's not even believed it.\nSpeaker E: Oh, no, it believes it.\nSpeaker E: No, no, but it might not even be true. I mean, it's a great idea.\nSpeaker E: I mean, one of the things that always disturbed me in the resurgence of neural nets that happened in the 80s was that a lot of people, because neural nets were pretty easy to use.\nSpeaker E: A lot of people were just using them for all sorts of things without looking at all into the linear versions of them.\nSpeaker E: And people were doing their neural nets, but not looking at our filters.\nSpeaker E: So I think, yeah, it's definitely a good idea to try it.\nSpeaker F: Yeah, and everybody's putting that on their systems now.\nSpeaker F: That's what made me wonder about.\nSpeaker E: Well, even putting them in their systems off in the out of ten years.\nSpeaker F: Yeah, what I mean is it's like in the Hub5 evaluations, you know, and you read the system descriptions.\nSpeaker F: And now they all have that.\nSpeaker F: Everybody's got LDA on their features.\nSpeaker B: It's the transformation that are estimating, but they are trained on the same data as the final nature.\nSpeaker B: Yeah, so it's different.\nSpeaker F: Yeah, exactly, because they don't have these mismatches that you guys have.\nSpeaker F: So that's why I was wondering if maybe it's not even a good idea.\nSpeaker F: I don't know.\nSpeaker F: That's about it.\nSpeaker E: I mean, part of why I think part of why you were getting into the KLT, you were describing to me at one point that you wanted to see if, you know, getting good orthogonal features was in combining the different temporal ranges was the key thing that was happening, or whether it was the scrumid thing, right?\nSpeaker E: So you were just trying.\nSpeaker E: I think you, I mean, this is, it doesn't have the LDA aspect, but as far as the orthogonalizing transformation, you were trying that, at one point, right?\nSpeaker E: Thank you.\nSpeaker G: Yeah.\nSpeaker E: That's something.\nSpeaker E: That's what is well.\nSpeaker E: Yeah.\nSpeaker A: So, I've been exploring a parallel VAD without neural network with, like, less latency using SNR and energy after the cleaning up.\nSpeaker A: So what I've been trying was, after the, after the noise compensation, I was trying to find a feature based on the ratio of the energy that is after clean and before clean.\nSpeaker A: So that if, if they are like pretty close to one, which means it's speech, and it is, if it is close to zero, which is, so it's like it's scaled nicely to a probability value.\nSpeaker A: So, just trying with full band and multiple bands, separating them to different frequency bands and deriving separate decisions on each band and trying to combine them.\nSpeaker A: The advantage being like it doesn't have the latency of the neural network if it, if it can, and it gave me like one point, one more than one percent relative improvements.\nSpeaker A: So from 53.6, it went to 54.8. So it's like, from the slightly more than a person improvements, it's like, which means that it's doing a slightly better job than the previous VAD.\nSpeaker A: I'd lower delay.\nSpeaker A: So, sorry, this is still of the median.\nSpeaker A: It's still as the median field.\nSpeaker E: So it was most of the, yeah.\nSpeaker A: The delay that's gone is the input, which is the 60 millisecond, the 40 plus 20.\nSpeaker A: At the input of the neural network, you have this nine frames of context plus the delta.\nSpeaker E: Oh, plus the author, right?\nSpeaker A: Yeah.\nSpeaker A: So that delay plus the LDA.\nSpeaker A: So the delay is only the 40 millisecond of the noise cleaning plus the 100 milliseconds, smoothing a output.\nSpeaker A: So, yeah, so the, because the problem for me was to find a consistent threshold that works well across a different databases, because I, I try to make it work on speech.car, and it fails on TI digits, or try to make it work on that's just the Italian or something it doesn't work on the finish.\nSpeaker A: So, so there was, there was like some problem in balancing the deletions and insertions when I try different thresholds.\nSpeaker A: So the, I'm still trying to make it better by using some other features from the, after the cleanup, maybe some correlation or the correlation or some additional features.\nSpeaker A: So mainly the improvement of the word.\nSpeaker A: I've been trying.\nSpeaker E: Now this, this, this, before and after clean, it sounds like you think that's a good feature that, that you think that the, it appears to be a good feature, right?\nSpeaker E: Yeah, what about using the neural net?\nSpeaker A: Yeah, so, yeah, so that's, yeah, so we've been thinking about putting it into the neural net also.\nSpeaker A: Yeah, because we did that itself.\nSpeaker A: You don't have to worry about the threshold.\nSpeaker A: Yeah, yeah, so that's the, yeah.\nSpeaker E: Yeah, so if we, if we can live with latency or cut the latency elsewhere, then that would be a good thing.\nSpeaker E: Anybody has anybody, you guys or, or Naren, somebody tried the second, second stream thing?\nSpeaker A: Oh, I just, I just put the second stream in place and, and one experiment, just like, just to know whether everything is fine.\nSpeaker A: So it was like, 45 kept stream plus 23 mile log mail.\nSpeaker A: Yeah.\nSpeaker A: And it was like, it gave me the baseline performance of the Aurora, which is like zero improvement.\nSpeaker A: Yeah.\nSpeaker A: So I just tried it on the talent just to know whether everything is, but I didn't expect anything out of it because it was like a weird feature set.\nSpeaker E: Yeah.\nSpeaker E: Yeah, well, what I think, you know, what, the more what you want to do is, is, is put into another neural net.\nSpeaker D: Yeah.\nSpeaker E: But yeah, we're not quite there yet.\nSpeaker E: So we have to figure out the neural net.\nSpeaker E: I guess.\nSpeaker E: Yeah.\nSpeaker A: The other thing I was wondering was, if the neural net has any, because of the different noise, unseen noise conditions for the neural net, like you train it on those four noise conditions, but you're feeding it with like additional some four plus some few more conditions, which it doesn't seem actually from the wild testing.\nSpeaker A: Yeah.\nSpeaker A: Instead of just having those clean up kept stream should be feed some additional information like the, we have the VAD flag and should be feed the VAD flag also at the input.\nSpeaker A: So that it has some additional discriminating information at the input.\nSpeaker A: We have the VAD information also available at the back end.\nSpeaker A: So if it is something, the neural net is not able to discriminate the classes.\nSpeaker A: Yeah.\nSpeaker A: I mean, because most of it is, when we have dropped some silence, we have dropped silence, we haven't dropped silence for him still.\nSpeaker B: Still not yet.\nSpeaker A: Yeah. So the biggest classification will be the speech and silence.\nSpeaker A: So by having an additional feature which says this is speech and this is non speech.\nSpeaker A: I mean, it certainly helps in some unseen noise conditions for the neural net.\nSpeaker F: What do you have that feature available for the test data?\nSpeaker A: Well, I mean, we have, we are transmitting the VAD to the back end, feature to the back end because we are dropping it at the back end after everything.\nSpeaker A: All the features are computer.\nSpeaker A: Oh, oh, I see.\nSpeaker A: So that is coming from a separate neural net or some VAD, which is certainly giving you that also.\nSpeaker A: Yeah.\nSpeaker A: So it's an additional discriminating information.\nSpeaker E: You could feed it into the neural net.\nSpeaker E: And the other thing you could do is just modify the output probabilities of the neural net based on the fact that you have a silence probability.\nSpeaker E: So you have an independent estimator of what the silence probability is and you could multiply the two things and re-normalize.\nSpeaker E: Yeah, I mean, you have to do the non-linearity part.\nSpeaker E: Do that.\nSpeaker E: I mean, go backwards once the non-linearity would be.\nSpeaker F: But, uh, maybe I went, but in principle, wouldn't it be better to feed it in unless the net do that?\nSpeaker E: Well, not sure.\nSpeaker E: I mean, let's put it this way.\nSpeaker E: I mean, you have this complicated system with thousands of thousand parameters and you can tell it learn this thing or you can say, it's silence, go away.\nSpeaker E: I mean, I think the second one sounds a lot more direct.\nSpeaker F: What if you, right?\nSpeaker F: So what if you then, uh, since you know this, what if you only use the neural net on the speech portions?\nSpeaker F: Well, well, it's, well, it's that's the same.\nSpeaker F: Yeah, I mean, you have to actually run it continuously, but it's just.\nSpeaker F: I mean, train the net only on.\nSpeaker E: Well, no, you want to train it on the non-speech also because that's part of what you're learning in it to generate that it's, you have to distinguish between.\nSpeaker F: You need to multiply the output and it by this other decision.\nSpeaker F: Uh, then you don't care about whether that makes that distinction.\nSpeaker E: Wait, but this other thing isn't perfect.\nSpeaker E: Uh, so that you bring in some information in the net itself.\nSpeaker F: That's a good point.\nSpeaker E: Yeah.\nSpeaker E: Now, the only thing that bothers me about all this is that I, I, the fact it's sort of bothersome that you're getting more deletions.\nSpeaker B: Yeah.\nSpeaker B: So I might maybe look at, is it due to the fact that, uh, the probability of the silence, the output of the network is, uh, is too high to I or.\nSpeaker B: Yeah.\nSpeaker B: So maybe it's okay.\nSpeaker A: Yeah, it's not really doing any distinction between speech and non speech or I mean different MN classes.\nSpeaker F: I'd be interested to look at the, yeah, for the, but if you look at the, um, high mismatch, the output of the net on the high mismatch case and just look at, you know, the distribution versus the, the other ones, you, you see more peaks or something.\nSpeaker B: Yeah, yeah, entropy of the output.\nSpeaker B: Yeah.\nSpeaker E: For instance.\nSpeaker B: But it seems that the VAD network doesn't, well, it doesn't drop, uh, too many frames because the, the number of deletion is reasonable.\nSpeaker B: But it's just when we had to tend them, the final MLB at them.\nNone: Yeah.\nSpeaker E: Now, the only problem is you don't want to take, I guess, wait for the output of the VAD before you can put something into the other system because I'll shoot up the latency a lot, right?\nSpeaker E: Am I missing something here?\nSpeaker B: Yeah, right.\nSpeaker E: Yeah. So that's maybe a problem with what I was just saying.\nSpeaker E: But, but I guess.\nSpeaker F: But if you were going to put it in as a feature, it means you already have it at the time you get to the tandem net, right?\nSpeaker A: Um, well, we, we don't have it actually because it's, it has a higher rate in the VAD as a.\nSpeaker E: Yeah.\nSpeaker E: Okay. It's kind of done, I mean, some of the things are not in parallel, but certainly it would be in parallel with a, with a tandem net in time.\nSpeaker E: So maybe if that doesn't work, um, but it'd be interesting to see if that was the problem anyway.\nSpeaker E: And, and, and I guess another alternative would be to take the feature that you're feeding into the VAD and feeding it into the other one as well.\nSpeaker E: And then maybe we just learn, learn it better.\nSpeaker E: Um, but that's, yeah, that's an interesting thing to try to see if what's going on is that in the highly mismatched condition, it's, um, causing deletions by having this silence probability up, up too high.\nSpeaker E: At some point where the VAD is saying it's actually speech. Yeah. So it's probably true.\nSpeaker E: Because well, the VAD said since the VAD is, is, is right a lot. Anyway, might be. Yeah, well, we just started working with it. These are some good ideas, I think.\nSpeaker B: Yeah, and the other thing, well, there are other issues, maybe for the tandem, like, what do we want to, we want to work on the targets.\nSpeaker B: Like, instead of using phonemes, using work on text dependent units.\nSpeaker B: But the time, yeah, I'm thinking also about dance work where we trained the network, not on phonemed targets, but on the HMM state targets.\nSpeaker B: And you know, it's giving slightly better results.\nSpeaker E: The problem is if you were going to run this on different tests, including large vocabulary. Yeah.\nSpeaker B: I was just thinking maybe about, like, generalized iPhones. Come up with a reasonable, not too large set of context dependent units.\nSpeaker B: Yeah. And then anyway, we would have to reduce this. Yeah.\nSpeaker E: Yeah. So, but. Yeah. Maybe. But I, it's all worth looking at, but it sounds to me like looking at the relationship between this and the speech and voice stuff is, is probably a key thing.\nSpeaker F: That and the correlation between stuff. So if, if the high mismatch case had been more like the other two cases, in terms of giving you just better performance, how would this number have changed?\nSpeaker B: It would be, yeah, around 5% better, I guess, if like, six if. Well, we don't know what's going to be a TI that it's, yeah, it's back.\nSpeaker B: If you extrapolate the speech that car will match in medium mismatch, it's around maybe five. So this would be 62.\nSpeaker B: So, around 60 must be. Right. Yeah. Well, it's around 5% because if everything is 5%.\nSpeaker B: All the other ones were 5%. I just have just speech that car. So, yeah.\nSpeaker B: So, I'm just running each, should have the result today during the afternoon.\nSpeaker E: Well. So, I won't be here for.\nSpeaker E: I'm leaving next Wednesday. May or may not be in the morning or the afternoon.\nSpeaker E: Are you, you're not going to be around the afternoon? Yeah. Oh, well, I'm talking about next week. I'm leaving, leaving next Wednesday.\nSpeaker E: This afternoon, oh, right for the meeting meeting. Yeah, that's just because of something on campus.\nSpeaker E: Ah, okay. Okay. But, yeah. So next week I won't. And the week after I won't. So, I'll be in Finland.\nSpeaker E: And the week after that I won't. By that time you'll be, you both become from here.\nSpeaker E: So, there'll be no, definitely no meeting on September 6th. What September 6th?\nSpeaker E: Ah, that's during your speech. Oh, so, Sonil will be in Oregon.\nSpeaker E: Stephanie and I will be in Denmark. Right. So, it'll be a few weeks really.\nSpeaker E: Before we have a meeting of the same cast of characters. But, I guess, just, I mean, you guys should probably meet and maybe bury.\nSpeaker E: I'll be around. And then, we'll start up again with Dave and, Dave and Barry and Stefan.\nSpeaker E: And us on the 20th. No. 13th.\nSpeaker F: So, you're going to be gone for the next three weeks or something. I've gone for two and a half weeks starting next Wednesday.\nSpeaker F: So, you won't get the next three of these meetings. Right.\nSpeaker E: I won't. I was probably four because of the three. See, 23rd, 36th. That's right. Next three.\nSpeaker E: And the third one probably won't be a meeting because, because, uh, Sonil, Stefan and I will not be here.\nSpeaker E: So, it's just the next two where there will be, there may as well be meetings, but I just won't be at them.\nSpeaker E: And then starting at the 13th. Uh, we'll have meetings again that we'll have to do without Sonil here somehow.\nSpeaker E: 31st. Yeah. Yeah. So.\nSpeaker F: When is the evaluation of November or something? Yeah, supposed to be November 15th. Does anybody heard anything different?\nSpeaker B: I don't know. The meeting is the five and six of December. So, yeah. Stentatively.\nSpeaker B: That's a proposed deal, I guess. Yeah. So, the evaluation should be a week before. Yeah.\nSpeaker E: But, no, this is good progress. So. Okay. Did it. Yeah.\nSpeaker F: That's good. L-352. 576456704693. 685091394648. 34427182. 187499845897. 1839.\nSpeaker F: 1839 01453629. 543 626673. 7151 60725942.\nSpeaker B: 888819818. Transcript L-353. 791126 542. 873 984 9646. 3574225961.\nSpeaker B: 599 97 985182. 75456653012. 9907 3926. 019 398 0350. 286 202 181.\nSpeaker A: Transcript L-354 296 8637605. 7156 1370 4256. 9537 0218 1863 987 111029.\nSpeaker A: 35334930315. 4086489503. 80289 791. 8991351804.\nSpeaker E: Transcript L-3508442322617. 1283199113. 4524596233. 3846552025.\nSpeaker E: 4693133646. 2846414464. 2994 3287 8742. 428207 480.\nSpeaker E: It's a wrap.\n"}, {"meeting_id": "IS1000c", "summary": "The meeting was mainly about conceptual design. First, the marketing gave some opinions about product requirements based on market surveys and the group discussed some specific ways to fulfill them. After that, the user interface designer proposed to make a universal controller with speech recognition and a blinking thing. A list of basic components were also given. When it came to the cost control, the group believed that they should do some compromise to keep the cost under 12.5 Euros and the specific design scheme would be decided at the next meeting.", "dialogue": "None: I'm making it out.\nSpeaker C: Rock this time, this time.\nNone: There we go.\nNone: There we go.\nSpeaker C: It's not that complicated, but I get it right every time.\nNone: I'm going to do this.\nSpeaker A: I guess we're just waiting for.\nSpeaker A: I'm not doing it. I'm making it with you.\nSpeaker D: I suggest we start a meeting without Matthew.\nSpeaker D: He's late for some reason.\nSpeaker D: Today we will talk about the conceptual design.\nSpeaker D: I hope we both did some work concerning the idea.\nSpeaker D: This will be the agenda for the meeting.\nSpeaker D: I will take some minutes again.\nSpeaker D: We will have to presentations of a few different team members.\nSpeaker D: Then try to come to decisions about the concepts you presented.\nSpeaker D: We have 40 minutes to complete this.\nSpeaker D: Who has this?\nSpeaker D: I think Matthew is important.\nSpeaker A: Matthew is here.\nSpeaker D: There is Matthew.\nSpeaker A: It's really a team project with a team.\nSpeaker A: If someone is not here, then we can make it okay.\nSpeaker A: It's good.\nSpeaker C: I'll just email you this file.\nSpeaker C: My presentation.\nSpeaker D: You have a presentation already?\nSpeaker C: I'm a CML IT.\nSpeaker B: I sent you the slides.\nSpeaker D: I see them.\nSpeaker B: TEACHS are coming up.\nSpeaker C: Okay, so this is just a presentation on the trends that we're going to use to make the product stand out from the rest of the products out there at the moment. So we have to work out a way what we can do with our product to make it stand out and make it so people want to buy it. This is to do this, I will not trim my microphone. We basically use some focus group surveys which I went through with you last time, the main results of that.\nSpeaker C: And some research of the current design and fashion trends throughout the moment. And as part of this, the important aspect is that it came out or things that were already discussed really. The most important by far was the look and feel of it. It needs to be something that's very different from everything else out there. It needs to stand out. It needs to be not functional like the rest of the things out there at the moment. Most people find remote control of boring at the moment. We need to have something that looks interesting, that looks exciting, that will stand out. People will want to buy it. That was twice as important as the next item on here which is that it has to be technologically innovative, has to have something else. Apart from just the look of it, people have to then think about it and say there's got something there that I want that's a really cool feature.\nSpeaker C: And it has to make them want to buy it again. Third on the list and again, innovative was twice as important as its last aspect. It has to be easy to use. So they have to be able to be able to look at it and have some intuitive idea of how to use it. Drawing on the fashion trends at the moment are fruit and vegetables. This is basically talking about just the feel of it. So probably not the smell of it. But the bright colours are catching really bold designs. And the sponge you feel. I had to talk to the design people about this but having a remote that's tactile, it feels different. That would be really cool. That will make it stand out.\nSpeaker A: So can you repeat and be more precise about what you just said?\nSpeaker A: It's a wonderful feel.\nSpeaker C: The feeling you make it not that it's spongy is the current thing. Spongy is the current texture. But basically there are no reports, no remotes at the moment which are spongy or tactile at all. So if you make it maybe fairy or soft or something, don't leave something that sits at a pan. Rather than just be a plastic which they all are at the moment.\nSpeaker C: So as far as design goes, the very most important aspect was the design to the customers. So going with the fruit and vegetable idea, we've got the bright colours. So makes it stand out. The oranges and the bright yellows and the fluorescent colours, part of the fruit and vegetables. Going back to the idea of taking inspiration from mobile phones. They've all got those, a lot of them have the tangible colours so you can choose what colour the outside is. That's one way of looking at it. Text with feel we just talked about. Maybe it's another way of doing that. So if it's part of the tangible colours, maybe they can choose a different texture. The sponge you want or a soft one or something like that.\nSpeaker C: So they can choose that as they want to.\nSpeaker C: Yeah, that's a very good idea.\nSpeaker C: To get in with their decor and their living room or just what they like. The sports team or whatever. And yeah, still taking the inspiration from the mobile phone design. So functionality, the way the mobile phones work, the way the keypad looks. Also, just the way that a lot of industrial designers going into mobile phones at the moment. The big telling items.\nSpeaker C: People put a lot of thought into that so we can leverage off that. And we can start using some of their ideas.\nSpeaker C: Back to technological innovation. Not quite as important but still a big issue. We talked about having a way of finding the remote control of this being lost. That's one thing we could look at. There are other aspects like LCD screens and speech recognition which weren't, I don't think my personal opinion going to be worth the extra expense and the extra effort that will go into them. I think we better do something basic like this which is very important and would be a really cool feature to put in. Easy use. I had no real specific ideas for this. Maybe just the basic idea of having your core functions big and at the top maybe by themselves. Maybe that you can. The finer details of buttons you don't use as much. Either hidden away or completely separate. And that's the presentation.\nSpeaker D: Okay, good. Very clear. Yeah. Very clear. So they don't have any comments or ideas in that? Maybe we, yes. Well, maybe can decide later on the look and feel of what's a good idea maybe to let the people choose. Yes, there are changeable coppers. But on the other hand, I don't know whether my superiors would be so glad with it because you have to introduce a complete new line of supplies that would be a very complicated organizational.\nSpeaker C: Well, we're selling so many units at this. This is going to be a mass marketer product.\nSpeaker A: We need to have two or three different designs at least. Yeah. A range of a set of three,\nSpeaker D: four different aspects. And of course, it will be, we get, if it works, we get after sales, that would be very good. I mean, those covers could go for three, five euro. It's very good idea. And then maybe we can go achieve presentation because then we could discuss later, like,\nSpeaker B: we can put all the ideas together together in the data. You might have some information\nSpeaker D: on the easy to use. Yeah. And your past is very related to mine. Yeah. When you suggest\nSpeaker B: something, then it has to be attributed. I'll go with that actually. So, then the idea of having a remote is generally you have different keys and different structures, different forms.\nSpeaker B: And they could be like buttons or like, triangle. And they could be of varying sizes if you want to, to basically emphasize a particular key more than other. And maybe like you can have different colors, for example, having the red for the on off, switching on and off the button. So this is a general trend to the method they do. So what I found was that conversely, the conversely, mostly that the TV, VCR, music system of rated ones actually, and they are very specific to each other. But there are some common keys. For example, if you want to forward a VCR and if you want to forward a strong track on the, see, they have this common thing actually you can have. And there is also speech recognition to store channel information names. Like you can basically, if you have multiple functionality, say TV, VCR or something, I say to the TV and the tech, the TV, and you can program the keys if you want to certain keys or even the channel information.\nSpeaker C: I like the idea though, having speech recognition for like the name of the channel like BBC, rather than remember the number of it on the keeper. It's a good idea.\nSpeaker B: Yeah, so you can just, because as more and more channels come, then you have more and more problems to remember the exact channel numbers exactly. Even if you arrange it by however you arrange it, you still have the problem to remember exactly.\nSpeaker A: So what functionalities do you suggest for that?\nSpeaker B: So it's like it should be limited one. In the present market, I saw that says something like they are looking for 80 word thing, 80 word. We shouldn't be that difficult to implement like 80 to 100 word. Basically, you want, you don't want to store all channels in the remote control. You want to store your favorite channel.\nSpeaker B: Yeah, some 10, 12 channel information, you know, you don't want to store all the 100 channel information into that. And basically, it depends like the remote with LCT display for browsing because you have multiple functionalities. For example, you are watching a movie and you are having a universal remote control and you want to, you don't know really which functionality is my reward in now. So I am using the TV. So every time I use it, it could be like, for example, I can use a simple toggle switch and a display. So I press it. So the display says, okay, I have an TV or DVD or whatever it is, instead of having three keys separately for four keys, tomorrow the functionalities will increase actually in for you and you might want, you don't want separate keys for all of them. You can, and well, there can be children friendly where you can program your remote so that they, they are not allowed to browse certain channels which you can block them and you can operate them. So these are the things presently which are seen in the market scenario. So present, I personally would look at things like having a universal remote is a good idea. Like instead of having image, we individual ones for all of them, you can think of having with multiple functionality, possibly with speech recognition, I got a mail from the coffee machine interface unit that they have integrated the speech recognition into the coffee machine. And so if you say hello coffee machine, it's the high Joe or something like that, you know, and the machine is not too many words of using with that. So it's a limited vocabulary thing and maybe isolated word and it's, it is interesting and basically storing the channel through voice or other ways of programming your keys on the display for the browsing which is again, maybe having something like a blinking thing, like it could indicate your, it could indicate what is called like the, whether you have in a battery in your, in your remote, the blinking.\nSpeaker B: At the same time, if it's a dark room, it can be used to locate the remote also.\nSpeaker A: So I really are. And you want, okay, for coming back to one point, you want to let the user programming the keys, some of them? Yeah, you can let them to do that. And isn't that too difficult for the, we want, I don't know if we still want the RC to be easy to use. That's the word.\nSpeaker B: No, but the, it depends on the easiness like the user, how much effort he can put. Like for example, I would like to store in certain way. So if you want to give the full freedom to the user or you want to keep some constraints and let the user use it with their constraint.\nSpeaker C: I think you can do it both ways. You can have it so it's easy. They can pick it up and use it straight away without doing any, without customizing it. Or if they want to, they have the option of\nSpeaker D: using these extra features. So, but I, maybe you can give him to us because I, I'm not sure whether that, that we can implement that for 12 euro and 50 cents. I'm sorry to have, so every time I have to come down on a price again.\nSpeaker D: So this might be a little limiting for your creativity, but it's, it's, it's real.\nSpeaker D: Yeah. We have to consider it. So do you think these ideas and my speech recognition, maybe it's possible for, for 12 euro, then it will be at cost of other functionality.\nSpeaker D: We might implement like the, the, the, the furry, uh, uh, case of the,\nSpeaker A: yeah, like I would say that for priming, uh, keys, you said, uh, it could be, uh, easily, uh, done within the, the package of 12 euros. Yeah. But for the ASR system, uh, I'm not sure if it's feasible\nSpeaker B: to have this. Well, we can still look at, we can talk with the coffee unit and you can check how much,\nSpeaker A: how much, how much, if it's a low vocabulary, it's already implemented. Yeah. I knew how much it's\nSpeaker B: cost. Maybe we can come, we can talk to them and we can come with that. You know, and also, well, you can think of having a, since you have a, you know, something more, and maybe if you add a little bit of even display, you might need the, to keep checking the battery. So anyway, you need some kind of indicator. So it could be a blinking option of a LED, which could actually be used to detect also if it's in a dark room, you can basically detect it also. Yeah. So a lot of the idea to\nSpeaker C: be able to use the remote in the dark. So either having the buttons so you can feel the difference\nSpeaker B: between them or they know, I think it's like, you know, it tells you, um, it can be for two purposes, like if you have an LCD display and all those things, it's not going to be the remote, which is having, we need just a six, six volt, uh, sorry, three volts, um, uh, DC. It may need more actually.\nSpeaker B: So you may need to check your battery usage. And then you need that some functionality to indicate the battery limit. And then if the battery limit is indicated, if it could be indicated, we're blinking something and it can change the color depending on your, how much is the battery.\nSpeaker B: Well, that is good enough to even locate it, even if you want to, you know.\nSpeaker A: Okay. Good. Yeah. So I don't know if, if I have time to talk about, uh, yes, you have\nSpeaker B: time some more. Yeah. Yes. Yes. You can. You can still. Sure. You can, you know. Okay.\nSpeaker A: So what I'm going to present here is very, um, yeah, basic knowledge about the, uh, or the, the components that are inside, uh, RC, and remote control. And how is it manufactured?\nSpeaker A: What is the process is to explain you? So the method is there is a set of components in, uh, in a remote control, like everywhere. And, uh, what caused the, the components in themselves do not cost a lot, but the way to assemble everything costs obviously. And I will show you my preferences, uh, at the end. So there are two, uh, different types of, uh, different ways of using the, the components for making a remote control. Uh, the basic way is to use a, an integrated circuit and some transistors, resonator resistors that aims at communicating, uh, uh, the message and to, to send the message to the, um, to the lead that will transmit to the receiver. And, uh, yeah, there are the components and the circuit board, boring buttons and for a lead, et cetera, for the components. Um, so it's not in finding just to say that the chip can detect, uh, when a key is pressed and then translate to the key to a sequence, something like Morse code, as you know, uh, with different sequence for each key. And, uh, and that's with the components we use, we'll have different, uh, messages, different sequences. And the chip sends that signal signal to the transistor that amplify to make it stronger. Um, so electronic parts are assembled onto printed boards, uh, because it's easier to mass produce an assemble. And, uh, uh, so I think for our design, we want some, uh, programmable, uh, you know, VLSI or FPGA, uh, high technology. And this is important. And so we'll use, uh, yeah, like in any, uh, high tech devices, uh, chip of five fiberglass to solve them and connect them.\nSpeaker A: Um, so my personal design, we need to find a solution. What, um, what is the material of the cover we want to use? If it's plastic or you said that, yeah, you had some ideas, uh, like foot veg or,\nSpeaker D: uh, maybe we can give the, uh, the case a very, uh, uh, normal, very normal case, but with a change book covers to fancy it up. So like a normal cheap plastic case, which can be covered up in, for instance, a wooden case. I'm just doing with cars, I think inside the car. Yeah.\nSpeaker A: Yeah. So they also emailed me that, uh, they have, uh, available a bunch of different buttons, squirrel wheels, integrated push buttons, such as a motor computer, mouse, and, uh, very cheap LCDs. So liquid crystal displays. So I'm wondering, I think we might be able to integrate LCD into our RC. And the final point, okay, is, um, we have, yeah, there are some, uh, compromise to, to do. So we have to know that the push button requires a simple chip, but the squirrel, uh, and that kind of higher high tech stuff needs more money, um, which is a higher price range, right?\nSpeaker A: And the display requires an advanced chip, which in turn is more expensive than the regular chip, but I think, uh, with 12 euros, um, any if it's, uh, made for four million, uh, items, then I think we could be able to under that. So to, to sum up, um, we need, yeah, so I just said that the components, the list of components, uh, has to be, uh, yeah, listed and, uh, and, um, assembly is an important process that has to be taken into account. And, uh, for the designing of the car, uh, cover layout, then it's better to, to maybe see that with, uh, the, your, uh, UI expert so that we can, it's really a team working. Uh, so I, I cannot design something without\nSpeaker B: your agreement, right? Sure. Of course. For example, uh, I wanted to know, like, if you want to have, uh, uh, if you want to have the LCD display over there, or if you want to store the programs with the keys, what kind of things you need inside your thing inside? What, what, yeah, it's kind of, um, simple\nSpeaker A: FPGA program, programmable device. Okay. And we have to insert, I think we could insert one that could\nSpeaker B: under the several functions of, okay. So in that case, you can even look at the technology, what the mobile phone is trying to use with the PGA card. Exactly. Yeah. Focused onizing. Yeah. They have actually how cost effective it would be to put that chip into it and do that. Yeah. So I think we,\nSpeaker D: we should come to some decisions now. Yeah. Uh, about this. Um, so I understand, uh, when we want the display, we need a expensive chip. But when we want the scrolling wheel, we also need to expensive chip. So can we use chain chip? So with one expensive chip, we can, uh, implement several\nSpeaker A: complex, or, uh, advanced future. Exactly. Yeah. That's a very good idea. We could have one main chip that could handle, uh, it's called FPGA chip that could handle both, uh, like scrolling wheels, as well as, uh, LCD. So when the more expensive chip you mentioned there is, is possible in the,\nSpeaker D: in the given budget, uh, maybe we should go for, for the more expensive chip. So all features, which you mentioned can be implemented based on the same chip. Yes. Well, do you think just\nSpeaker B: feasible? I don't know if it will fit into our cost of, uh, you know, you think, you think it's\nSpeaker A: possible? Is it possible to fit in? Yeah. I was thinking, I think both, uh, if we had a budget of 20, 20, uh, euros, it will be okay. But, um, but we need specific costings and actually do\nSpeaker C: no, two designs and then cost them out and see which one is going to fit in our budget better.\nSpeaker D: Yeah. That's an excellent idea. Yes. Well, when you make a, uh, a design, you can, next meeting, you can give an quite an exact cost price. Yeah. Yeah. Because right now I don't\nSpeaker A: have the price in, in head, but for next meeting, I'm sure you're, we are. Yeah. That's, that's\nSpeaker B: something which I wanted to ask you also, like what will be the each inventory, the cost of it. Yeah. For example, if you want to put wood, I wouldn't suggest for wood. Okay. And the big, because it's, I think it's much easier to use a plastic or anything like that. And rather than a wood, it will be much, much expensive though it's the most natural thing. But yes, but I think,\nSpeaker D: I think we can just use more cheap plastic for kind of basic addition that people can fancy it up with, with more expensive materials, which, which, which, which, with, we can give the preference to\nSpeaker B: them. But it is, but with plastic or the rubber or whatever it is, it's much better with that.\nSpeaker A: Rather than buying. Yeah. But it's a detailed, yeah. Plastic versus wood. And we need maybe to center our discussion on the, the really what buttons, what functionality you want to offer to the user. Yeah. And maybe with graphs, I don't know, as a user interface designer, you could maybe\nSpeaker D: help us on that. I think for next meeting, we, you two can present a real design. Okay. Okay. So drawing is on the board. Perfect. And then we now only have to, to decide the general function. So let's say next meeting, we produce two designs. Yeah, sure. One, one.\nSpeaker D: Yeah, we have one less advanced and one more advanced than with the cost price.\nSpeaker D: For the more we go for the, for the basic plastic case, which can be laid, fancy it up with a, with the additional, how do you call them? This like, like mobile telephones, you can put a cover over it. Yeah. But that can be done later. Yeah.\nSpeaker D: Yeah. Now we can concentrate on the, on the basic remote control. Okay. We can give them\nSpeaker B: smooth keys, you know, smooth keys with bigger eyes. So that, you know, the problem most of the time you see in the case is that it's small. Yeah. And every time you have to be very, but if we, if we, if we go to a different ways of designing those keys, then you can merge them\nSpeaker C: together. So is there any, is it you're looking at particular? Oh, you can actually, for example,\nSpeaker B: if you see they're, they're, they're quite small over here. And now you can, for example, as I was running, as you make them big, it may change the look of the thing also to the people at the same time, it is more, like it would be more interesting for people who are having these hours.\nSpeaker B: Yeah. Big, big keys. Big keys.\nSpeaker B: Better true for them, actually. I agree on. I'm not too, too many keys, of course.\nSpeaker C: No, no. Well, one I've had before a remote control we have at home is one has actually got a cover on the bottom. So the bottom bit is just covers half the keys most of the time. And you can start\nSpeaker D: the cover back to get to the, the more advanced case. Yeah. But then you have still have, when you don't use it, you have such a, an extent of your remote control, which you don't use. So maybe it's possible. I don't know whether you can, can indicate this that you can ask to, where open your remote control and on the inside, yeah, are buttons you don't use that much. I've seen that before, too.\nSpeaker C: Another, another, like the, it drips up and then you got another low button underneath.\nSpeaker B: Yeah. So it's something like this, the model, yes, where you can put the key.\nSpeaker C: That's what you mean. I've seen also with keys and buttons on the top of here as well.\nSpeaker D: That's what it means. So I mean something like, like a book.\nSpeaker C: I like this one. I like the shape of this one. Yeah.\nSpeaker C: Can we have, can we think about maybe having a normal tongue, not non-renticular one?\nSpeaker C: So it's not just the state little box. So they curved or something.\nSpeaker A: Maybe we need to also to make a decision on how, how big we want to be and how many buttons,\nSpeaker C: like we should. Is this for the next week? No, I think it might be our time, our time for this\nSpeaker B: meeting. Yeah, that, yeah, next meeting we should be. Just, just make two designs.\nSpeaker D: Yeah, that, that, turn up on a second. And then we, we can assign, design, swing those designs. Yeah. Okay. So anyone, any questions? No, no, no.\nSpeaker C: So is there anything else I need to do from a marketing point of view for the next meeting?\nSpeaker D: Yes, I come to that.\nSpeaker B: Maybe it would be interesting if you could look for the cost inventories of other devices if you're\nSpeaker D: using speech recognition or something like that. Yes, well, maybe, I don't know whether this is possible. Maybe you can start evaluating their work somehow. Okay. Or is this me designing\nSpeaker C: a way to evaluate it? So thinking about how to set up test scripts and things. I don't know whether\nSpeaker D: that's possible in the given time, but as far as possible. Okay. So you two will be together on two prototypes. Yeah. And further instructions will be, will be sent to you by two or three\nSpeaker B: prototypes. Two. One for like cost and the one is like higher than. Okay. So that, and then we can be recombining them. Find a compromise between both of them. Yeah. Perfect. Yeah. Okay. Let's call this to an end. Okay. Thanks. Thanks. So we have done.\n"}, {"meeting_id": "TS3009d", "summary": "The group mainly talked about the extra deciding of the product at this meeting including the presentation of the prototype assessment, the discussion of the requirements and trends of marketing and the product cost and quotation. Firstly, the Industrial Designer introduced the prototype of the product. It had not only the round basic shape which was made of hard plastics and titanium using different colors, but also the buttons like channel and volumn. Later, the group discussed some details and changes of redesigning the logo, buttons and screens. Next, the Marketing Expert mentioned the exterior of the product, the material attraction, and how easy it was to learn or use the basic functions of the product. Besides, the group discussed some details of the cost of components and made the product cheaper by replacing the titanium by hard plastics with similar color. They finally got an estimate of fourteen point one Euros, which was above the budget. In the end, they discussed the distinction between the product and other devices, and celebrated the completion of the project.", "dialogue": "None: You got a tan from all of them.\nNone: They put some...\nNone: You got a tan. You became brown on your own.\nNone: You don't say you became brown.\nNone: That's not the matter. Speak.\nSpeaker C: They put some stress.\nSpeaker C: I'm just relaxing.\nSpeaker B: You're going to stress out now, so I'm going to...\nNone: Because?\nNone: I'm going to do a...\nNone: We have to do some extra design.\nSpeaker C: You'll see. You'll see.\nSpeaker C: Fourth meeting.\nSpeaker D: We have to do what?\nSpeaker B: Extra deciding.\nSpeaker B: What?\nSpeaker C: Well, we'll see.\nSpeaker C: I'll show you the notes again.\nSpeaker C: Very interesting.\nSpeaker C: Well, you two will present your prototype.\nSpeaker C: Then, I guess that's your bit?\nSpeaker C: Yeah.\nSpeaker C: I didn't see anything about it, so I already thought you were dead to do that.\nSpeaker C: Then you were a telescope.\nSpeaker C: I'll show you how we're going to do with financing this design.\nSpeaker C: That's important to evaluate after we have redesigned it.\nSpeaker C: We'll see about the costs.\nSpeaker C: We'll evaluate our production.\nSpeaker C: Then, we can close.\nSpeaker C: The finance world will later.\nSpeaker C: Firstly, I'll show you the notes.\nSpeaker C: I don't think it's very interesting.\nSpeaker E: I think it is.\nSpeaker C: Oh, nay.\nSpeaker C: I know.\nSpeaker C: All right.\nSpeaker C: This is copy-based.\nSpeaker C: For me, of course.\nSpeaker B: Of course.\nSpeaker B: It's a very strange layout.\nSpeaker B: It's a nice course.\nSpeaker C: Well, we went to the antenna.\nSpeaker C: We had some presentations from you three.\nSpeaker C: I summarized what you said to us.\nSpeaker C: I don't think it's very interesting to go to it.\nSpeaker C: This is what we decided.\nSpeaker C: It's also copy-based from what we made together.\nSpeaker C: We still know that.\nSpeaker C: We can use the time better.\nSpeaker C: Next, you two will present the prototype for us.\nSpeaker B: We both will.\nSpeaker B: You're going.\nSpeaker B: If I make mistakes, you'll...\nNone: Right.\nSpeaker B: Well, this is how we're at the time.\nSpeaker B: It's pretty much like my draw drew in the during the last meeting.\nSpeaker B: With the different perspectives of it, we'll begin with the front.\nSpeaker B: We have, of course, the round shape, the round basic shape, with the upper part being the front.\nSpeaker B: So, this part, which is made of heartless.\nSpeaker B: We're using different colors.\nSpeaker B: Of course, for the launch, we use the basic upper colors.\nSpeaker B: Later, we'll put out more interesting colors with different patterns and pictures and everything.\nSpeaker B: But basically, different colors.\nSpeaker B: Bright colors, not black, too dark.\nSpeaker B: Fancy colors.\nSpeaker B: Then we have the lower part of the device, which is, of course, part of the back, actually, because it's also titanium.\nSpeaker B: You can see it also on the side view.\nSpeaker B: This part is the front.\nSpeaker B: The rest of it, the underside of it, the back side and the lower part of the front is, of course, titanium, made of titanium, and has the titanium core, of course, the look.\nSpeaker B: Then we have, on the back of the front, the logo in the upper corner, which is made, which is also part of the back, part of the titanium part.\nSpeaker E: It's a double R.\nSpeaker B: It's a double R, yeah, the logo.\nSpeaker B: It's very different.\nSpeaker B: It's a difficult to draw, it's too small, but it's our double R.\nSpeaker B: The logo is in there.\nSpeaker B: So, that's the logo in the upper right corner.\nSpeaker B: Then we have the buttons. It's difficult to draw again.\nSpeaker B: The little oval or round, I think oval will be better for the different channel buttons.\nSpeaker B: So, they're all oval. Those are here.\nSpeaker B: Then we have the...\nSpeaker E: The channel open volume.\nSpeaker B: Yeah, the volume and the channel controls in the middle here.\nSpeaker B: With kind of arrow shapes, which makes it also a bit more exciting and basic.\nSpeaker B: Round or square buttons.\nSpeaker B: And then also here are the two buttons we agreed on.\nSpeaker B: We have the OK button.\nSpeaker B: The OK button is here in the middle of the... of the operator, of the channel and the volume changers.\nSpeaker B: And then we have here the menu button and the...\nSpeaker E: Alright, menu for...\nSpeaker E: And the video button.\nSpeaker B: The button screen.\nSpeaker B: And of course, this is the low part. This is the LCD screen.\nSpeaker B: This is what we made of it. You can make suggestions if you want.\nSpeaker C: If I look at it, the side view...\nSpeaker B: Maybe we should finish first our talk and then you can...\nSpeaker B: Alright.\nSpeaker B: Maybe I don't want to...\nSpeaker B: I don't want to express you, but I'll finish this quick.\nSpeaker B: Go now.\nSpeaker B: OK, I've had everything, I guess, on the back.\nSpeaker E: On the back, yeah.\nSpeaker B: We thought about, yeah, the back is of course totally titanium.\nSpeaker B: And we thought about the logo back in the middle.\nSpeaker B: So again, the double R. We had the logo on front and on the back.\nSpeaker B: Maybe that's too much, but you have to say it if you think that way.\nSpeaker B: And the company slogan, we thought in a kind of arc shape...\nSpeaker B: Yeah.\nSpeaker B: Above the logo.\nSpeaker B: That's basically what we were thinking about.\nSpeaker E: And about the side view.\nSpeaker E: This, the front won't be as thick, but...\nSpeaker E: Well, I see that the drawing technique makes it very difficult to...\nSpeaker B: Oh, and before I get the voice recorder, it's at the bottom.\nSpeaker B: You can record it using a back-apart.\nSpeaker C: When I look at this side view, I think when I have that in my hand, it's terrible.\nSpeaker C: If you look, if this is thick, and this is...\nSpeaker C: Well, it fits...\nSpeaker B:...that it lies over your hand.\nSpeaker C: It fits the hand.\nSpeaker C: You mean...\nSpeaker B: I agree.\nSpeaker C: What I agree is that when you have a search of an arc in the middle, so that the end and the front is a bit thicker, so then it falls over your hand.\nSpeaker E: If you handle or remote, you usually don't have your hand straight in the middle.\nSpeaker D: It depends on the size.\nSpeaker D: If it's kind of small, this is great, but if it's larger, then you want to grab it.\nSpeaker D: And how large is it?\nSpeaker D: Yeah, that's the question.\nSpeaker D: That's the question.\nSpeaker B: Well, yeah.\nSpeaker B: What do you suggest?\nSpeaker B: It's what Mike's prototype, and you seem to agree on it.\nSpeaker C: Well, the sides I haven't seen yet.\nSpeaker C: The sides, well, the sides doesn't really matter.\nSpeaker B: The sides, the sides, the sides.\nSpeaker B: He drew the sides, but you weren't paying attention, as you were going.\nSpeaker B: Well, in any case, we'll discuss it now.\nSpeaker B: I think this is a pretty good idea.\nSpeaker D: Yeah, I agree with the LCD screen.\nSpeaker D: You have it in your palm like this, and you can watch the screen.\nSpeaker D: And if you have it in the middle, your hand might be over it.\nSpeaker B: But you hold it like this.\nSpeaker B: You're not holding it like this or something.\nSpeaker E: You don't grab it.\nSpeaker B: Yeah, how do you hold it?\nSpeaker B: Well, you don't have it like this, more like this.\nSpeaker B: You're using buttons this way, or if you're writing it.\nSpeaker D: So if you have a screen on it, you want to look at your screen.\nSpeaker E: Yeah, well, it falls into your hand, I think.\nSpeaker B: Okay.\nSpeaker B: Yeah, I agree with that.\nSpeaker B: And maybe you can grab it a bit higher.\nSpeaker D: No, I don't think so.\nSpeaker B: That's the reason to put the LCD screen, of course, on the upper side.\nSpeaker E: Well, as far as I can see, three of us agree.\nSpeaker C: Yeah, well, I think if you three agreed, and then that's it.\nSpeaker E: You're the project manager.\nSpeaker E: You can make the hard decision.\nSpeaker C: Yes, but can you leave it like this?\nSpeaker C: All right, if you think that that's the way you should tell it.\nSpeaker B: You said it was totally unusable.\nSpeaker C: But do you know, when I, my personal taste is that I wanted to fall over my hand with a thick,\nSpeaker B: but you don't think this is completely unusable, I guess? No, not totally.\nSpeaker C: Not totally.\nSpeaker C: For me, I would buy it.\nSpeaker B: Of course, you are also human.\nSpeaker B: We have to take every everyone into account.\nSpeaker D: You might be talking with customers.\nSpeaker B: Yeah.\nSpeaker B: Who else thinks like you?\nSpeaker B: We don't know.\nSpeaker B: Maybe you're a thousand people.\nSpeaker C: Maybe.\nSpeaker C: That's more market research.\nSpeaker C: Let it be like this at a space.\nSpeaker B: Let it be.\nSpeaker B: All right.\nSpeaker B: Okay.\nSpeaker B: That's that.\nSpeaker C: Any other suggestions?\nSpeaker C: No, I think it's great.\nSpeaker E: But what about the redesigning?\nSpeaker C: It comes to that later.\nSpeaker C: Okay.\nSpeaker C: You can do it.\nSpeaker C: You're very, very, very powerful.\nSpeaker C: You can do the evaluation criteria on this.\nSpeaker C: All right, great.\nSpeaker C: That's more useful.\nSpeaker C: Then just speaking.\nSpeaker D: Well, this is just a short intro.\nSpeaker D: I'm going to do the evaluation.\nSpeaker D: This is going to be done at the end of seven point skill criteria.\nSpeaker D: So I made a few questions on the hand of the most important requirements and trends.\nSpeaker D: And we have to look if our device is working correctly.\nSpeaker D: Well, I put some questions in a word file.\nSpeaker D: See if I can find them.\nSpeaker A: Wow.\nSpeaker D: I think you have all seen this kind of evaluation.\nSpeaker D: So I don't have to explain it.\nSpeaker D: The first question is, is the device good looking?\nSpeaker D: Because most people thought that earlier the devices were ugly.\nSpeaker D: Ugly.\nSpeaker D: 75% of them.\nSpeaker D: So what do we think?\nSpeaker B: Well, we designed it.\nSpeaker B: So of course we are very...\nSpeaker C: Yeah.\nSpeaker C: We're not quite objective about that.\nSpeaker C: No, we designed it.\nSpeaker D: We have to evaluate it.\nSpeaker D: So I have to take this questionnaire.\nSpeaker D: So we have to answer now.\nSpeaker C: Change your question.\nSpeaker C: I have to take this questionnaire.\nSpeaker C: No, no, no, no, no.\nSpeaker C: But we can go because of the time.\nSpeaker C: Pretty quick through this.\nSpeaker C: Do we find that you're looking?\nSpeaker C: Well, we think so.\nSpeaker C: Yeah, I think we designed it.\nSpeaker D: You know, we designed it to be perfect.\nSpeaker D: But we have to be critical about it.\nSpeaker D: And I have to take all these points and get an effort at the end.\nSpeaker D: So we know where we stand.\nSpeaker D: One.\nSpeaker D: But there are some things we might have forgotten.\nSpeaker C: So it's 0.4.\nSpeaker D: Right.\nSpeaker D: So let's start with the beginning.\nSpeaker B: Just one-on-one.\nSpeaker D: Right.\nSpeaker D: Is it good looking?\nSpeaker B: Well, I guess I think it's pointed towards the youth, of course.\nSpeaker B: If you look at the design and the colors and everything, although that was our target audience, of course.\nSpeaker B: But it's also not completely unacceptable for all the people, I guess.\nSpeaker B: It's not the type that you might be.\nSpeaker D: Yeah, that's for all the people.\nSpeaker B: It's more that classical look.\nSpeaker D: You put a black front on it or something?\nSpeaker B: No, they like black, of course.\nSpeaker B: But I think they think the titanium look of it is also...\nSpeaker B: I think it's also good for them.\nSpeaker B: So I think we both have considered youth and also a bit all the people.\nSpeaker B: Right.\nSpeaker B: It's very good looking and not only for youth, young people.\nSpeaker E: I think we shouldn't discuss any points that long.\nSpeaker D: Right.\nSpeaker E: No.\nSpeaker C: How many points there are?\nSpeaker C: Right.\nSpeaker C: We have to get on.\nSpeaker B: Is it easy to change channels?\nSpeaker B: Yeah, well, I think so.\nSpeaker D: So, last one is 7.\nSpeaker C: Easy to change channels?\nSpeaker C: No, it's false.\nSpeaker C: It's 1.\nSpeaker D: Oh, sorry.\nSpeaker D: Yeah, right.\nSpeaker C: Well, 2.\nSpeaker C: 10 channels?\nSpeaker C: Well, we have to go to it as well.\nSpeaker B: I think it's as easy as GEMI made.\nSpeaker B: Right.\nSpeaker B: So, how can you make it any easier?\nSpeaker D: The power channel and volume buttons are easy accessible.\nSpeaker D: Yeah, you would use it.\nSpeaker D: 2.\nSpeaker D: 2?\nSpeaker D: Right.\nSpeaker D: The device is easy to find if you lose it.\nSpeaker C: Well, no.\nSpeaker C: We don't have anything about that.\nSpeaker E: Yeah, that's easier to find than a normal black one or something.\nSpeaker E: Of course, you call it.\nSpeaker E: Well, 6.\nSpeaker B: Yeah.\nSpeaker B: 6.\nSpeaker B: But we don't have to defies that beeps when you lost it at all.\nSpeaker C: Functions easy to learn.\nSpeaker C: Well, we do.\nSpeaker C: But we have a...\nSpeaker C: We have so few films.\nSpeaker D: Less one.\nSpeaker D: Yeah.\nSpeaker C: I agree.\nSpeaker C: The device are as sensitive.\nSpeaker E: Well, I think 2 because the voice recorded isn't...\nSpeaker E: Yeah.\nSpeaker E: Self-learning.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: All right.\nSpeaker C: 2.\nSpeaker C: Yeah, but just do something.\nSpeaker C: Yeah.\nSpeaker C: Are we taking too much time?\nSpeaker C: I think this is two time consuming.\nSpeaker C: Not too much.\nSpeaker C: I agree.\nSpeaker C: Not too much this all.\nSpeaker C: This is how you have to put this to the customers.\nSpeaker D: Are you sensitive?\nSpeaker C: Well, a bit.\nSpeaker C: So, 4.\nSpeaker C: 4.\nSpeaker D: Yes.\nSpeaker B: Yeah, very much.\nSpeaker C: 1.\nSpeaker C: And features included also.\nSpeaker C: And...\nSpeaker C: Those 9 is defenci-looking.\nSpeaker C: We still think so.\nSpeaker C: Yeah.\nSpeaker C: Device-fencing.\nSpeaker B: And I think...\nSpeaker B: Yeah.\nSpeaker C: Because of the titanium back.\nSpeaker C: Yeah.\nSpeaker C: Are there enough technology?\nSpeaker C: Yeah.\nSpeaker C: Well, also we have 2.\nSpeaker D: Yeah, that's great.\nSpeaker C: Is it device easy to use?\nSpeaker C: Yes.\nSpeaker C: We have not many buttons.\nSpeaker B: So, yeah, maybe 2.\nSpeaker C: The voice recorded.\nSpeaker D: Yeah.\nSpeaker C: 3.\nSpeaker C: The difference about fruit and vegetables implemented well in our covers in our fronts.\nSpeaker C: Yes.\nSpeaker C: So, yes, 1.\nSpeaker C: 1 or 2.\nSpeaker D: 1 or 2.\nSpeaker D: 1 or 2.\nSpeaker D: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 1 or 2.\nSpeaker B: 4, I think.\nSpeaker B: 4.\nSpeaker B: If you look at this.\nSpeaker B: Well, it doesn't really resemble any fruit, that's true.\nSpeaker D: But we have the distracting fruit colors of course.\nSpeaker C: Ok.\nSpeaker C: And you can also have a front with...\nSpeaker C: Yeah, that's true.\nSpeaker C: With fruit lines.\nSpeaker C: With the attractive.\nSpeaker C: Right.\nSpeaker C: Ok.\nNone: Yeah.\nSpeaker E: This is the last meeting.\nSpeaker E: Yes, but we have to decide much more.\nSpeaker C: Because there was some irritating account manager coming to me.\nSpeaker B: I know things were going smoothly.\nSpeaker B: There had to be some kind of trouble along the way.\nSpeaker C: Yes, well, look at the costs at this point.\nSpeaker B: My God.\nSpeaker C: I had to fill it in.\nSpeaker C: It has to go through 12, right?\nSpeaker C: 12 and a half.\nSpeaker C: 12 and a half.\nSpeaker C: So, well, what costs a lot?\nSpeaker C: The sample speed cost 4.\nSpeaker C: The what?\nSpeaker C: The sample speaker.\nSpeaker C: Oh.\nSpeaker C: The sensor.\nSpeaker D: That's easy.\nSpeaker C: Kick it out.\nSpeaker C: Kick it out.\nSpeaker C: Yeah.\nSpeaker C: We have to go to 12 and a half.\nSpeaker C: The speaker costs fair by far the most.\nSpeaker C: Yeah, that's a bit of an optional option.\nSpeaker C: It isn't worth it.\nSpeaker C: No.\nSpeaker C: We could make two different versions, one with and one without.\nSpeaker C: But for this, kick it out.\nSpeaker C: So, zero.\nSpeaker C: Yeah.\nSpeaker C: Let me go to 14.6.\nSpeaker C: But more.\nSpeaker C: Well, the titanium I don't want to lose, actually.\nSpeaker C: No.\nSpeaker C: LC3.\nSpeaker C: Yeah, hand dynamo.\nSpeaker C: Come on.\nSpeaker B: Remote control has a battery.\nSpeaker B: Well, we have a bit of a problem, I think.\nSpeaker E: Why should we use a defense chip?\nSpeaker E: Well, with 14 LC3, you had said.\nSpeaker E: Yeah.\nSpeaker E: Come with a regular chip?\nSpeaker B: No.\nSpeaker B: Why not?\nSpeaker B: Because that, you know, what's the difference?\nSpeaker B: My information is regular.\nSpeaker C: Huh?\nSpeaker C: What's the difference between a simple and a regular chip?\nSpeaker C: Yeah.\nSpeaker C: And simple?\nSpeaker C: Simple.\nSpeaker B: Well, yeah, I read something about it, but.\nSpeaker B: Elementary.\nSpeaker C: Well, guys.\nSpeaker C: Yeah.\nSpeaker B: I read something about it, but it wasn't very clear.\nSpeaker B: I didn't include it in my report.\nSpeaker C: What happens if we do single?\nSpeaker B: Well, you have to use a chip, so.\nSpeaker B: Well, you have to use a defense chip.\nSpeaker B: One.\nSpeaker B: If you have the LCD screen.\nSpeaker E: We have very little options for the more.\nSpeaker B: But if you have a regular chip, you can't have the LCD screen.\nSpeaker D: Oh, well, we have to put that.\nSpeaker D: No, we need to have to do that.\nSpeaker C: That's a factor.\nSpeaker C: Well, we could say, well, this special color.\nSpeaker C: That isn't there because the fronts, they will buy it.\nSpeaker C: The special color.\nSpeaker C: Yeah.\nSpeaker C: But it's only one half.\nSpeaker D: Yeah, it doesn't.\nSpeaker D: Not relevant.\nSpeaker C: Yeah.\nSpeaker C: Then you only have one half left.\nSpeaker C: Yeah.\nSpeaker E: You must change the chip back in this.\nSpeaker B: Yeah.\nSpeaker B: But we don't have to lose the LCD screen.\nSpeaker D: No.\nSpeaker D: Then the whole concept is.\nSpeaker B: Yeah, I know.\nSpeaker B: That's what my information is.\nSpeaker B: I didn't put the advanced chip in there for five years.\nSpeaker D: No.\nSpeaker D: You have to make a chip.\nSpeaker D: But if you don't sell.\nNone: Yeah.\nSpeaker E: We have an advanced chip on print and we have an LCD display.\nSpeaker E: I think that's a bit double.\nSpeaker C: We don't need both.\nSpeaker C: The advanced chip is needed to have an LCD display.\nSpeaker C: Yeah.\nSpeaker C: Says his.\nSpeaker C: There was in the second meeting.\nSpeaker B: Well, what do they mean with curves again?\nSpeaker C: Because I have a single curve to, well, you said double curved.\nSpeaker C: Right.\nSpeaker C: Yeah.\nSpeaker C: Because I thought it's a bit cheaper already.\nSpeaker B: We can also make it flat.\nSpeaker B: No, no, no.\nSpeaker B: What do they mean with curves?\nSpeaker B: Is this curved?\nSpeaker B: Yes, that's curved.\nSpeaker B: One curve.\nSpeaker B: Yeah, there's actually two curves.\nSpeaker B: Yeah.\nSpeaker B: No, it's one curve.\nSpeaker B: One curve.\nSpeaker B: One curve.\nSpeaker B: Simple.\nSpeaker B: But then we have a huge problem, I think.\nSpeaker B: We can never get below the.\nSpeaker C: We have a big financial problem.\nSpeaker E: We make it more expensive to buy.\nSpeaker C: Well, then we have two dollars less profit.\nSpeaker C: Come on, if we make this 50 million, they won't test it to congratulate us.\nSpeaker D: If I make people brand aware, they are willing to pay more.\nSpeaker D: But because we use a brand at the front and at the back, if you have lots of marketing, people might buy.\nSpeaker C: Yeah, or if you make it cool to have.\nSpeaker C: By the way, we also have this one.\nSpeaker D: Oh, that's just great.\nSpeaker C: Oh, cost nothing.\nSpeaker C: Oh, that's nice.\nSpeaker C: Oh, right.\nSpeaker B: I think you'll agree that now we have this screen.\nSpeaker B: It's very not practical to consider this after we have designed the entire thing.\nSpeaker C: Yes, but I just got it.\nSpeaker C: Well, that's pretty.\nSpeaker A: That is pretty stupid.\nSpeaker E: It's a practical, plastic, like titanium.\nSpeaker E: Who?\nSpeaker E: Titanium, color, plastic.\nSpeaker C: You want to jump to titanium?\nSpeaker E: Yeah, well, we have to get cheaper.\nSpeaker C: And make all plastic.\nSpeaker C: Then we're there.\nSpeaker B: But I don't agree, I think.\nSpeaker E: But then we've got to read through the evaluation process again.\nSpeaker C: It's not going to be looking dumb.\nSpeaker C: No, no, no.\nSpeaker E: That no one will buy.\nSpeaker E: It's not ugly looking.\nSpeaker E: The looks remain the same.\nSpeaker E: No, I don't think so.\nSpeaker B: I think the titanium just provided the tough look.\nSpeaker C: And the feel that it is strong.\nSpeaker B: And also, the older people will like it because of that.\nSpeaker B: Yeah.\nSpeaker B: Because, all right, it's not our target audience, but it's useful.\nSpeaker B: We still, we had to put it for all people.\nSpeaker C: To get more people from the younger group, but not lose the one.\nSpeaker B: I think that titanium is very important.\nSpeaker E: All right, but then we get there.\nSpeaker B: Yeah, we have a problem.\nSpeaker B: But you can better dump the LCD screen.\nSpeaker E: You can dump the special color.\nSpeaker E: We use plastic.\nSpeaker E: In plastic case, or any color.\nSpeaker D: I don't think you should dump the LCD screen.\nSpeaker B: Well, what else?\nSpeaker B: I mean, yeah.\nSpeaker D: Nothing.\nSpeaker D: Or change the titanium or dump the LCD screen.\nSpeaker D: But I think you could better change titanium to a hard kind of plastic looking like titanium.\nSpeaker D: Yeah, then lose the LCD screen.\nSpeaker D: Because you have lots of functions in it too.\nSpeaker B: Yeah, well, yeah.\nSpeaker B: All right.\nSpeaker B: I agree with that.\nSpeaker B: So we use, unfortunately, I put in the report, titanium-colored plastic.\nSpeaker C: One is the lowest price you can make and remote for.\nSpeaker C: That's trendy.\nSpeaker B: Yeah, I agree.\nSpeaker B: So titanium-colored plastic for the back.\nSpeaker C: I'll talk to the managers.\nSpeaker E: No, no, no.\nSpeaker E: Titanium stays there.\nSpeaker C: Titanium, I think this is really good.\nSpeaker C: Yeah, this is good.\nSpeaker B: But it is not good enough.\nSpeaker B: So we have to use the ditch the titanium, I'm afraid.\nSpeaker C: Ah, those account managers.\nSpeaker C: What do they know?\nSpeaker C: Come on.\nSpeaker C: Right.\nSpeaker B: What do we know?\nSpeaker B: All we want is a fancy design, but we don't really have a set of costs.\nSpeaker E: Yes.\nSpeaker E: No, because we didn't know anything about it.\nSpeaker D: Yeah, right.\nSpeaker D: You can make it.\nSpeaker D: You can do it with it now.\nSpeaker D: If you don't have the money, you can make it.\nSpeaker D: So this is too expensive.\nSpeaker B: So we have to make it cheaper.\nSpeaker B: Titanium, gone and add plastic.\nSpeaker E: Right.\nSpeaker E: Yeah, but then we've got money.\nSpeaker B: And on plastic times, too.\nSpeaker B: And then we are there.\nSpeaker C: Well, no, it's just on all plastic.\nSpeaker C: All right.\nSpeaker B: Yeah, well, I'll stick.\nSpeaker B: I'll stick.\nSpeaker B: A lot of plastic.\nSpeaker B: It's just free money.\nSpeaker B: So now, too, for the to make it clear.\nSpeaker C: But then we can add a special color.\nSpeaker C: Yeah.\nSpeaker C: Because we have money over left.\nSpeaker C: And we still have money left.\nSpeaker C: What do we want, guys?\nSpeaker C: I want gold plating.\nSpeaker B: No, no, no.\nSpeaker B: Yeah, right.\nSpeaker B: I want chrome.\nSpeaker E: Well, I think the case is double curved.\nSpeaker C: Yes.\nSpeaker A: We have to.\nSpeaker A: Oh, no!\nSpeaker B: All right.\nSpeaker B: No, no, no, but that's safe.\nSpeaker B: Well, you can.\nSpeaker B: You can double curve if you don't have titanium.\nSpeaker B: And that we dropped.\nSpeaker B: So it's going to be done.\nSpeaker B: All right.\nSpeaker B: Well, that's pretty funny.\nSpeaker B: We do want to reach 12.5.\nSpeaker B: It is in bed to stay at 11.\nSpeaker E: It's entertaining, but it looks like it.\nSpeaker E: We get more salary if we make it to you.\nSpeaker B: Nice.\nSpeaker C: We have to jump out of titanium.\nSpeaker C: And we'll hate the managers for that.\nSpeaker C: But now we're going to object.\nSpeaker C: We evaluate our project.\nSpeaker C: Project.\nSpeaker C: Project.\nSpeaker C: Well, satisfaction.\nSpeaker C: Satisfaction.\nSpeaker C: For example, are we satisfactory about our creativity?\nSpeaker B: Well, I can get no satisfaction, but I think it's a, well, if we consider the cost, then this is the best.\nSpeaker C: I think it's terrible that we got those costs at the last moment.\nSpeaker C: Yeah, me too.\nSpeaker C: That's really bad.\nSpeaker C: That's a ridiculous.\nSpeaker B: That's a realistic.\nSpeaker C: That's a really, but also for creativity.\nSpeaker C: We have a nice design.\nSpeaker C: And then you get the cost and you have to jump all your creativity.\nSpeaker E: We use our creativity, but we just have to adapt it to the cost.\nSpeaker B: Which is very practical, but that's the way.\nSpeaker C: Well, all right.\nSpeaker C: Leadership.\nSpeaker C: Next.\nSpeaker C: Terrible.\nSpeaker C: Teamwork.\nSpeaker C: Leadership.\nSpeaker D: Well, it was very democratic.\nSpeaker C: Well, I feel that in the questionnaires each time.\nSpeaker C: So, well, the managers were terrible.\nSpeaker C: So, with their all the useless requirements.\nSpeaker C: But, all right.\nSpeaker B: Teamwork.\nSpeaker B: They didn't think of the requirements.\nSpeaker B: It's the requirements of the user, I guess.\nSpeaker C: Now they said, oh, we won't, we won't use TDX.\nSpeaker C: We won't use DVD.\nSpeaker B: But they based that on user specifications.\nSpeaker C: All right.\nSpeaker C: Teamwork.\nSpeaker C: Well, great.\nSpeaker C: I think.\nSpeaker C: Yeah.\nSpeaker C: Well, what do you think about the smart board?\nSpeaker B: It was a complete disaster.\nSpeaker B: Yeah.\nSpeaker B: That's a smart board.\nSpeaker E: Yeah.\nSpeaker E: And that's a digital pen.\nSpeaker E: It's also, it's more of a smart board.\nSpeaker B: That's the smart board.\nSpeaker C: I like this smart board, but I hate it.\nSpeaker C: This isn't the smart board.\nSpeaker D: Well, it's both a smart board.\nSpeaker D: Yeah, right.\nSpeaker D: But you, this is just a large, large television.\nSpeaker C: You, you used that.\nSpeaker C: It's both a smart board.\nSpeaker C: But this one is used for a desktop and that one is used to draw.\nSpeaker D: Yeah.\nSpeaker D: You could, this is just a beamer function.\nSpeaker D: And you could, you know, draw a picture and things like that.\nSpeaker C: Well, which one did you like?\nSpeaker C: Left or right?\nSpeaker C: That one.\nSpeaker D: That one is accurate.\nSpeaker D: It just doesn't work.\nSpeaker E: You can, I think this is meant by the digital pen.\nSpeaker C: Yes.\nSpeaker C: That is, so.\nSpeaker C: Yeah.\nSpeaker C: I think there's a big, smart board.\nSpeaker D: I think there's a big, smart board.\nSpeaker D: I don't need a smart board.\nSpeaker E: No, it's much more.\nSpeaker E: We, we use that one.\nSpeaker E: Yeah.\nSpeaker D: We need it.\nSpeaker D: Give me a beamer.\nSpeaker C: That's much cheaper.\nSpeaker C: Or install a laptop to a beamer or have this one standing here in an, I like it.\nSpeaker C: Right.\nSpeaker C: Yeah, I like it.\nSpeaker E: All right.\nSpeaker B: But you can, I said it about three times now.\nSpeaker B: A green board with chalk is much more useful than that.\nSpeaker C: Yeah.\nSpeaker C: It is.\nSpeaker B: The simple, cool board.\nSpeaker C: And the digital pen that, that you like that one.\nSpeaker B: No.\nSpeaker B: I didn't use it.\nSpeaker B: I wrote things down, but I didn't.\nSpeaker B: I used it too much.\nSpeaker C: I used it too much.\nSpeaker C: Check it out.\nSpeaker C: Yeah.\nSpeaker C: You can send that to anyone because you just scribbled something on a page for yourself.\nSpeaker C: And then you're going to send it.\nSpeaker C: Yeah.\nSpeaker B: Well, no, and then also you write things down and you can also bring your, your, your note block.\nSpeaker B: So what's the, what's the point of importing it into?\nSpeaker E: I drew this.\nSpeaker E: And I made a mistake, but it would have been useful if I could show this on the screen.\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker B: For drawings.\nSpeaker B: Yes.\nSpeaker E: Not for a personal note.\nSpeaker B: Well, I mean, you can bring your paper work along.\nSpeaker B: But we'll, of course, if you have a lot of paper.\nSpeaker D: Yeah.\nSpeaker C: If you agree on that.\nSpeaker C: If you offer notes, you have to put it in a strict, you have to put in names, then the date and all those things, a note for a meeting are very strict.\nSpeaker C: So if you would write them down for yourself and then put that in your computer, you still have to type it over to words.\nSpeaker C: Yeah.\nSpeaker C: So it doesn't, it doesn't have any.\nSpeaker C: That's definitely.\nSpeaker B: Yeah, I understand.\nSpeaker B: But I think this option is only useful if you've got a lot of paper work, but you can't, it's not very, not, well.\nSpeaker B: I don't think so.\nSpeaker D: A lot of documents are, yeah.\nSpeaker E: It's only useful if you have to draw something, but then it's really useful, I think.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: All right.\nSpeaker C: Yeah, right.\nSpeaker C: Et cetera.\nSpeaker D: Well, the laptops.\nSpeaker C: Yeah, great.\nSpeaker C: Of course, great.\nSpeaker C: Can we keep them?\nSpeaker C: Yeah.\nSpeaker C: You can buy them.\nSpeaker C: Wireless, right.\nSpeaker C: Thanks for the manager.\nSpeaker C: Other things we used here.\nSpeaker C: I hated the camera.\nSpeaker C: I hated the microphones.\nSpeaker C: Well, the cameras.\nSpeaker C: Well, thank you.\nSpeaker D: Really great.\nSpeaker B: Did you really take those in account?\nSpeaker B: No.\nSpeaker B: F of timeline.\nSpeaker B: Did you notice they were there?\nSpeaker C: I haven't looked one time directly at the camera and I don't care about it.\nSpeaker B: But we shouldn't talk about that because we're right.\nSpeaker C: Et cetera rather stands.\nSpeaker C: That stands.\nSpeaker C: Right.\nSpeaker C: New ideas found.\nSpeaker C: What kind of ideas for a future schedule of work with?\nSpeaker C: For future meetings.\nSpeaker E: You have got the option to communicate to email.\nSpeaker E: That's just an environment they set it up with.\nSpeaker B: All right.\nSpeaker B: But that's one new idea.\nSpeaker E: Yeah.\nSpeaker E: Well, new ideas found by this.\nNone: Not.\nNone: Nothing.\nSpeaker C: Well, we don't want this.\nSpeaker C: We hate this.\nSpeaker C: That's...\nSpeaker C: Digital pen is useless.\nSpeaker C: Oh, it isn't.\nSpeaker C: Well, for drawings.\nSpeaker B: Yeah.\nSpeaker C: So it isn't useless.\nSpeaker C: But then you have to have a lot of drawings because if I had a company and I'm going to buy those expensive, huge expensive things, and I have to pay those papers that are expensive, I'm not going to...\nSpeaker D: For people who sketch the old day, I can imagine.\nSpeaker C: Well, they still have a nice graphics program on the laptop because this is huge.\nSpeaker E: Well, you draw on a laptop like you paint or draw with your hand.\nSpeaker E: Yeah, I agree.\nSpeaker E: But if you...\nSpeaker D: Yeah.\nSpeaker D: Well, if this isn't working, if you're sketching, indeed.\nSpeaker B: But what if maybe this board, a smart board is malfunctioning or something, but suppose it was working correctly.\nSpeaker B: Would it be useful then?\nSpeaker B: If it wasn't off all the time.\nSpeaker C: Oh, no.\nSpeaker C: I hate it to draw like that.\nSpeaker C: You can draw anything neat.\nSpeaker D: Yeah, but he's saying if it is correct, and you can draw...\nSpeaker D: Anything you want?\nSpeaker B: Yeah.\nSpeaker B: Very precise.\nSpeaker C: Very precise.\nSpeaker C: It would be perfect following.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It doesn't, but maybe that thing is...\nSpeaker C: It's very expensive towards a green...\nSpeaker C: Yeah, a board.\nSpeaker E: Yeah, but then you can save it then instantly.\nSpeaker E: Yeah.\nSpeaker B: It does.\nSpeaker B: It's useful.\nSpeaker B: Yeah.\nSpeaker B: If it works, maybe this thing is just malfunctioning.\nSpeaker B: So if we get an... if you get another one and you make sure it does work, I think then it's pretty useful.\nSpeaker C: Yeah.\nSpeaker B: Because, yeah, you can draw things quickly and clearly for anyone in the discussion room.\nSpeaker B: And then you can save it immediately.\nSpeaker C: Well, they are now...\nSpeaker C: Hm.\nSpeaker D: Celebration.\nSpeaker C: It is.\nSpeaker C: So...\nSpeaker C: Did you type in...\nSpeaker C: Congratulations, true.\nSpeaker C: It's a celebration.\nSpeaker E: Well, we didn't go to the bar.\nSpeaker E: Finally, it might be a...\nSpeaker C: Money.\nSpeaker C: Well, that's it.\nSpeaker C: I think...\nSpeaker C: I don't know how long we still have...\nSpeaker C: I don't know how long we had for this last meeting.\nSpeaker C: But...\nSpeaker B: Maybe till four o'clock or something.\nSpeaker B: Well, I do know.\nSpeaker C: Well, yeah, it is a bit...\nSpeaker C: Well, we still have to make the end report and all those things.\nSpeaker C: I have to do that.\nSpeaker C: Yeah.\nSpeaker C: Well, you better get started.\nSpeaker C: Did you save this one in the folder?\nSpeaker C: Can you do that?\nSpeaker B: No, no, no.\nSpeaker B: We must save this thing, yeah.\nSpeaker B: In the Share Map map.\nSpeaker E: But, Neil, you've got some work left.\nSpeaker E: I don't know what you would have to do.\nSpeaker C: It's a good word to the folder.\nSpeaker B: Yeah.\nNone: You don't...\nSpeaker E: I have to go to...\nSpeaker E: Yeah.\nSpeaker E: You see your thing at the...\nSpeaker C: Oh, it worked.\nSpeaker C: Two times quick.\nSpeaker C: All right.\nSpeaker B: Shall we take you to the floor?\nSpeaker B: Yeah.\nSpeaker B: Oh, great man.\nSpeaker A: Gonna buy one more...\nSpeaker A: This is nice.\nSpeaker D: Buy one more...\nSpeaker D:... from a battery.\nSpeaker E: And a conch and that's okay.\nSpeaker E: I'm still...\nSpeaker C:... in my design.\nSpeaker B: Do you believe it yourself?\nSpeaker C: Oh!\nSpeaker C: He saved them all ten.\nSpeaker C: Well, all right.\nSpeaker E: They wanted everything they produced.\nSpeaker B: They also wanted to see my cat and his rabbit.\nSpeaker B: Well...\nSpeaker E: My big bird.\nSpeaker B: You're a big, beautiful bird.\nSpeaker C: Where is this move?\nSpeaker C: Maybe the pen is just broken and the board isn't...\nSpeaker B:... or the other way around them.\nSpeaker B: Maybe you are broken.\nSpeaker B: Yeah, I think...\nSpeaker B: So, too.\nSpeaker B: I think you have to make clear to her that we're away.\nSpeaker B: I think she's listening.\nSpeaker B: I also...\nSpeaker E: She always knows...\nSpeaker E: She's your talker.\nSpeaker D: Oh, you mean our big brother.\nSpeaker D: Coach.\nSpeaker C: Our personal coach.\nSpeaker B: I think you're a big brother.\nSpeaker B: I think you're a big brother.\nSpeaker B: Coach. Our personal coach.\nSpeaker B: Our manager.\nSpeaker B: She also our accountant.\nSpeaker B: Is she responsible for sending that information to the lathe?\nSpeaker B: I don't think so.\nSpeaker B: Because then we have to confront her with our...\nSpeaker B: So, as you can see, we're ready.\nSpeaker D: I feel watched.\nSpeaker C: All right.\nSpeaker C: We put the passion...\nSpeaker C:... actually to promote into...\nSpeaker C: But we couldn't because of the costs.\nSpeaker C: That's a time to move on.\nSpeaker A: That's a beautiful thing.\nSpeaker C: Yeah.\nSpeaker C: I like that.\nSpeaker C: Couldn't put the passion into the electronics.\nSpeaker D: It's going to blame our accountant.\nSpeaker D: Do that.\nSpeaker C: Yeah.\nSpeaker C: That's a nice title.\nSpeaker C: Yeah.\nSpeaker C: Fuck the good.\nSpeaker D: Very catchy.\nSpeaker D: Well, I'm sure management would like that.\nSpeaker C: Well, I'm going to resign after this project anyway.\nSpeaker C: So...\nSpeaker D: That's just great.\nSpeaker D: Yeah.\n"}, {"meeting_id": "TS3004c", "summary": "This meeting was about the functional design of the remote control. Firstly, Marketing gave a presentation on the remote control market trend and suggested a fancy look as well as technology innovation features. Secondly, User Interface presented on technical function design and recommended changeable fonts on remote buttons and a modern look with a handheld shape. Thirdly, Industrial Designer presented on component design and suggested a soft material case with flashy colors and transparent buttons. Industrial Designer thought the normal infrared LED should be sufficient. Lastly, the group discussed the remote's features and they decided to have features including normal battery, scroll wheel, round shape, and speech recognition.", "dialogue": "None: It won't wake up.\nSpeaker D: Yeah, I was a bit early.\nSpeaker A: Why?\nSpeaker D: No, I just came in here normally.\nSpeaker D: I was one of them.\nSpeaker E: Okay.\nNone: Come on.\nSpeaker D: Check, check, check.\nSpeaker D: Why won't it wake up?\nSpeaker D: Is it done?\nSpeaker E: It's probably that doesn't work.\nSpeaker E: You turn it off.\nSpeaker D: But huh?\nSpeaker D: No, I don't.\nSpeaker C: Okay.\nSpeaker C: I received an email with a few possibilities on the materials.\nSpeaker C: So I will discuss them.\nSpeaker C: Okay, we just came to the back later.\nSpeaker D: Yeah, I was...\nSpeaker B: Shall we go and talk about the conceptual model?\nSpeaker B: I'm sorry.\nSpeaker D: It's one of those models.\nSpeaker B: So that's me.\nSpeaker B: Okay, so...\nSpeaker B: Okay, so I'll just show you the...\nSpeaker B: The minute.\nSpeaker B: The minute.\nSpeaker B: What's it called?\nSpeaker D: I don't know.\nSpeaker D: Minutes.\nSpeaker B: Okay, so we just talked about...\nSpeaker B: Do you want me to show it there?\nSpeaker B: No, I just...\nSpeaker B: Okay, we just talked about it.\nSpeaker B: It looks...\nSpeaker B: It has a look nice.\nSpeaker B: Usability is very important.\nSpeaker B: People don't want to spend money on something that's similar to cheaper ones.\nSpeaker B: It has to be very basic, not too many buttons.\nSpeaker B: Light switches on if you use a button.\nSpeaker B: Textivity still has to be a possibility.\nSpeaker B: And it has to be easy to learn.\nSpeaker B: The other things I make minutes of.\nSpeaker B: And the function is our full-room channel to choose channels on, off a mute button and the textivity button that are the functions.\nSpeaker B: Is that right?\nSpeaker B: Yeah.\nSpeaker B: That's okay.\nSpeaker B: So I just want to give you Mike again the first presentation of your...\nSpeaker E: Shall I start?\nNone: No?\nSpeaker E: Okay, well, I received an email.\nSpeaker E: Okay.\nSpeaker E: I searched the web.\nSpeaker E: And I searched on this document.\nSpeaker E: We sent an investigation of the remote control market.\nSpeaker E: It has been done in Italy and in another country in Europe.\nSpeaker E: I forget it.\nSpeaker E: But they found out the most important aspect for remote controls happens to be a fancy look and feel instead of the current functional look and feel.\nSpeaker E: So it's very important for us to create something new.\nSpeaker E: So what Michael just said, it must be some very different from ordinary remote controls.\nSpeaker E: Fancy stands for an original look and feel of the case and interface.\nSpeaker E: This aspect is the most important one.\nSpeaker E: It came out of the research.\nSpeaker E: It is twice important as the following.\nSpeaker E: The second most important aspect is that a remote control should be technological innovative.\nSpeaker E: That stands for new technical features.\nSpeaker E: And then this is a point of discussion because we just decided that we don't make use of LCD or speech recognition.\nSpeaker E: But this is the second important aspect.\nSpeaker E: And I think we must use some of the new technology to be innovative.\nSpeaker B: So we already have the flashing light on that.\nSpeaker B: Maybe some new.\nSpeaker E: We have to discuss about it.\nSpeaker E: We'll get back.\nSpeaker E: Time of speech that do not exist in current remote controls.\nSpeaker E: And that's very hard I think.\nSpeaker E: The third one is the aspect that remote controls should be easy to use.\nSpeaker E: But that was an overall point.\nSpeaker E: We already discussed that.\nSpeaker E: I've got one picture.\nSpeaker E: Our target group we thought about was Yo and Trendy.\nSpeaker E: So I took that part of the web page.\nSpeaker E: And it says fashion watchers of a very friends in Italy.\nSpeaker E: I have detected the following trends.\nSpeaker E: This trends trend I think it's not that meaningful for us.\nSpeaker E: Maybe but it's about clothes and shoes.\nSpeaker E: But the next aspect also in contrast to last year the field of material is expected to be spongy.\nSpeaker E: Spongy.\nSpeaker E: What does spongy say?\nSpeaker E: Spongy.\nSpeaker C: So rubber kind of.\nSpeaker D: Soft materials.\nSpeaker B: Maybe that's definitely a good idea because you've dropped the remote control very often on the ground.\nSpeaker E: So it has to be flexible.\nSpeaker E: It's something that's standard.\nSpeaker E: I didn't know what it means.\nSpeaker E: So spongy means soft.\nSpeaker E: It's also a stress ball.\nSpeaker B: That's a good idea.\nSpeaker B: It's the bit like the borders of the remote control.\nSpeaker C: I have some material information but I will give you it later in my presentation.\nSpeaker E: Okay, what do I think?\nSpeaker E: Because a fancy look is the most important thing for remote control.\nSpeaker E: I think about changeable fronts.\nSpeaker E: Maybe I see through one in a fruit front because it's hot.\nSpeaker E: And some basic color fronts.\nSpeaker E: So we can make five different fronts to start with or something.\nSpeaker E: Maybe an extraordinary shape like a spongy.\nSpeaker E: Or just another shape than a normal remote control has just something round in it or maybe not.\nSpeaker E: Yeah, I don't know.\nSpeaker E: We have to discuss about that.\nSpeaker E: Yeah, the main point still is the technological innovative.\nSpeaker E: How do we do that?\nSpeaker E: Maybe speech?\nSpeaker E: We must have some kind of gadget.\nSpeaker D: I'll get back on that.\nSpeaker B: So it's very difficult to do it because it's only 12 and a half years you have to spend.\nSpeaker D: Yeah, no problem.\nSpeaker D: I got also an email from the technology department.\nSpeaker D: Maybe watch first the next research about it and even more possibilities now with speech.\nSpeaker D: So they recommend it using it.\nSpeaker D: Okay.\nSpeaker E: Let's check what the...\nSpeaker E: First watch, both presentation first then.\nSpeaker C: No, actually we don't have an idea how much is going to cost.\nSpeaker C: Maybe it's cheap and easy to implement.\nSpeaker D: They don't say how much it will cost.\nSpeaker D: But if we implement...\nSpeaker D: It's mass production, speech, classification.\nSpeaker D: I think it would be better to implement LCD as well.\nSpeaker D: Since you have to configure speech.\nSpeaker B: That's definitely more expensive than...\nSpeaker D: That's something I don't know.\nSpeaker E: But we're going to make many of those so we can start a mass production and then...\nSpeaker E: Yeah, but a telephone has...\nSpeaker E: Okay, yeah.\nSpeaker B: But a telephone also has LCD and...\nSpeaker B: Yeah, so we've got it.\nSpeaker B: And it's about 200 euros.\nSpeaker A: We have to decide on that.\nSpeaker E: Okay.\nSpeaker E: That was this?\nSpeaker D: No, no.\nSpeaker C: I got an email and it says the chip can be simple, regular or advanced.\nSpeaker C: And they say a display requires an advanced chip and this is more expensive than all the other chips.\nSpeaker C: So it smells the most expensive.\nSpeaker D: It smells the most expensive.\nSpeaker D: It smells the most expensive.\nSpeaker C: Yeah, it says an email.\nSpeaker C: The display requires an advanced chip.\nSpeaker C: Okay.\nSpeaker D: And speech are good, nation?\nSpeaker D: Yeah, probably too.\nSpeaker C: I haven't got anything about speech recognition but...\nSpeaker C: Oh, okay.\nSpeaker C: Well, that's the most expensive chip we need if we're doing a display.\nSpeaker D: Well, we can...\nSpeaker D: All right.\nSpeaker D: To make a sort of a design.\nSpeaker D: So I did some searching on the internet.\nSpeaker D: And what's the old remote controls and use remote controls?\nSpeaker D: I think we should... this is one of the modern remote controls of the moment.\nSpeaker D: I think we should go more to the iPod and the MPD players.\nSpeaker B: Yeah, just mobile phones.\nSpeaker D: More than more than but still a basic.\nSpeaker D: Yes.\nSpeaker D: You probably have to make it a bit bigger and a bit smaller because when you're in remote control you can see it here.\nSpeaker D: You have to reach both sides and you just have one few buttons.\nSpeaker D: So that's the main difference.\nSpeaker D: But it looks, I definitely think we should go like this and then changeable funds.\nSpeaker D: So that's the most important part I think.\nSpeaker D: But on the home basis, something like that, something simple.\nSpeaker D: Well, and then I just ordered the buttons a bit.\nSpeaker D: Basic buttons on, off, mute and maybe two others.\nSpeaker D: I don't know.\nSpeaker B: Text button.\nSpeaker D: Yeah, text button maybe there and there.\nSpeaker D: And then the color buttons if we want it on.\nSpeaker D: I don't find it very usable but I don't think it fits in the modern stream as well.\nSpeaker D: And then where you saw the display in the iPod, you can put the basic buttons.\nSpeaker D: One, two, three, four and above ten.\nSpeaker D: That's on the display.\nSpeaker D: No, no, there was no display there but it's on the place of the display.\nSpeaker D: And I think we should light this up much more than that part.\nSpeaker D: So focus is on these two parts.\nSpeaker D: So you don't see all the buttons you have.\nSpeaker B: But this one is very difficult to use because the volume and channel is on the bottom of it so you can use it down for it.\nSpeaker D: This is how it is now.\nSpeaker D: Here, well, volume, I think it is on the bottom too.\nSpeaker B: It's not the best.\nSpeaker D: Well, here I have also side scrolls.\nSpeaker D: I don't know if you can use that.\nSpeaker D: Yeah, I've got something on that.\nSpeaker B: For volume or channel?\nSpeaker D: For volume.\nSpeaker D: Well, then we can even simplify it more by just putting the volume on the side and just channel buttons here.\nSpeaker C: Yeah, I think the channel buttons are often used and you can use them now with your thumb because the thing is not easy to use.\nSpeaker C: Well, control.\nSpeaker D: Basically, that's here.\nSpeaker C: Yeah, okay.\nSpeaker C: Maybe we should put that on top and buttons we don't use on any buttons.\nSpeaker B: Yeah, just the older buttons like textivity.\nSpeaker B: You put that on the button.\nSpeaker C: Because you can't hold it.\nSpeaker B: You can't hold it alone.\nSpeaker B: Yeah, control and push the buttons.\nSpeaker B: Except from the on off button.\nSpeaker D: Oh, okay.\nSpeaker D: Well, yeah.\nSpeaker C: Well, I think you'll use the switch channels buttons more often than the normal channel buttons like one, two, three.\nSpeaker D: So maybe we can put that on the bottom.\nSpeaker D: Maybe I don't know.\nSpeaker C: Like zapping is just switching.\nSpeaker D: Maybe it's not easy if it's below.\nSpeaker D: It's harder to zap.\nSpeaker D: So I think it should be, it should be easy.\nSpeaker D: I think it's pretty standard.\nSpeaker D: These row buttons on the top.\nSpeaker D: Yeah, okay.\nSpeaker D: That's good.\nSpeaker D: But if you don't light them up, they don't, you don't seem very good.\nSpeaker D: I think it's moderate to light this area up and to light this area up.\nSpeaker D: So the focus gets on this part and not on there.\nSpeaker C: Yeah, okay.\nSpeaker D: But the position of course can be different.\nSpeaker D: We have to look at what's easy to use and how it's easier to use.\nSpeaker D: So we can switch these to, I don't know if it will look good if you put those on the bottom.\nSpeaker C: No, I think the top buttons are okay.\nSpeaker C: Those should be on top.\nSpeaker C: Maybe we can switch those to.\nSpeaker D: Do us to, yeah.\nSpeaker D: And yeah, you have to make sure it's easy to, yes, that has to be big enough so you can heal.\nSpeaker D: Right.\nNone: Okay.\nSpeaker D: Well, that's my finding.\nSpeaker D: So my personal opinion is that we should go more modern look and be player.\nSpeaker D: And well, if you want to put in a speak recognition or something, I don't think we should put it on top then.\nSpeaker D: I think that if you're going to put in more technology, that you need to be able to switch it open.\nSpeaker D: Do you use?\nSpeaker D: Yeah, that's cool.\nSpeaker D: So if you put in a speech recognition, you need so many more buttons which won't look good on the front side, I think.\nSpeaker C: Okay.\nNone: No.\nSpeaker D: So that's something we have to decide on.\nSpeaker C: Yeah, we have to keep it simple.\nSpeaker B: Yeah.\nSpeaker B: We have to decide this lecture.\nSpeaker B: What are we going to do?\nSpeaker B: 50 minutes.\nSpeaker A: Yeah.\nSpeaker B: I was just going to look.\nNone: Okay.\nSpeaker C: The component design.\nSpeaker C: I looked at some similar devices and my own common knowledge.\nSpeaker C: So this was on the website.\nSpeaker C: If you aim at a young public, you should use materials that are soft with primary colors like green, blue, red.\nSpeaker C: So flashy kind of colors.\nSpeaker C: Shapes should be curved.\nSpeaker C: So round shapes, not nothing square like.\nSpeaker B: Okay.\nSpeaker B: So, okay.\nSpeaker D: Your iPod is trendy.\nSpeaker D: And it is.\nSpeaker D: No curved square.\nSpeaker C: It's right.\nSpeaker C: Yeah, but it has round corners I think.\nSpeaker C: Okay.\nSpeaker C: So we shouldn't have two square corners.\nSpeaker D: Yeah, okay.\nSpeaker D: Not the old box look.\nSpeaker C: And sports and gaming divine style characteristics.\nSpeaker C: I don't know exactly what that means, but it should be, well, yeah.\nSpeaker C: Popular kind of.\nSpeaker B: Yeah, we have to put our real reaction logo as well on the, on the removed controls.\nSpeaker B: So the colors also, so we have made make it in black, black, yellow.\nSpeaker C: Yeah, black yellow.\nSpeaker B: Maybe the sides yellow and the top black.\nSpeaker C: Not that weird because we definitely want to make it kind of flashy to attract a young\nSpeaker D: public. But I don't think the colors black and yellow go well together.\nSpeaker B: Maybe you can put the yellow on the side and black on the front.\nSpeaker C: That's just a matter of taste.\nSpeaker C: But yeah, okay.\nSpeaker C: We have to use kind of flashy colors.\nSpeaker D: Can't we use different fronts with all with the logo on it?\nSpeaker D: Yeah.\nSpeaker C: Can we do that?\nSpeaker B: Like France in red and yellow and blue and still transparent.\nSpeaker D: Okay, but with all with logo on it.\nSpeaker C: Okay.\nSpeaker C: Well, this is a remote control, a very old one.\nSpeaker C: Then the components, the case is just the, here's black.\nSpeaker B: Maybe we have to make it from software material.\nSpeaker C: Sure.\nSpeaker C: Yeah, I don't know.\nSpeaker C: Maybe.\nSpeaker C: But anyways, it should be transparent, we decided that.\nSpeaker D: No one of the options.\nSpeaker D: You can, just like a mobile phone, you can make a perfect font on it.\nSpeaker D: So you can just replace them, I think.\nSpeaker D: Yeah, we could do that.\nSpeaker D: But I do just release one.\nSpeaker B: Give five with them just in the box.\nSpeaker D: Yeah, or just a different one.\nSpeaker C: You could make them like blue and transparent.\nSpeaker C: So you can still look through it.\nSpeaker C: The buttons, normal rubber, I think.\nSpeaker C: Like normal ordinary buttons.\nSpeaker C: Yeah, I don't know.\nSpeaker B: It could be like a Nokia, like the new.\nSpeaker D: Yeah, just a hard part.\nSpeaker D: It's better.\nSpeaker D: It's a rubber really as an old look.\nSpeaker E: Okay.\nSpeaker E: And then the new modern remote controls.\nSpeaker E: The buttons are part of the style, I think.\nSpeaker E: It's part of the remote control itself.\nSpeaker E: It's one out of one shape.\nSpeaker E: It doesn't, it's a button.\nSpeaker E: Yeah, it's all on one level.\nSpeaker C: It doesn't come out of the background.\nSpeaker E: It is in the remote control.\nSpeaker C: Yeah, I know what you mean.\nSpeaker C: So we have to keep it on one level.\nSpeaker C: Yeah, like mobile phones.\nSpeaker C: The top is, yeah.\nSpeaker D: Not the iPod.\nSpeaker C: Yeah, okay.\nSpeaker C: So that should be hard plastic than the buttons, I think.\nSpeaker D: Or maybe it's the kind of material that is.\nSpeaker B: Yeah.\nSpeaker B: Maybe you can make it around platform soft material.\nSpeaker B: Just only the basic remote control from normal plastic and the rounds of it from soft.\nSpeaker C: Okay.\nSpeaker C: Then the LED, the normal infrared LED, I think.\nSpeaker C: It's sufficient.\nSpeaker C: And backlight LEDs.\nSpeaker C: So yeah.\nSpeaker C: But I think we have to make the case transparent.\nSpeaker C: Otherwise the backlight won't work.\nSpeaker C: So if you put, you can just not put my run to buttons.\nSpeaker C: Yeah, that's right.\nSpeaker D: Okay.\nSpeaker D: Or run to the whole.\nSpeaker C: Yeah, but we can still make it transparent.\nSpeaker D: You can have transparent or just that it comes out a bit.\nSpeaker C: Okay.\nSpeaker C: Good.\nSpeaker C: And in green color, the backlight or?\nSpeaker D: Difference, I think also.\nSpeaker B: Blue.\nSpeaker D: Blue, red, whatever you want.\nSpeaker D: I think it depends on the color.\nSpeaker B: Maybe you can choose it when you buy it.\nSpeaker B: That's true.\nSpeaker C: Now I think there are multiple color LEDs.\nSpeaker C: So is it okay?\nSpeaker C: Yeah, but can you change the color?\nSpeaker C: I know.\nSpeaker C: I know.\nSpeaker D: I can't control you.\nSpeaker C: Maybe it's more important.\nSpeaker C: We can put some different ones.\nSpeaker C: I think it doesn't matter.\nSpeaker D: I have to remind, I have a blinking light on my phone and I can change the color of\nSpeaker B: it. Okay, cool.\nSpeaker B: Just make it some different color, blue, red and green.\nSpeaker C: Yeah, I don't know.\nSpeaker C: Maybe it's too expensive.\nSpeaker C: Well, I don't think so.\nSpeaker D: Put in any fancy technology yet.\nSpeaker C: Then some more technical things.\nSpeaker C: I don't know what it is, but it should be there, I think.\nSpeaker C: This is the normal circuit board, like chip board in a lot of things.\nSpeaker B: You have to hurry up a bit.\nSpeaker C: So we just need this and this transistors and resonators, this is all these kind of things.\nSpeaker C: They basically said that's almost the same on any remote control.\nSpeaker C: So I guess we just need that.\nSpeaker C: I don't know what they do or...\nSpeaker E: Yeah, you can change.\nSpeaker C: No, but they just said we needed the battery contacts, like normal batteries.\nSpeaker C: You can put it.\nSpeaker B: And recharge maybe.\nSpeaker B: Yeah, we just make sure.\nSpeaker B: We still want to have a recharge.\nSpeaker C: Yes, yes.\nSpeaker C: Yeah, but we don't want to have a...\nSpeaker C: How do you call it?\nSpeaker C: Aku.\nSpeaker C: Recharge.\nSpeaker B: Recharge.\nSpeaker C: Just batteries.\nSpeaker C: Rechargeable batteries.\nSpeaker D: Yeah, batteries, yes.\nSpeaker C: Rechargeable batteries.\nSpeaker C: Not a separate.\nSpeaker C: No, just rechargeable batteries.\nSpeaker C: And a chip, that's this one.\nSpeaker C: Then I received some possibilities.\nSpeaker C: For the energy source, we can use batteries or a kinetic, like with the pulse watch.\nSpeaker C: So it operates on your wrist.\nSpeaker C: Okay.\nSpeaker B: So if you have a normally remote control on the table or...\nSpeaker C: Yeah, I don't think it will work.\nSpeaker C: And we can also use solar cells, but you must use it indoors.\nSpeaker B: Just batteries, that's cheaper.\nSpeaker C: Yeah, and we can use the home station kind of thing.\nSpeaker C: Cases, flat, so uncurfed.\nSpeaker C: 2D curved is like front to the back.\nSpeaker C: And 3D curved is also in depth.\nSpeaker C: Okay.\nSpeaker C: So that's possible.\nSpeaker C: But with 3D curved remote controls, we must use rubber buttons.\nSpeaker C: So we can use the flat.\nSpeaker C: So we need 2D.\nSpeaker C: Yeah, I think.\nSpeaker C: Definitely.\nSpeaker C: These kind of materials can be used.\nSpeaker B: But it doesn't really matter.\nSpeaker B: We just make it plastic.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I think so too.\nSpeaker B: Scroll wheels, that's cool.\nSpeaker B: This is for the full view.\nSpeaker C: Yeah, scroll wheels.\nSpeaker C: Yeah.\nSpeaker C: Yeah, that's good.\nSpeaker C: We can use multiple scroll wheels too if we want to.\nSpeaker C: But I think just the volume is enough.\nSpeaker D: For channels, it's not handy.\nSpeaker D: Okay.\nSpeaker C: You scroll it fast.\nSpeaker C: And the LCD.\nSpeaker C: So we need the most expensive chip if we use an LCD.\nSpeaker B: I don't think that's an opportunity.\nSpeaker B: Just skip it.\nSpeaker B: Because we don't have time for that to.\nSpeaker B: Okay, done.\nSpeaker C: We must use the second most expensive chip.\nSpeaker C: So the regular chip.\nSpeaker C: Because we use scroll wheels.\nSpeaker C: Yeah.\nSpeaker C: And yeah, that was it.\nSpeaker C: Okay.\nSpeaker C: I guess.\nSpeaker C: Are we using a rubber case or?\nSpeaker C: Oh, I just decided.\nSpeaker B: Maybe we have to skip that one as well.\nSpeaker D: I don't think it will be a case.\nSpeaker D: It should be soft.\nSpeaker E: Something about the trend too.\nSpeaker E: The trend is spongy.\nSpeaker E: And fruit or fresh.\nSpeaker E: Fresh.\nSpeaker E: Okay.\nSpeaker E: Now we have nothing about those two.\nSpeaker D: Yeah, fruit and fresh can be just the covers.\nSpeaker D: Yeah.\nSpeaker D: So you can just off-line the spongy.\nSpeaker D: Yeah, I don't know.\nSpeaker D: I can't imagine.\nSpeaker D: No.\nSpeaker E: I don't like it.\nSpeaker E: No.\nSpeaker D: I just got to imagine it.\nSpeaker C: So just hard plastic?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker D: Titanium.\nSpeaker D: It's mentioned there.\nSpeaker C: Titanium, I think it's too expensive.\nSpeaker B: Yeah, maybe the form has to be a bit different.\nSpeaker B: Not the square form just a bit.\nSpeaker C: Yeah, you can make it curved or round.\nSpeaker C: And just in 2D, nothing adapts.\nSpeaker B: Yeah, that's right.\nSpeaker B: So if we have to decide which one we're going to choose from these, what exactly?\nSpeaker B: Because we have to know it.\nSpeaker B: So the energy is to recharge and you already know that?\nSpeaker B: Yeah, that's right.\nSpeaker B: The chip on prints is a normal one.\nSpeaker C: Yeah.\nSpeaker B: Okay, the case is just a plastic one.\nSpeaker B: Yeah, the chip is the regular one.\nSpeaker C: You have to simple one, regular and it's fine.\nSpeaker C: So it should be regular in the second.\nNone: Okay.\nSpeaker C: I think I'll just check it.\nSpeaker B: And we need a plastic case with a scroll wheel.\nSpeaker B: No matter the least.\nSpeaker C: Yes.\nSpeaker B: That's pretty much it.\nNone: Yeah.\nSpeaker B: And it actually lies.\nSpeaker B: So I'm not sure.\nSpeaker B: I don't know if we're expected to roll on this one at this moment.\nSpeaker D: I don't know.\nSpeaker C: Or should we do it in the next meeting?\nSpeaker D: ID and UID work together on prototype drawing on smartboard.\nSpeaker C: So we should do it here.\nSpeaker D: So we're staying here?\nSpeaker C: Or should we do it in the next meeting?\nSpeaker B: I think that's the next meeting.\nSpeaker B: Okay.\nSpeaker B: But you definitely get a specific instruction.\nSpeaker D: Okay, so I don't work through the minutes alone again.\nSpeaker B: But think about something that's more rounded and more...\nSpeaker B: Yeah, I didn't know.\nSpeaker D: But the iPod and etc.\nSpeaker D: And the player has mobile phones.\nSpeaker B: I bet just on the top one.\nSpeaker D: Yes, the bits currently.\nSpeaker D: Okay.\nSpeaker D: I'll see if I can see any of these.\nSpeaker B: Maybe the wheel can be like this.\nSpeaker B: If you draw it like this.\nSpeaker B: Get it.\nSpeaker B: What the fuck is it?\nNone: Okay.\nSpeaker B: Doesn't work.\nSpeaker B: You see what I mean?\nSpeaker B: If I draw here...\nSpeaker B: What?\nSpeaker B: It draws about four centimeters over the middle.\nSpeaker B: Okay.\nSpeaker B: Maybe you can make it like this.\nSpeaker B: This is all the wheel for a volume.\nSpeaker B: So you can do...\nSpeaker B: Like a very big scroll wheel.\nSpeaker B: Yeah, but just not on the top, but on the side of it.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: So we have this at the moment.\nSpeaker C: I think you'll get a lot of volume changing when it's not one.\nSpeaker B: Okay.\nSpeaker B: So we have this?\nSpeaker B: Is that okay?\nSpeaker A: Yeah.\nSpeaker A: That's good as well.\nSpeaker C: Yeah, but maybe we can make a plastic so that if you drop it, it won't change the volume.\nSpeaker D: Yeah, I don't know if you use your thing like this.\nSpeaker D: That's quite about crazy.\nSpeaker B: Like this.\nSpeaker B: And what's the channel choose?\nSpeaker B: Where do we put that?\nSpeaker C: I think in middle.\nSpeaker D: It's in the bottom or?\nSpeaker D: Yeah.\nSpeaker D: What is the middle part?\nSpeaker B: That's the numbers.\nSpeaker C: I think the numbers should be in the bottom and the switch channel in the middle.\nSpeaker D: Yeah, I agree as well.\nSpeaker B: It doesn't make a difference if you put the switch channels on side of each other on top of each other because you already have to volume here.\nSpeaker B: So you can also put it here one bit and the other one there next to each other.\nSpeaker B: Yeah, that's right.\nSpeaker B: Back in front of you.\nSpeaker B: So you can also put it all on the top and this.\nSpeaker B: You keep this empty because you have to hold it as well.\nSpeaker C: Or you could do the switch channel up button above the numbers and switch channel down button.\nSpeaker C: That's not the one too.\nSpeaker B: That's very quick.\nSpeaker D: I think zipping style is priority and then you use those.\nSpeaker B: Is this opportunity?\nSpeaker B: Yeah, of course.\nSpeaker C: Yeah.\nSpeaker C: But I think we should put them on top of each other.\nSpeaker C: So why?\nSpeaker C: Because then it's easy to know if I push the.\nSpeaker B: But still the next one.\nSpeaker C: Yeah, but the top button is like you switch channel up and down button is.\nSpeaker B: Yeah, but if you put them after writes exactly the same, it doesn't make a difference.\nSpeaker C: I think it's.\nSpeaker C: But I think left to right is more often associated with volume and top down is more with channel changing.\nSpeaker C: It's like almost remote controls.\nSpeaker C: So if we use that, they will probably have a long learning time.\nSpeaker B: I don't know.\nSpeaker B: You already have volume on the side.\nSpeaker B: So you can't make it.\nSpeaker B: You can't make it.\nSpeaker B: You can't make it.\nSpeaker D: You can't make it.\nSpeaker D: I think it's simple.\nSpeaker B: It's just.\nSpeaker B: Okay.\nSpeaker D: So what else do we have to discuss about?\nSpeaker B: I think we need to work three minutes again.\nSpeaker E: We have to get it.\nSpeaker E: It looks really new because we still hold on to the ordinary square.\nSpeaker E: Yeah.\nSpeaker E: We're not control because what I see, the only difference in fact is that we use a scroll wheel on the outside.\nSpeaker C: And the left.\nSpeaker C: Yeah, but it should be around in shape.\nSpeaker D: The current controllers are all black and plastic.\nSpeaker D: You have to look at that image of the iPod.\nSpeaker D: Yes.\nSpeaker D: More that kind of style.\nSpeaker D: And that you don't the old gray black.\nSpeaker E: Some kind of blingling.\nSpeaker D: Well, you can put a very, if we do it like that, we have below if we do a lot of room to put a nice logo.\nSpeaker B: Yeah.\nSpeaker B: The logo has to be on there.\nSpeaker E: And how many fonts do we put on the market then?\nSpeaker E: Five or something?\nSpeaker D: More or is it a customer?\nSpeaker D: Maybe you can buy separate ones.\nSpeaker E: And by the product you buy, you get one.\nSpeaker E: And you can choose one.\nSpeaker B: If you buy the project, that's your choice I think.\nSpeaker B: So you can put the bottom of the remote control in recharge.\nSpeaker B: Is that a good opportunity?\nSpeaker B: Yeah.\nSpeaker B: So I put it like that.\nSpeaker E: Okay.\nSpeaker E: Yeah.\nSpeaker E: You let it slide in the docking station.\nSpeaker C: Do we have to design that as well?\nSpeaker D: The docking station?\nSpeaker D: Yeah, I think so.\nSpeaker D: Yeah, we can.\nSpeaker D: Yeah, that can be for example.\nSpeaker D: Just a recharge.\nSpeaker D: Just a second.\nSpeaker D: Yeah, around something.\nSpeaker D: Like, yeah, we had one example.\nSpeaker C: But we have to make contacts on the remote control and the recharging as well.\nSpeaker E: Yeah.\nSpeaker E: But that's a round one.\nSpeaker E: Maybe we can choose them.\nSpeaker D: Here you just see one.\nSpeaker D: That's very round.\nSpeaker D: Oh, yeah.\nSpeaker D: Okay.\nSpeaker D: So I think that can be all kind of shapes.\nSpeaker B: But maybe you can just round up the corners a bit.\nSpeaker C: After remote control.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker D: So you don't want to, like the eye pops and fills, but more rounded.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I think it will just look like more like this one since this is also rounded.\nSpeaker D: No, just the corners.\nSpeaker D: Right here.\nSpeaker D: Okay.\nSpeaker D: Those are already a bit cornered.\nSpeaker C: Okay.\nSpeaker C: Yeah, but we can do all kinds of, as long as it is in 2D, we can use all kind of round shapes.\nSpeaker C: Yeah, okay.\nSpeaker D: But then we have to think of something totally new.\nSpeaker C: Yeah.\nSpeaker C: Yeah, if we want to make it kind of new.\nSpeaker D: I've had a lot of pictures of old ones and all curves have already been done.\nSpeaker B: It's a bit annoying, is it?\nSpeaker E: Yeah.\nSpeaker E: What do we do wrong?\nSpeaker E: Just more like this and not square.\nSpeaker E: Yeah, okay.\nSpeaker C: Yeah.\nSpeaker C: I could do a lot of, a lot more curving.\nSpeaker D: I would do it.\nSpeaker D: Yeah.\nSpeaker D: I know we can do a lot more, but yeah.\nSpeaker C: Like in this kind of shape.\nSpeaker E: Oh, sorry.\nSpeaker E: I know you.\nSpeaker C: Okay.\nSpeaker C: I don't know.\nSpeaker C: I don't know if it's handy.\nSpeaker D: I think it will only look more like the old remote controls.\nSpeaker C: This?\nSpeaker C: Yeah.\nSpeaker C: The old ones look like just a square thing.\nSpeaker D: Yeah, okay.\nSpeaker D: But I had a lot of pictures.\nSpeaker D: Oh, I can show you here.\nSpeaker D: What the old ones look like.\nNone: Mm.\nSpeaker D: Curves, curves.\nSpeaker D: Yeah.\nSpeaker D: Even more than there as well.\nSpeaker C: Yeah, okay.\nSpeaker D: This wasn't a very small one.\nSpeaker D: Yeah.\nSpeaker D: If you go very simple.\nSpeaker B: So we have to make a decision what kind of form it's going to have.\nSpeaker C: Yeah.\nSpeaker D: I don't know.\nSpeaker D: I think if, in my opinion, if we just take the iPod and the same look.\nSpeaker D: So light or just whatever color but the same light color.\nSpeaker D: And just together with the back lights, we'll look very new.\nSpeaker D: No rubber buttons or something.\nSpeaker D: Just together with the black lights, you'll get a totally new look.\nSpeaker D: More like the MP3 player.\nSpeaker D: Yeah.\nSpeaker B: And you have to scroll button and side.\nSpeaker B: Yes.\nSpeaker B: Why do we have to round it on the bottom then?\nSpeaker B: Skip that one as well.\nSpeaker D: Doesn't have to be?\nSpeaker B: Okay.\nSpeaker C: Yeah, it's cool.\nSpeaker C: Yeah, that's a trend if we want to make it.\nSpeaker C: But yeah, I'm not a trend watcher.\nSpeaker C: You are.\nSpeaker E: So the trend is?\nSpeaker B: The point G.\nSpeaker B: Spongy and fruity.\nSpeaker E: But yeah.\nSpeaker E: Spongebob.\nSpeaker E: It's not a lot of friends I find.\nSpeaker B: Okay, so we have still one minute left.\nSpeaker B: I think it's okay if you just give it a bit square.\nNone: Okay.\nSpeaker C: I don't know something about ergonomic kind of fits in hand.\nSpeaker B: But I think it's still for older people.\nSpeaker B: You still have older people.\nSpeaker B: It's only annoying if it's like that form like that.\nSpeaker B: Like whatever.\nSpeaker B: There is one.\nSpeaker B: Yeah, but we're aiming at the young.\nSpeaker E: Which is one overall important aspect is that we must make it fancy and it looks original.\nSpeaker E: And I hope we can make it look not like the iPod itself.\nSpeaker E: It must have an idea.\nSpeaker D: But if you look at the way remote controls are now.\nSpeaker D: Yeah, I don't know.\nSpeaker B: But it is already fancy because it's a light on the bottom of it.\nSpeaker B: Yeah, okay.\nSpeaker B: So that's already a very big change.\nSpeaker B: Maybe make the, what's it called?\nSpeaker B: Scroll wheel, make it in yellow or something.\nSpeaker B: Just like the corners of real reaction.\nSpeaker D: Yeah, yeah.\nSpeaker E: Well, now we could do that.\nSpeaker E: Could.\nSpeaker E: Yeah, but if you change the front, the scroll wheel will still be yellow.\nSpeaker D: Yeah, I think that's right.\nSpeaker D: The scroll wheel won't be very big since if you put it somewhere and chances that it will scroll are too big.\nSpeaker D: So it will just be a small, small scroll wheel.\nSpeaker D: So it won't stick out much.\nSpeaker B: Yeah, maybe the ones who are going to draw there, maybe have to ask to her if it can work better than this because it doesn't work properly.\nSpeaker B: No, maybe it's all sure.\nSpeaker D: Maybe you can just open images there and I'll paint and paint.\nSpeaker D: Okay.\nSpeaker E: Yeah, it'll do a very good job.\nSpeaker E: If you set it down, he will draw here.\nSpeaker E: It doesn't work.\nSpeaker B: Okay, so just finish it.\nSpeaker B: So we make it a bit like that one probably.\nSpeaker B: Yeah, is that okay?\nSpeaker B: Yeah, I'll see.\nSpeaker B: Yeah, I agree more.\nSpeaker B: Okay, only the color and the flashy light.\nSpeaker B: Yes.\nSpeaker B: We just skipped the photo or the speech recognition.\nSpeaker D: Yeah, do we skip that?\nSpeaker D: Yeah, I think.\nSpeaker B: Or keep that?\nSpeaker B: Okay, but you definitely need a.\nSpeaker C: Yeah, I don't know.\nSpeaker C: Yeah, I think so.\nSpeaker C: We have to build in a microphone and.\nSpeaker C: Well, that's very easy.\nSpeaker D: We already have the beeping of the.\nSpeaker C: Yeah, I don't know anything about that.\nSpeaker C: I didn't receive any information on speech recognition.\nSpeaker E: Yeah, that's hard.\nSpeaker E: The information about that.\nSpeaker E: The main points I just said, do you have to be original in technological, innovative?\nSpeaker E: Go, yeah.\nSpeaker B: Just put a speech recognition in it as well, okay?\nSpeaker D: So we clap it open then?\nSpeaker D: Yeah.\nSpeaker D: So we can put all the.\nSpeaker B: I think she'll have that in there.\nSpeaker B: And we need a.\nSpeaker B: Probably we need a.\nSpeaker B: At phones chip then.\nSpeaker C: Yeah, we probably do.\nSpeaker C: But it doesn't say anything about it, does it?\nSpeaker C: No.\nSpeaker E: We don't have any information about the cost.\nSpeaker E: No.\nSpeaker E: We started with information about.\nSpeaker C: I just received.\nSpeaker C: I have some information about the cost, but just about the chip.\nSpeaker E: And how much is the chip?\nSpeaker C: I don't know how much.\nSpeaker C: You are the business.\nSpeaker C: Just something new.\nSpeaker C: Expense information.\nSpeaker D: The integrated program, simple, simple, speaker unit.\nSpeaker D: This is a very small electronic unit.\nSpeaker D: We'll give a standard answer after it recognizes the question.\nSpeaker E: How does it work?\nSpeaker D: It doesn't say.\nSpeaker D: You say record followed by your question sample and after a few seconds the answer is simple.\nSpeaker D: The question.\nSpeaker D: So it works like good morning remote control and then remote control says good morning.\nSpeaker B: Does that say anything?\nSpeaker B: No.\nSpeaker B: Does it say something bad?\nSpeaker B: Okay, we have to stop it now.\nSpeaker D: Well, that's integrated into chips.\nSpeaker D: So if you use speech recognition.\nSpeaker D: That's a separate chip.\nSpeaker D: I don't know.\nSpeaker D: But if you use speech recognition that will be in it as well.\nSpeaker C: Okay.\nSpeaker C: Yeah, I don't know anything about this.\nSpeaker B: We just decide not to put it in because it's too difficult.\nSpeaker D: I don't know.\nSpeaker D: It would be a good feature.\nSpeaker B: Yeah.\nSpeaker B: Okay, we just put it in.\nSpeaker D: No worries about the cost, etc.\nSpeaker E: Okay, just stop it.\nNone: Okay, fine.\nNone: 30 minutes?\nNone: Yeah.\nSpeaker E: Okay.\nSpeaker E: Ciao.\nSpeaker E: Okay.\nNone: Ciao.\nSpeaker C: Hello?\nSpeaker D: Okay.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I should make it on.\nSpeaker A: Okay.\nSpeaker C: Yeah, good thanks.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker B: Okay.\nSpeaker B: Coffee.\nSpeaker B: Yeah, okay.\nSpeaker B: Okay.\n"}, {"meeting_id": "ES2016c", "summary": "The group met to decide which features were desired by users and to decide how to design the shape and appearance of the remote control. Marketing and User Interface presented on what features were desired by remote control users, and Industrial Designer presented on the required internal components of a remote control. They decided not to pursue speech recognition and settled on designing a rounded one-handed remote control with minimal buttons.", "dialogue": "None: Reid?\nNone: some Ioddy Okay.\nSpeaker C: Alright.\nSpeaker C: Okay.\nNone: Alright.\nSpeaker C: Is everyone here?\nSpeaker C: Yep.\nSpeaker C: Okay.\nSpeaker C: This is our conceptual design meeting.\nSpeaker C: I'll just take a few minutes and go through the previous minutes.\nSpeaker C: Then each of you will have your presentation.\nSpeaker C: And then we will need to make a decision on the concept for the remote control.\nSpeaker C: And then we'll have 40 minutes for finishing up.\nSpeaker C: And then we'll go through the minutes first.\nSpeaker C: We just refreshed our goal of making the finest remote control available.\nSpeaker C: We decided that we know that we need to use company colors, company logo.\nSpeaker C: And our marketing expert gave us some information from interviews with 100 different remote users with some statistics that backed basically what we were thinking before.\nSpeaker C: People thought the remotes were ugly.\nSpeaker C: That remotes zap a lot.\nSpeaker C: They only use a finite amount of buttons.\nSpeaker C: And that they often lose, that's easy to lose remote.\nSpeaker C: Which we're all things we were thinking we would want to make it simple.\nSpeaker C: And some sort of locator, a button or tracking device.\nSpeaker C: And that it should look different than what's out there.\nSpeaker C: Kind of mixed response on the speech recognition.\nSpeaker C: The younger people said they wanted it, the older people did not.\nSpeaker C: I think we decided that the expense was not necessarily worth it.\nSpeaker C: And that it was probably a gimmick that would increasingly wear on the consumers' nerves.\nSpeaker C: Then the user interface designer explored some of the technical functions of the remote.\nSpeaker C: The simple versus the complex, the simple one being better for a user, the complex better for an engineer.\nSpeaker C: And some personal preferences that were found in that would be that it should be a user oriented remote, something simple.\nSpeaker C: And that we didn't want to go with a universal remote because increasing cost and increasing complexity.\nSpeaker C: We would just have a TV remote.\nSpeaker C: And that we should also focus on the appearance of the remote, have it be something that looks different.\nSpeaker C: And finally, our industrial designer gave us a rundown of how the remote will work from energy source, what we would use batteries because we don't want to have a cable.\nSpeaker C: How that would power the remote and the lamp.\nSpeaker C: If we were to have one, the user interface then would connect to a chip, which would work with the infrared controls to send the signal to the TV.\nSpeaker C: I believe then we came up with a couple ideas for what we think the design of the remote will be.\nSpeaker C: And we're going to be able to get into easily into someone's hand and with a just a few buttons, just the basics and with a scrolling function.\nSpeaker C: We're going to be able to get that marketing, we're watching trends.\nSpeaker B: Okay, so I was looking at trend watching.\nSpeaker B: Unfortunately, I wasn't given too much information.\nSpeaker B: I was given a brief executive summary and then an update on some recent fashion trends that we might like to look at.\nSpeaker B: And then I'll just tell you some personal preferences that I got from that.\nSpeaker B: Okay, the most important finding was that the fancy look and feel seems to be twice as important to the users as the current functional look and feel design, which I think we kind of already discussed before.\nSpeaker B: The second most important finding was that the remote should be technologically innovative.\nSpeaker B: And again, these are all things we've kind of already come up with on our own, but this just backs it up.\nSpeaker B: And thirdly, the remote should be easy to use.\nSpeaker B: As far as fashion update, we have learned that fruits and vegetables will be the most important theme for cloths, shoes and furniture.\nSpeaker B: And that might be a bit of a challenge to incorporate this into our remote, but we can try. And also, as opposed to last year, this year, the material is expected to be spongy and feel.\nSpeaker B: Okay, so from that, as we've already said, we need to focus on a fancy look and feel.\nSpeaker B: I think we've already discovered that it's kind of hard to go away from the traditional rectangular design, but I think that even if it's very subtle, we need to kind of trick our consumers so they at least get the idea that they're getting something that's new and modern and sleek and similar to the shape or the colors or all of that.\nSpeaker B: And so, we need to incorporate this fashion trend of fruits and vegetables.\nSpeaker B: So, I guess the two options are if we had a remote in the shape of the fruit or natural.\nSpeaker B: Or with exterior designs, but my question is, I mean, stereotypically speaking, you kind of picture males with their remote controls, and I'm not sure how they'd feel about having fruity logos on the outside, so maybe we could have something that's somewhat removable or, I don't know, different options for female male target and then the spongy feel.\nSpeaker B: I guess we could look at mobile phones and other technology that's out there and look at different types of material that might please our users.\nSpeaker B: What's spongy feeling? Remote controls.\nSpeaker B: What's that?\nSpeaker C: Possibly, sorry, just a button for a second, possibly a cover like the head for mobile phones.\nSpeaker C: One with a flag and one with a banana and one that's a spongy feel to it.\nSpeaker C: We could leave that to the cover department.\nSpeaker A: So they've got nothing to do all day.\nSpeaker A: I talked about the interface concept and how the customer relates to the consumer use the actual device.\nSpeaker A: So I've looked at some of the stuff I've sent, try and get some inspiration, but keep in mind that our own ideas that we had.\nSpeaker A: I was sent some information from the company saying that the technology department have devised a new speech recognition technology where you could program questions into such devices.\nSpeaker A: They gave an example of a coffee machine where you program a question, you program the answer, and the machine responds accordingly.\nSpeaker A: There's different ways of a user can use products like remote.\nSpeaker A: There's a graphical use where you look at pictures and on a screen.\nSpeaker A: A command line where you obviously type things in and you get a response.\nSpeaker A: That's just to point out the inconsistent interface in remote.\nSpeaker A: So they had a look at new products that are on the market, not necessarily remote controls, but ones that you recognize.\nSpeaker A: This is the voice recognition and remote control which can control multiple devices, the service-based channels with your voice.\nSpeaker A: It's all 80 speech samples, controls for devices, TV, cable satellite, VCR, DVD and audio.\nSpeaker A: You can record your own verbal labels that are connected to remote control functions.\nSpeaker A: The one on the left is very similar to what we drew upon the board in the previous meeting, where it has scroll down functions on the side, you can sort of just make those out.\nSpeaker A: And then on the right is obviously an iPod, which is possible, the simplest thing to use out there.\nSpeaker A: And all that is is just a nice big scroll menu that you sort of go through.\nSpeaker A: That is a possibility and nothing simple really.\nSpeaker A: Then there's things like this, which is a kid's remote, where the parents have the facility to control and program what children can watch before.\nSpeaker A: So, the remote control only allows them to access the channels that the parents want them to watch, and it means that children have a novelty of having their own remote control.\nSpeaker A: So, I don't know if there's a possibility of having one remote control.\nSpeaker A: We just had two components, maybe we could have more components, you know, different than those.\nSpeaker A: The point made at the end there is that you just have to sort of be clear on your devices as to what, you know, things you use sometimes.\nSpeaker A: An arrow pointing down, which may suggest volume down, could be completely used just as a V for volume, just little things like that, which would need to be made clear in the design.\nSpeaker A: I think you can't afford what you said, and usually friendly remote with minimum buttons, maybe we've sort of suggested this two part thing, where if it was to have a speech recognition thing, maybe control that on the program or that on the control bit, and then just have the simple sort of handheld thing that we sort of devised earlier as the actual remote.\nSpeaker A: It could be a graphical display, the actual remote control port, maybe could have like an iPod, we just sort of control through the menus, stuff like, it gets more and more complicated, and then the handheld bit should be ergonomic and designed.\nSpeaker A: And that is it.\nSpeaker A: Just, just a song show you.\nSpeaker A: And this is a limb got things like that, huge things, which is, that's one.\nSpeaker A: 19 year old, ground of my eye.\nSpeaker A: Industrial design.\nSpeaker D: Yeah, cream stuff like that.\nSpeaker C: I noticed the giant dog bone shape.\nSpeaker C: Also good for animals.\nSpeaker A: Well, let's see.\nSpeaker D: I'm going to bore you with a couple of descriptions of the interior.\nSpeaker D: Just to make it more obvious what we have to fit in there, and that we do have to fit the stuff in there.\nSpeaker D: I have more information on possible materials as well, what we can and cannot do.\nSpeaker D: Just wait for this to load up and I'll show you.\nSpeaker D: What we're talking about here.\nSpeaker D: The details of the components design, as you can see there, what we have is the board, the main board of the remote control, the underside.\nSpeaker D: It's a pretty cheap piece of technology.\nSpeaker D: You can see the chip, which is the device to recognize the signals, the input, and it passes it on to a row of the transistors and stuff like that on the right side that actually amplified the signal, which later on.\nSpeaker D: This is the transfer to an infrared lamp, which then, of course, shines infrared light onto the television terminal, recognize what signal it's getting and what you tell it.\nSpeaker D: So much for the workings of the remote control itself, its job is to wait for you to press a key, then to translate that key press into infrared light signals that are received by the television.\nSpeaker D: When you press the key, you complete a specific connection, the chip senses the connection, knows what button you press, that produces a Morse code line signal specific to that button.\nSpeaker D: You can see the input and the signal, which is the signal, is the signal to the signal, which translates the signal to the infrared light, the sensor in the TV, you can see the infrared light, and seeing the signal reacts appropriately.\nSpeaker D: The green greenish board is what we saw in the first slide just flipped over. You can see the circuit board itself, it's the cheapest way to make electronic connections basically on the market.\nSpeaker D: If you do, as you don't have cables, but you have the connections actually in these lines on the board. These are the actual keys that are being pressed to close the electric circuit that then sends the signal to the chip on the other side that will be behind here, which sends it over to the transistors and all that stuff, amplified the signal and all that has been sent to the infrared lamp up there.\nSpeaker D: Now, you can see this is the rubber button version of it, the way it works instead, the keys here, the rubber button has a little metal plate on the other side, which closes the circuit here, and thus gives on the signal. Now, this is the simple version.\nSpeaker D: We are talking, this is the simple and cheapest version at the same time. We're talking something more complicated, of course, it's going to be more expensive as well.\nSpeaker D: I'll tell me that. We also restricted the use of our outer shell or in the material that we could use for our outer shell. I've gotten some information that we could use for the case material plastic, rubber as well, rubber that is used in these anti-stress balls.\nSpeaker D: It's pretty squishy that would serve that purpose. We could also use wood or titanium.\nSpeaker C: What's the approximate cost per 100,000 for the titanium?\nSpeaker D: I don't have an information on that. However, our company obviously can provide us with the titanium. I was given an OK to use it. It certainly is an expensive material. I'm aware of that.\nSpeaker D: I was given an OK, but there are certain restrictions to certain materials. Now, let's first go through the list with materials. What we can use is plastic, rubber, wood, and titanium.\nSpeaker D: We can mix these as for the energy source. We're talking about that shortly in the meeting. What we could use is what I was offered, what we could use as a basic battery.\nSpeaker D: Interestingly enough, we could use solar cells or a device that was not further specified that provides kinetic energy, such as watches.\nSpeaker D: We could just move them, move the actual device and provide some energy. Obviously, I personally have to say that the dynamo is out of question.\nSpeaker D: You don't want to wind up your remote control before you can use it.\nSpeaker D: Solar cell is interesting. May fail, though, every time they're. Yeah, or you do use it, advice behind the couch for a weekend.\nSpeaker D: Works well in Arizona, but not so.\nSpeaker D: The kinetic energy thing might work, but same problem. You leave it lying around. First, I have to shake it before it starts to work.\nSpeaker D: So I'd say what we're stuck with really is the basic battery, which also makes a base station basically obsolete.\nSpeaker D: Our interface options are push buttons in which in the production of which an official which company is expert.\nSpeaker D: However, we've discussed that scroll wheels are a better option and they are possible. We have an okay for scroll wheels.\nSpeaker D: However, when it comes to the scroll wheel of the iPod, I have one big objection and that is that we have to fit an LCD into the remote control as well.\nSpeaker D: This however may exclude certain materials.\nSpeaker D: We have a squishy kind of remote control and LCD screen may be affected by the movement, so we might not be able to put it in there.\nSpeaker D: There's also restrictions to when it comes to the chip. If we have a more sophisticated scroll wheel rather than this very basic setup that we've just presented, the chip has to be more sophisticated and that's more expensive as well.\nSpeaker D: I don't have any details to the cost but it will be significant difference. I'd rather say drop the titanium and therefore that's how it's more sophisticated chip but that's not up to me to decide really.\nSpeaker D: So that's for the scroll wheel. It limits our choice and squishy is hip so I'd say, rather not go for that. Let's see now.\nSpeaker D: Solar cells cannot be used on a curved or latex surface or remote control but obviously that's not a problem since we have decided against solar cells I assume right?\nSpeaker D: Yeah, anybody still? Yeah, I think batteries are probably the way to go.\nSpeaker D: With the titanium we cannot make it a curved design, just be able to make it flat and straight design pretty much which I assume would exclude.\nSpeaker A: With the sort of spongy and plastic thing you can get those mobile phones and initially have a is plastic but then they have sort of a cover on it which is just sort of soft stuff.\nSpeaker A: So I don't know if that will still be possible to have it open in plastic but then whether people hold it still be sort of spongy.\nSpeaker B: And we can have the fruits and vegetables on this one too far until we can remove it.\nSpeaker A: So was it a possibility of having a graphical display on it like a screen?\nSpeaker D: I can have an LCD screen but therefore no rubber will be used.\nSpeaker D: So the plastic is titanium, yes but this will of course influence the form of plastic as I understand it.\nSpeaker D: And titanium also seems to be tricky when it comes to the form. So the way to go is if you want to scroll wheel you either make it flat and angular, add an LCD screen and then you can basically choose either plastic or titanium.\nSpeaker D: If you want to make it a particular shape use plastic and the LCD screen add a scroll wheel that will be fine or make it just put buttons.\nSpeaker D: Basically plastic gives you the biggest variety of options, maybe not the nicest feel, not much of the analogy.\nSpeaker D: So the rubbery we can shape it however we want, for the rubbery we cannot.\nSpeaker D: With the rubbery we could pretty much shape it the way we wanted it but we cannot add swirl wheels and we cannot add an LCD screen.\nNone: That's the tricky thing.\nSpeaker A: Could we not have a shape with a scroll on the screen and then the initial shape we had which is sort of banana esque.\nSpeaker A: That's the whoever thing we did it yellow and you just stick on just some other things and that's not a quick from it.\nSpeaker A: They wouldn't have any, they're just on the exterior, they wouldn't be necessary to the actual shape of the thing.\nSpeaker C: Is that an option, a plastic shell with a rubbery coating on a certain spot?\nSpeaker D: Certainly can be done.\nSpeaker D: Yes.\nSpeaker D: That doesn't affect the functional side of it all.\nSpeaker D: Like say just the underside of it can be done.\nSpeaker D: I assume.\nSpeaker D: So the fruit design, how about affecting the surface of the actual remote control.\nSpeaker D: Say we don't make it the particular fruit shape obviously but a good surface of an orange banana whatever you name it.\nSpeaker C: What about a smell?\nSpeaker C: It's the remote.\nSpeaker A: You could just sell it in different colours as well.\nSpeaker A: Great.\nSpeaker A: We're supposed to stick to the company cost.\nSpeaker D: Yellow and grey.\nSpeaker D: So what have we lemon banana?\nSpeaker D: Grapefruit.\nSpeaker D: Grapefruit is what we go for in terms of the author appearance perhaps.\nSpeaker C: I would say if I were to make a decision I would probably put the fruit aspect at the lower end of the spectrum of importance.\nSpeaker B: I think having a shape could be a little ridiculous.\nSpeaker D: Well we haven't been out of it already.\nSpeaker C: Perhaps the implied shape will be enough to lower that fruit minded.\nSpeaker A: It's yellow.\nSpeaker A: It's curved.\nSpeaker D: So why not add a couple of grey stripes?\nSpeaker C: Couple of grey stripes.\nSpeaker C: We could put the grey stripes on the bottom so that person could turn it over.\nSpeaker C: The grey stripes would look like a banana just sitting on their table.\nSpeaker A: And then you could actually put the banana shape thing on the fruit bowl on the coffee table.\nSpeaker A: Maybe the holder.\nSpeaker C: If we were to have a holder it could be shaped like a fruit bowl.\nSpeaker C: It could be an ape or a fruit bowl.\nSpeaker C: Do you have more to your presentation?\nSpeaker D: That's pretty much it.\nSpeaker D: I've been informed about the materials.\nSpeaker D: What the materials are.\nSpeaker C: I'm going to plug in here real quick.\nSpeaker C: Like I said we have to make a decision on a couple of these items here.\nSpeaker B: So is the two piece idea out or will we not decide it?\nSpeaker A: We sort of might want to look at that because if we're going to use a battery then the base station might not be necessary.\nSpeaker D: We can still design a two piece remote without having one of them be a base station but just have it be an optional big remote with lots of functions or you take out a smaller piece.\nSpeaker D: We can still do that.\nSpeaker D: However of course it would be like designing two remotes.\nSpeaker D: Which then design would probably limit the limit again to use of certain materials because they would be too expensive.\nSpeaker D: They like a scroll wheel and on both of them or have an LCD screen and so on.\nSpeaker D: You'd probably have to stick rather with a traditional rubber button design which we saw there.\nSpeaker C: Could be done.\nSpeaker C: So these are the decisions that we do need to make by the end of this meeting.\nSpeaker C: For our components concept we need to come up with the energy source chip on print and the case.\nSpeaker C: Probably case material and probably a shape also.\nSpeaker C: And then for the user interface concept we need to decide what the type is and what kind of supplements will have.\nSpeaker C: Energy source I think we've decided batteries although not exciting are probably our best bet.\nSpeaker D: We have five minutes.\nSpeaker D: The more advanced features you want the fancier the chip has to be and the more expensive.\nSpeaker C: So I guess we should pick the case then.\nSpeaker C: Then the chip on print is still kind of we could have either or we could have a complex one or a non complex.\nSpeaker B: But did we decide that the rubbery feel is important enough to us.\nSpeaker C: Okay so we would have the LCD screen.\nSpeaker C: So I guess the case would be plastic with.\nSpeaker C: Perhaps that's not even enough rubber to qualify as being part of it.\nSpeaker B: It's more of a supplement maybe.\nSpeaker B: So we're going to go to the scroll are we going for the iPad type.\nSpeaker C: Which will require a more expensive chip on print right.\nNone: But I guess that is that about it.\nSpeaker C: So we have a good idea of what we're going to need to do on this.\nSpeaker C: So we will have another meeting in 30 minutes.\nSpeaker C: Here's what's going to be going on.\nSpeaker C: Ryan you'll be working on the user interface design.\nSpeaker C: And you'll be working on the look and feel design.\nSpeaker C: Corinne will want a product evaluation.\nSpeaker C: And the two of you get to play with the modeling components and maybe get us a prototype.\nSpeaker C: It should go along well with your look and feel design and your interface.\nSpeaker C: So that basically will just be working on the prototype.\nSpeaker C: We'll accomplish your other two actions.\nSpeaker C: All right.\n"}, {"meeting_id": "ES2010d", "summary": "This was the last meeting on the remote project. Industrial Designer and User Interface first gave the prototype presentation with a model remote they made based on previous team decisions, showing the design of colours, buttons, and shapes. Marketing then led the discussion on the evaluation criteria of their product, after which the team evaluated the remote in various aspects. As for finance, the team calculated the costs of each component of the remote, finding the total cost under their budget and thus allowing the remote to go into production. Finally, the team talked about their experience in the project process, digressing into a casual discussion on computers, cell phones, and ideas for commercials. The team members thanked each other for their effort.", "dialogue": "Speaker C: I want to find out if our work works.\nSpeaker C: Me too.\nSpeaker C: Oh.\nSpeaker C: So I...\nSpeaker B: What do you think?\nSpeaker B: Here's the agenda for our last meeting.\nSpeaker B: After the opening, we're going to have a prototype presentation.\nSpeaker B: Then we're going to discuss the evaluation criteria and the financing of our remote.\nSpeaker B: Then we're going to evaluate the product and I think the whole production process.\nSpeaker B: And then we're going to close it up and we have 40 minutes.\nSpeaker B: So...\nSpeaker B: Sorry.\nSpeaker B: Oh.\nNone: Nope.\nSpeaker B: Let's have a prototype presentation.\nSpeaker B: Okay.\nSpeaker B: Ready?\nSpeaker D: Sure.\nSpeaker D: You are me.\nSpeaker D: Yee.\nSpeaker A: You read that stuff since you wrote it.\nSpeaker D: Well, since our materials aren't exactly what we were going for.\nSpeaker D: I'm just going to translate what this all means for you.\nSpeaker D: The base is going to be gunmetal-glory, which is what we decided.\nSpeaker D: It's going to be plastic.\nSpeaker D: Then there's the latex cover, which is what you see as red.\nSpeaker D: Because it can be replaceable, we just kind of went with the color.\nSpeaker D: Right.\nSpeaker D: And then the buttons are actually kind of poking through rather than on top.\nSpeaker D: And the buttons will be much lighter blue.\nSpeaker D: Almost see-through.\nSpeaker D: And the whole thing lights up, you should press any button rather than just that one button or a light up.\nSpeaker D: And then at the bottom we have our logo.\nSpeaker D: Bright yellow is sort of designed with the RR, which will actually look like our logo.\nSpeaker D: And then on the side you have the buttons.\nSpeaker D: And they're one button, but they kind of push up and down.\nSpeaker D: Okay.\nSpeaker D: And then they're screwing.\nSpeaker A: Nope.\nSpeaker D: They're just buttons, right?\nSpeaker D: Yeah.\nSpeaker D: And then the buttons.\nSpeaker A: On-off switch will be here.\nSpeaker A: And as you've noticed on our prototype, they've ended up with a curvature kind of biconcave sort of thing, because I can't see underneath.\nSpeaker A: So I'm hoping that when we get to production we can actually make them like that, because they're very nice to stock.\nSpeaker A: Stick your fingering.\nSpeaker A: The two squared buttons are to probably least used menu, mute.\nSpeaker A: And then these are the numbers.\nSpeaker A: So our channel and our volume will be on either side.\nSpeaker D: And then the last thing is just that it'll be black labeling on top.\nSpeaker D: Just...\nSpeaker B: And did you determine the curvature of the bottom part of it, or the hand is going to be a single or a double?\nSpeaker B: It's a single, single, single, single, single, because it's not big enough to really constitute\nSpeaker A: a double. It's only actually the size of my hand.\nSpeaker B: Great.\nSpeaker B: I think you did an awesome job.\nSpeaker B: Yeah, I think it's beautiful.\nSpeaker B: It is beautiful.\nSpeaker B: And it's everything that we discussed.\nSpeaker B: Yeah.\nSpeaker C: Good job, you guys.\nSpeaker B: Good job.\nSpeaker B: That was really good.\nSpeaker B: All right.\nSpeaker B: What's next in our agenda?\nSpeaker B: We're going to discuss the evaluation criteria, and that's with Courtney.\nSpeaker C: Okay.\nSpeaker C: It's PowerPoint presentation.\nSpeaker C: I don't really know exactly what we should talk about.\nSpeaker C: It's under evaluation.\nSpeaker C: Right.\nSpeaker C: All right.\nSpeaker C: So these are the criteria we're going to ask.\nSpeaker C: Is it easy to use?\nSpeaker C: Is it fashionable?\nSpeaker C: Yeah, I guess we should write these down.\nSpeaker C: We should reference them.\nSpeaker D: Feel good meaning or?\nSpeaker C: Like, does it feel good?\nSpeaker C: Physically.\nNone: Yeah, physically.\nSpeaker C: That's just for current trends.\nSpeaker C: Right.\nSpeaker C: It doesn't really count, you guys.\nSpeaker A: Yeah, it's a little difficult to incorporate the cover with the cherry fruit on it.\nSpeaker C: Yeah.\nSpeaker C: But so we do have removable covers, right?\nSpeaker C: Yes.\nSpeaker C: Yeah, well then that's correct.\nSpeaker C: And so we can, everybody have that?\nSpeaker C: Oh, wait.\nSpeaker C: Yeah, she's got it.\nSpeaker C: Yeah.\nSpeaker C: Okay, so we're using the criteria for some point scale.\nSpeaker C: And so we need to discuss how we feel it falls within this range.\nSpeaker C: So for easy to use.\nSpeaker C: Do we feel it's very easy to use?\nSpeaker C: Are we going to?\nSpeaker B: I say we individually, right?\nSpeaker B: What do you guys just really?\nSpeaker B: Why not?\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Easy to use.\nSpeaker B: I vote six.\nSpeaker B: Oh, wait.\nSpeaker A: That's false.\nSpeaker A: Two.\nSpeaker A: Okay.\nSpeaker A: I'd say two as well.\nSpeaker A: Yeah, two.\nSpeaker C: Two.\nSpeaker C: That's what I say.\nSpeaker C: Hello.\nSpeaker C: We're great.\nSpeaker C: Okay.\nSpeaker C: Fashionable.\nSpeaker C: At the moment?\nSpeaker B: No.\nSpeaker C: No, I mean like no.\nSpeaker C: I think it's very fashionable.\nSpeaker C: Very chic.\nSpeaker C: I would give it a one.\nSpeaker A: One.\nSpeaker A: I'll give it a two because at the moment it's not looking that way.\nSpeaker C: Well, that's just like a prototype.\nSpeaker C: That's a clay.\nSpeaker D: It's a prototype.\nSpeaker D: What do you think?\nSpeaker D: I don't know if that's a life.\nSpeaker D: Give it like three or four.\nSpeaker B: Okay.\nSpeaker B: So the average is about a two.\nSpeaker B: Yeah.\nSpeaker D: Two.\nSpeaker C: Two points.\nSpeaker C: That's okay.\nSpeaker C: Yeah.\nSpeaker A: All the customers we have.\nSpeaker B: Does it feel good?\nSpeaker B: Imagine.\nSpeaker B: Since we obviously don't have that.\nSpeaker A: I feel like the shape of it actually does.\nSpeaker B: And it's it is very ergonomically designed.\nSpeaker B: It's going to be curved.\nSpeaker B: Yeah, that's going to be thicker.\nSpeaker B: Yeah.\nSpeaker C: I think it feels good.\nSpeaker C: I think so too.\nSpeaker C: I'll give it a two.\nSpeaker B: Okay.\nSpeaker B: Two.\nSpeaker B: I'll give it a one.\nNone: What do you say?\nSpeaker A: I'd say two.\nSpeaker A: I can average just two.\nSpeaker B: Is it technologically innovative?\nSpeaker B: I'm sorry, I'm taking over your job.\nSpeaker B: Oh no, it's fine.\nSpeaker B: You're writing it.\nSpeaker B: You're a project manager.\nSpeaker C: Yeah.\nSpeaker C: I mean, does it have voice?\nSpeaker C: I mean, the phrase recognition on it?\nNone: Yes.\nNone: Right.\nSpeaker B: We were able to do it with that kind of chip.\nSpeaker B: Oh, right.\nSpeaker A: We could do it with the chip.\nSpeaker A: Yes.\nSpeaker A: It wasn't.\nSpeaker A: We had no reflection of it.\nSpeaker A: There's no reflection of it on the prototype.\nSpeaker A: Right.\nSpeaker A: That's because it's way too dimension.\nSpeaker B: And we discussed that.\nSpeaker C: Yeah.\nSpeaker C: Right.\nSpeaker C: So then yes, then I would, well, it doesn't, what else would it need to be technologically innovative?\nSpeaker B: Well, we don't have the, you know, we can't say channel.\nSpeaker B: It changes the channel.\nSpeaker C: And it doesn't cover anything other than TV.\nSpeaker C: Right.\nSpeaker C: So I'd probably give it a three.\nSpeaker A: Yeah.\nSpeaker C: Okay.\nSpeaker C: Even though it is for just a TV remote, it's very advanced.\nSpeaker C: But it is just a TV remote.\nSpeaker B: Yeah.\nSpeaker B: I'd go for a three or four on that one.\nSpeaker B: Okay.\nSpeaker B: Let's go for a three point five.\nSpeaker B: Three and a half.\nSpeaker B: All right.\nSpeaker B: And the last criteria, is it, is it.\nSpeaker B: Squishy and fruity.\nSpeaker B: Well, we've covered that with these.\nSpeaker B: We'll try and do you base it.\nSpeaker C: Yeah.\nSpeaker C: So I'd give it a two.\nSpeaker A: It's capable of being.\nSpeaker A: Oh, very, very beautiful.\nSpeaker C: Squishy and fruity.\nSpeaker B: Okay.\nSpeaker B: And it's very important.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker C: Okay.\nSpeaker C: Next.\nSpeaker C: So, our model slightly resembling a giant delicious cookie appears to be winner.\nSpeaker C: And hopefully we'll sell millions.\nSpeaker C: Good job team.\nSpeaker A: How did you get that in there?\nSpeaker A: What?\nSpeaker A: Slightly resembling a giant delicious cookie.\nSpeaker A: It does.\nSpeaker B: It was good.\nSpeaker B: Thanks.\nSpeaker B: Very good.\nSpeaker B: All right.\nSpeaker B: Let's go back to this.\nNone: No.\nSpeaker B: That's it.\nSpeaker B: Oops.\nNone: Okay.\nSpeaker B: So now we're moving on to finance.\nSpeaker B: Okay.\nSpeaker B: I'm going to show you an Excel spreadsheet.\nSpeaker B: And we're going to fill it in together based on what components we're including in our remote and see if it's under 1250 euro.\nSpeaker B: If so, we can proceed.\nSpeaker B: If not, we need to go back to the drawing board a little bit.\nSpeaker B: Okay.\nSpeaker B: Let me bring that up.\nSpeaker B: Here we go.\nSpeaker B: All right.\nSpeaker B: It's not a hand on.\nSpeaker B: It's powered by battery.\nSpeaker B: Yep.\nSpeaker B: So we give it a two number of components you plan to use.\nSpeaker B: Did I just put quantity being one battery or?\nSpeaker B: Do you want to go for this is where we need to make the final call.\nSpeaker B: If it's a lithium or you want to go triple A's because triple A's we're going to have to do more than one battery.\nSpeaker B: Oh, let's just go for it.\nSpeaker B: Lithium.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Let's do a lithium.\nSpeaker B: I think we're going to be able to purchase this are going to be technologically advanced.\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Down to the electronics section.\nSpeaker B: We're going to need this kind.\nSpeaker B: Correct.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Voice sensor.\nSpeaker B: So one of those.\nSpeaker B: It is a single curve.\nSpeaker B: So one of those.\nSpeaker B: Oh, what's that?\nSpeaker B: Yeah, that's correct.\nSpeaker B: Okay.\nSpeaker B: Down here.\nSpeaker A: Voice material.\nSpeaker B: Plastic.\nSpeaker B: And special color.\nSpeaker B: Okay.\nSpeaker B: Down here.\nSpeaker B: Interface type.\nSpeaker B: We're going to have the integrated scroll.\nSpeaker B: Scroll.\nSpeaker A: Oh, those are just regular.\nSpeaker A: But it's a push button too.\nSpeaker A: Right there.\nSpeaker A: Yeah.\nSpeaker A: So integrated scroll wheel or push button.\nSpeaker A: We're really having just push button interface.\nSpeaker A: Okay.\nSpeaker B: So we can just go.\nSpeaker B: Okay.\nSpeaker C: Okay.\nSpeaker C: But will we.\nSpeaker C: Actually, we'll need to won't we one for the top and then one for this one for each side.\nSpeaker A: But that just covers the type of button we're having.\nSpeaker A: Because we're not doing a scroll on the side.\nSpeaker A: So push button.\nSpeaker D: I think that 29 means like you have both scrolls and push button.\nSpeaker B: Right.\nSpeaker B: I think what Courtney's talking about is do we need to put like here.\nSpeaker C: Because there's like one interface right here and then because it's not going to be on the same plane when you press the button.\nSpeaker C: You have to be additional signals on the side.\nSpeaker C: Okay.\nSpeaker C: So is that going to be an extra one on each side?\nSpeaker C: I don't know.\nSpeaker B: That might put us well.\nSpeaker B: Let's just.\nSpeaker D: Two or three because of one each side and one on top.\nSpeaker C: I mean, it's fine.\nSpeaker C: It comes out the same as 29.\nSpeaker C: We'll less than 29.\nSpeaker B: Okay.\nSpeaker B: And we're going to button supplements.\nSpeaker B: The buttons are not their special color.\nSpeaker C: Okay.\nSpeaker C: They're special form.\nSpeaker C: They're indented.\nSpeaker C: Right.\nSpeaker C: And there's special material.\nSpeaker B: Well, we're under costs then.\nSpeaker B: Over.\nSpeaker B: Over.\nSpeaker B: 12.5 is our limit.\nSpeaker B: Got 11.2.\nSpeaker A: So we can go to production.\nSpeaker A: We can go to.\nSpeaker B: I don't know what I just did.\nSpeaker B: Okay.\nSpeaker B: Now we're going to talk about the project process and whether or not we're satisfied with the whole process and the result.\nSpeaker B: Did we have a lot of room for creativity?\nSpeaker B: Did we have a lot of room for individual leadership?\nSpeaker B: Teamwork and the means, meaning technology that we use to produce our little guy there?\nSpeaker B: And if we found a new ideas.\nSpeaker B: Now, the question is how do we do this?\nNone: Go back.\nSpeaker C: I think we just discussed it.\nSpeaker B: Discuss, sure.\nSpeaker B: All right.\nSpeaker B: Who would like to go first?\nSpeaker A: I think we got stifled for creativity by the company itself.\nSpeaker A: In restricting assembly to using a TV remote.\nSpeaker A: Initially.\nSpeaker D: Oh, that's true.\nSpeaker D: And no internet.\nSpeaker D: Yeah.\nSpeaker C: No, yeah, that's a good point.\nSpeaker C: Because I've forgotten that that was an art decision.\nSpeaker C: Yeah.\nSpeaker B: And how did you feel about the whole process though?\nSpeaker C: Oh, overall, I mean, I thought we did a good job.\nSpeaker C: Like, we got to choose basically a control over minus of being just merely a TV remote.\nSpeaker C: We got to choose what we wanted to do with it.\nSpeaker B: Right.\nSpeaker B: And we got to say over how technologically advanced it should be and also how fashionable.\nSpeaker B: Which I kind of am more fashionable for like technology.\nSpeaker B: Yep.\nSpeaker B: That's right.\nSpeaker B: What about the teamwork aspect?\nSpeaker B: How did you guys enjoy making the model the prototype?\nSpeaker B: I think we did well.\nSpeaker B: Yeah.\nSpeaker B: I think you did.\nSpeaker B: Did you work well together in there?\nSpeaker B: Well, no, there was someone fighting.\nSpeaker A: They were scratching it.\nSpeaker A: Oh, my.\nSpeaker B: And we've all been a pretty congenial team here.\nSpeaker B: I think we have many many.\nSpeaker C: I mean, I'm fine with you guys being like, what is it?\nSpeaker C: The survey annoying or irritating?\nSpeaker C: Wow, that's a definitely a strong one.\nSpeaker A: Okay.\nSpeaker A: The means the whiteboard didn't work.\nSpeaker B: Yeah.\nSpeaker B: I'll go into that.\nSpeaker B: Have to knock that one down.\nSpeaker B: Yeah, no, I'll just.\nSpeaker A: And her friend here really feels strongly about the internet.\nSpeaker A: And they're so much available.\nSpeaker C: The digital panels.\nSpeaker B: Yeah.\nSpeaker B: Digital panels.\nSpeaker B: They were really fine.\nSpeaker B: Yeah.\nSpeaker C: They're fun.\nSpeaker C: You know, I'm not really sure what I could do with them, but they are awesome.\nSpeaker A: These are the laptops for receiving everything.\nSpeaker A: Right.\nSpeaker A: Wireless too.\nSpeaker A: Wireless.\nSpeaker B: And that we have a shared network where we can put all of them.\nSpeaker A: Yeah.\nSpeaker A: And let's not forget sexy dual microphones.\nSpeaker A: Yeah.\nSpeaker B: And big brother.\nSpeaker B: Big brother.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Have we found any new ideas through this process?\nSpeaker C: We are really going to sell this.\nSpeaker B: For something that looks cool and also has what I want it to do technologically.\nSpeaker B: And that's your right brain taking over wanting the artistic, the fashionable, the hip.\nSpeaker B: You know, if we all just went out and bought useful things, I don't think.\nSpeaker D: I mean, that's why I don't like a Max or apples, just because I look at it.\nSpeaker D: I know it's probably a very good computer, but I look at it and I'm taking back to elementary school because they look the same.\nSpeaker D: They look like they did when I was in elementary school.\nSpeaker A: Yeah, because they're pretty and just like the term of district school would only use his Max with their kids exactly.\nSpeaker D: So I associate them with like really low tech, really cheap back font.\nSpeaker B: Yeah.\nSpeaker B: But I do like I pods go for you.\nSpeaker C: Yeah.\nSpeaker C: No, I pods.\nSpeaker A: They want all those words for you.\nSpeaker A: I pods are now quite trendy and they come in different colors.\nSpeaker B: Exactly.\nSpeaker B: I mean, how many people went out and bought a Nokia phone back when we were like in high school, just they could get the changeable faceplates.\nSpeaker B: Oh, yeah.\nSpeaker B: Everybody.\nSpeaker B: Anyway, so that.\nSpeaker D: I didn't have a phone until you got one issue.\nSpeaker D: It's just like the whole it's for TV only.\nSpeaker D: I was like, who's going to buy a remote just for the TV unless they've lost theirs.\nSpeaker C: Look at it.\nSpeaker C: That is a fashionable sheet.\nSpeaker B: People.\nSpeaker C: Whoa.\nSpeaker C: Working director says, yeah.\nSpeaker C: Special people.\nSpeaker D: Working has to actually create the desire for it.\nSpeaker D: Oh, I will.\nSpeaker D: That's okay.\nSpeaker A: We can create a commercial where they think that all their needs will be met.\nSpeaker A: This will help them find the one.\nSpeaker A: They'll be sexy with it.\nSpeaker C: We have a type of commercial and that's the fig leaf.\nSpeaker C: Okay.\nSpeaker C: That'll sell.\nSpeaker A: And so the serpent says use our remote.\nSpeaker D: Yeah.\nSpeaker B: No.\nSpeaker B: Okay.\nSpeaker B: We're going to wrap this up now.\nSpeaker B: The costs are within the budget.\nSpeaker B: We evaluated the project and now we're going to complete the final questionnaire and meeting summary and then we're going to have a big giant party parent like four to this.\nSpeaker B: So.\nSpeaker B: Great.\nSpeaker B: Thank you, team.\nSpeaker A: Great job.\nSpeaker B: It was lovely working with you.\nSpeaker A: You too.\nSpeaker A: Yay.\nSpeaker A: Thanks to the project leader.\nSpeaker A: Yeah.\nNone: We knew.\n"}, {"meeting_id": "Bed010", "summary": "The meeting discussed various topics such as data collection and updates on the German parser. They discussed XML modifications and future thoughts on the ontology. Additionally, object representations will include an EVA vector. This can be incorporated in the database entry for a particular building or inherited from the ontology of the building type. The actual number of the inputs can create a combinatorial explosion when setting the probabilities. In any case, further to fulfilling the basic requirements (translating the parser and the generator into english), the project is entirely open-ended in terms of focus of research.", "dialogue": "Speaker D: Okay, how many batteries do you go through?\nNone: You have no idea if there are a lot of kids.\nNone: All right, you're recording.\nSpeaker G: Thank you.\nNone: All right.\nNone: Good.\nNone: So there's open for Nancy?\nSpeaker C: Yeah.\nSpeaker C: Okay, so let's get started. Nancy said she's coming.\nSpeaker C: That means she will be.\nSpeaker C: My suggestion is that Robert and Jono sort of give us a report on last week's adventures to start.\nSpeaker C: So everybody knows there were these guys from Hyde Ober, actually from the FKI, part of the German SmartCon project who were here for the week.\nSpeaker C: And I think got a lot done.\nSpeaker E: Yeah, I think so too.\nSpeaker E: We got to the point where we can now speak into this SmartCon system and it'll go all the way through and then say something like Roman numeral one Amsmartakos.\nSpeaker E: It actually says Ruhmisch, Amsmartakos, which means it's just using German synthesis module for English sentences.\nSpeaker E: Okay.\nSpeaker E: So...\nSpeaker E: There's no I.\nNone: Okay.\nSpeaker H: I am not going to get this out of the way.\nSpeaker E: The synthesis is just a question of hopefully it's just a question of exchanging a couple of files once we have them.\nSpeaker E: And it's not going to be a problem because we decided to stick to the so-called concept to speech approach.\nSpeaker E: I'm going backwards now.\nSpeaker E: So synthesis is where you sort of make these sounds.\nSpeaker E: And concept to speech is feeding into this synthesis module, giving it what it needs to be said and the whole syntactic structure.\nSpeaker E: So it can pronounce things better, presumably, than just with text to speech.\nSpeaker E: And John O. learned how to write XML tags and did write the three joining grammar for some sentences now, right here.\nSpeaker D: Yeah, so the way the dialogue manager works is it dumps out what it wants to know or what it wants to tell the person to an XML.\nSpeaker D: And there's a conversion system for different to go from XML to something else.\nSpeaker D: And so the knowledge base for the system that generates the syntactic structures for the generation is in a list of like, the knowledge base is in a list of like form.\nSpeaker D: And then the thing that actually builds these syntactic structures is something based on prologue. So you have basically a goal and it says, OK, well, I'm going to try to do the greet the person goal.\nSpeaker D: So it just starts, it binds some variables and it just decides to do some such goals.\nSpeaker D: Basically, it just means build the tree.\nSpeaker D: And it passes the tree on the generation module.\nSpeaker E: But I think the part is that out of the 12 possible utterances that the German system can do, we've already written that the syntax trees for.\nSpeaker D: Yeah, so the syntax trees are very simple.\nSpeaker D: It's like most of the sentences in one tree.\nSpeaker D: And instead of breaking down to small units and building big up, they basically take the sentences and basically cut them in half or into thirds or something like that and made trees out of those.\nSpeaker D: And so, a Tillman wrote a little tool that can take a list of notation and generate an XML tree structure from the list.\nSpeaker D: And so basically you just say, you know, noun goes to, you know, or, I don't, I've never been good at those. So there's like the VP goes to, and those things, and list, and I'm January for you.\nSpeaker E: And because we're sticking to that structure of these synthesis modules, we need to be changed. So all that fancy stuff.\nSpeaker E: And the Texas speech version of it, which is actually a simpler version, is going to be done in October, which is much too late for us.\nSpeaker E: So this way we worked around that.\nSpeaker E: The system, I can show you the system, I actually want at least, maybe you should be able to start it on your own if you want to play around with it in the future.\nSpeaker E: Right now it's brittle and you need to start it up and then make 20 changes on 17 modules before they actually can stomach it, anything.\nSpeaker E: And send in a couple of side queries on some dummy sender setup program so that it actually works because it's designed for this CVIT thing where you have the gesture recognition running with a Siemens virtual touch screen that we don't have here. And so we're doing it via mouse, but the whole system was designed to work with this thing.\nSpeaker E: It was a lot of engineering stuff. No science in there whatsoever, but it's working now, and that's a good news.\nSpeaker E: So everything actually did prove to be language independence except for the parsing and the generation.\nSpeaker D: Well, I did need to generate different trees than the German ones.\nSpeaker D: I mean, because the gerund in German is automatically taken care of with just a regular verb. So I'd have to have am walking or I'd have to have a little stand for the am.\nSpeaker D: Okay. And when I built the tree.\nSpeaker H: Yeah, I noticed that some of the examples they had had, you know, non-English word order isn't so on, you know, and all that good stuff.\nSpeaker C: Yeah. So it might be worth Keith, you looking at this.\nSpeaker H: Yeah. Well, tell me, I still don't really understand, like, I mean, we sort of say, you know, I still don't exactly understand, sort of, the information flow in this thing or what the modules are and so on.\nSpeaker H: So, you know, like, just such and such module decides that it wants to achieve the goal of greeting the user. And then magically it sort of, I mean, how does it know which syntactic type of stuff is on that?\nSpeaker C: Yeah. So I think it's not worth going over in the group, but sort of when you get free and you have the time, either Robert or John O'Reilly can walk you through it.\nSpeaker C: Okay. And you can ask all the questions about how this all fits together.\nSpeaker C: It's a messy, but once you understand it, you understand that it's, there's nothing really complicated about it.\nSpeaker H: Okay. And I remember one thing that came up in the talk last Wednesday was this, I think he talked about the idea of like, he was talking about these lexicalized tree adjoining grammars where you sort of, for each word, you, for each lexical item, the lexical entry says what all the trees are that it can appear in. And of course, that's not, that's the opposite of constructional. That's, you know, that's, that's HPSG or whatever.\nSpeaker C: Right. Now, we're not committed for our research to do any of those things.\nSpeaker C: So, we are committed for our funding.\nSpeaker C: Right. Okay. To make that stuff fit to that.\nSpeaker C: No, to just get the demos they need.\nSpeaker C: Okay. So between us all, we have to get them the demos they need.\nSpeaker C: If it turns out, we can also give them lots more than that by tapping in the other things we do. That's great. But it turns out not to be an entity.\nSpeaker C: It's not to be an entity of the contracts. Okay. And deliberately.\nSpeaker C: So the reason I'd like you to understand what's going on in this demo system is not because it's important to the research. It's just for closure so that if we come up with a question of, could we fit this deeper stuff in there or something, you know what the hell? Right. We're talking about fitting in.\nSpeaker C: Okay. So it's just, same, same actually with the rest of us.\nSpeaker C: We just need to really understand what's there. Is there anything we can make use of?\nSpeaker C: Is there anything we can give back beyond the sort of minimum requirements?\nSpeaker C: But none of that has a short time fuse. Okay.\nSpeaker C: So the demo requirements for this fall are sort of taking care of as of later this week or something.\nSpeaker C: And then so it's probably 15 months or something until there's another serious demo requirement. I mean, we don't think about it for 15 months.\nSpeaker C: It means we cannot think about it for six months. Right. Yeah.\nSpeaker C: So the plan for this summer really is to step back from the applied project.\nSpeaker C: Keep the context open, but actually go after the basic issues.\nSpeaker C: Okay. And so the idea is there's this other subgroup that's worrying about formalizing the notation, getting a notation, but sort of in parallel with that, the hope is that in particular you will work on constructions in English and German for this domain.\nSpeaker C: But not worry about parsing them or fitting them into smart com or any of the other, any other constraints for the time being. It's hard enough to get it semantically and syntactically right and get the constructions in their form and stuff.\nSpeaker C: And I don't want you feeling that you have to somehow meet all these other constraints.\nSpeaker C: And similarly, the parsing, we're going to worry about parsing the general case, you know, construction parser for general constructions. And if we need to cut down version for something or whatever, we'll worry about that later.\nSpeaker C: Okay. So I'd like to, for the summer, turn into science mode.\nSpeaker C: Okay. And I assume that's also your plan as well.\nSpeaker H: So I mean, the point is that like the meetings so far that I've been at have sort of been geared towards this demo.\nSpeaker H: Yeah. It's going to go away pretty soon. Right. Okay. And then sort of shift gears.\nSpeaker E: Yeah. Fairly substantially. Yeah. What I think is a good idea that I can show to anyone who's interested. We can even make a sort of an internal demo.\nSpeaker E: When I show you what I do, I speak into it and you hear a talk.\nSpeaker E: Okay. And I can sort of walk through the information flow. This is like a half hour, 45 minutes, just fun. Okay. And so you, when somebody on the streets comes up to you and asks you what a smart code is right here. Give a sensible answer.\nSpeaker C: So we could set that up as actually an institute wide thing.\nSpeaker C: Just give a talk in the big room and so people know what's going on.\nSpeaker C: When you're ready. Absolutely. Yeah. I mean, that's the kind of thing.\nSpeaker C: That's a level at which we can just invite everybody and say, this is a project that we've been working on and here's a demo version and that's stuff like that.\nSpeaker E: Okay. But we do want to have all the bugs out where you have to sort of pipe in extra XML messages from left and right before you. Indeed.\nSpeaker E: Okay. Makes sense.\nSpeaker C: But it's clear then, I think, actually roughly starting, let's say next, next meeting because this meeting we have one other thing to tie up besides the tripp report. Okay. But starting next meeting, I think we want to flip into this mode where there are a lot of issues.\nSpeaker C: What's the ontology look like? You know, what the constructions look like?\nSpeaker C: What's the execution engine look like? Mm-hmm.\nSpeaker C: Lots of things. But more focused on an idealized version than just getting the demo out.\nSpeaker C: Now, before we do that, let's get back in. Oh, but it's still, I think, useful for you to understand the demo version enough so that you can see what it is that it might eventually get retrofitted into or something.\nSpeaker C: And John has already done that.\nSpeaker C: Looked at the smart com stuff.\nSpeaker D: What parts?\nSpeaker C: What's the other parts? Yeah. Okay. Anyway, so the trip, the report on these last, we sort of interrupted you guys telling us about what happened last week.\nSpeaker E: Maybe that was just amazing to see how unstable the whole thing is.\nSpeaker E: And if you just take the, and I got the feeling that we are the only ones right now who have a running system, I don't know what the guys in Kaiser's Cloud 10 have running because the version that is the full version that's on the server does not work.\nSpeaker E: And you need to do a lot of stuff to make it work.\nSpeaker E: And so it's an even term that involves sort of said, yeah, that never was a really working version that did it without all the shortcuts that were able to do for the October version. So we actually maybe ahead of this is TeamCrop by now, the system, the integration group.\nSpeaker E: And it was fun to some extent.\nSpeaker E: But the outcome that is sort of a scientific interest is that I think both Ralph and Tillman, I know that they enjoyed it here.\nSpeaker E: And they liked a lot of the stuff they saw here, what we have been thinking about.\nSpeaker E: And I'm willing to cooperate by all means.\nSpeaker E: And part of my responsibility is to use our internal group where server at EML make that open to all of us and them so that whatever we discuss in terms of parsing and generating and constructions, we sort of put it in there and they put what they do in there.\nSpeaker E: And maybe we can even get some overlap, get some synergy out of that.\nSpeaker E: And if I find someone at EML that is interested in that, I may even think that we could take constructions and generate from them because the tree-drone in grandma's that Tillman is using, as you said, nothing but a mathematical formalism. You can just do anything with it.\nSpeaker E: It's syntactic trees, each piece of g-like stuff, or whether it's construction.\nSpeaker E: So if you ever get to the generation side of constructing things, there might be something of interest there, but the moment we're, of course, definitely focused on the understanding pipeline.\nSpeaker C: Any other visit reports for the stories?\nNone: There So we now know I think what the landscape is like and so we just push on and do what we need to do. And one of the things we need to do is the, and then I think it's relatively tight, totally constrained, is to finish up this belief net stuff. So, and I was going to switch to start talking about that unless there are other more general questions.\nSpeaker C: Okay, so here's where we are on the belief net stuff as far as I understand it. Going back I guess two weeks ago, Robert had laid out this belief net missing only the connections.\nSpeaker C: Right? So I put all the dots down and we went through this and I think more or less convinced ourselves that at least the vast majority of the nodes that we needed for the demo level we were thinking of were in there. Yeah, we may run across one or two more, but of course the connections weren't. So Boschern and I went off and looked at some technical questions about were certain operations sort of legitimate belief net computations and was there some known problem with them or had someone already solved how to do this and stuff. And so Boschern tracked that down. The answer seems to be no, no one has done it, but yes, it's perfectly reasonable thing to do if that's what you set out to do. And so the current state of things is that again starting now we'd like to actually get a running belief net for this particular subdomain done in the next few weeks. So Boschern is switching projects as the first of June and he's going to leave us an inheritance which is a hopefully a belief net that does these things. And there are two aspects to it, one of which is technical getting the coding right and making it run and stuff like that. And the other is the actual semantics. What are the considerations in them and what are the ways in which they relate. So he doesn't need help from this group on the technical aspects or if he does we'll do that separately. But in terms of what are the decisions and stuff like that, that's something that we all have to work out. Is that right? I mean that's both you guys\nSpeaker B: understanding where we are. Okay. So I guess is there like a latest version of the belief\nSpeaker E: net or the proposed belief net? We had decided, we didn't decide we wanted to look into maybe getting it the visualization a bit clearer but I think if we do it sort of a paper version of all the nodes and then the connections between them. That should suffice.\nSpeaker C: Yeah, that's a separate problem. We do in the long run want to do better visualization\nSpeaker D: and all that's separable. Yeah. I did look into that in terms of you know exploding the nodes out and downing. Right. Java based on support that I can imagine a way of hacking at the code to do that. He'd probably take two weeks or so to actually go through and do it. And I went through all the other packages on Kevin Murphy's page. Right. And I couldn't find the necessary mix of free and with the gooey and with this thing that we can pay.\nSpeaker C: If we can pay, you know, it's paying a thousand dollars or something we can do that. Okay.\nSpeaker C: So don't view free as a absolute constraint. Okay. So then I'll go back and look at the\nSpeaker B: one. Okay. And you can ask Kevin. The one that people seem to use is a hugin or whatever. Hugin. Yeah. That's free. I don't think it's is it free because I've seen it advertised\nSpeaker C: in places. So maybe free documents like I don't know. I have a copy that I downloaded. So at one point it was free. But I know people do use hugin. So how do you spell that?\nSpeaker C: Hugin. And Boston can give you a pointer. Not in any case. But being, you know, if it's probably for university, it's going to be real cheap anyway. But you know, it's $50,000.\nSpeaker E: We aren't going to do it. I was supposed to just not to spend two weeks and change you know. The Java basically. I will send you a pointer to a Java app that does that. It's sort of a fish. You have a node and you click on it and it shows you all the connections.\nSpeaker E: And you click on something else that moves away. That goes into the middle. And maybe there is an easy way of interfacing those two. If that doesn't work, it's not a problem we need to solve right now. But I'm what my job is I will give you the input terms of the internal structure. Maybe node by node or something like this. Or should I collect it all?\nSpeaker B: That's a matter. Just any rough representation of the entire belief net. It's probably\nSpeaker E: less. And you're going to be around, again, always two stays and three stays after noonish\nSpeaker B: as usual. What were that change? I mean, yeah, I can, this week I guess I kind of have a lot of projects and stuff. But after that, I'll generally be more free. So yes, I can be around. I mean, generally, if you email me, I can be around other days. Yeah,\nSpeaker C: this is not a crisis that, I mean, you do everybody who's student should, you know, do their work. Of course, there's no one good shape. And then we'll dig down on this.\nSpeaker E: No, that's good. That means I have, I can spend this week doing it. How do you go about\nSpeaker H: this process of deciding what these connections are? And there's an issue of how to weight the different things too. You just sort of gas and see if it sort of.\nSpeaker C: Well, there's two different things you do. One is you design and the other is you learn.\nSpeaker C: Okay. So what we're going to do initially is do design. And if you will, guess. Okay.\nSpeaker C: That is, you know, use your best knowledge of the domain to hypothesize what the dependencies are and stuff. If it's done right and if you have data, then there are techniques for learning the numbers given the structure. And there are even techniques for learning the structure, although that takes a lot more data and it's not as, and so forth and so on.\nSpeaker C: So, but for the limited amount of stuff we have for this particular exercise, I think we'll\nSpeaker E: just design it. Hopefully as time passes, we get more and more data from Hydeburg and people actually using it and stuff. So, but this is the long run. But to solve our problems,\nSpeaker H: mediocre design will do. Yeah, that's great. And by the way, speaking of data, are there I could swear, I could swear I saw it sitting on someone's desk at some point. But is there a transcript of any of the sort of initial interactions of people with the system? Because, you know, I'm still sort of itching to look at what, look at the stuff and see what people\nSpeaker C: are doing. Yeah, make yourself. And of course, Keith would like the German as well as the English. So, whatever you guys can get. Oh, yeah, of course, German. You're native language. I can read that one. Okay. That's important. So, he'll get you some data. Okay.\nSpeaker H: Yeah, I mean, I sort of found the audio of some of those and it kind of sounded like I didn't want to trudge through that. You know, it's just strange. But, yeah. We probably\nSpeaker E: will not get those to describe because they were trial runs. But we have data in English\nSpeaker C: and German already transcribed. Okay. Okay. Okay. So, while we're still at this sort of top level, anything else that we ought to talk about today? How was your finger?\nSpeaker H: Oh, I just wanted to, like, mention as an issue, you know, last meeting I wasn't here because I went to a linguistics vocarium on the fictive motion stuff and that was pretty interesting and, you know, I mean, it seems to me that that will fairly obviously be of relevance to what we're doing here because, you know, people are likely to give descriptions like, you know, what's that thing right where you start to go up the hill or something like that, you know, meaning a few feet up the hill or whatever from some reference point and all that stuff. So I mean, I'm sure in terms of, you know, people trying to state locations or, you know, all that kind of stuff is going to be very relevant. So, um, that was, the talk was about English versus Japanese, which obviously the Japanese doesn't affect us directly except that some of the construction, what he talked about was, you know, in English we say things like, you know, your bike is parked across the street and we use these prepositional phrases, you know, well, if you work to move across the street, you would be at the bike. But in, in Japanese, the more conventionalized tendency is to use a sort of a description of where one has crossed to the river. There's a tree.\nSpeaker H: You know, you can actually say things like, there's a tree where one has crossed the river, but no one has ever crossed the river or something like that. So the idea is that this really, you know, that's supposed to show that it's really fictive and so on.\nSpeaker H: But the point is that that kind of construction is also used in English, you know, like right where you start to go up the hill or just when you get off the train or something like that to indicate where something is. So we'll have to think about how much is that used in German?\nSpeaker E: Well, I was on a different side. I mean, the deep map project, which is undergoing some renovation at the moment, but this is a three language project German English Japanese.\nSpeaker E: And we have, I have taken care that we have the Japanese generation and stuff. And so I looked into a special description so we can generate special descriptions how to get from A to B and information on objects in German English and Japanese.\nSpeaker E: And there is a huge project on special descriptions differences in special descriptions. Well, if you're interested in that. So how, how, I mean, it does sort of go all the way down to the conceptual level to some extent.\nSpeaker C: So where is this huge project?\nSpeaker E: It's closed. It's the beta felt generation of special descriptions and whatever.\nSpeaker C: Well, that may be another thing to keep us.\nSpeaker E: But I think we should leave Japanese constructions maybe outside of the scope for now, but definitely it's interesting to look at it across the border there.\nSpeaker A: So, I think it's a bit of any tension to the relative position of direction, relative to a speaker, for example, that some differences between Hebrew and English, we say, back in front of the car as you come here, you drop behind the car.\nSpeaker A: In Hebrew, it means, back behind the car because the front of the car you can find it.\nSpeaker A: Interesting.\nSpeaker A: Well, in English, the front of the car is the absolute front of the car.\nSpeaker A: So, the canonical direction of motion determines where the front is.\nSpeaker E: I think did you ever get to look at the red paper that I sent you on that problem in English in German, Carol 93?\nSpeaker E: There is a study on the differences between English and German on exactly that problem.\nSpeaker E: They actually say the monkey in front of the car, worse than monkey. And they found statistically very significant differences in English and German.\nSpeaker E: It might be, since there are only a finite number of ways of doing it, that the German might be more like Hebrew in that respect.\nSpeaker E: The solution they proposed was that it was due to syntactic factors.\nSpeaker E: That syntactic factors do play a role there, whether you're more likely to develop choices that lead you towards using intrinsic versus extrinsic reference frames.\nSpeaker H: It seems to me that you can get both in English, depending, you know, like in front of the car, here's the car sideways to me and between me and the car is in front of the car.\nSpeaker H: I can also give you a point of view of the paper of mine, which is the ultimate taxonomy of reference frames.\nSpeaker E: I'm the only person in the world who actually knows how it works.\nSpeaker E: Not really.\nSpeaker C: Great. No, I've not seen that.\nSpeaker A: It's called reference frames.\nSpeaker E: It's special reference frames. You actually have only, if you want to have a, this is usually, this should be an L.\nSpeaker E: Well, actually, you have only have two choices. You can either do a two point or a three point, which is, you familiar with the, with the orygo, whether it's a center orygo is the center of the frame of reference.\nSpeaker E: And then you have the reference object and the object to be localized.\nSpeaker E: In some cases, the orygo is the same as the reference object.\nSpeaker E: So that would be origin.\nSpeaker E: Orygo is a terminus technique in that sense. It's even used in the English literature. Orygo.\nSpeaker E: Okay. All right. I've heard it. Okay. And so this videotape is in front of me.\nSpeaker E: I'm the orygo, and I'm also the reference object. Those are two point.\nSpeaker E: And three point relations is, if something has an intrinsic front side like this chair, then your shoe is behind the chair.\nSpeaker E: And reference object. And no, from my point of view, your shoe is left of the chair.\nSpeaker H: You can actually say things like, it's behind the tree from me or something like that, I think, in certain circumstances in English, right?\nSpeaker H: From where I'm standing, it would appear that.\nSpeaker D: So it's a bit like break it back for a time.\nSpeaker C: Yeah, it sounds like it doesn't.\nSpeaker E: And then here you, on this scale, you have it either be ego or allocentric.\nSpeaker E: And that's basically it. So ego-centric, two point, ego-centric, three point, or you can have allocentric.\nSpeaker E: So as seen from the church, the town hall is right of that fire station. Hardly ever used, but it's worth.\nSpeaker C: Yeah, see this is getting into Ami's thing. He's very interested in that.\nSpeaker C: So, yeah, well, why didn't you just put it on the webpage? There's this EDU, right?\nSpeaker E: Yeah, it's also a lot of my home page at e-mail. And then that to me of a special description, or I'll send it link.\nSpeaker C: Yeah, but just put it link on. Yeah.\nSpeaker C: By the way, something that I didn't know until about a week ago or so is apparently there are separate brain areas for things within reach and things that are out of reach.\nSpeaker C: So there's all this linguistic stuff about near and far or yawn and so forth.\nSpeaker C: So there's this linguistic facts, but apparently, here's the way the findings go.\nSpeaker C: That they do MRI and if you've got something within reach, then there's one of your areas lights up.\nSpeaker C: And if something's out of reach, a different one. But here's the amazing result. They say, you get someone with a deficit so that they have perfectly normal ability at distance things.\nSpeaker C: So the typical task is subdivisions. So there's a line on the wall over there and you give them a laser pointer and you say, where's the midpoint? They do fine.\nSpeaker C: If you give them a line and if they touch it, they can't. They're just that part of the brain isn't functioning so they can't do that.\nSpeaker C: Here's the real experiment. Same thing on the wall. Give them a laser. Where is it? They do it.\nSpeaker C: Give them a stick. Long stick and so they do it. They can't do it. So there's a remapping of distance space into nearby space.\nSpeaker A: So they don't get this within reach?\nSpeaker C: It's not within reach. And you use it within reach. So I'll dig up this reference.\nSpeaker C: First of all, I'll explain something that I've always wondered about. And I'll do this test on you guys as well.\nSpeaker C: I have had an experience not often, but a certain number of times. When, for example, I'm working with a tool, a screwdriver or something, for a long time, I start feeling the tip directly.\nSpeaker C: Not indirectly, but you actually can feel the tip. And people who are accomplished by Linus and stuff like that claim they also have this kind of thing where you get a direct sensation of physical sensation of the end of the tool.\nSpeaker H: What's going on at the end of the tool?\nSpeaker A: The extension of the tool. Yeah. Right. Have you had this? I think so. I mean, it's not exactly the same thing, but it's getting close to it.\nSpeaker C: What does it feel like?\nSpeaker C: It feels like you're as if your neurons had extended themselves out to this tool and you're feeling forces on it and so forth. And you deal directly with it.\nSpeaker A: I was playing with those devices that allow you to manipulate objects when it's dangerous to get close.\nSpeaker A: Right. You're being set your hands, something, and there's a correspondence.\nSpeaker A: So I've played with it after a while. You don't feel the difference anymore.\nSpeaker A: You stop back and suddenly it goes away.\nSpeaker A: Right.\nSpeaker C: So this was the first actual experimental evidence I've seen that was consistent with this anecdotal stuff.\nSpeaker C: And of course, it makes a lovely story about why languages make this sensation.\nSpeaker C: Of course, their behavioral differences too. Things you can reach are really quite different things you can't.\nSpeaker C: But there seems to be an actual really deep embodied neural difference.\nSpeaker C: So in addition to the...\nSpeaker C: This is more proximal distal. Yeah, exactly. So in addition to ego and allocentric, which appear all of this place, you also apparently have this proximal distal thing which is very deeply embedded.\nSpeaker E: Well, Dan Montello sort of... he does the cognitive map, Guro, done in Santa Barbara.\nSpeaker E: And he always talks about these... he already, probably most likely without knowing this evidence, is talking about these small scale spaces that you can manipulate versus large scale.\nSpeaker C: Well, there's a lot of behavioral things on this. But it was the first neurophysiological thing I saw.\nSpeaker C: Anyway, yeah. So we'll look at this.\nSpeaker C: And so all of these issues are now starting to come up. So now we're done with demos. We're starting to do science, right?\nSpeaker C: So these issues about reference, spatial reference, discourse reference, all this sort of stuff.\nSpeaker C: Dikesus, which is part of what you were talking about. So all of this stuff is coming up essentially starting now.\nSpeaker C: So we're going to do all this. So there's that. And then there's also a set of system things that come up.\nSpeaker C: So okay, we're not using their system. That means we need our system right in it follows.\nSpeaker C: And so in addition to the business about just getting the linguistics right and the formalism and stuff, we're actually going to build something.\nSpeaker C: And John O. is point person on the parser analyzer, whatever that is.\nSpeaker C: And we're going to start on that in parallel with the grammar stuff.\nSpeaker C: But to do that, we're going to need to make some decisions like ontology.\nSpeaker C: So this is another thing we're going to get involved.\nSpeaker C: And makes relatively early, I think, make some decisions on, is there an ontology API?\nSpeaker C: There's a sort of standard way of getting things from ontologies and we build the parser and stuff around that.\nSpeaker C: Or is there a particular ontology that we're going to standardize on?\nSpeaker C: And if so, for example, is there something that we can use there?\nSpeaker C: It does either the smart come project or one of the projects at EML have something that we can just pull out for that.\nSpeaker C: So there are going to be some things like that which are not science but system.\nSpeaker C: But we aren't going to ignore those because we're not only going, the plan is not only to lay out this thing but actually build some of it.\nSpeaker C: And how much we build and so forth.\nSpeaker C: Part of it, if it works right, is it looks like we're now in a position that the construction analyzer that we want for this applied project can be the same as the construction analyzer that Nancy needs for the child language modeling.\nSpeaker C: So it's always been out of phase but it now seems that there's a good shot at that.\nSpeaker C: So we've talked about it and the hope is that we can make these things the same thing.\nSpeaker C: And of course it's only, in both cases, it's only one piece of a bigger system.\nSpeaker C: But it would be nice if that piece were exactly the same piece. It was just this construction analyzer.\nSpeaker C: And so we think we have a shot at that.\nSpeaker C: So to come circle on that, this formalization task is trying to get the formalism into a shape where it can actually be of use to someone who's trying to do this.\nSpeaker C: Yeah, where it actually is, it covers the whole range of things. And the thing that got Mark in the worst trouble is he had a very ambitious thing he was trying to do.\nSpeaker C: And he insisted on trying to do it with a limited set of mechanisms that turned out inherently not to cover the space.\nSpeaker C: And it just was just terribly frustrating form. And he seemed fully committed to both sides of this irreconcilable thing.\nSpeaker C: And John O. is much more pragmatic. This is true, isn't it? Yes. So there's sort of really deep emotional commitment to certain theory being complete.\nSpeaker C: It certainly hasn't been observed. Now you do, but that's okay. Because I don't have to implement that.\nSpeaker C: Actually, the thing is you do, but the thing you have to implement is so small that...\nSpeaker C: I think get something done. But to try to do something of scale and purist, particularly if what you're purist about doesn't actually work.\nSpeaker C: And then the other thing is while we're doing this, Robert's going to pick a piece of this space for his absentee thesis. I think you all know that you can just, in Jeremy, almost just send him to the office.\nSpeaker C: I'm just driving through this. I'm driving through. You put in your credit card as well. But anyway, so that's also got to be worked out hopefully over the next few weeks.\nSpeaker C: So that it becomes clear what piece Robert wants to jump into. And while we're at this level, there's at least one new doctoral student in computer science who will be joining the project either next week or the first of August depending on the blindishments of Microsoft.\nSpeaker C: And her name is Eva. Nobody believes this.\nSpeaker C: Is this person someone who's in first year this year? No, first year coming. So she's now out here. She's moved. And she'll be a student as of that.\nSpeaker C: And probably she'll pick up from you on the belief net stuff. So she'll be chasing you down and stuff like that.\nSpeaker C: Against all traditions. And actually I talked today to a undergraduate who wants to do an honors thesis on this class. No, interestingly enough.\nSpeaker C: You always get these people who are not in the class. Some of them. Yeah. So anyway, but she's another one of these ones. It was the 3.9 average and so forth and so on.\nSpeaker C: So I've given her some things to read. So we'll see how this goes. Oh, there's yet another one of the incoming first incoming first year graduate students to rest into. So we'll see how that goes.\nSpeaker C: Anyway, so I think as far as this group goes. It's certainly worth continuing for the next few weeks to get closure on the belief net and the ideas that are involved in that.\nSpeaker C: And what are the what are the concepts we'll see whether it's going to make sense to have this be separate from the other.\nSpeaker C: Bigger effort with the formalization stuff or not. I'm not sure. It probably depends on what your thesis turns out to be and how that goes.\nSpeaker C: So I was and then I'm and you can decide, you know, how much time you want to put into it. And it's beginning to type shape shape. So.\nSpeaker C: And I think you will find that if you want to look technically at some of the your traditional questions in this light, Keith, whose building constructions will be quite happy to see what you envision as the issues and the problems and how they might get reflected in constructions.\nSpeaker C: I suspect that's right. Yeah.\nSpeaker A: I have to go to prison and for June or July.\nSpeaker C: And if it's useful, we can probably arrange for you to drop by and visit either at Heidelberg or at the German AI Center while you're in the neighborhood.\nSpeaker A: Actually, I'm invited to do some consulting with a bank in Geneva, which has an affiliation with research institute in Geneva, which I forgot the name of.\nSpeaker C: Well, we're connected to this. There's a very significant connection between we'll go through this. I see a sign and EPFL, which is the.\nSpeaker C: Germany's got two big technical institutes. There's one in Zurich, ETH, and then it's one French speaking one in Loseon, OK, which is EPFL.\nSpeaker C: So find out who they are associated with in Geneva. Probably we're connected to them.\nSpeaker C: Yeah. And so anyway, we can undoubtedly get to give a talk at the email or something like that.\nSpeaker E: I think the one you gave here a couple of weeks ago would be your interest.\nSpeaker C: A lot of interest. Actually, either place, DFKI or.\nSpeaker C: So, and if there is a book that you'll be building up some audience for it, and you'll get feedback from these guys, because they've actually these DFKI guys have done as much as anyone over the last decade in trying to build them.\nSpeaker C: So we'll set that up.\nSpeaker C: OK, so unless we want to start digging into the belief net and the decisions now, which would be fine.\nSpeaker E: It's probably better if I come next week with the version 0.9.\nSpeaker C: So how about a few two guys between now and next week come up with something that is partially proposed and partially questions.\nSpeaker C: Here's what we think we understand. Here are the things we think we don't understand.\nSpeaker C: And that we as a group will try to finish it.\nSpeaker C: What I'd like to do is shoot for finishing all this next Monday.\nSpeaker C: OK, these are the decisions. I don't think we're going to get lots more information. It's a design problem.\nSpeaker C: And let's come up with a first cut at what they should look like.\nSpeaker C: And then finish it up. Does that make sense?\nSpeaker E: And this semester will be over next week, but then you have projects for one more week to come?\nSpeaker B: No, I think I'll be done by this weekend.\nSpeaker D: Same with you, no?\nSpeaker D: Well, I have projects within my professor. One of my classes also has a final that he's giving us.\nSpeaker D: And he's giving us five days to do it, which means it's going to be hard.\nSpeaker C: Oh, it's taken on fine.\nSpeaker C: Yeah, he's doing this.\nSpeaker C: Akin.\nSpeaker C: That would have been in my guess.\nSpeaker C: But anyway, yeah.\nSpeaker C: Pretty soon.\nSpeaker D: OK, so I guess that's definitely the last day.\nSpeaker D: Like it or not for me.\nSpeaker C: So let's do this. And then we'll be making some separate.\nSpeaker C: These guys are talking.\nSpeaker C: We have a group on the formalization, Nancy and John O'Neil are going to talk about parsers.\nSpeaker C: So there are various kinds of, of course, nothing gets done even in a meeting of seven people, right?\nSpeaker C: So two or three people is the size in which actual work gets done.\nSpeaker C: Yeah.\nSpeaker C: So we'll do that.\nSpeaker C: Great. Well, the other thing we want to do is catch up with Ellen and see what she's doing because the image schemas are going to be important.\nSpeaker C: Yeah.\nSpeaker C: But we want those, right?\nNone: Yeah.\nSpeaker C: And we want them formalized and stuff like that.\nSpeaker C: Yeah.\nSpeaker C: So let me make a note to do that.\nSpeaker C: OK.\nSpeaker H: Yeah, I'm actually probably going to be in contact with her pretty soon anyway, because various of us students were going to have a reading group about.\nSpeaker H: Oh, right.\nSpeaker C: Right. Right. Right. Right. That's great.\nSpeaker C: Yeah. I should have mentioned that.\nSpeaker C: OK.\nSpeaker C: Although she said it's a secret.\nSpeaker C: Right.\nSpeaker C: The faculty.\nSpeaker C: The faculty are posting that.\nSpeaker C: I wouldn't say as much.\nSpeaker C: But I'm sufficiently clueless that I count as eight.\nSpeaker C: Yeah. Right.\nSpeaker H: It's as if we didn't tell anyone at all.\nSpeaker H: I'll ask her.\nSpeaker H: I'm going to get a bit of a man to say it about.\nNone: Yeah.\nNone: I was like a little bossery.\nNone: You could say you were using a little boss.\nSpeaker F: OK.\nSpeaker F: Fine. You could say.\nSpeaker F: You could say. You could say.\nSpeaker F: And I hope you.\nNone: I don't know my question, but I'll give it my all.\nNone: Yeah.\n"}, {"meeting_id": "Bed004", "summary": "The team began the meeting by discussing the logistics of setting up the interface for data collection. Some members ran a trial of it earlier and found someone who would make a suitable wizard. The team shared concern about how they would recruit non-university student participants. Grad D introduced the team to the second iteration of the bayes-net model and its schemas. Then, the discussion moved onto controlling the size of the bayes-net as it would otherwise be based on too much information. The team ended the meeting by delving into how the method of creating a Bayes-net in different scenarios could itself be abstracted, i.e. narrowing the input and output factors and the intermediate representation.", "dialogue": "Speaker B: You're not supposed to be drinking it here, did?\nSpeaker C: Transcript number three, two, seven, one, three, two, nine, oh.\nSpeaker C: Three, two, nine, oh.\nSpeaker C: Three, two, six, seven, six, four.\nSpeaker C: Seven, eight, nine, zero, one, zero, one, one, eight, one, three, one, two, three, oh, one.\nSpeaker C: Two, seven, five, three, four, nine, nine, five, six, oh, nine, eight, two, nine, five, four, two, one.\nSpeaker B: We have to read them that slowly.\nSpeaker B: Okay. Sound like a robot.\nSpeaker B: This is transcript three, eight, nine, one, three, nine, one, zero.\nSpeaker B: Three, three, six, zero, zero.\nSpeaker B: Four, two, zero, zero, one, one, seven.\nSpeaker B: Five, six, six, nine, seven, eight, oh, three.\nSpeaker B: Zero, five, three, one, five, six, six, two, eight, seven, six, five, three, four, five, zero.\nSpeaker B: Seven, two, seven, nine, oh, three, eight, eight, three, nine, nine, nine, oh, eight, eight, nine, two.\nSpeaker B: Zero, eight, two, eight, one, two, zero, seven, three.\nSpeaker A: This is transcript three, nine, five, one, three, nine, seven, oh.\nSpeaker A: Five, six, zero, five, one, one, eight, one, eight, nine, five, oh, five, seven, oh, seven, oh, seven, zero.\nSpeaker A: One, two, three, zero.\nSpeaker A: Five, one, two, three, four, six, five, two, five, eight, six, oh, seven, six, six, six, zero.\nSpeaker A: Eight, nine, one, seven, nine, oh, zero, zero, seven, two, two, one, four, three, three, zero, two, two, four, three, five, eight, zero, three, seven, four, five.\nSpeaker G: So I think the one thing that you had in the start is the data that you met, talked to you, but it's so useful in your college practice. It's recording now, so just keep it pointing.\nSpeaker G: And then it's one, two, a big favor, thinking about how to do it as a thing, where the black gear will be done, but make sure not to be under recording anymore.\nNone: Because if I turn it straight, I'm going to be at the meeting at night, when I was on the board, and then when you leave, just turn off the microphone, and the battery will be set to be down, and it's great.\nSpeaker G: And if Jerry and Anne start rolling, three of the numbers, that would be very best of all.\nSpeaker A: So you think how would you do that?\nSpeaker B: When you read the numbers, it kind of reminded me of beat poetry.\nSpeaker C: It's a practical for the coming sort of feeling, but.\nSpeaker B: 3, 3, 6, zero, zero, four, two, zero, zero, one, seven.\nSpeaker B: That's what I think of when I think of beat poetry.\nSpeaker B: You ever seen so I married an axe murderer?\nSpeaker B: Bought to it.\nSpeaker B: There's parts when he's doing beat poetry.\nSpeaker B: Yeah.\nSpeaker B: He talks like that.\nSpeaker B: That's why I think.\nSpeaker B: That probably is why I think of it that way.\nSpeaker C: I didn't see that movie.\nSpeaker C: Who didn't? Who made that?\nSpeaker B: Mike Myers is the guy.\nSpeaker B: It's his cute romantic comedy.\nSpeaker B: That's his cute romantic comedy.\nNone: Yeah.\nSpeaker B: I'm feeling this real funny.\nSpeaker B: I'll spoil it for you.\nSpeaker B: He works in a coffee shop in San Francisco.\nSpeaker B: And he's sitting there on this couch and they bring him this massive cup of espresso.\nSpeaker B: Like, excuse me, I ordered the largest espresso.\nSpeaker C: We're having...\nSpeaker C: Tiramisu tasting contest this weekend?\nSpeaker B: Are you trying to decide who's the best taste or of tiramisu?\nSpeaker C: No. There was a fierce argument that broke out over who's tiramisu might be the best.\nSpeaker C: We decided to have a contest where those people who claimed to make good tiramisu make them.\nSpeaker C: Then we got a panel of impartial judges that were faced to a blind taste.\nSpeaker C: And then vote.\nSpeaker C: Should be fun.\nSpeaker B: It seems like you could put a magic special ingredient in so that everyone would know which one was yours.\nSpeaker B: Then if you were to bribe them, you could...\nSpeaker C: Well, I was thinking, if you guys have plans for Sunday, we're not...\nSpeaker C: It's probably going to be this Sunday, but we're sort of working with the weather here, because we also want to combine it with some barbecue activity where we just fire it up and whoever brings whatever...\nSpeaker C: You can throw it on there.\nSpeaker C: So only that tiramisu is free.\nSpeaker B: I'm going back to visit my parents this weekend, so I'll be out of town.\nSpeaker C: So you're going to the West Bay then?\nSpeaker B: It's half Bay, yeah.\nSpeaker A: I should be free.\nSpeaker C: I'm sorry I was looking for you guys down the stairs.\nSpeaker E: So are we recording that song?\nSpeaker D: We are.\nSpeaker B: Is Nancy going to show up?\nSpeaker E: She's got a student like in that, probably not.\nSpeaker B: These things ever emit a very...\nSpeaker B: Like piercing screeds right in your ear.\nSpeaker C: They are going to get more comfortable headsets.\nSpeaker C: Or the other...\nSpeaker C: Okay, let's get started.\nSpeaker C: The...\nSpeaker C: Should I go first with the...\nSpeaker C: Data, can I have the remote control?\nSpeaker F: Okay.\nSpeaker C: So on Friday we had our wizard test, data test, and these are some of the results.\nSpeaker C: This was the introduction.\nSpeaker C: I actually, even though this was kind enough to offer to be the first subject, I felt that she knew too much.\nSpeaker C: So I asked Latonya just on the spur of the moment and she was kind enough to serve as the first subject.\nSpeaker C: So this is what she saw as part of...\nSpeaker C: Introduction. This is what she had to read aloud.\nSpeaker C: That was really difficult for her.\nSpeaker C: And...\nSpeaker A: All the names, you mean?\nSpeaker C: The names.\nSpeaker C: This was the first three tasks she had to master after she called the system.\nSpeaker C: And then of course the system broke down and those were...\nSpeaker C: I should say the system was supposed to break down and then these were the remaining three tasks that she was going to solve with a human.\nSpeaker C: There are the results.\nSpeaker C: And I will not...\nSpeaker H: We will skip the reading now.\nSpeaker H: The five playhouses, city theater, the two are three.\nSpeaker H: Is it our part of the city?\nSpeaker C: And the reading was five minutes exactly and now comes the...\nSpeaker C: This is the phone and face office.\nSpeaker G: And if we're calling high to work to this information, I'm going to help you.\nSpeaker A: Great, now I have a question.\nSpeaker A: Hi.\nSpeaker A: So there is no system, right?\nSpeaker A: There was a wizard for both devices.\nSpeaker C: Yeah, it was both times the same person.\nSpeaker C: One time pretending to be a system, one time pretending to be a user, which is actually not pretending.\nSpeaker A: And she didn't...\nSpeaker A: I mean, isn't this kind of obvious when it says, okay, now you're talking to a human and the human is the same voice?\nSpeaker C: No, no, no, you wait.\nSpeaker C: Okay, good question, but you just wait and see.\nSpeaker C: You're going to learn.\nSpeaker C: And the wizard sometimes will not be audible because she was actually...\nSpeaker C: There were some laps in the wireless.\nSpeaker C: We have to move her closer.\nSpeaker H: We'll be deciding on going to the concert or the Hendobergs, or the Hendobergs, especially the orchestra.\nSpeaker H: I want to buy some elegant, a wireless redesign or new piece, which is located on February's rebirth.\nSpeaker H: Out.\nSpeaker G: In order to get to the Friedrich-erbott-Anlanger, we're going around and go towards the church.\nSpeaker G: Next, you have to add the church into the Friedrich-erbott-Anlanger.\nSpeaker B: Is she mispronouncing Anlaga?\nSpeaker B: Is it Anlaga or Anlanger?\nSpeaker C: They're mispronouncing everything but the church.\nSpeaker G: And some can castle differs from the regular sort of castle in that it is built in a flat region rather than on a mountain.\nSpeaker H: Oh, I don't know if you could tell me about the stories of Fidi and the old Stunian prison.\nSpeaker H: And I want to know how to get there.\nSpeaker G: In order to get to the Stunian prison from the University Square, turn right and walk one block.\nSpeaker C: This is the sister-breaking doughnut, actually.\nSpeaker G: Hello?\nNone: Yes.\nSpeaker G: I'm sorry, something seems to have happened to the system.\nNone: See if I can help you.\nSpeaker H: Are you a live person?\nSpeaker H: I'm a live person.\nSpeaker H: Okay.\nSpeaker G: No, I'm a live person.\nSpeaker G: I need to know where you are, first of all.\nNone: Where am I?\nSpeaker H: I don't know.\nSpeaker G: See, you're at Aus?\nSpeaker H: I see a building that's under construction.\nSpeaker G: Okay.\nSpeaker G: That's probably the University Square.\nSpeaker G: Okay. So what do you need to do?\nSpeaker H: Can you give me a description of the powder tower of the Hengelburn Castle?\nSpeaker G: Yeah, in order to get there you have to take a bus.\nSpeaker G: You've got to take the number 42 to the Town Hall station.\nSpeaker G: And then you have to transfer to the Venezuelan train and take the castle exit and turn right.\nSpeaker G: And after about half a mile you'll see the powder tower on your right hand side.\nNone: Okay.\nSpeaker G: How much is it for the train?\nSpeaker G: You can get a transfer on the bus for the train for about four miles.\nSpeaker G: Okay. That's all I need to do?\nSpeaker G: Sure. Bye-bye.\nSpeaker H: I don't know, recording.\nSpeaker H: I heard that there was a recording that you could go with the thing.\nSpeaker H: You could use the right hand.\nSpeaker H: Did I call Europe?\nSpeaker C: Did I call Europe?\nSpeaker C: This is it.\nSpeaker C: So if we have heard Putin's recommendation,\nSpeaker E: maybe she had two hours later on when the first person was recording on us. So are you trying to record this meeting?\nSpeaker C: There was a strange reflex.\nSpeaker C: I have a headache. I'm really stiff out of it.\nSpeaker C: The lessons learned, the reading needs to be short of five minutes.\nSpeaker C: That was already anticipated by some people suggested that if we just have bullets here, they're going to not, there are subjects are probably not going to go into follow the order.\nSpeaker C: And she did not.\nSpeaker C: She jumped around quite a bit.\nSpeaker E: So if you just number them one, two, three, it's...\nSpeaker C: We need a better introduction for the wizard that is something that Faye actually thought of in the last second that the system should introduce itself when it's called.\nSpeaker C: And another suggestion by Liz was that we, through subjects, switched the tasks.\nSpeaker C: So when they have tasks one with a computer, the next person should have tasks one with a human and so forth.\nSpeaker C: So we get nice data for that.\nSpeaker C: We have to refine the tasks more and more, which of course we haven't done it all so far, in order to avoid this rephrasing.\nSpeaker C: So even though we don't tell the person ask blah, blah, blah, blah, blah, they still try, or these, the Tanya try to repeat.\nSpeaker C: It's much of that text as possible.\nSpeaker C: And my suggestion is, of course, we keep the wizard because I think she did a wonderful job.\nSpeaker C: Great.\nSpeaker C: In the sense that she responded quite nicely to things that were not asked for, how much is it passed ticket and the transfer.\nSpeaker C: So this is going to happen all the time.\nSpeaker C: You can never be sure.\nSpeaker C: Jono pointed out that we have maybe a grammatical gender problem there with wizard.\nSpeaker B: So I wasn't sure whether wizard was the correct term for not a man.\nSpeaker A: But there's no female equivalent of, are you sure?\nSpeaker B: No, I don't know.\nSpeaker C: Well, there is Richard Warlock.\nSpeaker C: Yes, I was thinking with.\nSpeaker C: Right.\nSpeaker C: And so some work needs to be done, but I think we can, and this, in case you hadn't seen it, this is what Littonia looked at during the year, while taking the data collection.\nSpeaker E: Okay, great.\nSpeaker E: So first of all, I agree that we should hire, say, and start paying her, probably pay her for the time she's already put in as well.\nSpeaker E: You know exactly how to do that, or is Lila, I mean, you know, what exactly do we do to put her on a payroll in some way?\nSpeaker C: I'm completely clueless.\nSpeaker E: Okay.\nSpeaker E: I would like to learn.\nSpeaker E: Well, you have to, right.\nSpeaker E: So anyways, so why don't you ask Lila and see what she says about, you know, exactly what we do for someone in the student type worker?\nSpeaker E: Well, yeah, she's not a student she just graduated, but anyway.\nSpeaker E: So if, yeah, I agree, she sounded fine.\nSpeaker E: She actually was more present and stuff than she was in conversation.\nSpeaker E: So she did a better job than I would have guessed from just talking to her.\nSpeaker E: So I think that's great.\nSpeaker C: This is sort of what I gave her.\nSpeaker C: So this is, for example, how to get to the student prison.\nSpeaker C: Yeah.\nSpeaker C: And I didn't student spell it out here.\nSpeaker C: And in some cases, I spell it out a little bit more thoroughly.\nSpeaker C: This is the information on the Sunk and Castle and the M5 theater that never came up.\nSpeaker C: And so if we give her even more instruments to work with, I think, the results are going to be...\nSpeaker C: Oh, yeah.\nSpeaker E: And of course, as she does it, she'll learn this.\nSpeaker E: So that's great.\nSpeaker E: And also if she's willing to take on the job of organizing all the subjects and stuff, that would be wonderful.\nSpeaker E: And she's actually... she's going to graduate school in kind of an experimental paradigm.\nSpeaker E: So I think this is all just fine in terms of her learning, things she's going to need to know to do her career.\nSpeaker E: So my guess is she'll be quite happy to take on that job.\nSpeaker C: Yeah, she did explicitly state that.\nSpeaker C: Great. And I told her that we're going to figure out a meeting time in the near future to refine the tasks and look for the potential sources to find people.\nSpeaker C: She also agrees that if it's all just going to be students, the data is going to be less valuable because of that.\nSpeaker E: Well, I was going to say there's this set of people next door.\nSpeaker C: It's not hard to...\nSpeaker C: However, we may run into a problem with a reading task there.\nSpeaker C: And we'll see.\nSpeaker E: Yeah, we could talk to the people who run it and see if they have a way that they could easily tell people that there's a task, a base ten box or something, but you have to be comfortable reading relatively complicated stuff.\nSpeaker E: And they'll probably be self-selection to some extent.\nSpeaker E: So that's good.\nSpeaker E: Now, I signed us up for the Wednesday slot and part of what we should do is this.\nSpeaker E: So my idea on that was partly we'll talk about system stuff for the computer scientists.\nSpeaker E: Apparently I did want it to get the linguists involved in some of this issue about what the task is and what the dialogue is and what's going on linguistically.\nSpeaker E: Because to the extent that we can get them contributing, that'll be good.\nSpeaker E: So this issue about reformulating things.\nSpeaker E: Maybe we can get some of the linguists sufficiently interested that they will help us with it.\nSpeaker E: Other linguists, if you're a linguist, but any case.\nSpeaker E: The linguistic students and stuff.\nSpeaker E: So my idea on Wednesday is partly to...\nSpeaker E: I mean, what you did today is just fine. You just do this as we did and here's the thing and here's some of the dialogue and so forth.\nSpeaker E: But then the other thing, of course, is we should give the computer scientists some idea of what's going on with the system design and where we think the belief nets fit in and where the pieces are and stuff like that.\nSpeaker E: Does this make sense to everybody?\nSpeaker E: So I don't think it's worth a lot of work, particularly on your part to make a big presentation.\nSpeaker E: I don't think you should make any new PowerPoint or anything.\nSpeaker E: I think we've got plenty of stuff to talk about.\nSpeaker E: And then just see how the discussion goes.\nSpeaker C: Sounds good.\nSpeaker C: The other two things is we have Jon Oat.\nSpeaker C: He'll tell us a little bit about this.\nSpeaker C: Great.\nSpeaker C: And we also have a little bit on the interface, M3L enhancement.\nSpeaker C: And then that was it, I think.\nSpeaker B: And what I did for this is a pedagogical belief net because I was trying to conceptually do what you're talking about with the nodes that you could expand out.\nSpeaker B: So what I did was I took, I made these dummy nodes called detector and introductory out that would isolate the things related to the detector and then the things with the source and the path and the goal.\nSpeaker B: And I separated them out. And then I did similar things with the context and the discourse and whatnot.\nSpeaker B: So we could sort of isolate them or whatever in terms of the top layer.\nSpeaker B: And then the bottom layer is just the mode.\nSpeaker E: So let's, yeah, I don't understand it. Let's go slide all the way up to we see what the very bottom looks like. Is that it?\nSpeaker B: There's just one more node and it says mode, which is the decision between the.\nSpeaker B: Okay, great. All right.\nSpeaker B: So basically all it was I took the last belief net.\nSpeaker B: And I group things according to what how I thought they would fit into image scheme is they would be related.\nSpeaker B: So I did a projector landmark and a source path goal as initial ones.\nSpeaker B: And then I said, well, the projector would be the person in this case probably.\nSpeaker B: You know, we have, we had the concept of what their intention was whether they were trying to tour or do business or whatever.\nSpeaker B: They were hurried that's kind of related to that.\nSpeaker B: And then in terms of the source, the things, the only things that we had on there, I believe were whether.\nSpeaker B: I kind of, I might have added these because I don't think we talked too much about the source and the old one, but whether the where I'm currently at is a landmark might have a bearing on whether or the landmark is the where I currently at.\nSpeaker B: And usefulness is basically means is it an institutional facility like a town hall or something like that that's not something that you would visit for tourists tourism, say, or whatever travel constraints would be something like, you know, maybe they said they can they only want to take a bus or something like that.\nSpeaker B: So that would determine whether we could take, we would be telling to go to the bus stop or versus walking there directly.\nSpeaker B: Goal similar things as the source, we also added whether the entity was closed and whether they have somehow marked that it was the final destination.\nSpeaker B: And then if you go up.\nSpeaker B: Yeah, so in terms of context, what we currently said was whether they were a businessman or a tourist or some other person.\nSpeaker B: This course was related to whether they'd asked about open hours or whether they asked about where the entrance was or the admission fee or something along those lines.\nSpeaker B: I'm not really sure what the prosody means in this context, so I just made up, you know, whether what they say is or how they say it is that the parse would be what verb they chose and then maybe how they modified it in the sense of whether they said I need to get here quickly or whatever.\nSpeaker B: And in terms of world knowledge is just basically be like opening and closing times of things, the time of day it is and whatnot.\nSpeaker B: Tourbook that would be I don't know the landmarkiness of things.\nSpeaker B: Whether it's in the tourbook or not.\nSpeaker E: So I understand what you got. I don't yet understand how you would use it. So let me see if I can ask.\nSpeaker B: This is not a working Bayes net.\nSpeaker E: Now I understand that, but so what let's slide back up again and see it started at the bottom.\nSpeaker E: So you could imagine.\nSpeaker E: Go ahead, you were about to go up there and point of center.\nSpeaker B: Well, okay, just say what you're going to say.\nSpeaker B: I was going to wait until.\nSpeaker E: So if we want to make it into a real Bayes net that is, you know, we've filled, you know, actually fill it in, then.\nSpeaker B: So we have to get rid of this and connect these things directly to the mode.\nSpeaker E: Well, I don't that's an issue.\nSpeaker B: Because I don't understand how it would work otherwise.\nSpeaker E: Well, here's the problem.\nSpeaker E: And the bus guy I was talking about this a little earlier today is if we just do this, we could wind up with a huge.\nSpeaker E: Commentator input to the mode thing.\nSpeaker E: And.\nSpeaker B: Well, yeah, I understand that.\nSpeaker B: I just it's hard for me to imagine how we could get around that.\nSpeaker E: So that's what we have to do.\nNone: Okay.\nSpeaker E: Okay.\nNone: So.\nSpeaker E: There are a variety of ways of doing it.\nSpeaker E: Let me just mention something that I don't want to pursue today, which is there are technical ways of doing it.\nSpeaker E: I slipped the paper to buscarer about noisy ors and noisy maxes.\nSpeaker E: And there are ways to sort of back off on the purity of your Bayes net in this.\nSpeaker E: So if you could have met and now I don't know that any of those actually apply in this case.\nSpeaker E: But there is some technology you could try to apply.\nSpeaker B: So it's possible that we could do something like a summary note of some sort.\nSpeaker B: Yeah.\nSpeaker E: Okay.\nNone: Yeah.\nSpeaker E: And.\nSpeaker E: So.\nSpeaker B: So in that case, the summer we'd have we I mean these wouldn't be the summer knows we'd have summer knows like.\nSpeaker B: Whether things were I guess maybe think if things were related to business or.\nSpeaker E: So what I was going to say is maybe a good idea at this point is to try to informally.\nSpeaker E: I mean not necessarily in this meeting, but to try to informally think about what the decision variables are.\nSpeaker E: So if you had some bottom line decision about which mode, you know what are the most relevant things.\nSpeaker E: And the other trick, which is not a technical trick, it's kind of an all engineering trick is to make the.\nSpeaker E: Each node sufficiently narrow that you don't get this comment of torque.\nSpeaker E: So that if you decided that you could characterize.\nSpeaker E: The decision is a trade off between three factors, whatever they may be.\nSpeaker E: Okay, then you could say, aha, let's have these three factors.\nSpeaker E: Okay, and maybe a binary version for each or some relatively compact decision node just above the final one.\nSpeaker E: And then the question would be if those are the things that you care about.\nSpeaker E: Can you make a relatively compact way of getting from the various inputs to the things you care about.\nSpeaker E: So that you know you can sort of try to do a knowledge engineering thing given that we're not going to screw with the.\nSpeaker E: Technology and just always use sort of orthodox base nets.\nSpeaker E: Then we have a knowledge engineering problem of how do we do that.\nSpeaker E: And.\nSpeaker B: So what I kind of need to do is take this one and the old one and merge them together.\nSpeaker E: Well, something.\nSpeaker E: I mean, so Robert has thought about this problem for a long time because he's had these examples kicking around.\nSpeaker E: So he may have some good intuition about, you know, what are the crucial things.\nSpeaker E: And I understand where this, the.\nSpeaker E: This is a way of playing with this source of ethical trajectory abstraction and sort of displaying it in a particular way.\nSpeaker E: I don't think our friends on Wednesday are going to be able to.\nSpeaker E: Well, maybe they will let me think about whether whether I think we could present this to them or not.\nSpeaker C: Well, I think this is still I mean, I talk this is sort of the second version. And I look at this maybe just as a, you know, whatever you am, a diagram or, you know, just a screenshot, not really as based on.\nSpeaker B: We could actually get drawn a different way in the sense that we make it more abstract.\nSpeaker C: But the the nice thing is that you know, just is a visual aid for thinking about these things which has clearly have to be specified more carefully.\nSpeaker E: So, sorry, well, let me think about this more and see if we can find a way to present this to this language group that is helpful to them.\nSpeaker C: So, ultimately we may, we regard this as sort of an exercise in thinking about the problem and maybe a first version of a module, if you want to call it that, that you can ask, that you can give input and it will throw the dice for you, throw the die for you because I integrated this into the existing smart com system.\nSpeaker C: In the same way as much the same way we can sort of have this thing.\nSpeaker C: So, if this is what m3l will look like and what it will give us, the very simple thing we have an action that he wants to go from somewhere, which is some type of object to someplace.\nSpeaker C: And this change now only, it's doing it twice now because it already did it once, we'll add some action type, which in this case is approach and could be more refined in many ways or we can have something where the goal is a public place.\nSpeaker C: And it will give us then of course an action type of the type enter. So, this is just based on this one feature and that's about all you can do.\nSpeaker C: So, if this play, if the object type here is a landmark, of course it will be vista.\nSpeaker C: And this is about as much as we can do if we don't, if we want to avoid a huge combinatorial explosion where we specify, okay, if it's this and this but that is not the case and so forth, it just gets really really messy.\nSpeaker E: So, it was much too quick for me. Okay, so let me see if I understand what you're saying. So, I do understand that you can take the m3l and add not, and you need to do this for sure.\nSpeaker E: You have to add, you know, not too much about object types and stuff. And what I think you did is add some rules of the style that are already there that say if it's of type landmark, then you take, you're going to take a picture of it.\nSpeaker E: Exactly. Full stop. I mean, that's what you do. Every landmark, you take a picture of. Every public face, you enter. Every statue, you want to go as nearest possible. You approach.\nSpeaker E: And certainly you can add rules like that to the existing smartcom system and just did.\nSpeaker E: What do we know about that? Well, so, well, let's think about this. That's another kind of baseline case. It's another sort of thing. Okay, here's another kind of minimal way of tackling this.\nSpeaker E: Add extra properties, a deterministic rule for every property. You have an action. You do that.\nSpeaker E: Then the question would be, now, if that's all you're doing, then you can get the types from the ontology. Okay, because you're all you're using is this type of types in the ontology and you're done.\nSpeaker E: We don't use the discourse. We don't use the context. We don't do any of those things. That's okay. Again, one minimal extension of the existing things.\nSpeaker E: And that's something the smartcom people themselves with. That's a sure that's no problem. No problem to add types to the right.\nSpeaker C: And this is just in order to exemplify what we can do very, very easily is we have this silly interface. And we have the rules that are as banal as we just saw.\nSpeaker C: And we have our content. Now, the content, which is sort of what we see here, which is sort of the Mr. Schema, Stutt, Pascal, whatever.\nSpeaker C: This will be a job to find ways of writing down image schema, X schema constructions in some form and have this be in the content, loosely called constructicon.\nSpeaker C: And the rules we want to throw away completely. And here is exactly where what's going to be replaced with our base net, which is exactly getting the input feeding it here. This decides whether it's in whether action, the enter the vista or the whatever approach you call it.\nSpeaker C: That's what you said. Yeah, that's fine. Construction should be active rated, but it's not construction there. It's action. Construction is a different story.\nSpeaker E: This is a, to what we'd be generating would be a reference to a semantic like parameters for the, for the X schema. Yes. Okay. Yeah. So that that if you had the generalized go X schema, and you want to specialize into these three ones, then you'd have to supply the parameters.\nSpeaker E: And then, although we haven't worried about this yet, you might want to worry about something that would go to the GIS and use that to actually get detailed route planning.\nSpeaker E: So, you know, where do you go to take a picture of it and stuff like that. But that's not, it's not the immediate problem.\nSpeaker E: But presumably that's that that functional is there when, when the media problems just deciding which aspects of the X schema to add.\nSpeaker E: Yeah. So the problem, the media problem is is back to what you were, what you were doing with the belief net. You know, what are we going to use to make this decision?\nSpeaker E: Right. And then once we've made the decision, how do we put that into the content? Right. Well, that, that actually is relatively easy in this case.\nSpeaker E: The harder problem is we decide what we want to use. How do we going to get it? And that, that's the hardest problem.\nSpeaker E: So the hardest problem is how are you going to get this information from some combination of the what the person says in the context and the ontology.\nSpeaker E: So I think that's the hardest problem at the moment is, is where you're going to get this information.\nSpeaker E: And that's so getting back to here. We have a technical problem with the belief nets that we don't want all the factors.\nSpeaker E: Too many factors if we allow them just go combinatorially. So we want to think about which ones we really care about and what they really most depend on. And can we clean this, this up to the point where.\nSpeaker B: So what we really want to do is, because this is really just a three layer that we want to make it expanded out into more layers.\nSpeaker E: We might. I mean, that that's certainly one thing we can do. It's true that the way you have this, a lot of the times you have what you're having is the values rather than the variable.\nSpeaker E: So, right.\nSpeaker B: Okay. So it should really be just be intention as a node instead of intention business or intent.\nSpeaker E: Yeah. Right. And then it would have values to our business or heard. But then it's still some knowledge designed to do about how do you want to break this up? What really matters.\nSpeaker B: I mean, it's fine. We have this iterative. We're going to have to work with it some. I think what was going through my mind when I did it was someone could both have a business intention and a touring intention and the probabilities of both of them happening.\nSpeaker E: Well, you could do that. And it's perfectly okay to insist that you know, they add up to one. But that there's that it doesn't have to 100.\nSpeaker E: Okay. So you could have the conditional. So the each of these things is going to be a probability. So whenever there's a choice.\nSpeaker E: So like landmarkness and usefulness. I don't think those would be mutually. Seems like something could both be absolutely right.\nSpeaker E: And so that you might want to then have those then they may have to be separate. They may not be able to be values of the same very. So that's bidding and this is this is the sort of knowledge design you have to go through.\nSpeaker B: This is just more me taking the last the last one. Right. How can I pull it? It's great. It's is you know, it's one step toward.\nSpeaker E: To where we want to go.\nSpeaker C: It also strikes me that we we may want to approach the point where we can sort of try to find the specification for some interface here that takes the normal m3L looks at it.\nSpeaker C: And then we discussed in our pre-edu meeting how to ask the ontology what to ask the ontology.\nSpeaker C: The fact that we can pretend we have one make a dummy until we get the real one. And so we we way want to decide we can do this from here.\nSpeaker C: But we also could do it. You know if we have a belief net interface. So the belief net takes as input a vector right of stuff.\nSpeaker B: And it.\nSpeaker C: Yeah. And the output is whatever as well. But this information is just m3L and then we want to look up some more stuff in the ontology and we want to look up some more stuff in the maybe want to ask the real world.\nSpeaker C: Maybe want to look some up in the D.I.S. But also we definitely want to look up in the dialogue history. Some some stuff based on we we have.\nSpeaker C: I was just made some examples from the ontology. And so we have for example information there that the town hall is both a building.\nSpeaker C: And it has doors and stuff like this. It is also institutions. So it has a mayor and so forth and so forth. We get these relations out of it. And once we have them we can use that information to look in the dialogue history.\nSpeaker C: Were any of these things that are part of the town hall as an institution mentioned were any of these that make the town hall a building mentioned and so forth.\nSpeaker C: And maybe draw some inferences on that. So this may be a sort of a process of two to three steps before we get our vector that we feed into the belief net.\nSpeaker E: Yeah, I think that's I think that's exactly right. There will be rules that they aren't rules that come to final decisions.\nSpeaker E: There are rules that gather information for a decision process. Yeah, I think that's that's just fine. Yeah, so they'll presumably they'll be a threat of process or something that agent.\nSpeaker E: What everyone want to say, yeah, that is rule driven and can can can do things like that. And there's an issue about whether there will be that'll be the same agent is the one that then goes off and carries out the decisions. It probably will.\nSpeaker E: My guess is it'll be the same basic agent that can go off and get information running through this belief net that turn the crank into belief net that'll come out with more of another vector, okay, which can then be applied at what we would call the simulation or action end.\nSpeaker E: We now know what you're going to do and that may actually involve getting more information. So once you pull that out, it could be that that says, ah, now that we know that we're going to go ask the ontology something else.\nSpeaker E: Now that we know that it's a bus trip, okay, we didn't we didn't need to know beforehand how long the bus trip takes or whatever, but but now that we know that's the way it's coming out, then we got to go.\nSpeaker E: Find out more so I think that's okay.\nSpeaker C: So this is actually if we were to build something that is.\nSpeaker C: It needs to do. Yeah, I think we I can come up with a code for a module that we call the cognitive dispatcher, okay, which does nothing but it looks a complex object trees and decides how there's parts missing that need to be filled out.\nSpeaker C: There is this is maybe something that this module can do something that this module can do and then collect sub objects and then recombined them and put them together. So maybe this is actually some useful tool that we can use to rewrite it.\nSpeaker E: Okay, I confess I'm still not completely comfortable with the overall story.\nSpeaker E: I'm just not a complaint. This is a promise to do more work. So I'm going to have to think about it some more in particular. See what we'd like to do and this has been implicit in the discussion is to do this in such a way that you get a lot of reuse.\nSpeaker E: What you're trying to get out of this deep cognitive linguistics is the fact that if you know about source, so it's pathas and goals and all this sort of stuff that a lot of this is the same for different tasks and that there's some important generalities that you're getting so that you don't take each and every one of these tasks and have to redo it.\nSpeaker E: And I don't yet see how that goes.\nSpeaker E: What are the primitives and how do you bring this? I mean, just saying, I know how to do any individual case. Right, but I don't yet see what's the really interesting question is can you use a deep cognitive linguistics to get powerful generalizations.\nSpeaker C: And should we add them the what's this domain? How do I get to X? Then we also have the what's this domain? Could we get some slightly different.\nSpeaker C: And John actually does not allow us to call them intentions anymore. So he dislikes the term. Well, I don't like the term either. So I have.\nSpeaker E: But I'm sure the what is questions also create some interesting esquema aspects could be. I'm not I'm not particularly opposed to adding that or any other task. I mean, eventually we're going to want a whole range of them.\nSpeaker E: I just saying that I'm going to have to do some sort of first principles thinking about this. I just at the moment don't know.\nSpeaker E: Do you mean in terms of the way we can generalize the base. No, no, the base the base. The base. The base. That's what we just specific for each decision. But what I'd like to be able to do is to have the way that you extract properties that will go in the different base nets be general.\nSpeaker E: So that if you have sources of trajectors and stuff like that, and there's language for talking about trajectors, you shouldn't have to do that differently. For going to something then for circling it for telling someone else how to go there, whatever it is.\nSpeaker E: The decision processes are going to be different. What you really like of course is the standard thing you'd always like, which is that you have a kind of intermediate representation, which looks the same over a bunch of inputs and a bunch of outputs.\nSpeaker E: So all sorts of different tasks and all sorts of different ways of expressing them use a lot of the same mechanism for pulling out what are the fundamental things going on.\nSpeaker E: That would be the really pretty result. And pushing it one step further, when you get the construction grammar and stuff, what you'd like to be able to do is say you have this parser, which is much fancier than the parser that comes with smart com that actually uses constructions and is able to tell from this construction that there's something about the intent, you know, the actual what people want to do or what they're referring to and stuff.\nSpeaker E: So the independent of whether about what is this or where is it or something that you could tell from the construction, you could pull out deep semantic information, which you're going to use in a general way.\nSpeaker E: You might be able to say that this is the kind of construction in which the construction implies that this thing is being viewed as a container.\nSpeaker E: So just from this local construction, you know that you're going to have to treat it as a container and you might as well go off and get that information and that may affect the way you process everything else.\nSpeaker E: So you say how do I get into the castle.\nSpeaker E: Or, you know, what is there in the castle or so there's all sorts of things you might ask that involve the castle as a container and you'd like to have this orthogonal so that any time the castle is referred to as a container, you crank up the appropriate stuff.\nSpeaker E: Independent of what the goal is and independent of what the surrounding language is.\nSpeaker E: So that's the thesis level.\nSpeaker C: It's important also that English sort of get rid of most of its spatial adverbs because they're really fancy than for these kinds of analysis.\nSpeaker C: Well, they have a prepuditional phrases.\nSpeaker C: Yeah, but they're easier for parses.\nSpeaker C: Right.\nSpeaker C: Parses can pick those up with the spatial adverbs.\nSpeaker C: They have a tough time because the mathematics are very complex and that's right.\nSpeaker C: Okay.\nSpeaker C: I don't remember.\nSpeaker C: Just put that it again.\nSpeaker C: Oh, yeah, but an architecture like this would also enable us maybe to throw this way and replace it with something else or whatever so that we have.\nSpeaker C: So this is sort of the representational formats where we're talking about that are independent of the problem that generalize over those problems and are of a higher quality than any actual whatever.\nSpeaker C: Leave net or X that we may use for the decision making ultimately.\nSpeaker C: Should be decoupled.\nSpeaker E: Okay.\nSpeaker E: So are we going to be meeting here from now on?\nSpeaker E: I'm happy to do that.\nSpeaker E: We had talked about it because you have the display and everything.\nSpeaker E: That seems fine.\nSpeaker C: Yeah, and this also asks whether we're going to have presentations every time.\nSpeaker C: I don't think we will need to do that, but it's so far.\nSpeaker C: I think it was nice as a visual aid for some things.\nSpeaker E: Oh, yeah, no, I think it's worth it to meet here to bring this and assume that something may come up that we want to look at.\nSpeaker E: Yeah, why not?\nSpeaker C: And...\nSpeaker B: I think it would be presentation.\nSpeaker B: She was good.\nSpeaker E: She was good.\nSpeaker E: Latanya was good.\nSpeaker C: She was definitely good in the sense that she showed us some of the weaknesses.\nSpeaker C: Right.\nSpeaker C: And also the fact that she was a real subject.\nSpeaker C: Right.\nSpeaker E: And she took it seriously and...\nSpeaker E: No, I was great.\nSpeaker C: Yeah.\nSpeaker C: So I think that...\nSpeaker C: I mean, just looking at this data, listening to it, what can we get out of it in terms of our problem?\nSpeaker C: For example, she actually said...\nSpeaker C: She never spoke about entering.\nSpeaker C: She just wanted to get some place and she said for buying stuff.\nSpeaker C: So this is definitely interesting.\nSpeaker C: Yeah.\nSpeaker C: And in the other case, where she wanted to look at that stuff at the graffiti.\nSpeaker C: Also, of course, not in the sense how do you get there was pretty standard, except that there was nice enough for pointing at what she talked about before.\nSpeaker C: And there she was talking about looking at pictures that were painted inside a wall on walls.\nSpeaker C: Actually, you need a lot of word knowledge.\nSpeaker C: This would have been a classical tango, actually, because graffiti is usually found on the outside, not on the inside, but...\nSpeaker C: Okay.\nSpeaker C: So the mistake would have made a mistake.\nSpeaker D: This is the movie made a mistake.\nSpeaker D: Click.\nSpeaker D: Right.\nNone: Thanks for watching.\nNone: Thanks for watching.\n"}, {"meeting_id": "Bmr006", "summary": "In this meeting, the speakers addressed issues that related to their meeting annotation progress. They talked about the types and variability of the future meetings. This led to them speaking about how to use their recording equipment if they got groups outside of their own departments to record meetings. Additionally, they discussed how to handle the storage of meetings on disk. They eventually address how to handle speech overlap and speaker anonymity, which were two crucial issues.", "dialogue": "Speaker F: Okay, now we're on it. It seems to be working.\nSpeaker E: So it seems like it's been sitting for a long time.\nSpeaker F: I don't know what it is, but all I know is that it seems like every time I am up here after a meeting and I start it, it works fine.\nSpeaker F: And if I'm up here and I start it and we're all sitting here waiting to have a meeting, it gives me that error message and I have not yet sat down with being able to get that error message at a point where I can sit down and find out where it's occurring in the code.\nSpeaker F: Yeah, we will.\nSpeaker F: One of these days.\nSpeaker F: Was it on pause or something?\nSpeaker D: No.\nSpeaker D: So the new procedural change that Scott suggested, I think is a good idea, is that we do the digit recording at the end.\nSpeaker D: And that way, if we're recording somebody else's meeting and the number of participants have to run off to some other meeting who don't have the time, then they can run off.\nSpeaker D: Then we'll get somewhat fewer sets of digits, but I think that way we'll cut into people's time.\nSpeaker D: There's someone's on strict time.\nSpeaker D: Less.\nSpeaker D: So I think we should start doing that.\nSpeaker D: So let's see, we were having discussion the other day.\nSpeaker D: I mean, we should bring that up.\nSpeaker D: The nature of the data that we're collecting.\nSpeaker D: That we should have a fair amount of data that is collected for the same meeting so that we can.\nSpeaker D: I know.\nSpeaker D: What was on the point again about that?\nNone: Well, okay.\nSpeaker A: I'll back up.\nSpeaker A: Yeah.\nSpeaker A: At the previous, at last week's meeting, this meeting, I was brave.\nNone: I didn't have about wanting to get more data.\nSpeaker A: I talked about this with Jane and Adam, and was thinking of this mostly just so that we could do research on this data since we'll have a new student, does want to work with us.\nSpeaker A: That was at the last meeting.\nSpeaker A: Great.\nSpeaker A: And he's already funded part time, so we'll be paying him only for half of the normal part time.\nSpeaker F: What did he?\nSpeaker F: Yeah.\nSpeaker F: And what's he interested in specifically?\nSpeaker A: He comes from a signal processing background, but I like to unlock because he's very interested in higher level things like language and disfluencies and all kinds of, maybe, positive.\nSpeaker A: So he's just getting his feedback in that.\nSpeaker A: Anyway, I thought, okay, maybe we should have enough data so that he starts, he'd be starting in January next semester, that we'd have enough data to work with.\nSpeaker A: But Jane and Adam brought up a lot of good points that just posting a note to Berkeley\nNone: people that have them come down here has some problems in that. We need to make sure that the speakers are who you want and that the meeting type is what you want and so forth.\nSpeaker A: So I thought about that and I think it's still possible.\nSpeaker A: But I'd rather try to get more regular meetings of types that we know about in here than sort of a mishmash of a bunch of one time.\nSpeaker A: Yeah, just because it would be very hard to process the data in all senses, both to get the figure out what type of meeting it is and to do any kind of higher level work on it like while I was talking to Morgan about things like summarization or what's this meeting about.\nSpeaker A: I mean, it's very different if you have a group that's just giving a report on what they did that week versus coming to a decision and so forth.\nSpeaker A: So then I was talking to Morgan about some new proposed work in this area sort of a separate issue from one student would be working on where I was thinking of doing some kind of summarization of meetings or trying to find cues in both the utterances and in the utterance patterns like in numbers of overlaps and amount of speech sort of raw cues from the interaction that can be measured from the signals and from the different microphones that point to\nNone: sort of hotspots in the meeting or things where stuff is going on that might be important for someone who didn't attend to listen to.\nSpeaker A: And in that regard, I thought we definitely will need, it'd be nice for us to have a bunch of data from a few different domains or a few different kinds of meetings.\nSpeaker A: So this meeting is one of them, although I'm not sure I could participate if I would feel very strange being part of a meeting that you were then analyzing later for things like summarization.\nSpeaker A: And then there are some others that Morgan mentioned, like the front end meeting, maybe a networking group meeting.\nSpeaker F: We're hoping that they'll let us start recording regularly.\nSpeaker A: So if that were the case, then I think we'd have enough.\nSpeaker A: But basically for anything where you're trying to get a summarization or some kind of meeting out of the meeting, it would be too hard to have 50 different kinds of meetings where we didn't really have a good grasp of what does it mean to summarize, but rather we should have different meetings by the same group, but hopefully that have different summaries.\nSpeaker A: And then we need a couple of, we don't want to just have one group because that might be very specific to that particular group.\nNone: But three more times.\nSpeaker F: Here we have a overlap between this meeting and the morning meeting.\nSpeaker A: See, that I've never listened to the data for the front end meeting.\nSpeaker F: We've only had three.\nSpeaker A: But maybe that's enough.\nSpeaker A: So in general, I was thinking more data, but also data where we hold some parameters, constant or fairly similar like a meeting about people doing a certain kind of work where at least have to participate in each time or the same.\nSpeaker D: Now let me just give the other side to that because I don't disagree with that.\nSpeaker D: But I think there is a complimentary piece to it too.\nSpeaker D: For other kinds of research, particularly the acoustic oriented research, I actually feel the opposite need.\nSpeaker D: I'd like to have lots of different people.\nSpeaker D: As many people here and talking about the kind of thing that you're just talking about, it would have too few people from my point of view.\nSpeaker D: I'd like to have many different speakers.\nSpeaker D: So I think I would also very much like us to have a fair amount of really random scattered meetings or somebody coming down from campus.\nSpeaker D: And I mean, sure, if we can get more from them fine, but if we only get one or two from each group, it still could be useful acoustically just because we have close and distant microphones with different people.\nSpeaker B: Okay, can I say about that?\nSpeaker B: The issues that I think Adam and I raised were more a matter of advertising so that you get more native speakers.\nSpeaker B: Because I think if you just say, and in particular, my suggestion was to advertise to linguistics grad students because there you have people who would have proficiency enough in English that it would be useful for purposes.\nSpeaker B: But I think I've gathered data from undergrad students at an on campus.\nSpeaker B: You just post randomly to undergrad students.\nSpeaker B: I think you'd get such a mix back.\nSpeaker B: It would be hard to know how much conversation you'd have at all.\nSpeaker B: Well, you want- The English you'd have with the language models would be really hard to build because it would not really be- it would be an inner language with that.\nSpeaker D: Well, okay.\nSpeaker D: First place, I don't think we just want to have random people come down and talk to one another.\nSpeaker D: There should be a meeting that has some goal and point because I think that's what we're\nSpeaker A: investigating. It needs to be a pre-existing meeting.\nSpeaker A: Yeah.\nSpeaker A: Right.\nSpeaker A: That would always happen.\nSpeaker D: So I was thinking more in terms of talking to professors and senior doctoral students who are leading projects and offering to them that they have the whole their meeting down here.\nSpeaker D: That's the first point.\nSpeaker D: One point is I think that for some time now going back through birth, I think that we have had speakers that we've worked with who had non-native access.\nSpeaker B: Oh, I'm not saying access.\nSpeaker B: I think that- The access is not the problem.\nSpeaker B: Okay.\nSpeaker B: No, it's more a matter of proficiency, just simply fluency.\nSpeaker B: I mean, ideal for people on campus who I think sometimes people undergraduates in computer science have language skills that make, you know, they're balancing the writing skills.\nSpeaker B: Oh, you're not talking about the poor language at all.\nSpeaker B: Yeah, I just think- I just think- But, you know, it's like when you get into the graduate level, no problem.\nSpeaker B: I mean, I'm not saying access.\nSpeaker B: Yeah, that's what we're saying.\nSpeaker B: It's the same.\nSpeaker D: It's the same.\nSpeaker D: It's the same.\nSpeaker D: So that the habits are already burnt in.\nSpeaker F: Well, I think that- I think the only thing that we should say in the advertisement is that the meeting should be held in English.\nSpeaker F: Yeah.\nSpeaker F: And I think if it's a pre-existing meeting and it's held in English, I think it's probably okay if a few of the people don't have particularly good English skills.\nSpeaker F: Can I-\nSpeaker B: Can I say the other aspect of this from my perspective, which is that there's this- This issue you have a corpus out there it should be used for- for multiple things because it's so expensive to put together.\nSpeaker B: Right.\nSpeaker B: And if people want to approach- So I know- you know, this is the idea of computational linguistics and probabilistic grammars and all may not be the folks in this group.\nSpeaker B: But the idea of language models, which are fun, you know, generally speaking, you know, in terms of like the amount of benefit per dollar spent on our invested in preparing the data, if you have a choice between people who are more proficient in- more fluent, more close to being academic English, then it would seem to be a good thing.\nSpeaker D: I guess- Maybe.\nSpeaker B: Because otherwise you don't have the ability to have- so if you have a bunch of media like- that's the worst possible case.\nSpeaker B: If you have people who are using English as an interlanguage, because they don't- they can't speak in their native languages, but their interlanguage isn't really a match to any existing language model.\nSpeaker B: This is the worst case scenario.\nSpeaker D: Well, that's pretty much what you're going to have in the networking group.\nSpeaker D: Right?\nSpeaker D: Because the network group is almost entirely Germans and Spaniards.\nSpeaker B: But the thing is, I think that these people are of high enough level in their language proficiency.\nSpeaker B: And I'm not objecting to accents.\nSpeaker B: I'm just thinking that we have to think at a higher level view, could we have a language model, a grammar, a grammar basically, that would be a possibility?\nSpeaker B: So if you wanted to bring in a model like Dandjer asks you to model and do some top-down stuff to help up the bottom of the merge of the things or whatever, it seems like I don't see that there's an argument.\nSpeaker B: What I think is that why not have the corpus since it's so expensive to put together useful for the widest range of central corp things that people generally use corp-brough for and which are used in computational linguistics.\nSpeaker B: That's my point.\nSpeaker B: Okay.\nSpeaker B: So you include both top-down and bottom-up.\nSpeaker B: Okay.\nSpeaker D: Well, let's see what we can get.\nSpeaker D: I mean, I think that if we're aiming at groups of graduate students and professors, if they were talking about things together, and it's from the Berkeley campus, probably most of it.\nSpeaker D: Exactly.\nSpeaker B: And my point in my note to Liz was, I think, under graduate, are any iffy pocket-related for me to agree with that?\nSpeaker A: I mean, for this person.\nSpeaker F: Well, not to mention the fact that I would be hesitant, certainly, to take anyone under 18, probably even in anyone under 21.\nSpeaker F: So, what's that?\nSpeaker F: What's that?\nSpeaker F: Well, the 18 is because of the consent form.\nSpeaker F: We have to get to find their parent to sign for them.\nSpeaker A: That's true.\nSpeaker A: Yeah.\nSpeaker A: I have a question.\nSpeaker A: Well, Morgan, you were mentioning that Murray may not use the equipment from IBM if they found something else because...\nSpeaker D: Yeah, they're assessing whether they should do that or you do something else, hopefully, with the next few weeks.\nSpeaker A: Because I mean, one remote possibility is that if we inherited that equipment, if she weren't using it, could we set up a room in the linguistics department?\nSpeaker A: And maybe a lot more or in psych, or wherever, in another building where we could record people there.\nSpeaker A: I think we'd have a better chance with it.\nSpeaker F: I think we'd need a real motivated partner to do that.\nSpeaker A: Right.\nSpeaker A: We need to find someone on campus who is interested in this.\nSpeaker A: If there were such a...\nSpeaker A: I mean, it's a remote possibility, then one of us could go out there and record the media or something, rather than bring all of them down here.\nSpeaker A: So this is the...\nSpeaker A: Well, the other thing...\nSpeaker A: The end of not using.\nSpeaker D: Yeah, and the other thing that I was hoping to do in the first place was to turn it into some kind of portable thing.\nSpeaker D: Right.\nSpeaker D: So you could wheel it around.\nNone: But...\nSpeaker F: I know that space is really scarce on at least NCS, to actually find a room that we could use regularly might be very difficult.\nSpeaker A: You may not need a separate room.\nSpeaker A: That's true.\nSpeaker A: You know, if they have a meeting room and they can guarantee that the equipment will be safe and so forth, and if one of us is up there to record the meeting once a week,\nSpeaker D: or something. Yeah.\nSpeaker D: Well, maybe Janu, that is pretty good for now with the other person.\nSpeaker A: Yeah.\nSpeaker A: I think it's not out of the question.\nSpeaker F: Yeah.\nSpeaker F: Yeah, I think it would be interesting because then we could regularly get another meeting, another type of meeting.\nSpeaker F: Right.\nSpeaker C: Right.\nSpeaker C: I think you need another portable team, another portable equipment to do...\nSpeaker C: More easier the recording process out from next thing.\nSpeaker C: Right.\nSpeaker C: And probably I don't know.\nSpeaker C: If you want to record a seminar or a class in the university, it could be very difficult to put a lot of headphones.\nSpeaker C: Yeah, but...\nSpeaker C: If you want to record a recording with this kind of...\nSpeaker F: If we want to just record with the tabletop microphones, that's easy.\nSpeaker F: Right.\nSpeaker F: That's very easy, but that's not the corpus that we're collecting.\nSpeaker D: Actually, that's an interesting point that came up in our discussions.\nSpeaker D: Maybe we're through a meeting.\nSpeaker D: We realized that when we're talking about this, that, okay, there's these different things we want to do with it.\nSpeaker D: So it's true that we want to be selective in some ways, the way that you're speaking about with not having an interlingua and these other issues.\nSpeaker D: But on the other hand, it's not necessarily true that we need all of the corpus to satisfy all of it.\nSpeaker D: So, as for the example, that we want to have a fair amount that's done with a small...\nSpeaker D: We recorded with a small...\nSpeaker D: Tight number of types of meetings.\nSpeaker D: But we can also have another part that's just one or two meetings of each of a range of them.\nSpeaker D: That's okay, too.\nSpeaker D: We realized in discussion that the other thing is, what about this business of distant and close microphones?\nSpeaker D: I mean, we really want to have a substantial amount recorded this way.\nSpeaker D: That's what we did it.\nSpeaker D: But what about for these issues of summarization, a lot of these higher level things, you don't really need the distant microphone.\nSpeaker A: Right.\nSpeaker A: I mean, I don't really need the close microphone.\nSpeaker A: You actually don't really need any.\nSpeaker A: You just need some microphone somewhere.\nSpeaker A: You can use a found data.\nSpeaker A: You can.\nSpeaker A: I mean, I can use a found data.\nSpeaker A: But I think that any data that we spend a lot of effort to collect, each person who's interested, we have a bunch of different slants and perspectives on what it's useful for, they need to be taking charge and making sure they're getting enough of the kind of data that they want.\nSpeaker A: Right.\nSpeaker A: And so, in my case, I think there's enough data for some kinds of projects and not enough of any other projects.\nSpeaker A: So I'm looking and thinking, well, I'd be glad to walk over and record people and go forward to help my interest.\nSpeaker A: And other people need to do that for themselves.\nSpeaker A: Right.\nSpeaker A: So that discusses that we can find some optimal.\nSpeaker D: But I think I'm raising that because I think it's relevant exactly for this idea up there that if you think about, well, gee, we have this really complicated setup to do.\nSpeaker D: Well, maybe you don't.\nSpeaker D: Maybe if really all you want is to have a recording that's good enough to get a transcription from later, you just need to grab a tape recorder and go up and make a recording.\nSpeaker D: I mean, we could have a fairly, we could just go to that machine.\nSpeaker A: Well, I agree with Jane, though on the other hand, that that may be true.\nSpeaker A: You may say, for instance, summarization or something that sounds very language oriented.\nSpeaker A: You may say, well, oh yeah, you just do that from transcripts of a radio show.\nSpeaker A: I mean, you don't even need to miss speech.\nSpeaker A: But what I was thinking is long-term, what would be needed to be able to pick up on, I suppose you just had a distant microphone there and you really wanted to be able to determine this.\nSpeaker A: There's lots of cues you're not going to have.\nSpeaker A: So I do think that long-term, you should always try to satisfy the greatest number of interest and have this parallel information, which is really what makes this special powerful.\nSpeaker A: Otherwise, you know, lots of other sites can propose.\nSpeaker D: I agree.\nSpeaker D: So.\nSpeaker D: But I think that the, we can't really underestimate the difficulty.\nSpeaker D: It shouldn't really underestimate the difficulty of getting a set of like this up.\nSpeaker D: And so it took quite a while to get that together and to say, oh, we'll just do it up there.\nSpeaker D: If you're talking about something simple, we throw away a lot of these dimensions, then you can do that right away.\nSpeaker D: I'm talking about something that has all of these different facets that we have here.\nSpeaker D: It won't happen quickly.\nSpeaker D: It won't be easy.\nSpeaker D: And there's all sorts of issues about keeping the equipment safe or else, hauling it around and all sorts of things.\nSpeaker A: So then we'll try to bring people here.\nSpeaker D: I think your first priority should be to try to get people to come here.\nSpeaker D: We're set up for it.\nSpeaker D: The room is really underused.\nSpeaker D: I thought the free lunch idea was a great idea.\nSpeaker F: Yeah, I felt so too.\nSpeaker A: Free lunch is good.\nSpeaker A: I think we can get people to come here, but the issue is you definitely want to make sure that the kind of group that you're getting is the right groups that you don't waste a lot of your time in the overhead of bringing people down.\nSpeaker E: No crunchy food.\nSpeaker A: Well, I was thinking lunch afterwards, right?\nSpeaker A: And they have to do their digits.\nSpeaker D: Yeah, they have to do their digits or they don't get their food.\nSpeaker F: I spoke with some people up at Hoss Business School who volunteered.\nSpeaker F: Should I pursue that?\nSpeaker A: Oh, definitely.\nSpeaker F: Yeah.\nSpeaker F: So they originally, they've decided not to go into speech.\nSpeaker F: So I'm not sure whether they'll still be so willing to volunteer, but all of a sudden we're about the free lunch.\nSpeaker F: I'll tell them about the free lunch.\nSpeaker F: And they'll say there's no such thing.\nSpeaker A: I'd love to get people that are not linguists or engineers.\nSpeaker A: Right.\nSpeaker A: They need a wider sampling.\nSpeaker F: The problem with engineers is be.\nSpeaker D: They make funny sounds.\nSpeaker D: The other thing is that we're talking about is giving them a burn and extra CD run and give them.\nSpeaker A: So if they want an audio record up there, I thought he meant giving them a music CD.\nSpeaker A: I guess it depends on what audience you're talking to.\nSpeaker A: I personally would not want to see my meeting.\nSpeaker A: But maybe.\nSpeaker D: If you're having some planning meeting of some sort, it would just be fun if nothing else.\nSpeaker D: But it also I think builds up towards the goal.\nSpeaker D: We're saying, look, you're going to get this.\nSpeaker D: Isn't that neat?\nSpeaker D: Then you're going to go home with it.\nSpeaker D: It's probably going to be pretty useless to do.\nSpeaker D: But you'll appreciate where it's useful and where it's useless.\nSpeaker D: And then we're going to move this technology so it'll become useful.\nSpeaker A: I think that's a great idea, actually.\nSpeaker E: What if you could tell in the award and send it to the transcripts when they come back?\nSpeaker E: Oh, yeah.\nSpeaker F: Really anyone can have the transcripts.\nSpeaker F: So I have to add a good point to that.\nSpeaker B: So you can see what's it concerned about doing, given the CD immediately, because of these issues of this kind of stuff.\nSpeaker B: Good point.\nSpeaker B: That's a very good point.\nSpeaker B: So we can.\nSpeaker B: So we can.\nSpeaker B: Right.\nSpeaker A: That's right.\nSpeaker A: That's the same thing that we just reviewed publicly, right?\nSpeaker A: Right.\nSpeaker B: Otherwise, you're not allowed to go or like you're not allowed to go up or anyone.\nSpeaker B: So after the transcripts screen phase.\nSpeaker B: Yeah, that's true.\nSpeaker B: Otherwise we need to do lawyer's day.\nSpeaker B: Yeah, it's right.\nSpeaker B: You say, yeah, I got the CD and your honor, I.\nSpeaker A: That's a good point.\nSpeaker D: Yeah, so let's start with Haaz.\nSpeaker A: Sorry, I have to leave.\nSpeaker A: I will be your full time.\nSpeaker F: Okay, see you.\nNone: Okay.\nSpeaker G: Yeah.\nNone: Sorry.\nSpeaker G: See you.\nSpeaker G: Okay.\nSpeaker G: So, let's see.\nSpeaker D: So that was that topic.\nSpeaker D: And then I guess another topic would be where are we in the whole disc resources question for?\nSpeaker F: We are slowly, slowly getting to the point where we have enough room to record meetings.\nSpeaker F: So I did a bunch of archiving and still doing a bunch of archiving.\nSpeaker F: I am in the midst of doing the P files from broadcast news and it took 11 hours to copy it and it will take another 11 to do the clone.\nSpeaker F: Well, it's Abbott.\nSpeaker F: It's Abbott, so it's just, but it's a lot of data.\nSpeaker D: It's copying from one place to another place to another place to another place.\nSpeaker D: Tape.\nSpeaker D: Oh, I did not want to tape.\nSpeaker F: So I'm archiving it and then I'm going to delete the files.\nSpeaker F: So that will give us 10 gigabytes of free space.\nSpeaker F: We are archiving for a long time.\nSpeaker F: And so that will be done in about two hours and so at that point we will be able to record five more meetings.\nSpeaker F: So.\nSpeaker B: One thing the good news about that is that once it's archived it's pretty quick to get back.\nSpeaker B: I mean, the other options fast, but the instructions are really slow.\nSpeaker F: Well, especially because I'm generating a clone also.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: And it takes a while.\nSpeaker F: Generating a clone?\nSpeaker F: Two copies.\nSpeaker F: Right.\nSpeaker B: Oh.\nSpeaker B: Now, what about is the plan to, so stuff will be saved.\nSpeaker B: It's just that you're relocating it.\nSpeaker F: I mean, so we're going to get more disk space or did I know that these are the P files from broadcast news, which are regeneratable, regeneratable, if we really need to, but we have a lot of them.\nSpeaker F: And for the full 140 hour sets.\nSpeaker F: And so they were two gigabytes per file and we had six of them or something.\nSpeaker D: We are getting my space.\nSpeaker D: We are getting another disk rack and 436 gigabyte disks.\nSpeaker D: So, but that's not going to happen in some time.\nSpeaker D: Or maybe six.\nSpeaker F: Maybe six?\nSpeaker F: The sun takes more disks than the Andataco one did.\nSpeaker F: The sun rack takes.\nSpeaker F: One took four and one took six or maybe it was eight and twelve.\nSpeaker F: Whatever it was, it was, you know, 50 percent more.\nSpeaker F: Was there a difference in price?\nSpeaker F: Well, what happened is that we bought all our racks and disks from Andataco for years according to Dave.\nSpeaker F: And Andataco got bought by another company and doubled their prices.\nSpeaker F: And so we're looking into other vendors.\nSpeaker F: By we, of course, I mean Dave.\nSpeaker E: So, I've been looking at the Aurora data.\nSpeaker E: And first, first look at it, there were basically three directories on there that could be moved.\nSpeaker E: One was called Aurora.\nSpeaker E: One was Spanish, which was Carmen Spanish stuff.\nSpeaker E: And the other one was Spine.\nSpeaker E: And so I wrote to Dan and he was very concerned that the Spine stuff was moving to a non-backed up disk.\nSpeaker E: So I realized that what probably not all of that should be moved just the CD-ROM type data, the static data.\nSpeaker E: So I moved that and then I asked him to check out and see if it was okay for I actually deleted the old stuff, but I haven't heard back.\nSpeaker E: I told him he could delete it if he wanted.\nSpeaker E: I haven't checked today to see if he's deleted it or not.\nSpeaker E: And then Carmen's stuff, I realized that when I had copied all of her stuff to XA, I had copied stuff there that was dynamic data.\nSpeaker E: And so I had to redo that one and just copy over the static data.\nSpeaker E: And so I need to get with her now and delete the old stuff off of the disk.\nSpeaker E: And then I looked, haven't done any of the Aurora stuff.\nSpeaker E: I have to meet with Stefan to do that.\nSpeaker D: So but you're figuring you can record in other five meetings or something with the space that you're clearing up from broadcast news.\nSpeaker D: But we have some other disks, some of which you're using for Aurora, but do we have some other?\nSpeaker D: Yep.\nSpeaker F: So we have space on the current disk right now where Meeting Recorder is.\nSpeaker F: And that's probably enough for about four meetings.\nSpeaker F: Yeah.\nSpeaker F: Is that the one that has, is that DC?\nSpeaker F: Yep.\nSpeaker F: Okay.\nSpeaker F: No, no.\nSpeaker F: Well, to wherever the Meeting Recorder currently is, I think it's DI.\nSpeaker F: Okay, but I don't remember.\nSpeaker E: I'm moving from Aurora's on the DC disk that we, I think it's DC.\nSpeaker F: It's whatever that one is.\nSpeaker F: Okay.\nSpeaker F: I just don't remember it.\nSpeaker F: It might be DC.\nSpeaker F: And that has enough room for about former meetings right now.\nSpeaker F: I mean, we were at 100% and then we dropped down to 86 for reasons I don't understand.\nSpeaker F: Someone deleted something somewhere.\nSpeaker F: And so we have some room again.\nSpeaker F: And then with the broadcast news, that's five or six more meetings.\nSpeaker F: So you know, we have a couple of weeks.\nSpeaker F: So I think we're okay until we get the new disk.\nSpeaker E: So should, one question I had for you was, we need, we probably should move the Aurora and all that other stuff off of the Meeting Recorder disk.\nSpeaker E: Is there another backed up disk that you know of?\nSpeaker F: We should put it onto the broadcast news one.\nSpeaker F: That's probably the best thing to do.\nSpeaker F: And that way we can solidate Meeting Recorder onto one disk rather than spreading the mic.\nSpeaker E: Do you know what happened to know what disk guy is on?\nSpeaker F: Nope.\nSpeaker F: I mean, I can tell you I just don't know off the top of my head.\nSpeaker F: I'll find out for that.\nSpeaker F: We could just do that at the end of today once the archive is complete and I verified it.\nSpeaker F: Okay.\nSpeaker F: Is that what gives us plenty of disk?\nSpeaker D: Okay.\nSpeaker D: So then I guess the last thing I had in my agenda was to just hear her not paid on what Jose has been doing.\nSpeaker D: Okay.\nSpeaker C: I have the result of my work during the last days.\nSpeaker C: Thank you for the information because I read the last days.\nSpeaker C: I work in my house in the database and thinking within a different thing about the Meeting Recorder project.\nSpeaker C: And I have some ideas.\nSpeaker C: This information is very useful because you have the distribution.\nSpeaker C: For me, it's interesting because here is the demonstration of the overlap problem.\nSpeaker C: It's a real problem, a frequently problem because you have overlapping sums all the time.\nSpeaker C: By the moment I have the IDETA mark all the overlap zone in the Meeting Recorder with a set mark.\nSpeaker F: Oh, you did that by hand.\nSpeaker F: That's a jet.\nSpeaker C: Can I see that?\nSpeaker C: Yeah, but I can't because why?\nSpeaker C: My idea is to work.\nSpeaker C: I don't know if it will be possible because I have enough time to work on this as you know.\nSpeaker C: But my idea is very interesting to work in the line of automatic segment.\nSpeaker C: But in my opinion, we need a reference session.\nSpeaker F: Yes, absolutely.\nSpeaker F: So are you planning to do that or have you done that already?\nSpeaker F: No, sorry.\nSpeaker F: Have you done that or are you planning to do that?\nSpeaker C: No, I plan to do that.\nSpeaker C: Okay, I plan.\nSpeaker C: I plan.\nSpeaker C: But the idea is the following.\nSpeaker C: Now I need to delete all the overlapping sums exactly.\nSpeaker C: I will talk about the in the in the lab.\nSpeaker C: This information with the exactly time marks for the overlapping sums.\nSpeaker C: Overlapping zone and speaker, pure speech, speaker, song.\nSpeaker C: I mean, songs of speech of one speaker without any noise, any acoustic event that is not speech.\nSpeaker C: I need to do a silence for that because my idea is to study the set of parameters.\nSpeaker C: What are more discriminant to classify the overlapping sounds in comparison with the speech sounds.\nSpeaker C: The idea is to use a not sure yet, but my idea is to use a cluster algorithms or a PerseStron neural nets to study what is the property of the different feature to classify speech and overlapping speech.\nSpeaker C: My idea is to have a control set.\nSpeaker C: My control set will be the silence, silent without any noise.\nSpeaker B: Which means that we still use the different sounds.\nSpeaker B: With the backgrounds.\nSpeaker C: I mean noise, clubs, tape, clips, the difference.\nSpeaker C: Which has a hard effect of the distortion in the.\nSpeaker F: So you intend to hand mark those and exclude them?\nSpeaker C: I have mark in that not in all the file.\nSpeaker C: Only I have a, I don't remember what is the quantity, but I have marked enough speech and all the overlapping sounds.\nSpeaker C: I have 230 more or less overlapping sounds and it is similar to this information.\nSpeaker C: Because with the problem I cross the information of the gene with my experimentation by hand.\nSpeaker C: And it is more similar.\nSpeaker C: Exactly.\nSpeaker C: But, sorry.\nSpeaker C: My idea is to.\nSpeaker F: I should get digital camera.\nSpeaker C: To classify.\nSpeaker C: I need the set mark of the different sound because I want to put for each frame a level indicating it's a supervised and a class in process.\nSpeaker C: I put for each frame a level indicating what is the type, what is the class which belong.\nSpeaker C: I mean the class overlapping speech, the class speech.\nSpeaker E: And the class will be assigned by hand based on the.\nSpeaker C: I put the mark by hand because my idea is in the first session.\nSpeaker C: I need to be sure that information can be in validation.\nSpeaker C: Sure.\nSpeaker C: It's right because if not I will return to the speech file to analyze what is the problem.\nSpeaker C: And I would prefer to have this level automatically.\nSpeaker C: You need to.\nSpeaker B: I ask you the difference between the top two.\nSpeaker B: By speech do you mean one person only?\nSpeaker B: One, two, three.\nSpeaker B: One, two, three.\nSpeaker B: One speaker in a breath overlapping.\nSpeaker B: Someone else in the back.\nSpeaker B: By the moment.\nSpeaker B: Clicking overlapping speech.\nSpeaker C: That's all those possibilities in the top.\nSpeaker C: In the first moment because I have information of the overlapping sounds.\nSpeaker C: Information about the overlapping sound is from speech, clear speech from a true speaker or three speaker.\nSpeaker C: It's a song where the impregnance of a speaker overlaps onto a speech.\nSpeaker B: A speech with something overlapping which could be speech but doesn't need to be.\nSpeaker C: You know, especially overlapping speech from different speakers.\nSpeaker D: No, but I think she's saying where do you in these three categories?\nSpeaker D: Where do you put the instances in which there is one person speaking and other sounds which are not speech?\nSpeaker C: Which category do you put that?\nSpeaker C: Here I put a speech from one speaker without any events.\nSpeaker D: So where do you put speech from one speaker that does have a non-speech event at the same time?\nSpeaker C: Which category?\nSpeaker C: Which category?\nSpeaker C: No, by the moment no.\nSpeaker C: Oh, you see, not marked.\nSpeaker C: No, not marked.\nSpeaker C: Because I want to meet the study.\nSpeaker C: Fine, so even for the all of the data.\nSpeaker B: So you're ignoring overlapping events unless their speech was speech.\nSpeaker C: Yeah, that's fine.\nSpeaker C: Why?\nSpeaker C: What's the reason?\nSpeaker C: Because it's the first study.\nSpeaker C: Oh, no, no.\nSpeaker D: It's perfectly sensible way to go.\nSpeaker D: We just wanted to try and understand what you're doing.\nSpeaker B: Because you've talked about other overlapping events in the past.\nSpeaker C: So in the future, the idea is to extend the class.\nSpeaker C: Is to consider all the information you mentioned.\nSpeaker D: Yeah, I don't think we're asking for that.\nSpeaker C: I don't know what we would have.\nSpeaker E: Is your silence category pure silence or?\nSpeaker C: Is there a door slammer?\nSpeaker E: No, no, it's pure silence.\nSpeaker C: Is the control set?\nSpeaker C: OK.\nSpeaker C: Is the control set?\nSpeaker D: What you will be silent with the majority of the world.\nSpeaker D: I think what you mean is that it's non-speech segments that don't have impulsive noises, right?\nSpeaker D: Because what you're calling a vent is somebody coughing or clicking a wrestling paper or hitting something, which are impulsive noises.\nSpeaker D: But steady state noises are part of the background, which it will be included in that.\nSpeaker C: Right?\nSpeaker C: Here.\nSpeaker B: Yeah.\nSpeaker C: So it's like a signal noise.\nSpeaker C: I think there are some kind of noises that don't water to be in that control set.\nSpeaker C: But I prefer the silence with this kind or the off.\nSpeaker D: Right.\nSpeaker D: It means background might be better word than silence.\nSpeaker D: It's just sort of the background of the acoustic.\nSpeaker D: Right.\nSpeaker C: So it's only going on.\nSpeaker C: OK.\nSpeaker C: And with this information, the idea is I have a level for each friend.\nSpeaker C: And with a cluster algorithms, I am sorry.\nSpeaker C: And I am going to prepare a set of features, a structure, molds.\nSpeaker C: Right.\nSpeaker C: And maybe it's, tell me whatever.\nSpeaker C: So I have a pitch structure yet.\nSpeaker C: I have to test.\nSpeaker C: But you have your own?\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I have to prepare.\nSpeaker C: It's a modified version of a pitch tracker from a Stanford universe.\nSpeaker C: Stanford, no, from Cambridge.\nSpeaker C: Oh, what's it really?\nSpeaker C: I don't remember what is the name of the author.\nSpeaker C: Because I have several library tools from a festival, from Edinburgh, from Cambridge, and from our department.\nSpeaker C: And I have to, because in general, the pitch tracker, that's a waste.\nSpeaker F: Right.\nSpeaker F: Very well.\nSpeaker F: Because the feature it might be OK.\nSpeaker F: Yeah.\nSpeaker C: So we don't know.\nSpeaker C: This is an idea is to attain, for example, a different, a grid number of emphesies, for example, 25, 30, 30 parameter for each one.\nSpeaker C: In the first step in the research, my idea is try to prove what is the performance of the difference parameter to classify the difference, what is the performance approach to classify the difference frames of each class.\nSpeaker C: What is the error about it?\nSpeaker C: This is the first idea.\nSpeaker C: And the second is try to use some ideas similar to the linear discriminant analysis.\nSpeaker C: Similar.\nSpeaker C: Because the idea is to study what is the contribution of each parameter to the process of classify the different parameter.\nSpeaker C: What sort of classifier?\nSpeaker C: The classifier, by the moment, is a similar classifier used in Victoria, Quantified, is used to some distance to put a vector in a class difference.\nSpeaker F: Unimodal?\nSpeaker F: So is it just one cluster?\nSpeaker C: It's only two clusters using a Kinear SNF or similar.\nSpeaker C: Another possibility is to use a neural network.\nSpeaker C: But what is my idea?\nSpeaker C: What is the problem?\nSpeaker C: I see.\nSpeaker C: If you use the neural network.\nSpeaker C: And this kind of cluster inaugurated to can test to can a share of what happened.\nSpeaker C: You can't analyze it.\nSpeaker C: Right.\nSpeaker C: You can't analyze it.\nSpeaker C: You use a neural network, a good idea.\nSpeaker C: But you don't know what happened in the interior of the neural network.\nSpeaker D: Well, actually, you can do sensitivity analyses, which show you what the importance of different parts of pieces input are.\nSpeaker D: What's going on internally?\nSpeaker D: But it's actually not that hard to analyze it and figure out the effects of different inputs, especially if they're all normalized.\nSpeaker F: Well, using something simpler first, I think.\nSpeaker D: Well, if you really wonder what different...\nSpeaker D: Decision tree.\nSpeaker D: Yeah, then, it's decision tree is really good.\nSpeaker D: But the thing is, he's not like he has one, a bunch of very distinct variables like pitch and this, he's talking about all these capture all coefficients.\nSpeaker D: So for the much case, any reasonable classifier is going to be a mess.\nSpeaker D: Right.\nSpeaker D: It's going to be hard to figure out what...\nSpeaker C: I would include you to the differential.\nSpeaker C: Yeah, that would be...\nSpeaker D: I mean, I think the other thing, I mean, this is, I think, a good thing to do to sort of look at these things at least.\nSpeaker D: See what...\nSpeaker D: Let me tell you what I would do.\nSpeaker D: I would take just a few features.\nSpeaker D: Instead of taking all the MFCCs or all the EPLPs or whatever, I would just take a couple.\nSpeaker D: Okay, like C1, C2, something like that, so that you can visualize it.\nSpeaker D: And look at these different examples and look at scatter plots.\nSpeaker D: Okay, so before you build up any kind of fancy classifiers, just take a look in two dimensions at how these things are split apart.\nSpeaker D: That I think will give you a lot of insight of what is likely to be a useful feature when you put it to a more complicated classifier.\nSpeaker D: Yeah.\nSpeaker D: And the second thing is, once you actually get to the point of building these classifiers, what this lacks so far is the temporal properties.\nSpeaker D: So if you're just looking at a frame at a time, you don't know anything about the structure of it over time.\nSpeaker D: And so you may want to build a mark off a model of some sort or else have features that really are based on some bigger chunk of time.\nSpeaker D: But I think this is a good place to start.\nSpeaker D: But don't...\nSpeaker D: Anyway, this is my suggestion.\nSpeaker D: Don't just throw in 20 features at it, the delta is in the delta, and all that into some classifier, even if it's K-nearest neighbors, you still won't know what it's doing.\nSpeaker D: You know it's not an Earl Matt.\nSpeaker D: I think to know what it's...\nSpeaker D: Have a better feeling for what it's doing, you want to look at it.\nSpeaker D: So you want to look at some picture that shows you, here's these things are offered some separation.\nSpeaker D: And in LPC, the thing that particularly look at is, I think, is something like the residual, the energy and the residual.\nSpeaker B: Can I ask, it strikes me that there's another piece of information that might be useful, and that's simply the transition.\nSpeaker B: So if you go from a transition of silence to overlap versus transition from silence to speech, it's going to be a big informative area there, it seems to be.\nSpeaker C: But it's my ambition of the project.\nSpeaker C: The meeting record project for me has a two, several parts, an adjective because it's a good thing to do.\nSpeaker C: At the first, in the acoustic parts of the project, I think we have two main adjectives.\nSpeaker C: One of these is to detect the chain, the acoustic chain.\nSpeaker C: For that, if you don't use a speech recognizer, a pro-class or not pro-class, to try to level the different friends, I think the Ike criterion, or big criterion, will be enough to detect the chain.\nSpeaker C: And probably, I would like to prove.\nSpeaker C: Probably, when you have the transition of speech or silence to overlap song, this criterion is enough with probably with this kind of, the more use a regular parameter, NCC, you have to find, you can find the mark, you can find the good chain.\nSpeaker C: But I understand that your adjective is too classified, to know that that song is not only a new song in the file, but you have to know that it is an overlap song because in the future, you will try to process that song with a no-regular speech recognizer, you will pretend to process the overlapping song with another claim, because it is very difficult to obtain the trackition from using a regular, normal speech recognizer.\nSpeaker C: I think it is the idea, the system will have two models.\nSpeaker C: A model to detect, the most, the most, the most, the most, the mark, the chain, and another model, or several models, to try, but several models, rowing models, simple models, to try to classify the different class.\nSpeaker F: I am sorry, I didn't understand you, what model?\nSpeaker C: The classifies, to detect the different class to the different song before trying to recognize, to track life with a speech recognizer.\nSpeaker C: The idea is to use, for example, a neural net with the information we obtain from this stage, the parameter with the selected parameter to try to put a class of each frame for the different song, you have to obtain it in the first step with the, for example, BK, Heterium, Compere-Mole.\nSpeaker D: I think in any event we are great that the first step is, because what we had before for speaker change detection did not include these overlaps.\nSpeaker D: So the first thing is for you to build something, you will detect the overlaps.\nSpeaker D: So again, I think the first thing to do to detect the overlaps is to look at these, and in, well, again, the things that you have written up there, I think are way too big.\nSpeaker D: If you are talking about, say, 12th order NFCC or something like that, it is way too much, you will not be able to look at it.\nSpeaker D: All you will be able to do is put into a class of fire and see how well it does.\nSpeaker D: I think if you pick one or two dimensional things, or three, if you have some very fancy display, and look at how the different classes separate themselves out, you will have much more insight about what is going on.\nSpeaker D: What are you doing?\nSpeaker D: Well, you will get a feeling for what is happening.\nSpeaker D: So if you look at, suppose you look at first and second order, capture coefficients for some of these kinds of things, and you find that the first order is much more effective than the second, and then you look at the third and there is not too much there, you may just take first and second order capture coefficients.\nSpeaker D: With LPC, I think LPC per se isn't going to tell you much more than the other, maybe.\nSpeaker D: On the other hand, the LPC residual, the energy in the LPC residual, would say how well a loader LPC models fitting it, which should be pretty poorly for two or more people speaking at the same time, should be pretty well for one.\nSpeaker D: And so again, if you take a few of these things that are promising features, and look at them in pairs, I think you will have much more of a sense of, okay, I now have, doing a bunch of these analyses, I now have ten likely candidates, and then you can do decision\nSpeaker C: countries or whatever to see how they combine. I've got a question.\nSpeaker C: Sorry, but I don't know the first way to do that, and I would like to know what's the European Union.\nSpeaker C: All these studies in the first moment, I will pretend to do with the equalizes pitch.\nSpeaker C: The misses pitch, the mix, the mix, the mix, the mix, the speech.\nSpeaker C: Why?\nSpeaker C: Because the spectral distortion is a lot clearer, very much clearer if we compare with the PDA.\nSpeaker C: PDA is pitchfile, it will be difficult.\nSpeaker B: It's messier, the PDA is messier.\nSpeaker C: Because the signal to know is relation is slow.\nSpeaker C: I think that's a good way to start.\nSpeaker C: I don't know that the result of the study within this speech, the mix speech, would be interesting that sounds to see.\nSpeaker C: It's actually with the PDA file.\nSpeaker C: Well, I think that would be an interesting result.\nSpeaker C: I mean, what is the effect of the low signal to know is relation?\nSpeaker D: Well, I think it's not at all unreasonable, it makes sense to start with a simpler signal, because if you have features which aren't even helpful in the high signal noise ratio, then there's no point putting them into the low signal ratio, one would think.\nSpeaker D: And so if you can get, again, my prescription would be that you would with the mixed signal, you would take a collection of possible features, look at them, look at how these different classes that you've marked separate themselves, and then collect in pairs, and then collect ten of them or something, and then proceed with a bigger classifier.\nSpeaker D: And then if you can get that to work well, then you go to the other signal.\nSpeaker D: And you know, they won't work as well, but how much, and then you can re-optimize.\nSpeaker F: But I think it would be interesting to try a couple with both, because it might be interesting to see if some features work well with close mix.\nSpeaker D: And that's true, that it also could be useful to do this exploratory analysis where you're looking at scatter plots and so on in both cases.\nSpeaker C: Sure.\nSpeaker C: I think the parameter we found, the word with both speech file, but what is the relation of the performance when you use the speech file, the BDA, you speech file, you don't know, but I think it will be important, because people, different groups, have experience with this kind of problem, it's not easy to solve, because if you, I have seen the speech file from a PDA and some parts, but it's difficult because you don't see the spectra.\nSpeaker C: That's another reason why very simple features, things like energy and things like car\nSpeaker D: density and residual energy are you here? Yeah, are better to use than very complex ones because the marble libel, Chuck was going to ask something.\nSpeaker E: Yeah, maybe this is a dumb question, but I thought it would be easier if you use the PDA because, couldn't you like use beam forming or something to detect speaker overlaps?\nSpeaker F: I mean, well, if you use the array rather than the signal from just one.\nSpeaker D: Yeah, no, you're right, that in fact, if we made use of the fact that there are two microphones, you do have some location information, which we don't have with the one.\nSpeaker D: And so that's not allowed with this.\nSpeaker D: Well, no, I mean, we don't have any rules, really.\nSpeaker E: I mean, given the goal, I mean, is that the violation of that?\nSpeaker D: I think it's an additional interesting question.\nSpeaker D: I mean, I think you want to know whether you can do it with one because it's not necessarily true that every device that you're trying to do this with will have two.\nSpeaker D: If on the other hand, we show that there's a huge advantage with two, then that could be a real point.\nSpeaker D: But we don't even know yet what the effect of detecting, having the ability to detect overlaps is.\nSpeaker D: Maybe it doesn't matter too much.\nSpeaker D: So this is all pretty early stages.\nSpeaker D: But no, you're absolutely right.\nSpeaker D: That's a good thing to consider.\nSpeaker B: There's a correlation, though.\nSpeaker B: And that is, a person turns their back to the PD8, then some of the positional information goes away.\nSpeaker B: Well, it does not.\nSpeaker B: It does not.\nSpeaker B: That's so much.\nSpeaker B: The issue is that the...\nSpeaker B: And everyone on the axis, on the axis of it, that's the other thing I was saying.\nSpeaker B: You mentioned this last time.\nSpeaker B: Yeah, we had to put it on a little turn table.\nSpeaker B: If you just write down the midline, then the left right is going to be different.\nSpeaker B: And in his case, he's closer to it anyway.\nSpeaker B: It seems to me that it's not a...\nSpeaker B: But it's another source of information.\nSpeaker B: It's a topology of it is.\nSpeaker B: I don't know.\nSpeaker C: I think...\nSpeaker C: I think because the reason between the two microphones, microphone in the PDI's brain here, it's from my opinion, it's an interesting idea to try to study the brain role, a problem with the question.\nSpeaker C: Yes, because I found a difference between the speech from each microphone in the PDI.\nSpeaker C: Yeah, it's timing difference.\nSpeaker D: It's an important issue, right?\nSpeaker B: I mean, it means the cells.\nSpeaker B: I know, that's very important.\nSpeaker B: Right.\nSpeaker B: But I'm just saying that the way we're seated around the table is not the same with respect to each person with respect to the PDA.\nSpeaker B: So we're going to have a lot of differences with reference.\nSpeaker B: But that's fine.\nSpeaker D: That's fine.\nSpeaker D: That's...\nSpeaker D: So I think the issue is, is there a clean signal coming from only one direction?\nSpeaker D: If it's not coming from just one direction, if there's a broader pattern, it means there's more likely these multiple people speaking.\nSpeaker D: Wherever they are.\nSpeaker D: So it's sort of like how...\nSpeaker D: Is it a...\nSpeaker D: Is there a narrow beam pattern or is it a distributed beam pattern?\nSpeaker D: Does it distribute a beam pattern?\nSpeaker D: Then it looks more like it's multiple people.\nSpeaker D: Wherever you are, even if he moves around.\nSpeaker B: Okay.\nSpeaker B: It just seemed to me that this isn't the ideal type of separation.\nSpeaker B: I think it's...\nSpeaker D: Oh, ideal would be to have the wall filled with him.\nSpeaker D: But I mean, the thing is just having too much...\nSpeaker D: If you looked at that thing on Dan's page, it was when there were two people speaking, it looked really, really different.\nSpeaker D: Oh, yeah, yeah, yeah.\nSpeaker D: Okay.\nSpeaker D: What looked different?\nSpeaker D: Well, basically, he was looking at correlation.\nSpeaker D: Just cross-correlation between two sides.\nSpeaker D: So cross-correlation is very sensitive.\nSpeaker E: I'm not sure what Dan's page is.\nSpeaker E: If you mean he was looking at the...\nSpeaker D: He took a signal from the two microphones and he crossed...\nSpeaker D: And you cross-correlate them with different blacks.\nSpeaker D: Okay.\nSpeaker D: And one person is speaking, then wherever they happen to be at the point where they're speaking, then there's a pretty big maximum right around that point in the like.\nSpeaker D: So whatever angle you are...\nSpeaker D: So if there's two...\nSpeaker D: At some like corresponding to the time difference between the two there, you get this boost in the cross-correlation.\nSpeaker F: And if there are multiple people talking, you'll see two peaks.\nSpeaker F: It's spread out.\nSpeaker B: Let me ask you, if both people were over there...\nSpeaker B: Yeah.\nSpeaker B: It would be less effective than if one was there and one was across Catechord.\nSpeaker B: Oh, I'm sorry.\nSpeaker D: If they're right next to another one.\nSpeaker E: If I was here and Morgan was there and we're both talking right, it wouldn't work.\nSpeaker B: Yeah.\nSpeaker B: Versus you versus...\nSpeaker B: And we're Catechord across the table and I'm farther away from this one and you're farther away from the other one.\nSpeaker B: Or even if that would be strong.\nSpeaker B: If people were saying right across from each other, you could tell them from either.\nSpeaker B: Cross-correlate, same axis, you don't have as much difference.\nSpeaker B: Well, we don't have a third dimension there.\nSpeaker B: Yeah, so...\nSpeaker B: It's differentially very valuable.\nSpeaker B: I mean, it's not to say...\nSpeaker B: I mean, I certainly think it's extremely bad.\nSpeaker B: And we humans depend on...\nSpeaker B: Yeah, but it's almost...\nSpeaker D: It's almost... but it's almost a... I think what you're talking about is there's two things.\nSpeaker D: There's a sensitivity issue and then there's a pathological or issue.\nSpeaker D: So the one where someone is just right directly in line is sort of pathological here.\nSpeaker D: If someone just happens to be sitting right there, then we won't get good information from it.\nSpeaker B: Okay, and if they're close...\nSpeaker B: If they're close...\nSpeaker D: Same subject.\nSpeaker D: It's just a question of the sensitivity.\nSpeaker D: So if the sensitivity is good enough and we just don't have enough experience with it to know how...\nSpeaker B: Oh, I'm not trying to argue against using it by any means.\nSpeaker B: I just wanted to point out that weakness that it's top of the level.\nSpeaker B: And I think Dan is still working on it.\nSpeaker F: So he actually... he wrote me about it a little bit.\nSpeaker F: Great.\nSpeaker D: No, I don't understand.\nSpeaker D: I mean, the other thing you can do...\nSpeaker D: I mean, we're assuming that it would be a big deal just to get somebody...\nSpeaker D: Come in somebody to put two microphones on the PDA.\nSpeaker D: But if you have put a third in, you could put in the other axis and then you got to...\nSpeaker D: You've got to.\nSpeaker D: Yeah, then you pretty much could cover.\nSpeaker E: What about just doing it from these mics, you know?\nSpeaker C: Yeah.\nSpeaker C: I mean, it's actually more interesting to study the PCN because...\nSpeaker C: Say, Parisian, anything.\nSpeaker D: But that's...\nSpeaker D: And they're much broader.\nSpeaker D: I mean, we can...\nSpeaker D: We'll be...\nSpeaker D: All of this is there for us to study.\nSpeaker D: Yeah, we can do whatever we want.\nSpeaker D: But the thing is...\nSpeaker D: One of the...\nSpeaker D: At least one of the things I was hoping to get out of this is what can we do with what we think would be the normal situation if some people get together and one of them has a PDA.\nSpeaker D: That's what I was asking about.\nSpeaker D: What are the constraints?\nSpeaker D: Yeah.\nSpeaker D: Well, that's the constraint of one question and I think both Adam and I were interested in.\nSpeaker D: But, you know, if you can instrument a room, this is really minor league compared with what some people are doing.\nSpeaker D: Right?\nSpeaker D: Some people...\nSpeaker D: Eight micro-round.\nSpeaker D: Yeah, brown and...\nSpeaker D: And...\nSpeaker D: Didn't they have the main cape?\nSpeaker D: And...\nSpeaker D: And Cape...\nSpeaker D: They both have these, you know, bigger rays on the wall.\nSpeaker D: And, you know, if you can...\nSpeaker D: You've got microphones all over the place.\nSpeaker D: You know, tens of microphones.\nSpeaker D: Oh, I saw it.\nSpeaker D: I don't know.\nSpeaker D: And if you do that, then you can really get very nice...\nSpeaker D: Kind of selectivity.\nSpeaker F: So, I saw one that was like 100 microphones.\nSpeaker E: Yeah.\nSpeaker E: 10 by 10, right?\nSpeaker E: And you could...\nSpeaker E: And you could...\nSpeaker E: And you could...\nSpeaker E: And you could...\nSpeaker F: Very, very...\nSpeaker F: They could have all kinds of noises and you can zoom right in...\nSpeaker F: Pretty much.\nSpeaker F: It was all in software and you could pick out an individual beam and listen to it.\nSpeaker F: Yeah.\nSpeaker D: That's cool.\nSpeaker D: But, I was interested.\nSpeaker D: The reason why I haven't focused on that is the first concern is because I'm interested in what happens for people, random people out in some random place where they're having an improv to discussion.\nSpeaker D: And you can't just always go, well, let's go to this heavily instrumented room that we spent tens of thousand dollars to set up.\nSpeaker E: No, what you need to do is you have a little fabric thing that you enroll and hang on a wall.\nSpeaker E: It has all these mics and it has a plug-in jack to the PDA.\nSpeaker D: The other thing actually gets it this little bit of something else I'd like to do is what happens if you have two PDAs.\nSpeaker D: Yeah.\nSpeaker D: And they communicate with each other.\nSpeaker D: And then, you know, they're in random positions.\nSpeaker D: The likelihood, I mean, basically, there wouldn't be any...\nSpeaker F: That's likely to be any kind of network.\nSpeaker F: You have two.\nSpeaker F: If you're three or four.\nSpeaker F: All sorts of interesting things you can do with that.\nSpeaker F: I mean, not only can you do microphone arrays, but you can do all sorts of multi-band as well.\nSpeaker E: I still like my rug on the wall, I guess.\nSpeaker E: But it happens.\nSpeaker B: In terms of...\nSpeaker B: Yeah.\nSpeaker F: In terms of the research, it's really...\nSpeaker F: Whatever the person who is doing the research wants to do.\nSpeaker F: So if Jose is interested in that, that's great.\nSpeaker F: But if he's not, that's great, too.\nSpeaker D: I would actually kind of like us to wind it down, see if we can still get the end of the birthday thing there.\nSpeaker F: Well, I had a couple things that I didn't want to bring out.\nSpeaker F: One is, do we need to sign new...\nSpeaker F: Well, it's slightly different.\nSpeaker B: So I would say a big idea.\nSpeaker B: Oh, they need?\nSpeaker B: Yeah.\nSpeaker D: Oh, this morning we didn't sign anything because we said if anybody had signed it already, we have to.\nSpeaker F: I should have checked with Jane first, but the form has changed.\nSpeaker F: So we may want to have everyone sign the new form.\nSpeaker F: I had some things I wanted to talk about with the thresholding stuff I'm doing, but if we're in a hurry, we can put that off.\nSpeaker F: And then also anonymity, how we want to anonymize the data.\nSpeaker F: Well, should I...\nSpeaker B: I mean, I have some results to present, but I guess we'll have time to do that this time.\nSpeaker B: But it seems like the anonymization is also something that we might want to discuss and greatly.\nNone: I mean, what...\nSpeaker D: We're about to wind down.\nSpeaker B: I think what I would prefer is that we delay the anonymization thing to...\nSpeaker B: I think, until next week, and I would like to present the results that I have on the\nSpeaker E: overlap. We still have to do this, too, right?\nSpeaker E: Right.\nSpeaker E: Digits.\nSpeaker F: Right.\nSpeaker F: No, well, we don't have to do digits.\nSpeaker D: Well, I mean...\nSpeaker D: So, okay.\nSpeaker D: It sounds like there were a couple technical things people would like to talk about.\nSpeaker D: Why don't we just take a couple minutes to briefly do them?\nSpeaker D: And then...\nSpeaker D: Go ahead, Jane.\nSpeaker B: I would prefer to have more time for my results.\nSpeaker B: Could I do that next week, maybe?\nSpeaker B: Oh, yeah, sure.\nSpeaker B: That's what I'm asking.\nSpeaker B: Oh, yeah.\nSpeaker B: And I think the anonymization, if you want to proceed with that.\nSpeaker B: Now, I just think that that's a discussion, which also really deserves...\nSpeaker B: Because...\nSpeaker B: You know, more than just a minute, I really do think that because you raised a couple of possibilities yourself.\nSpeaker B: You and I have discussed it previously.\nSpeaker B: And there are different ways that people approach it.\nSpeaker F: Right.\nSpeaker F: We're just...\nSpeaker F: We're getting enough data now that I'd sort of like to do it now before I get overwhelmed with once we decide how to do it.\nSpeaker F: Well, okay.\nSpeaker B: It's just...\nSpeaker B: Okay, I'll give you the short version, but I do think that it's an issue that we can't resolve in five minutes.\nSpeaker B: Okay, so the short thing is we have...\nSpeaker B: Table recording...\nSpeaker B: Sorry, digitized recording.\nSpeaker B: Those we won't be able to change if someone says, hey Roger, or so and so.\nSpeaker B: So that's going to stay that person's name.\nSpeaker B: Now in terms of like the transcript, the question becomes, what's simpler?\nSpeaker B: Are you going to put in there for everybody's name?\nSpeaker B: And whether you're going to put it in the text where he says, hey Roger, are we going to put that person's anonymized name in instead?\nSpeaker F: No, because that would give you a mapping.\nSpeaker F: And you don't want to have a mapping.\nSpeaker B: So the first decision is we're going to anonymize the same name for the speaker identifier and also in the text whenever the speaker's name is mentioned.\nSpeaker F: No.\nSpeaker F: Because that would give you a mapping between the speaker's real name and the tag we're using.\nSpeaker B: And we don't want...\nSpeaker B: I think you're going to see what I said.\nSpeaker B: Okay.\nSpeaker B: So within the context of an utterance, someone says, so Roger, what do you think?\nSpeaker B: Okay.\nSpeaker B: And it seems to me that...\nSpeaker B: Well, maybe it seems to me that if you change the name, the transcript is going to disagree with the audio video.\nSpeaker B: We don't want to...\nSpeaker F: We don't want to...\nSpeaker F: We don't want to.\nSpeaker F: We want the transcript to be Roger.\nSpeaker F: Because if we made the transcript be the tag that we're using for Roger, someone who had the transcript and the audio would then have a mapping between the anonymized name and the real name.\nSpeaker B: And we want to avoid that.\nSpeaker B: Okay.\nSpeaker B: But then there's this issue of if we're going to use this for a discourse type of thing.\nSpeaker B: And you know, Liz is mentioning stuff in previous meeting about things like Gays direction and who's the address C and all.\nSpeaker B: Then to have Roger be the thing in the utterance and then actually have this speaker identify who was Roger be Frank, that's going to be really confusing and make it pretty much useless for discourse in those.\nSpeaker B: Yes.\nSpeaker B: It's a good point.\nSpeaker B: Now, if you want to, you know, I mean, in some cases, I know that Susan Irvingtrip and some of hers actually did do a filter of the signal-worth person's name was mentioned, and I saw, I mean, the question then becomes one level back.\nSpeaker B: How important is it for a person to be identified by first name versus full name?\nSpeaker B: Well, on the one hand, it's not a full identity.\nSpeaker B: We're taking all these precautions and they'll be taking precautions, which are probably even the more important ones to, they'll be reviewing the transcripts, see if there's something they don't like.\nSpeaker B: Okay.\nSpeaker B: So, maybe that's enough protection.\nSpeaker B: On the other hand, this is a small pool and people who say things about topic X who are researchers and well known in the field, they'll be identifiable simply from first name.\nSpeaker B: However, taking one step further back, they'd be identifiable anyway, even if we changed all the names.\nSpeaker B: Right.\nSpeaker B: So, is it really, you know, now in terms of like, so I did some results which I'll report on next time, which do mention individual speakers by name.\nSpeaker B: Now, there, the human subjects community is very precise.\nSpeaker B: You don't want to mention subjects by name and published reports.\nSpeaker B: Now, it'd be very possible for me to take those data, put them in a study and just change everybody's name for the purpose of the publication.\nSpeaker D: Yeah, once you get to the publication, you can certainly do that.\nSpeaker D: You know, Z.\nSpeaker D: Yeah, exactly.\nSpeaker B: Doesn't matter.\nSpeaker D: Yeah, I mean, it doesn't, I mean, I'm not knowledgeable about this, but it certainly just bother me to have someone's first name in the transcript.\nSpeaker D: I think you don't want their full name to be listed.\nSpeaker B: And in the form that they sign, it does say your first name may arise in the course of the meetings.\nSpeaker B: Yeah.\nSpeaker D: So again, the issue is if you're tracking discourse things, if someone says, Frank said this and then you want to connect it to something later, you've got to have this part where that's Frank going, right?\nSpeaker B: Yeah, and even more immediate than that, just being able to, well, it just seems like to track from one utterance to the next utterance, who's speaking and who's speaking to whom, because that can be important.\nSpeaker B: You raise the point so and so it's kind of nice to be able to know who you are.\nSpeaker B: I'm thinking too much.\nSpeaker B: I remember, you remember last time we had this discussion of how I was sort of avoiding mention people's names.\nSpeaker B: I was too.\nSpeaker B: And we made the decision that was kind of artificial.\nSpeaker B: So well, I mean, if we're going to step in after the fact and change people's names in the transcript, we've basically done something once to have worse.\nSpeaker F: Yeah.\nSpeaker F: Well, I don't want to change the name in the transcript, but that's because I'm focused so much on the acoustics instead of on the discourse.\nSpeaker F: And so I think that's a really good point.\nSpeaker F: You're right.\nSpeaker F: This is going to require more thought.\nSpeaker D: Yeah.\nSpeaker D: Let me just back up this to make a brief comment about the, what we're covering in the meeting.\nSpeaker D: I realized when you're doing this that I didn't realize that you had a bunch of things that you wanted to talk about.\nSpeaker D: And so I was proceeding somewhat of random, frankly.\nSpeaker D: So I think what would be helpful would be, you know, I'll mention this to Liz and Andreas too, that before the meeting, if anybody could send me any agenda items that they were interested in, and I'll take the role of organizing them into the agenda.\nSpeaker D: But I'd be very pleased if everyone else completely make up the agenda.\nSpeaker D: No, sorry to make it up, but if no one's told me things, then I'm just proceeding from my guesses.\nSpeaker D: And yeah, I'm sorry, and it's up with you in the out here.\nSpeaker D: Time to, I mean, I'm just always asking, you know what he's doing, you know what?\nSpeaker D: And so it's, there's a, there's a lot of other things going on.\nSpeaker F: How would the person who is doing the transcript even know who they're talking about?\nSpeaker F: Do you know what I'm saying?\nSpeaker E: The person who's doing the transcript, the IBM people?\nSpeaker F: Yeah.\nSpeaker F: I mean, so, so how is that information going to get labeled anyway?\nSpeaker B: How do you mean what they're talking about?\nSpeaker F: I mean, so if I'm saying in a meeting, oh and Bob, by the way, wanted to do so and so, they're just going to write Bob wanted to do so.\nSpeaker F: They're just going to write Bob.\nSpeaker B: And so they won't be able to change it themselves.\nSpeaker F: If you're, if you're doing discourse analysis, how are you going to do any of this?\nSpeaker D: Yeah, really.\nSpeaker B: I'm bet we're going to have huge chunks that are just totally un, I mean, they're going\nSpeaker D: to say speaker one or speaker two.\nSpeaker F: Do you speak, I mean, well, the current one, they don't do speaker identity because in naturally speaking or excuse me in via voice, it's only one person.\nSpeaker F: And so in their current conventions, there are no multiple speaker conventions.\nSpeaker D: So it may just be one long transcript of a bunch of words.\nSpeaker B: Yeah, I think that my understanding for me is it in Changs, how are you pronouncing\nSpeaker D: me? You change, you change.\nSpeaker B: You change, you change.\nSpeaker B: It was that they will adopt the part of the conventions that we discussed where they put speaker identifier down, but you know, they won't know these people.\nSpeaker B: So I think it's well, they'll adopt some convention, but we haven't specified to them.\nSpeaker B: So they'll do something like speaker one, speaker two is one of it.\nSpeaker B: But I'm betting there will be huge variations in the accuracy of their labeling of speakers.\nSpeaker B: We'll have to review the transcripts in any case.\nSpeaker D: And it may very well be, I mean, since they're not going to sit there and worry about it being the same speaker, they may very well go.\nSpeaker D: So the first time it changes to another speaker, that will be speaker two.\nSpeaker D: And the next time it will be speaker three, even if it's actually speaker one.\nSpeaker B: That would be a very practical solution on that part.\nSpeaker B: But then we need to label it.\nSpeaker F: Yeah, we could probably regenerate it pretty easily from the close talking mics.\nSpeaker F: I was thinking the attempt, the attempt, the attempt, the analysis, what changes.\nSpeaker F: That doesn't answer the question.\nSpeaker F: It's a good point.\nSpeaker F: Which is what you do for this course tracking.\nSpeaker C: You don't need to know what is the identification of the speaker.\nSpeaker C: You don't need to know.\nSpeaker F: For acoustics you don't, but for discourse you do.\nSpeaker F: For the question.\nSpeaker D: Yeah, if someone says what is Jose doing and then Jose says something, you need to know that that was Jose responding.\nSpeaker B: And let's be adopted different set of norms, which is to not to make a point of not identifying people by name, which then leads you to be more contextual explicit.\nSpeaker E: That would be hot.\nSpeaker B: Well, people are very flexible.\nSpeaker B: You know what I mean?\nSpeaker B: So when we did this last week, I think that's what I'm saying.\nSpeaker B: I felt that, you know, Andreas, he sometimes people think of something else at the same time and they miss a sentence or something.\nSpeaker B: And because he missed something, then he missed the initial introduction of who we were talking about and was unable to do the tracking.\nSpeaker B: But I felt like most of us were doing the tracking and knew who we were talking about and we just weren't mentioning the name.\nSpeaker B: So people are really flexible.\nSpeaker E: But you know, like at the beginning of this meeting or you, I think said, you know, Liz said something about, is Morgane used the equipment?\nSpeaker E: I mean, how would you say that?\nSpeaker E: I mean, you have to really think, you know, about what you're saying.\nSpeaker D: Does you know who up and you know where?\nSpeaker D: Yeah.\nSpeaker E: I think it would be really hard if we made a call for you and saying that.\nSpeaker F: I was going to say is that the other option is that we can't leave out the names.\nSpeaker F: But then again, that kills your discourse now.\nSpeaker F: Yeah.\nSpeaker E: I think the, I think, I don't know, my own two-sense worth is that you don't do anything about what's in the recordings.\nSpeaker E: You only anonymize to the extent you can.\nSpeaker F: Well, but that said, that works great for the acoustics, but it hurts you a lot for trying to do discourse.\nSpeaker F: Yeah, why?\nSpeaker F: Because you don't have a map of who's talking versus their name that they're being referred\nSpeaker E: to.\nSpeaker F: I thought we were going to be referred to. Sure, but then you have to know that Jose is Speaker One.\nSpeaker D: Then you have to know who's being referred to.\nSpeaker D: Okay, so suppose someone says, well, I don't know if I really heard what, what, what Jose said.\nSpeaker D: And then Jose responds.\nSpeaker D: And part of your learning about the dialogue is Jose responding to it, but it doesn't say Jose, it says Speaker Five.\nSpeaker D: Okay.\nSpeaker E: So, oh, I see you want to associate the word Jose in the dialogue with the fact that then he responded.\nSpeaker D: Right.\nSpeaker D: And so if we pass out the data to someone else and it says Speaker Five there, we also have to pass them this little guide that says Speaker Five is Jose.\nSpeaker D: And we're going to do that.\nSpeaker F: We might as well give them Jose, say, and that violates our privacy issue.\nSpeaker B: Now, I think that we have these two phases in the data, which is the one which is our US University of Washington's use, IBM SRI.\nSpeaker B: And within that, it may be that it's sufficient to not change the, to not incorporate an automatization yet, but always, always in the publications we have to.\nSpeaker B: And I think also when we take it that next step and distribute it to the world we have to.\nSpeaker B: I don't, that's a long way for now.\nSpeaker B: And it's a matter between now and then.\nSpeaker B: I have to say.\nSpeaker B: It makes a decision.\nSpeaker B: You know, it may be that we will need to do something like actually ex out that part of the, for the public one, the audio and just put in bracket speaker one.\nSpeaker F: You know, what we could do also is have more than one version of release, one that's public and one that requires licensing.\nSpeaker F: And so the license one would, we could, it would be a sticky limitation.\nSpeaker F: You know, like, well, we can talk about that later.\nSpeaker B: I think that the public should be the same.\nSpeaker B: I think that when we do that world release it should be the same.\nSpeaker B: I agree with Jane.\nSpeaker D: I think that we, we have a need to have a consistent licensing policy as I'm sort.\nSpeaker B: I also think it consists of licensing.\nSpeaker E: Well, one thing to, to take into consideration is, are there any, for example, the people who are funding this work, they want this work to get out and be useful for discourse?\nSpeaker E: If we all of a sudden do this and then release it to the public and it's no longer useful for discourse, you know.\nSpeaker F: Well, depending on how much editing we do, you might be able to still have it useful.\nSpeaker F: Because for discourse, you don't need the audio, right?\nSpeaker F: So you could bleep out the names in the audio and use the anonymized one through the transcript.\nSpeaker F: Excuse me.\nSpeaker F: But if you release the audio for discourse, but, excuse me, you could bleep out just the names.\nSpeaker D: No, but she's saying from the argument before she wants to be able to save someone said Jose in their, in their thing and then connect to, right?\nSpeaker F: But in the transcript, you could say everywhere they said Jose that you could replace it with speaker seven.\nSpeaker F: Oh, I see.\nSpeaker B: Yeah, but I also want to see you.\nSpeaker F: And then it wouldn't match the audio anymore, but it would be still useful.\nSpeaker F: But both of those are publicly available.\nSpeaker D: But they, right.\nSpeaker D: And the other thing is if Liz were here, what she might say is that she wants to look at things that cut across between the audio and the dialogue.\nSpeaker B: And so, yeah, I think we have to think about how, I think this can't be decided today.\nSpeaker B: Yeah, okay.\nSpeaker B: It was good to introduce the thing in.\nSpeaker F: When I wrote you that email, I wasn't thinking it was a big can of worms, but I guess it\nSpeaker B: is.\nSpeaker F: Yeah, a lot of discourse.\nSpeaker B: Well, discourse, you know, also, I'm going to make the point that discourse is going to be more than just looking at a transcript.\nSpeaker B: It's going to be interesting.\nSpeaker B: Oh, yeah, sure.\nSpeaker B: And prosoprosotic stuff is involved.\nSpeaker B: And that means you're going to be listening to the audio.\nSpeaker B: And then you come directly into this, confronting this problem.\nSpeaker E: Maybe we should just not allow anybody to do research on discourse.\nSpeaker E: I'm really glad.\nSpeaker B: I wish you just marketed to non-English speaking countries.\nSpeaker D: Maybe we should only have meetings between people who don't know one another and who are also amnesiax who don't know their own names.\nSpeaker B: Did you mean the paper and yours?\nSpeaker B: We have little labels.\nSpeaker B: I want to introduce my reservoir dog solution again.\nSpeaker B: Mr. White.\nSpeaker B: Mr. White, Mr. Pink.\nSpeaker F: Did you read the paper a few years ago where they were reversing the syllables?\nSpeaker F: They had utterances and they would extract out the syllables and they would play them backwards.\nSpeaker E: But the syllables were in the same order.\nSpeaker E: Everything was in the soil.\nSpeaker F: But the individual syllables were played backwards.\nSpeaker F: And you could listen to it.\nSpeaker F: And it would sound the same if people had no difficulty interpreting it.\nSpeaker F: So what we need is something that's the reverse.\nSpeaker F: That a speech recognizer works exactly the same on it, but people can't understand.\nSpeaker F: Oh, well, that's easy way to do that.\nSpeaker D: Just play it all backwards.\nSpeaker D: Oh, right.\nSpeaker D: That speech recognizer is totally symmetric.\nSpeaker D: Was the speech recognizer correct?\nSpeaker D: No, anyway.\nSpeaker D: How do we do digit zero?\nSpeaker F: We already missed the party.\nSpeaker F: Okay.\nSpeaker F: Reading 1471-1490-9015-2430-39425456084833936-03702-08348302 Strike that 086311823053900628375168729039989276\nSpeaker D: Transcript 1531-155012050830536778 9896 0 0 0 131 2 1 0 35 4 2914 5702 6 7 8 0 308 0 0 1 0 1 14 9 1 0 0 3\nSpeaker E: Transcript 1631-16505 0 7 2 8 5 9 8 6 8 7 0 8 3 0 9 1 1 0 8 4 1 2 0 4 3 5 2 6 5 8 7 7 2 7 8 9 0 0 8 0 3 1 3 1 2 6 3 3 4 7 1 5\nSpeaker B: Transcript 1331-1350305396804 6 2 7 6 8 8 9 0 9 7 6 9 9 4 6 0 0 0 3 2 1 0 5 3 4 4 2 7 5 0 4 6 7 3 5 8 9 5 7 2 9 8 2 3 6 7 8 8 0 8 8 0 2 9 0 4 7 1 8 0 9\nSpeaker C: 9 2 6 5 3 903. 7 9 7 6 3 0 0 7 0 0 4 0 8 6 8 7 1 3 1 2 3 0 3 0 5 3 3 0 6 6 5 8 7 8 2 0 7 0 0\nSpeaker E: Okay, go off here. I think it would be fun sometime to read them with different intonations, like is if you were talking like 9 8 6 8 7.\nSpeaker B: Well, you know, in the one I transcribed, I did find it, I found one instance of contrastive stress where it was like the string had a, so it was like 9 8 2 4 9 9 2 4.\nSpeaker B: Oh, really? So they were like looking ahead, huh?\nSpeaker B: I mean, they differed. I mean, at that, that session, I did feel like they did it more sentences, but sometimes people do it as phone numbers.\nSpeaker B: I mean, I, I'm interested in, and sometimes, you know, and I, I never, when I do it, I, I ask myself what I'm doing.\nSpeaker E: Yeah, well, I was thinking that it must get kind of boring for the people who are going to have to transcribe this.\nSpeaker B: Well, except Rowan's interesting information.\nSpeaker B: I like your question, that's very funny. I haven't heard that one.\nSpeaker F: We have the transcript. We have the actual numbers they're reading, so we're not necessarily depending on that.\nSpeaker F: Okay, I'm going to go off.\n"}, {"meeting_id": "IS1006d", "summary": "It was the last meeting of the group and focused on detail design. First, the user interface designer displayed the appearance and the texture of the product with a working prototype. After that, the industrial designer gave some technical specifications on the product. In the product evaluation that included several criteria, the controller got an average score of two point one seven, which implied it was good. When it came to the quotation of product components, the original one was seventeen dot eight dollars, which was apparently over budget. After some discussions about battery and operations on the calculator, the group finally cut the cost down to under twelve dollars. Finally, they did some self-assessment and celebrated the completion of the project.", "dialogue": "None: Oh that rope Okay\nSpeaker A: So, it's the detailed design meeting. So we're going last meeting.\nSpeaker A: So first, Mark and Rama are going to present the prototype.\nSpeaker A: Then Samu will propose some criteria to evaluate this prototype.\nSpeaker A: Then we are going to do some finance to see if it is feasible.\nSpeaker A: Then at the end we will evaluate ourselves as a team.\nSpeaker A: So first, let's see the prototype.\nSpeaker C: Here we have our prototype model.\nSpeaker A: And you have some slides?\nSpeaker A: Yeah, some slides.\nSpeaker G: So in which... in participant 3 prototype.\nSpeaker G: So this is our remote control.\nSpeaker G: It's a working prototype.\nSpeaker G: You can use it now by switching all these buttons.\nSpeaker G: So first I present as we came to this perfect model and then we'll give some technical specifications.\nSpeaker G: So let's start.\nSpeaker G: Please, next slide.\nSpeaker G: We analyzed all the fruits and contacted NASA and made some real good.\nSpeaker G: Yeah, if you can see this and the stars are showing that we are doing right and society will accept it, for sure.\nSpeaker G: Making some analysis of different fruits, we choose the ultimate form, ultimate colors and ultimate smell of it.\nSpeaker G: Please, next slide.\nSpeaker G: But we still didn't want to go far from our titanium idea because it's the most of the model, the ever-murdering material we can select.\nSpeaker G: And it's practical and it still sits for our needs.\nSpeaker G: So please press something.\nSpeaker G: And as I said...\nSpeaker G: Hey!\nSpeaker G: Hey, this person, everyone is really, really glad to obtain this such a device.\nSpeaker G: So you can touch it with your own hand.\nSpeaker A: Can I?\nSpeaker A: Sure.\nSpeaker A: Yes.\nSpeaker E: What do you say?\nSpeaker E: It says...\nSpeaker F: No.\nSpeaker D: I'll buy it.\nSpeaker F: What do you say if I need to?\nSpeaker F: Hopefully my daughter will like it.\nSpeaker G: Okay.\nSpeaker G: And we got the answer.\nSpeaker G: Yes, of course.\nSpeaker F: Of course.\nSpeaker G: Of course.\nSpeaker G: Please, next slide.\nSpeaker G: This is the prototype you can have a look at it.\nSpeaker G: And that's all I wanted to say.\nSpeaker G: Now it's technical specification by our colleague.\nSpeaker D: So there's a button missing.\nSpeaker D: Okay.\nSpeaker D: This is really flexible.\nSpeaker D: I have your button.\nSpeaker C: Option.\nSpeaker C: Yeah.\nSpeaker C: So function.\nSpeaker C: So as we discussed, we have to switch on, switch off whenever we want.\nSpeaker C: And so we have buttons and using LCD or like you can use this jock wheel and select which are option on LCD and then do on and off.\nSpeaker C: Then you will have volume control.\nSpeaker C: So you can press this button to increase or decrease your volume.\nSpeaker C: And we have some LCD controls like switching the LCD display.\nSpeaker C: If you want to use LCD or you don't want, you can just use normal button.\nSpeaker C: And we have speech recognizer.\nSpeaker C: Here you have microphone and then it records your voice and then try to recognize and it converts into the action and location finder.\nSpeaker C: And we want to do the location basically using speech recognizer.\nSpeaker C: You can just say, where is my remote control or you can just give some nickname to your remote control.\nSpeaker B: And then Bob, you can see it.\nSpeaker B: You are tired then.\nSpeaker B: You can use it.\nSpeaker C: So our theme is no fruits, mainly strawberry.\nSpeaker C: So you can have, oh these are strawberries.\nSpeaker C: And then you can see the look LCD and all the switches.\nSpeaker A: Colorful.\nSpeaker C: Yeah.\nSpeaker C: So in the material, we want to stick to titanium.\nSpeaker F: And you want to root spelting, spelky, titanium.\nSpeaker F: I didn't know it exists but that's great.\nSpeaker C: So we want to have simple and perfect shapes like I shown in these fonts.\nSpeaker C: You can have your own designs and you can feel simple designs.\nSpeaker C: And you can choose colors on your day for each day or even many colors.\nSpeaker A: You mean we can change the color.\nSpeaker A: That's for the LCD or for the titanium.\nSpeaker C: For the LCD.\nSpeaker C: Oh, okay.\nSpeaker C: For the LCD.\nSpeaker C: We have titanium.\nSpeaker C: It's still working on titanium.\nSpeaker G: Okay.\nSpeaker G: And you can start with LCD, you can ask Bob, it's Tuesday.\nSpeaker B: Bob, please.\nSpeaker B: Hey, you know you're being today.\nSpeaker C: Tuesday color.\nSpeaker C: Even you can configure your colors for each day, depending on your mood or black.\nSpeaker C: You can have many colors on weekend.\nSpeaker A: And wait, what are the strawberries for?\nSpeaker C: These are like sensors.\nSpeaker C: Of course, what do you think?\nSpeaker E: That's strawberry.\nSpeaker A: They're useful.\nSpeaker A: Okay.\nSpeaker A: Strawberry.\nSpeaker C: So after this meeting, we will propose a party for our success.\nSpeaker C: For large meeting.\nSpeaker C: So if you are vegetarian or you have any options, please let us know.\nSpeaker G: Yeah, and we can just treat some strawberries first.\nSpeaker B: Thank you.\nSpeaker B: Thank you.\nSpeaker B: Thank you.\nSpeaker A: We'll see you in the financial party.\nSpeaker A: That's what we get.\nSpeaker G: Let's make a party first.\nSpeaker C: Then we can discuss.\nSpeaker C: We can...\nSpeaker A: With the five, 50 million, we first make a party.\nSpeaker C: Then we can have all of its plans, money is left.\nSpeaker A: What a design.\nNone: Okay.\nSpeaker A: So...\nSpeaker A: It's my turn.\nSpeaker F: Yeah, let's see if you meet the evaluation criteria.\nSpeaker F: So evaluation, please.\nSpeaker F: So you made a very nice prototype and I think we now need all together to try to evaluate it, to see if it makes sense to do it, if it fulfills what we wanted to and things like that.\nSpeaker F: So...\nSpeaker F: Next slide, please.\nSpeaker F: As you know, before going and producing these strawberries, remote control, it's very important to first verify if it makes sense, if we have a chance to sell it.\nSpeaker F: So we need to evaluate it and try to do it in a quantitative way as much as we can.\nSpeaker F: So what I propose is that we are going to have this scale from one to seven.\nSpeaker F: One, meaning that yes, it fulfills the criterion, whatever it is.\nSpeaker F: And seven, meaning no, it doesn't fulfill at all.\nSpeaker F: And while going to list all the criteria, I'm going to go to that next slide.\nSpeaker F: And together try to evaluate this according to this criterion from one to seven.\nSpeaker F: And then we're just going to have an average which will give us the value of our remote control.\nSpeaker F: Maybe we can have a look at the criteria.\nSpeaker F: So these are the criteria I thought were important.\nSpeaker F: Of course, this can be discussed, but let's see.\nSpeaker F: So let's suppose, so we have fancy here.\nSpeaker F: And we have this scale from one to seven.\nSpeaker F: We have four in the middle.\nSpeaker C: So, what do you think?\nSpeaker F: Is it fancy?\nSpeaker A: No, I think that fancy we can say it is fancy.\nSpeaker F: It is very, very fancy.\nSpeaker F: Have you ever seen something like that?\nSpeaker A: I am not the only one who thinks.\nSpeaker F: What do you think?\nSpeaker F: Feel the weight.\nSpeaker F: The weight is later.\nSpeaker F: I think it is quite fancy.\nSpeaker F: It is one or two.\nSpeaker F: Two, maybe two.\nSpeaker F: So here, two.\nSpeaker F: Then we have technology.\nSpeaker F: So what about technology?\nSpeaker F: We have speech recognition, location based.\nSpeaker F: We have a CD.\nSpeaker F: Change color.\nSpeaker F: I think it is one for that.\nSpeaker C: Robustness.\nSpeaker F: So let's suppose my daughter take it and throw it away.\nSpeaker F: Do you think it makes sense that it is going to live again?\nSpeaker F: Maybe not the prototype.\nSpeaker F: Let's try.\nSpeaker F: Oh my god.\nSpeaker G: We just lost one strawberry.\nSpeaker G: No, but it is still working.\nSpeaker F: You have to go to a bonus.\nSpeaker F: So it is not so bad.\nSpeaker F: It is looks like it is robust.\nSpeaker F: It is useful.\nSpeaker F: The question is, does it have the minimum requirement of remote control?\nSpeaker F: I don't know.\nSpeaker F: These buttons are not clear, but you have at least next.\nSpeaker C: What is the next challenge?\nSpeaker A: This is volume control and channel change.\nSpeaker C: You can do it on LCD using this.\nSpeaker F: Scrolling all the options.\nSpeaker F: Because LCD is not touch control, touch screen, you cannot go to channel 25.\nSpeaker G: You can by using the...\nSpeaker G: You go...\nSpeaker G: So it is a basic mode.\nSpeaker G: You have just two buttons and a drug dial.\nSpeaker G: If you go to the side, it is channel up, channel down.\nSpeaker G: If you want to make 25, you push on this.\nSpeaker G: You select 20, you select 5.\nSpeaker F: It is much longer than tapping 25.\nSpeaker F: Don't you think so?\nSpeaker F: We can go.\nSpeaker F: It is nice because people don't go there.\nSpeaker F: So what do you think for it?\nSpeaker F: It seems to be useful.\nSpeaker C: If we need to address, we only address two main functions here and the other functions will be on LCD.\nSpeaker A: Let me understand why, because I am not sure.\nSpeaker A: This one is two direction.\nSpeaker A: Up, down or left, right.\nSpeaker A: Which one is that?\nSpeaker G: The job is to control the cursor on the LCD.\nSpeaker C: You will have blocks and you will select which one.\nSpeaker F: I would say that...\nSpeaker F: No, it looks useful.\nSpeaker F: Two or three?\nSpeaker F: Two or three?\nSpeaker F: So size and weight.\nSpeaker F: Is it the effective size and weight?\nSpeaker F: Is it real size, real weight?\nSpeaker F: Size is almost size.\nSpeaker C: Size is going to be that.\nSpeaker C: The weight will be a bit lighter.\nSpeaker C: Sure, without titanium.\nSpeaker G: Titanium is going to be lighter.\nSpeaker G: It is going to be lighter.\nSpeaker F: This seems to be very heavy.\nSpeaker F: From my daughter, for instance.\nSpeaker F: I am not sure if she can use it.\nSpeaker C: The size should be okay.\nSpeaker F: I will put a three for that.\nSpeaker F: I have not seen the weight.\nSpeaker F: Color and shape.\nSpeaker F: Color seems to have several colors for the LCD.\nSpeaker F: It is not very clear what is the color of the case.\nSpeaker F: The case is too light.\nSpeaker F: It is going to be too light.\nSpeaker A: That is nice.\nSpeaker A: I think it is good.\nSpeaker A: What about the strawberries and top?\nSpeaker A: Maybe I am not trendy.\nSpeaker F: It is this fruit and vegetable year.\nSpeaker A: They are not useful.\nSpeaker F: I think it is useful as I just have to remind you that usefulness is much less important than fanciness.\nSpeaker F: Whether it is fancier or not now.\nSpeaker F: We have to decide.\nSpeaker A: I would have found more fancy that the fruits are useful.\nSpeaker F: That maybe the fruit may be here.\nSpeaker C: Then it is difficult to use.\nSpeaker C: We are just giving the fruits for more fanciness and more and more.\nSpeaker C: Maybe we can have rubber or some sponches and stuff for strawberries.\nSpeaker F: It seems to be even so clear.\nSpeaker A: I am not sure if it was like this.\nSpeaker C: It is not really good anymore.\nSpeaker C: These are kind of rubber things.\nSpeaker C: We can provide many different colors.\nSpeaker G: It covers all the angles.\nSpeaker G: Even if it is very rounded.\nSpeaker G: But still we have some rubber fruit here.\nSpeaker G: It is completely secure to live with children.\nSpeaker A: You feel like it is something of protection for the remote control.\nSpeaker C: We have sensors here and here.\nSpeaker C: Even if you do not put it on, this is really...\nSpeaker C: I suggest a 3.\nSpeaker F: Everybody does not seem to be convinced.\nSpeaker F: Although it is great.\nSpeaker F: You have good arguments.\nSpeaker F: The last one is adaptive.\nSpeaker F: This is not as important as the other one.\nSpeaker F: But can we adapt it to each personal use?\nSpeaker G: Sure, just look at it.\nSpeaker H: It is fully adaptive.\nSpeaker F: So you can fit it to your palm.\nSpeaker F: Do you think you are going to be able to do that with Titan?\nSpeaker A: It is fudge Titan.\nSpeaker F: If you are able to do that, then I think it deserves a one.\nSpeaker F: Let's go for one.\nSpeaker F: Now we have to do the average.\nSpeaker F: It is 2.17.\nSpeaker F: 2.17.\nSpeaker F: That is nice.\nSpeaker F: I think we have a good thing.\nSpeaker F: That is all I have to say about evaluation.\nSpeaker F: It is a good evaluation.\nSpeaker C: It seems to be good.\nSpeaker A: Thanks.\nSpeaker A: Now it has to fulfill the financial criteria.\nSpeaker F: How many batteries do we need?\nSpeaker F: One battery?\nSpeaker F: Two.\nSpeaker C: This is one.\nSpeaker E: No.\nSpeaker F: Two batteries are one.\nSpeaker A: No, number is one.\nSpeaker A: No, no, no, no.\nSpeaker E: You never use an XL?\nSpeaker E: What is the limit?\nSpeaker G: It is okay that I do not know.\nSpeaker G: 12 bucks.\nSpeaker E: Check the number also.\nSpeaker F: We have a sample chip.\nSpeaker C: One.\nSpeaker C: And then we have the sample speaker sensor for speech recognition.\nSpeaker F: One or two?\nSpeaker F: So the case, which one is it?\nSpeaker C: I think we will go for a single.\nSpeaker E: It is flat.\nSpeaker E: I thought you can curve it.\nSpeaker E: Maybe there is a superman for that.\nSpeaker A: I think the price is this one.\nSpeaker H: Don't you?\nSpeaker F: I think we tried.\nSpeaker F: What is it?\nSpeaker A: That is expensive.\nSpeaker F: I think it is five.\nSpeaker A: You know why?\nSpeaker A: I think it is a special color.\nSpeaker C: There is no color here.\nSpeaker A: How many buttons?\nSpeaker A: Two.\nSpeaker B: Is there the score?\nSpeaker C: One.\nSpeaker A: This one.\nSpeaker A: That is not.\nSpeaker A: We choose this one and not this one.\nSpeaker F: I think no.\nSpeaker A: Is it a scroll wheel and pinch button?\nSpeaker A: How do you do to select?\nSpeaker F: You select with the other two buttons.\nSpeaker A: You go on the location with your screen.\nSpeaker C: Then you automatically, you can just do like if you go.\nSpeaker C: You have all of these.\nSpeaker A: She is very hard on this.\nSpeaker E: Maybe not this one.\nSpeaker F: No.\nSpeaker F: We have buttons.\nSpeaker F: Nothing special.\nSpeaker A: We are at 17.8.\nSpeaker A: The color is in the LCD.\nSpeaker C: I think we can just use this red.\nSpeaker A: It is already too expensive.\nSpeaker F: I am not ready.\nSpeaker F: I am always supposed to cut things out now until we get 12.\nSpeaker A: Think of what we can cut here.\nSpeaker F: If I look at what is the most expensive thing, it is the LCD and the speaker.\nSpeaker A: Apparently we have to choose one or the other.\nSpeaker G: As you may know, there is some research done in the field of producing energy from mechanical energy.\nSpeaker G: The point is that when you take a device and push the button, you produce enough energy to make electricity.\nSpeaker G: It is something like hand dynamo, but a real high-tech version of it.\nSpeaker A: But it is like the hand dynamo.\nSpeaker C: Maybe the jack wheel can be lit.\nSpeaker F: If we select the hand dynamo, it is okay.\nSpeaker F: We win one.\nSpeaker G: That is already there.\nSpeaker A: Let's do that.\nSpeaker G: I propose about cheaps, advanced, cheap on print.\nSpeaker G: Put minus one there.\nSpeaker D: That is illegal.\nSpeaker A: Let's try.\nSpeaker A: Maybe minus three.\nSpeaker C: What is the result?\nSpeaker F: It is not changing.\nSpeaker F: Maybe put minus two.\nSpeaker E: Is it not?\nSpeaker H: Anyway, the minus three.\nSpeaker F: Nobody will know.\nSpeaker F: It is not recorded.\nSpeaker F: We are on time.\nSpeaker C: We can increase the volume.\nSpeaker A: We can put hand dynamo on the battery.\nSpeaker F: Now we are exceeding.\nSpeaker G: Let's add one instead of two.\nSpeaker C: I think we are exceeding now.\nSpeaker F: It is better.\nSpeaker F: I think the accounting is maximum.\nSpeaker F: We will remove one of them.\nSpeaker A: We are on target.\nSpeaker A: Target reached.\nSpeaker F: I am curious to see this.\nSpeaker A: I would say it is Russian.\nSpeaker F: It is a Russian trick.\nSpeaker A: It is English.\nSpeaker G: It is a Russian trick.\nSpeaker G: I am not sure who was programming this.\nSpeaker G: I wonder if we put A or B somewhere.\nSpeaker D: Let's finish this.\nSpeaker E: We can discuss all these things in our party.\nSpeaker A: What else?\nSpeaker E: We have to make a self-assessment.\nSpeaker F: Are we a good team?\nSpeaker F: We have to listen to everybody.\nSpeaker G: Is there enough room for creativity?\nSpeaker F: When we see the results, no doubt.\nSpeaker F: Maybe lack of leadership.\nSpeaker D: Teamwork is strong.\nSpeaker F: It means whiteboard digital pens.\nSpeaker G: What was good?\nSpeaker E: Digital pens useful.\nSpeaker A: We have to make a new idea.\nSpeaker A: So lack, but good.\nSpeaker G: We can play a good team.\nSpeaker F: It is somewhere in the shade.\nSpeaker F: We are on time.\nSpeaker F: We have to finish before.\nSpeaker E: We made a far meeting.\nSpeaker F: Not the best time.\nSpeaker C: We have a new idea, speech recognition, location, finding, new materials, new materials, new services, fancy strawberry design.\nSpeaker A: New ways of doing finance.\nSpeaker F: Just wondering if I can make the pinkness of that.\nSpeaker A: They are working on a pink titanium.\nSpeaker C: We are great.\nSpeaker E: No other words for that.\nSpeaker E: We are the best real reaction.\nSpeaker A: We are really nice.\nSpeaker A: Finish?\nSpeaker A: I think it is.\nSpeaker F: Are the cost we need the budget?\nSpeaker C: Is the project evaluated?\nSpeaker F: Yes, we got two.\nSpeaker G: We can even forecast all the day proposals, like celebration, we could forecast it.\nSpeaker F: To the whole company?\nSpeaker F: I am the one proposing the celebration.\nSpeaker A: I am the program manager.\nSpeaker F: Let's celebrate.\nSpeaker H: I think the meeting is over.\nSpeaker C: We have to.\nSpeaker A: We go to the party.\nNone: Thank you.\nNone: We are going to go to the party.\nNone: We are going to go to the party.\n"}, {"meeting_id": "Bmr012", "summary": "The group discussed recognition results generated for 20 minutes of close-talking microphone data. Recognition performance was very good, indicating promising results for forced alignment procedures and the ability to analyze other important signal information, e.g. prosody and overlapping speech. It was decided that close-talking data should be downsampled and fed to the SRI recognizer to compare recognition performance, and that data from the far-field microphones should be tested on the recognizer as soon as possible. The group also discussed recording setup and equipment issues. ", "dialogue": "None: Like that that wasn't picked up on the Well just to convert it to the\nSpeaker C: Where am I? What's the best placement for these lip column lights?\nSpeaker C: It's close to your mouth now. Is this channel one?\nSpeaker H: Okay, yes\nSpeaker J: Stay in the light For people wearing the wireless mics like like this one. I find the easiest way to wear it is sort of this sort of It's actually a lot more comfortable than if you try to put it over your temples\nSpeaker I: So you do it higher\nSpeaker J: And then also for all of them with your boom is adjustable Oh, it should be towards the corner of your mouth. It wasn't using the proper and about Uh, it wasn't a damn thing. I'm gonna have distance away from you about so about like I'm wearing it now\nSpeaker I: So why didn't you get the same result?\nSpeaker C: So Jane you could actually do even closer. Why didn't you get the same result as the end of it? Oh because when it's the\nSpeaker J: Oh Channel five speak again We could we could get up again slightly if you wanted to whose channel B\nSpeaker F: but\nSpeaker J: Channel B is probably Liz\nSpeaker F: No, yeah, yeah, you're gonna be so can you talk of it?\nSpeaker J: All right, so again, isn't real good Okay, so we are recording Everyone should have at least two forms possibly three in front of you depending on who you are We we're doing a new speaker form and you only have to speak fill out the speaker form once But everyone does need to do it and so that's the name, sex, email, etc We we had a lot of discussion about the variety of English and so on so If you don't know what to put just leave it blank It's like I I designed the form and I don't know what to put for my own region\nSpeaker C: So California California, may I make one suggestion instead of age put data a year of birth because age will change Oh your first changes it stays the same usually Perth Well the thing is if ten years from now you look at this form Yes, but we care about a C-A-T\nSpeaker I: But there's no other day on the form\nSpeaker F: Well, I don't know\nSpeaker J: Well, I guess depends on how long the corpus is going to be collected for I think age is all right and then there will be attached to this a pointer to these forms Uh, so that you'll be able to extract the data off that So anyway, and so then you also have a digits form Which needs to be filled out every time the speaker form only once the digit form every time Even if you don't read the digits you have to fill out the digit form so that we know that you are at the meeting Okay, and then also if you haven't filled one out already you do have to fill out a consent form And that should just be one person whose name I don't know Oh, sure Thank you So should we do agenda items?\nSpeaker H: Oh, that's a good idea. I shouldn't run the meeting\nSpeaker J: Well, I have I want to talk about new microphones and wireless stuff And I'm sure Liz and Andreas want to talk about recognition results Anything else\nSpeaker F: I guess what time do you have to leave? 330 yeah, why don't you go first then yeah good idea Well, I sent out an email a couple hours ago so With Andreas's help Andreas put together a sort of no frills recognizer which is a gender dependent but like no adaptation no crossword models no Trigrams a bi-graph recognizer and that's a train on switchboard, which is telephone conversations um And thanks to don's help who don took the first meeting that Jane had transcribed and You know separated use the individual channels we segmented it into the segments that Jane had used and Don simple dad so Okay, and then we ran up to I guess the first 20 minutes up to sync time of one two zero zero So that's my minutes or so um, yeah, because I guess there's some and Don can talk to Jane about this There's some bug in the actual sync time file that I'm we're not sure where it came from but stuff after that was a little messier Anyway, so it's 20 minutes and I actually\nSpeaker J: I was that did that did that recording have the glitch in the middle oh I see there's there's a\nSpeaker F: So that actually Was 20 minutes in then it was interesting\nSpeaker C: I don't know when it is the the overall air raid when we first ran and was like 80 It's said But looking at the first sensors looked much better than that and then suddenly it turned very bad And then we noticed that the reference was always one off with\nSpeaker F: Yeah, actually Also that was just no actually it was yeah, it was a complicated bug because they were sometimes one off and then sometimes totally random So yeah, I was too certain that it worked up until that yeah So that's what we have but that that will be completely gone if the so we have everything rationalized but we scored only the first\nSpeaker C: Whatever not to the time\nSpeaker J: There's an acoustic glitch that occurs where um the channels get slightly asynchronous So that that problem has gone away in the original driver I believe you're not when the SSH key gen ran the driver paused for fraction of a second And so the channels get a little asynchronous and so if you listen to it in the middle There's a little part where it starts doing doing click sounds\nSpeaker F: And is it only once that that happens? Yeah, right once in a while there's the previous page has some more information about sort of what was run\nSpeaker J: So I'm surprising that that should affect anything\nSpeaker F: Ah Oh, it happens is it actually affects the script that don't I mean if we know about it then I guess I guess it's well the acoustic one shouldn't do anything\nSpeaker J: I don't know exactly what effect to the file\nSpeaker G: I have about it. I'll show you\nSpeaker E: Yeah, I don't affect my transcription. I do remember hearing it and having time\nSpeaker J: But I do remember seeing once transcriber produced an incorrect XML file where one the sink numbers was in\nSpeaker F: Correct the sink time the sink numbers had more significant digits. Yeah, I mean they should right there's things that are in smaller increments than a frame Okay, so then I mean you look at that and it's got you know more than three significant digits in a sink time Then it can't be right. Yeah, it's like a bug anyway. It's it's just that's why we only have 20 minutes\nSpeaker G: But there's a significant non zero on of there like more the other one I saw was attacked on just because of the way the script ran I mean yeah, there was a point\nSpeaker J: Yeah, that was fine the other one I saw was non non monotonic sink times and that definitely\nSpeaker F: Decorate well that would really be a problem. Yeah, so anyway, these are just the ones that are the pre-bug for one meeting So that's very encouraging which this is really encouraging because this is free recognition There's no I mean the language model for switchboard is totally different. So you can see some like trend lot trend lot Which I mean these are funny ones. There's a lot of perfect ones and good ones and all the references I mean you can read them and we get more results\nSpeaker J: I can look through and see but and as I said I would like to look at the lattices because it sounded like even the ones It got wrong it sort of got right well\nSpeaker C: So I guess we can generate there are a number of errors that are you know what got the plural as wrong\nSpeaker J: Yeah, there were lots of of course the uh-uhs in on's no those are actually a lot of the errors\nSpeaker F: I think are out of vocabulary so it's like pzm is three words pzm. There's nothing right there's no language model for pzm Or did you say there's no language for pzm no language model. I mean\nSpeaker J: Do you mean so every time someone says pzm? It's an error. Maybe we shouldn't say pzm in these\nSpeaker F: Well, this all kinds of other stuff like jimlet Yeah, that's right in jimlet Anyway, well we don't even know what that means But this is really encouraging because so I mean the bottom line is even though it's not a huge amount of data It should be Reasonable to actually run raking dish and be like within the scope of Of reasonable you know switchboard. This is like about how well we do on switchboard to data with a switchboard one trained or trained recognizer and switchboard two in got sort of a different population of speakers and a different topic They're talking about things in the news that happened after switchboard one so there was\nSpeaker H: That's great. Yeah, so we're better. We were safe when we had the 93 workshop and we were all getting like 70% error on switchboard\nSpeaker F: Oh, yeah, I mean this is really and thanks to Andreas. I mean this is a yes, especially for the very first run\nSpeaker J: I mean the first run I ran a switchboard. I got 120% where there\nSpeaker F: So and what also this means is that um I mean there's a bunch of things in this node to various people especially I guess With Jane that that would help since we have this new data now in order to go from the transcripts more easily to Just the words that the recognizer would use for scoring. I had to do of some of it by hand But I think a lot of it can be automated. Oh, I guess I didn't get so yeah the language model was straight from\nSpeaker H: From biogram from switchboard the acoustic models were also from switchboard. Yeah, yeah\nSpeaker F: So they didn't have anything from this no and actually we actually Use switchboard telephone bandwidth models We just get a yeah, yeah, that's the only Acoustic training data that we have a lot of and I guess Ramana so a guy at SRI said that There's not a huge amount of difference going from it's not like we probably lose a huge amount But we won't know because we don't have any full band models for conversational speech\nSpeaker I: Probably not as bad as going using full band models on telephone band speech. Oh, yeah, right. Yeah\nSpeaker H: But for broadcast news when we played around between the two there wasn't a huge right yeah, so I was\nSpeaker C: I'll say the the language model is not just switchboard. It's also I mean there's Actually more data is from broadcast news, but with a less weight Like trend lot must have been for switchboard was By the way just for fun we also ran I mean our complete system starts by doing a gender detection So just for the heck of it and it might be reassuring to everybody to know that it got all the Yeah\nSpeaker J: It got all too\nSpeaker F: Jane and Adam have about equal performance and And that's interesting because I think the their language models are quite different So and I'm pretty sure from listening to Eric that you know given the words he was saying and given his Pronunciation that the reason that he's so much worse is the lapel right then yeah, so it's nice now If we can just sort of eliminate the lapel when when we get new I was that would be worse\nSpeaker H: I would bet on that because he certainly in that when as a as a burp user He was he was a pretty strong sheep\nSpeaker F: Yeah, he sounded to me just from he sounded like a what's it a sheep or a sheep sheep Kind of good Right so um so I guess the good news is that and again, this is without a lot of the sort of bells and whistles that we can do with the SRI system We'll have more data and we can also start to Maybe adapt the language models once we have enough meetings So this was like 20 minutes of one meeting with no no I mean clearly there\nSpeaker C: Um was just a small amount of actual Meeting transcriptions It's thrown into the language model you can probably do quite a bit better because or just to compare the vocabulary especially Not that much of a vocabulary actually I think um well we have to see but\nSpeaker F: Yeah, it's pretty good So then I had to try it on the far field Mike. Yeah, easy. I mean then there's things like for the transcription like I when someone has a digit in the transcript I don't know if they said you know one one or eleven and I don't know if they said tickle or tcl This things like that were you know the We probably have to ask the transcribers to indicate some of those kinds of things but In general, it was really good and I'm hoping and this is This is good news because that means the force alignments should be good and if the force alignments I mean it's good news anyway, but if the force alignments are good we can get all kinds of information For example about you know exotic information and speaker overlaps and so forth directly from the aligned times So that'll be something that Actually in order to assess the force alignment We need some linguists or some people to look at and say or these boundaries in about the right place Because it's just gonna give us a done that for one meeting For force alignment\nSpeaker J: Uh oh not for words. I'm sorry. Just for overlaps\nSpeaker F: Right so this would be like if you take the words um, you know enforce some line them on all the individual close talking close talking mics\nSpeaker J: Then how good are these for in reality and then I was so we might want to take 20 minutes and do a closer word level transcription Maybe actually mark the word boundaries\nSpeaker F: Or you have someone look at the alignments maybe a linguist who can say You know roughly if these are okay and how far away they are Yeah But I think it's got to be pretty good because otherwise a word recognition would be really bright on me It wouldn't necessarily be the other way around if the word recognition was crummy the alignment might be okay But if the word recognition is this good the alignment should be pretty good So\nSpeaker I: I wonder if this is a good thing or a bad thing though. I mean if we're starting so well Yeah, if we're producing a database\nSpeaker J: Don't worry about everybody's gonna do one that's the close talking mics try it on the pzm\nSpeaker I: So the real value of the database is the\nSpeaker H: But I don't know but I mean there's still just the the percentage is not I mean Is we've talked about before there's probably overlaps There's probably overlaps in in Fair number in switchboard as well, so But there's other phenomena. It's a meeting. It's a different thing And there's lots of stuff to learn with close talking mics but Yeah, certainly I'd like to see as soon as we could I mean maybe Get some of the glitches out of the way, but as soon as we could how well it does would say Well the pzm's are maybe right and see if it's you know is it 120% or maybe is not maybe if With some adaptation you get this down to 50% or 45% or something and then if the pzm at 70 or something like that That's actually something that sort of\nSpeaker F: Worked with a little bit. Yeah. No, I think it's really I mean this way we at least have a baseline We know that for instance the transcripts are very good So once you can get to the words of the recognizer Which is its whole subset of the things you need to understand the text Either pretty good so and then converting automatically from the xml to the Chopping up the wave forms and so forth. It's not the case that the end of one Adirance is in the next segment and things like that which we had more problems with and switchboard So that's good and see there was one more thing I wanted to To mention Um, sorry, I can't remember Anyway, well, it was I mean I really didn't do this myself. So Andreas set up this recognizer By the way, they've recognized her all the files. I'm moving to SRI and running everything there So I brought back just these result files and people can look at them\nSpeaker C: We talked about setting up the SRI recognizer here That's You know if there are more machines He or plus people can could run their own You know variance of the recognition runs certainly doable. Yeah\nSpeaker H: Yeah, certainly for the recognition is most training. Yeah\nSpeaker E: Yeah, which is um so this issue of the legalistic aspects of the Pre-sett, you know pre adapted so what I mean is um the The data that you take in SRI first question you're maintaining it in a place that wouldn't be hard to readable like that kind of stuff. I Um the outside world or like people who are not associated with this project\nSpeaker F: It's human subjects issues. I told you about that. Oh, well, okay. We have no names\nSpeaker J: Although I that's not the issue. It's just the audio data itself until people have a chance to\nSpeaker F: And so well, I can I can protect my director's they're there right now. They're not they're in the speech group director's which So I will I didn't know that actually. Yeah, so we just have to go through this process. Yeah, okay\nSpeaker E: Yeah, okay, and then it could because the other question is gonna ask you see if we're having Um, this this meeting that you have no problem because I I well, I speak for myself, but but I think that we wouldn't be concerned about it with respect to that. I'll know we should clear with Of course, but these results are based on data which haven't had the Haven't had the chance to be that's true on the subjects and I don't know how that stands I mean if you get fantastic results and it's Nothing data which which later ended up being lessened by You know, I don't know but I want a transition. Well, we mean once we get all this streamline\nSpeaker H: Maybe hopefully\nSpeaker J: I used to work on a system for doing that approval so that we can send people to transcripts and get back any beliefs that they want\nSpeaker H: Yeah, hopefully this is gonna be a rare thing\nSpeaker C: I actually had a question about the down sampling um, I don't know who I mean how this was done But is there other any Issues with down sampling because I know that the recognize it Um, that we use can do it sort of on the fly um, so we wouldn't have to have it you know do it explicitly beforehand and And is there any um other other Is there more than one way to do the down sampling where one might be better than another\nSpeaker G: They're also a lot of ways to the down sampling. Okay, um different filters to\nSpeaker C: Right, okay, so so the I don't think we even know which one\nSpeaker J: I assume you're using some cat using our sound sample\nSpeaker G: SND Resample our example. Yeah, and dance archaic acronyms. I don't know\nSpeaker F: Missing all the balls. Yeah, not all of them almost\nSpeaker C: So the other thing we should try is to just take the original waveforms. I mean segment them but not down sample them And see them to feed them to the SRI recognize her and see if the SRI front ends\nSpeaker J: That's sort of premature optimization, but sure can try it. I only done one line\nSpeaker C: Right, and it doesn't is no more work for um\nSpeaker F: Well, they're just bigger trans for that's why\nSpeaker C: Well, but they're only twice as big so\nSpeaker F: Well, I mean that was if it's the same then we\nSpeaker C: Yeah, I mean it would be you know it would probably take about You know minus the transfer time. Yeah, it would take you know 10 minutes to try and 50 minute drive right Well, it takes more time. And if for some reason we see that it works better than we might investigate why and you know\nSpeaker G: So it's a guy's own it can download itself Yeah, yeah\nSpeaker H: Yeah, just different filters. Yeah, I can imagine whether the filters right\nSpeaker F: So we could try that with this particular 20 minutes of speech and sort of see if it's any different\nSpeaker C: But you know at some point someone might have optimized whatever filtering is done for the actual recognition um Performance so in other words\nSpeaker J: Right, it just seems to me that you know small changes to the language model and the vocabulary will so swamp that that it may be premature to worry about that I mean, so one is a half a percent better than the other. I don't think that gives you any information\nSpeaker F: Well, it's just as easy to to give you the 16k individual. It was just more just space, you know, right? Are you are you using\nSpeaker H: uh No, capstone or phoenix?\nSpeaker H: Capstone\nSpeaker F: Oh, we could trust\nSpeaker C: That's what I would assume, but you never know\nSpeaker H: No, I recently said this phoenix uses other aggressive filtering modeling and so We can be sensitive to the kind of filtering that you do, but No capstone Expect to be so much\nSpeaker F: We can try it if you generate like the same set of files just up to that point where we stopped anyway Yeah, it's just sticking somewhere in all the time. No, don't stop\nSpeaker C: Don't stop at that point because we're actually using the entire conversation to estimate the speaker parameters So shouldn't use you should you know, yeah, I mean, I have to do this\nSpeaker G: The reference spiral to stay the same. It's just the individual segment would be Right, right, and I can just replace them with a bigger one\nSpeaker F: So I mean I corrected all I mean I hand edited the whole the whole meeting so that can be run. It's just Once we get\nSpeaker E: I am the impression from this from this meeting that I transferred that um that there was already automatic down sampling and curing Is that I thought that in on two so there's one level that's already happening right here\nSpeaker H: This is being recorded 48 kilohertz right and it gets down to after the 16 years\nSpeaker F: And that's actually said in your meeting that's how I know that\nSpeaker H: Like are we down something to it's a digital audio orientation\nSpeaker F: Think that it's not more than that and I have no idea what filter it's using\nSpeaker G: So is it killer? It's generally accepted as like standard telephone. Yeah, that's it. I mean like\nSpeaker H: So it's it's just that they were operating from switchboard which was\nSpeaker J: So 16 seems to be pretty typical for\nSpeaker H: Right, so it's more common for for broadband stuff that isn't that isn't music\nSpeaker F: And I guess if you're comparing like If you want to run recognition on the pzm stuff you would want You don't want a down sample. Why is that?\nSpeaker F: I mean if it's any better\nSpeaker H: No, actually, I would think that you would get better and you'd get better high frequencies in the local mind But who knows yeah\nSpeaker J: You want to find all this stuff out? We're gonna have plenty of low frequency on the pzm's with the fans. Yeah\nSpeaker F: Well, yeah, there was just one more thing I wanted to say was just totally unrelated to the recognition except that um well Sort of related but good news also I got well Chuck Frillmore agreed to record meetings, but he had too many people in his meetings And that's too bad because they're very animated But Jerry also agreed so Uh, we're starting on their last Well, but he has fewer he won't have more than eight and it's a meeting on even deeper understanding EDU so that sounds interesting As a compliment to our front-end meeting and So that's gonna start Monday and one of the things that I was realizing is It would be really great if anyone has any ideas on some kind of Time synchronous way that people in the meeting can make a comment to the person who's gonna transcribe it or or put a Push a button or something when they want to make a note about oh boy. You should probably erase those last few or Wait, I want this not to be recorded now or\nSpeaker H: Something like that do something with a pad at one point. Yeah, we could do it with cross paths\nSpeaker F: I was thinking you know if if the person who sets up the meeting isn't there and it's a group that we don't know And this came up talking to to Jerry Also that you know is there any way for them to indicate to make sure that the requests that they have that they make explicitly get So I don't know if anyone has ideas or you can even write down. Oh, it's about three twenty five\nSpeaker H: Well, what I was just suggesting is we have these cross paths. Yeah, and use that for this motivator Yeah, that'd be great. I think you're gonna be great The other thing is I don't know if you know this surface a question for me to mail to Dan, but Is this thing of two H-anl boards a maximum?\nSpeaker J: Or could we go to a third board? Oh send me all the data and ask I think That it's the maximum we can do without a lot of effort because it's one board with two digital channels The eight each so it takes two fibers in to the one board And so I think if we want to do that more than that we'd have to have two boards And then you have the synchronization issue\nSpeaker H: But that's a question because that would if it was possible because it is you know already we have a Group people in this room. It's not all be mine right what not just because we haven't been to the store\nSpeaker I: What is the limit on each of those fiber channels is it the eight? It's just it's eight channels coming in does it have to do with sampling right? I have no idea\nSpeaker J: But each each fiber channel has eight eight channels and There are two two fibers that go into the card\nSpeaker H: It might be hard limitation. I mean one thing is the whole thing as I said is all structured in terms of 48 killer So that pushes yeah\nSpeaker J: I'm just wondering if I mean then we all have to get another 8-d and we can drive that mixer and all that sort of stuff So I'll send a mail to Dan asking Okay on the is that are we done with that?\nSpeaker J: So the other topic is getting more mics and different mics. So I got a quote We can fit we have room for one more wireless and the wireless this unit here is 350 350 dollars it I didn't realize but we also have to get a tuner the receiver the other and that's 430 um and then also\nSpeaker F: I mean the tuner is 430 for each yep, and we just need one more so so we got the good ones. Yeah\nSpeaker J: So that's you know something like 780 bucks for one more of these. Okay, I mean then also um It turns out that the connector that this thing uses his proprietary of Sony Believe it or not and Sony only sells this headset So if we want to use a different set headset the solution that the guy suggested and they apparently lots of people have done Is Sony will sell you the jack with just wires coming out the end and then you can buy a headset that has Pigtail and solder it yourself And that's the other solution and so the jacks are 40 bucks a piece and And he recommended um Crown CM311 AE headset for 200 bucks a piece\nSpeaker H: There isn't this some sort of thing that plugs in you actually have to go and do the soldering\nSpeaker J: So because the reason is the only only thing you can get that will plug into this is this mic or just the\nSpeaker H: The reason I asked is to sort of handmade wiring jobs fall apart in use So the other thing is to see if we can get them\nSpeaker I: To do a custom job put it together. Oh, I'm sure they would they would just charge us. Well, and they probably want quantity to\nSpeaker H: Well, no, I just charge it as one\nSpeaker J: So it's so so my question is should we go ahead and get net nine identical head mounted crown mics\nSpeaker H: Not before having one come here and have some people try it out. Okay Because there's no point in doing that if it's not gonna be any better\nSpeaker J: So why don't we get one of these with the crown with a different headset? Yeah, and and see if that works see if it's preferable\nSpeaker F: Because I mean I think the microphones are okay, right. It's just they're not comfortable to wear\nSpeaker J: Give maker on headband and he said they don't have any of these in stock that they have them in LA And so we'll take about a week to get here So okay, just go ordered Yeah, it's a lot of money for you is the contact if I want to do an invoice because I think that's how we did it before It's a long time to get to himself. Okay And then nine channels is the next one we can do so\nSpeaker H: Okay, so we're gonna start for the days of change so that's 15 instead of 16 and there's six on the table right\nSpeaker F: Can I have some really dumb questions? Is there anyway we can have you know like a wireless microphone that you pass around to the people who You know the extra people for the time they want to talk That's a good idea\nSpeaker J: Well, I just not sure how we would hand like you know Jerry Springer\nSpeaker F: Yeah, there might be a way to see that there are gonna be these different people to nail the chairs down and yeah\nSpeaker H: If we can't get another board name it for the few times that you might want to have that Let's figure that we have eight which are set up and then there's a ninth which is pass around Yeah, that's a good idea infinite expansion Right We're left\nSpeaker F: Or also for you know, we can just hand around the lapel\nSpeaker J: Rather than get it you want to\nSpeaker H: Really, you know better yes, okay, I don't know but I know the lapel is also it depends on the handheld But hand of many handhelds are built with sort of anti shock sort of things so that it is less susceptible to hand noises If you hold the lapel mic you just give me the sorts of jump\nSpeaker F: I mean the ones they really pass around must be so I wonder if they have one that will hook up\nSpeaker J: What I wonder if they have one that will hook up to this or whether you would want to\nSpeaker I: Very just wanted to hook into the receiver in the other room, right?\nSpeaker H: No, that's a good one. It's not built into the mic. Oh you say get it get a different radio\nSpeaker F: Yeah, just these ones that they Yeah, I'm with no you know wireless Have a little antenna come up\nSpeaker J: It's gonna be much easier to get one of these and just plug in a mic isn't it?\nSpeaker C: But then the mic has to but you have to hand it around and if you have two pieces of\nSpeaker H: So right, so this is a good point. So yeah, you have these these mics with a little antenna on you\nSpeaker J: Okay, and do you think you would be able to use the same receiver?\nSpeaker H: You know, it's just a frequency But that's that's a great idea and then just sort of have that as the and then you can have groups of 20 people Yeah, because there's only\nSpeaker F: But it out actually I think in the large the larger the group the less Interact less people are talking Over each other just there might be a lot of people that speak ones or twice\nSpeaker J: Yeah, okay, so I guess people who have to leave can leave and do we have anything else to discuss or should we just do digits?\nSpeaker E: I thought of some extra One of them is So as of last night I assigned 12 hours and they'd finish nine and my goal was to have them 11 done by the end of the month I think that by tomorrow will have 10 It's great\nSpeaker F: I got this email from jaynet like two in the morning or something. So it's really great\nSpeaker E: Thanks And then um also an idea for another meeting which would be to have the transcribers talk about the day Uh, it's sort of a that's a great idea super right very interesting. I'd love to hear what they have say\nSpeaker I: So that we can remember all the little so if we got them to talk about this meeting it would be a meta meta\nSpeaker H: The rest is you know you have a transcribers and there's 10 of us. So how do we do this? The only thing or just have them have their own meeting\nSpeaker E: Well, that's what I'm thinking. Yeah, yeah, that would be great And they and they've made observations to me like they say Uh, you know, this meeting that we think has so much overlap in fact it does But there are other groups of similar size that have very little you know as part of it It's known that the group and all that and they have various observations that would be that's a great idea\nSpeaker H: Yeah, I'd like to hear what they said say so maybe we could they could have a meeting More or less without us that to do this and we should record it and then maybe one or two of them could come to one of these meetings Give us a status. You can tell us about it. Yeah\nSpeaker F: They will get to transcriber on meetings, but they also get paid for having a break and I think that's a good idea. Yeah They're really good at them involved It's a great idea really sorry. I have to\nSpeaker E: I have to go as well. Okay, and then I wanted to also say something about the Fiskus John Fiskus visit tomorrow And which is to say that it'll be from nine to one that I'm going to I offer the organization allow him to Adjust it if he wishes, but to be basically in three parts the acoustic part coming first which would be basically the room engineering aspects Other things and he'll be also presenting what NIST is doing and then number two would be sort of a The transcription process so this would be a focus on like pre segmentation and the modifications to the multi-trans interface Which allows more refined encoding of the beginnings and ends of the overlapping segments Which Dave Hillbart spend doing and then Of course the pre segmentation to you has been doing and then the third part would and again He has some stuff this relevant with respect to NIST and then the third one would be focus on transcription standards So at NIST he's interested in this establishment of a global encoding standard and I guess I can say I want to you know See what they're doing and also present what what we've chosen as ours and and discuss that kind of thing And so but he's only here till till one and actually we're thinking of noon being lunchtime So basically hoping that we can get as much of this done as possible before Okay, good and everybody wants to attend this welcome Oh, where are you going to meet here mostly but I've also reserved the barcode room To figure out how that works in terms of like maybe having a live demonstration\nSpeaker H: Keep it the nine o'clock. I'll be here. Yeah. I assume we're not going to try to record it. Oh, I think\nSpeaker J: Yeah, all right\nSpeaker H: So maybe do digits and unless there's anything else do digital\nSpeaker J: Uh should we make him wear and race his mic or would that just be too confusing? No, I don't think it's confusing\nSpeaker H: Well\nSpeaker J: Mess up the forms channel change right? Yeah, I just don't know how we do that. So\nSpeaker I: I mean other than on three forms here on the top there\nSpeaker J: Just clip it to your collar\nSpeaker B: Okay, my name is uh Espen Ericsson. I'm an origin. I'm uh this is my second semester at Berkeley currently I'm taking uh my first graduate level courses in DSP And then when I come back to Norway, I'm going to continue with the more research project work kind of work So this semester I'm starting up with the with a small project Through uh David Gail Bart which I'm taking a course with I couldn't touch with him. He told me about this project So with the help of uh Dan Ellis I'm going to do Small project associated to this What I'm going to try to do is uh use use echo cancellation to um To handle the periods where we have overlapping talk To try to do something about that. So currently I'm uh Just reading up on echo cancellation Looking into the theory behind that and then Hopefully get some results So it's a it's a project goes over the course of one semester All right, so I'm just here to date introduce myself Till that'll be I'll be working on this\nSpeaker J: And are you staying at Berkeley or are you just here or semester?\nSpeaker B: This is my second semester around the last It's so\nSpeaker H: Yeah, he's in the course to do two five decals. Yeah, I'm in Mullins course here\nSpeaker E: And then you go back to Norway this year\nSpeaker G: We're just talking about something like this yesterday Or yeah yesterday with Liz So this is\nSpeaker J: Legends Okay, this is transcript three four three one three four five oh 55591 681 4303 778 6030 8 9 0 040 5 0 1442477 257 8266 345 6082 8104832 99 0316 819 09300 192 2 305 112\nNone: 133 Trash green set 339 3 4 1 0 3 4 5 0 712 849 972 6 8 8 1 OZ 0 12 427 329 7 550 654 4 778 2 8 9 9 0 1133 1 8 7 2 4 0 2 3 7 2\nSpeaker G: Transfer 337 1-3390 2 309 5 2 403 1 8 654 0 0 11 764 259 4 8 8 9 0 0 252 3 4 5 0 4 4 2 2 0 5 8 6 7 8 0 0 1 6 5 0 0 4 3 2 1 7 2 9 6\nSpeaker B: Transcript number 351 1 Dash 353 0 8 0 3 2 1 0 7 0 5 9 1 9 2 3 6 5 2 4 5 0 752 9 8 3 8 2 9 3 0 0 8 7 8 0 6 0 1 2 4 1 3 5 3 7 6 4 7 8\nSpeaker I: Transcript 349 1-3510 7 0 9-258 443 0 5 4 9 0 8 1-5-0-2-1 2 3 4 0 0-8-8 6 6 4 9-6 7 5 7 7 8 0-7 8 5 9 0 0 2-49 9-8-2-0 3-50 4 2-8-0-0-9 5 6 7\nSpeaker H: Transcript 347 1-349-0 7 1 8 7 7 3 9 6 0 8 0 1 2 4 2 1 3 6 0 0 5 4 8 7 5 7 2 6619 167 7 8 9 0 1 4 2 2 3 7 0 2 4 5 7 1 6\nSpeaker E: Transcript 345 1-347 0 6 7 7 7 8 8 8 7 9 8 4 8 5 6 9 0 0 2 2 5 2 5 3 0 3 8 4 4 9 4 4 1 5 9 0 3 6 8 0 1 5 7 0 5 8 0 4 7 2 6 0 5 4 1 5 9 2 3 4 5 0\nSpeaker J: 4 5\nNone: 4 5 5 5 5 5 5 5 5 5\n"}, {"meeting_id": "ES2006d", "summary": "This meeting was about the detailed design. Firstly, User Interface presented on the prototype, including the shape, the size of the remote and the user interface including the buttons and the scrolling wheel. Next, the group had an evaluation criterion on the prototype. They scored an overall six point five average on the prototype and thought they had done well on it. Then, the group had a discussion about project finance. The group decided not to have a sample sensor and the locator because they had a production cost of 12.5 Euros. Lastly, the group had a discussion about the project process, including the use of a pen, the whiteboard and the PowerPoint. Also, the group was satisfied with Project Manager's leadership and Project Manager also mentioned that their costs were within the budget as well as some admin work.", "dialogue": "Speaker C: Okay, this is our final meeting, the detailed design meeting.\nSpeaker C: And again, I'll take minutes, but what we have to get through in this meeting is, firstly, the prototype presentation for you two, so you can show us what you've been looking at diligently. There's my way call.\nSpeaker C: Then Cat's going to present the evaluation criteria that we're going to be evaluating this against.\nSpeaker C: Then I need to say since a few things about finance, because we have to check that it's within the finance criteria.\nSpeaker C: And then we'll be making sure that our product fits both the evaluation criteria, Cat and the financial limit.\nSpeaker C: And then we will have a brief evaluation of the whole process of production and design.\nSpeaker C: So we've got 40 minutes.\nSpeaker E: And then we get to make it a little more comfortable.\nSpeaker C: Because we missed out.\nSpeaker C: So, it's now, I guess that was the start of 15.35.\nSpeaker C: So we've got until 4.\nSpeaker B: How much do we have?\nSpeaker B: 40 minutes.\nSpeaker C: Yeah, about 4.15.\nSpeaker A: So, yeah.\nSpeaker A: Okay, go for it.\nSpeaker D: Do you want to?\nSpeaker D: So, is it, are we starting with a, yeah.\nSpeaker D: So, will you maybe start with like the shape and things?\nSpeaker D: And then I will explain the user interface.\nSpeaker D: Things like the buttons and the screw.\nSpeaker B: Okay, so, basically going without a trend of vegetables.\nSpeaker B: And we selected the color and approximate shape of banana.\nSpeaker E: So, basically it's the flip open thing again.\nSpeaker C: So, we have the, okay.\nSpeaker B: Marika would explain user user interface here.\nSpeaker B: And it flips open on the side.\nSpeaker B: So, it opens like that.\nSpeaker B: And we have the user interface.\nSpeaker B: In here and the LCD and the screw on the inside.\nSpeaker B: Well, everything else is probably using the face.\nSpeaker C: Yeah, things made of rubber.\nSpeaker D: Oh, yeah.\nSpeaker D: It is, yeah.\nSpeaker E: Is it yellow? Do you think you make it a bit smaller?\nSpeaker D: It could be made a bit smaller.\nSpeaker D: And of course it would be shiny.\nSpeaker D: Yeah, but one thing we actually kind of forgot while designing it, one side was supposed to be rounder.\nSpeaker D: So, is it the backside round?\nSpeaker B: Yeah.\nSpeaker B: But it's in the smith of rubber anyway.\nSpeaker B: I think it's good.\nSpeaker E: Yeah, it does look like the fence like curvy and then the whole shapes curvy.\nSpeaker E: I would say that this curvy is quite like a vegetable.\nSpeaker D: Yeah.\nSpeaker D: And it's spongy as well.\nSpeaker D: I would say, Keenan.\nSpeaker D: So, are they using the face as we discussed last time on the, on the cover, we just have the very basic things.\nSpeaker D: So, we have the channels here starting from one, two, three.\nSpeaker D: There would be numbers on the actual one.\nSpeaker D: So, it's four up to four up to seven up to nine and zero here.\nSpeaker D: And zero here.\nSpeaker D: Yeah.\nSpeaker D: And then, well, this is on off button.\nSpeaker D: It's quite standard place for it and also the color is quite often red.\nSpeaker D: So, it's kind of user friendly.\nSpeaker D: And then this one would be for flipping the canals back in like the previous one and the next one.\nSpeaker D: And we would also have a little thing saying here, previous and then, a preven, preven next.\nSpeaker A: So, where's the volume?\nSpeaker D: The volume is scrolling on the side.\nSpeaker D: You just did it like this.\nSpeaker D: And, and it's on the back cover or back lid because if you flip it open, you can still do the scrolling here.\nSpeaker D: Okay, Keenan.\nSpeaker D: See.\nSpeaker D: So, the volume is, you just scroll, but then once you flip it open, okay, there you have the screen.\nSpeaker D: Yeah.\nSpeaker D: And you have the spinning wheel with options to choose.\nSpeaker D: You can move back and forth and then if you need to choose something on the screen, you just push the send in the middle button.\nSpeaker E: Cool.\nSpeaker E: Oh, the thing we forgot was like a mute button.\nSpeaker D: A mute button.\nSpeaker B: Yeah.\nSpeaker B: We would not put that in.\nSpeaker D: We'll have this on the screen, on the display.\nSpeaker B: So, on the cover, we have the bare essentials.\nSpeaker E: On the wheel, if you peel down, then it will.\nSpeaker B: On the LCD, you know, the main one you will have with this option.\nSpeaker D: But the, yeah, the scrolling is kind of, you have to scroll on the way to make it mute, right?\nSpeaker D: But if you're all the scrolling click, isn't it?\nSpeaker E: That's the scrolling click.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Cool.\nSpeaker D: So, that's why I was a whole mutitian.\nSpeaker D: Yeah.\nSpeaker D: And okay, so, so the voice recognition is also just part of it.\nSpeaker D: You can't really see it in the face.\nSpeaker D: Yeah, it's hidden in there.\nSpeaker D: And we do have the logo.\nSpeaker D: Very good.\nSpeaker C: So, I think, and it's with the black and yellow and the red and the black color.\nSpeaker C: Yeah.\nSpeaker D: So, yeah, I think we could do the logo in gray as it is on the website.\nSpeaker D: Yeah.\nSpeaker B: You can have a question.\nSpeaker B: Yeah.\nSpeaker B: Let's have a look.\nSpeaker C: Let's have a look.\nSpeaker C: Okay.\nSpeaker C: So, it's sort of intermediate color.\nSpeaker E: Yeah.\nSpeaker E: Oh.\nSpeaker E: It is beautiful.\nSpeaker E: Cold and slimy.\nSpeaker E: I hate K-Day.\nSpeaker E: It's just mingled.\nSpeaker E: Yeah.\nSpeaker E: Oh, that's cool.\nSpeaker E: Very good.\nNone: Okay.\nSpeaker C: So, maybe if we go on to evaluation criteria and then we'll...\nSpeaker C: Yeah.\nSpeaker C: I suspect we're going to have a couple of minor finance issues.\nSpeaker C: I'm sure we can get around them somehow.\nNone: Okay.\nSpeaker C: We'll just send all of our manufacturing to some nice poor country.\nSpeaker C: We'll just send all of our manufacturing to some of the places that we're in.\nSpeaker E: Wales.\nSpeaker E: Wales, for example.\nSpeaker E: Cool.\nSpeaker E: Okay.\nSpeaker E: Right.\nSpeaker E: Okay.\nSpeaker E: Cool.\nNone: Okay.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nNone: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: Cool.\nSpeaker E: So, we're going to be using the seven points scale.\nSpeaker E: So, one is, you know, yes, it totally meets what that requirement.\nSpeaker E: And seven is, no, it really doesn't.\nSpeaker E: We need to go back and start again.\nSpeaker E: You know, basically what I did was I went through all the like user requirements and things that we've done and we've worked on and like made a list of them.\nSpeaker E: You know, so that we can evaluate each one and like, so it's about going back to the start and saying, oh yeah, we did manage to do that or I know we really forgot that.\nSpeaker E: Cool.\nSpeaker E: So, these are what they are.\nSpeaker C: So, for each of these, we need to give it a one to seven.\nSpeaker C: Yes.\nSpeaker E: I did have a BC and D down here, but it seems to have turned into like just blood on.\nSpeaker E: Never mind.\nSpeaker E: Okay.\nSpeaker E: But if you can imagine that they say ABC and D, that would be really good.\nSpeaker B: I guess we'll give it maximum points and everything.\nSpeaker D: Yeah, it's definitely attractive.\nSpeaker D: Yeah, I agree.\nSpeaker D: That was a locatable thing we actually forgot.\nSpeaker E: Well, I thought you'd kind of said that you'd have a little thing to stick on the TV.\nSpeaker D: Yeah.\nSpeaker C: Shall I just prepare one now?\nSpeaker C: Yeah.\nSpeaker D: I would be right too.\nSpeaker E: Cool.\nSpeaker E: Okay.\nSpeaker E: So, be attractive to look at.\nSpeaker E: That's this one.\nSpeaker E: What do you all say?\nSpeaker D: Seven or something, the maximum, yeah?\nSpeaker D: Yeah.\nSpeaker D: I go for seven.\nSpeaker D: Oh, yeah.\nSpeaker D: It's terribly good.\nSpeaker E: We're also proud of the palette.\nSpeaker E: Okay.\nSpeaker E: So, that'll be a seven.\nSpeaker E: Okay.\nSpeaker E: Oh, no, you can't walk.\nSpeaker E: That's up there.\nSpeaker E: Okay.\nSpeaker E: What I've done on the next page is I've set it up.\nSpeaker E: So, we just put the marks in.\nSpeaker E: Okay.\nSpeaker C: Excellent.\nSpeaker C: Except we can't.\nSpeaker C: Yeah.\nSpeaker C: But that's all right.\nSpeaker C: I can take notes.\nSpeaker C: Yeah.\nSpeaker C: I'll take notes.\nSpeaker E: It's fine.\nSpeaker E: Okay.\nSpeaker E: So, we're all agreeing on seven.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: Okay.\nSpeaker E: Does it match the operating behavior of the user?\nSpeaker E: Oh, I think it does.\nSpeaker A: Yeah.\nSpeaker B: The only thing that we were considering was that this thing is kind of more for right hundred people than for left hundred people.\nSpeaker B: So, if you're left hundred, you kind of left.\nSpeaker B: Yeah.\nSpeaker D: And we might want to do like another model, another version which is like exactly the mirror image of this one.\nSpeaker D: Yeah.\nSpeaker C: That's going to be a problem because you don't always have all left-handed or all right-handed in a family.\nSpeaker C: Yeah.\nSpeaker B: I think it's not a huge problem because it is a breakable.\nSpeaker D: But then I think left-handed people are already used to discrimination anyway.\nSpeaker E: So, they just, yeah.\nSpeaker E: But I mean, because it's not like it's a pen, you know, left-handed people can't normally write, right-handed, but they can normally do most things, right-handed.\nSpeaker E: Right.\nSpeaker E: I would say it's not such a big issue.\nSpeaker D: Yeah, because I mean, anyway, right-handed people would be able to start with it.\nSpeaker D: Yeah.\nSpeaker D: So, if you've got majority, right-handed it's-\nSpeaker E: I mean, you can't mean your finger to scroll other than your thumb. Yeah.\nSpeaker E: So, I mean, that does kind of negate the whole RSI issue.\nSpeaker E: So, maybe we need to put that needs a little bit of investigation.\nSpeaker E: I'll give it a five, I would say.\nSpeaker E: What do you, what do you think?\nSpeaker D: Or maybe six because it's just one, one, one of them.\nSpeaker C: Yeah.\nSpeaker C: I think, I mean, most people are right-handed.\nSpeaker C: So, in terms of our greatest target group, I think it's pretty good.\nSpeaker C: Yeah.\nSpeaker C: One more thing is that- The target for management might want to- Okay.\nSpeaker B: It might be a little clumsy when it opens up, right?\nSpeaker B: It opens in the side.\nSpeaker D: No, yeah, but we have it nicely with a hinge or a wood-hole.\nSpeaker A: Yeah.\nSpeaker D: So, it won't be a problem, it will be- And it will be- it won't be heavy.\nSpeaker E: I think the alternative is flippings in the top.\nSpeaker E: Yeah.\nSpeaker D: Yeah, but it's a bit long.\nSpeaker D: It's a little bit long.\nSpeaker D: Well, I mean, it can be open like this, of course.\nSpeaker E: But you were thinking about making it smaller, yeah?\nSpeaker A: This kind of makes more sense.\nSpeaker B: So, you have to keep- I have to interfere with the IR channel.\nSpeaker D: Yeah, but if we flip it open only as much as that.\nSpeaker D: Okay.\nSpeaker E: So, it works like a mobile phone flipping, but, you know, as long as that side is flat, then that will work.\nSpeaker B: Right.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: So, see, are we admitting defeat on sea or are we saying we're going to sticky-locate on the table?\nSpeaker E: No, we have a look here.\nSpeaker C: There's a locator.\nSpeaker C: Cool.\nSpeaker E: So, that means- I just checked on TV.\nSpeaker E: That doesn't mean you need to speak on it, though.\nSpeaker E: Doesn't it?\nSpeaker E: Let's make it beep.\nSpeaker D: Yeah, well, but the sample speaker is included, so it has some capacity to do- Yeah.\nSpeaker D: To make some sounds.\nSpeaker D: So, yeah.\nSpeaker E: So, you see that?\nSpeaker E: Seven, yeah.\nSpeaker E: So, I'm going to go to the table.\nSpeaker E: Fabulous day.\nSpeaker B: Intuitive, completely intuitive.\nSpeaker B: Yeah.\nSpeaker D: If this means intuitive, if it means the way people kind of are used to finding things and it's- Yeah.\nSpeaker D: I think it's-\nSpeaker C: It's six. Because the- I mean, the standard layout for numbers is three, three, three, and one, rather than the way you've got it.\nSpeaker C: I really like the way you have it.\nSpeaker C: Yeah.\nSpeaker C: It's not the immediate thing that you're used to.\nSpeaker E: Yeah.\nSpeaker B: I mean- And even the scroll- It's a new technology, so it might be a little more difficult for people to get used to in the beginning.\nSpeaker B: So, it is kind of not very intuitive, but it's a good technology.\nSpeaker C: But it's something that they will be experiencing in a lot of different places.\nSpeaker E: Yeah.\nSpeaker E: So, should we maybe say a five and say it is intuitive, but it's different.\nSpeaker E: So, yeah.\nSpeaker E: Do you know, I mean, it's obvious how to use it, but you might have to think about it first.\nSpeaker E: So, we give that one a five, do you think?\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: I'm glad you're accepting this.\nSpeaker E: I have taken a little while, haven't it?\nSpeaker E: Intuitive.\nSpeaker E: But- Sorry, it's really hard to write on this.\nNone: I just went a bit mad.\nNone: Okay.\nNone: Cool.\nNone: E.\nSpeaker E: Okay.\nSpeaker E: I guess this comes back from this whole beat.\nSpeaker E: Links in here.\nSpeaker E: So, possibly for left-handed.\nNone: Investigate.\nSpeaker E: Yep.\nSpeaker C: But otherwise it's superb.\nSpeaker C: So, she give it a six.\nSpeaker E: Six.\nSpeaker E: Six.\nSpeaker D: The agronomic design, well, I mean, I don't know.\nSpeaker D: I mean, the repetitive stress things, but then who would be really pushing the button so much on the remote control anyway?\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Unless you are all the time sitting.\nSpeaker B: Yeah.\nSpeaker D: So, it's kind of- Yeah.\nSpeaker D: I think it is a text today.\nSpeaker E: You know, and I never got prepared in Australia, and I enjoy from that.\nSpeaker E: So, I find it quite hard to believe.\nSpeaker B: And moreover, it has LCD and everything so that we minimize the pressing of the buttons.\nSpeaker B: Okay.\nSpeaker E: So, we give that a six.\nSpeaker E: Okay.\nSpeaker E: F.\nSpeaker D: Yeah.\nSpeaker D: That's awesome.\nSpeaker D: Absolutely.\nSpeaker E: I don't know how to make showing up with the things there, but it only has bullet points there.\nSpeaker C: Oh, that's the second one.\nSpeaker C: So, you must have changed it on this one where it's got score, but not on the previous slide.\nSpeaker E: Oh, okay.\nSpeaker E: Cool.\nSpeaker E: Right.\nSpeaker E: It has voice control.\nSpeaker E: Yes.\nSpeaker E: That's a seven, man.\nSpeaker B: The most sophisticated mode that I've ever seen.\nSpeaker C: Yeah.\nSpeaker C: But in terms of the actual technology, none of it is actually new.\nSpeaker D: We have the sample speakers as well.\nSpeaker E: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I mean, all of the components have been used in other things before.\nSpeaker D: But at the same time, they all relatively new.\nSpeaker E: They've never been using remote control before any link.\nSpeaker C: Yeah.\nSpeaker C: What do you reckon?\nSpeaker E: Five, six.\nSpeaker E: Yeah.\nSpeaker E: What do you think?\nSpeaker D: Six.\nSpeaker D: Six.\nSpeaker D: I mean, how far can you go with a remote control?\nSpeaker D: Really?\nSpeaker D: Well, that's it.\nSpeaker D: It still has to do with what it has to do.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: But I mean, everything has been used in space before it gets to anyone else really, haven't it?\nSpeaker E: See?\nSpeaker E: I don't think many of you have space remote.\nSpeaker C: They can take you to put fashion in electronics.\nSpeaker C: Absolutely.\nSpeaker B: Yeah.\nSpeaker B: Isn't it special?\nSpeaker B: Yeah, sure.\nSpeaker B: The Carrick banana.\nSpeaker B: It's the maximum fashion.\nSpeaker D: We give it through the vegetables with our fashionable base days.\nSpeaker D: There we go.\nSpeaker E: So I think we've done very well.\nSpeaker D: Very good.\nSpeaker E: We need the average here.\nSpeaker E: So you go...\nSpeaker D: The average is about six and something.\nSpeaker D: A little bit over six.\nSpeaker D: Seven.\nSpeaker D: There are how many?\nSpeaker D: No, wait.\nSpeaker E: A little bit under six.\nSpeaker E: Twenty-eight.\nSpeaker E: Three-sixes.\nSpeaker E: And one-five.\nSpeaker E: Eighteen.\nSpeaker E: Oh, three-sixes.\nSpeaker E: Okay.\nSpeaker B: Thirty-eight.\nSpeaker B: Thirty-eight.\nSpeaker B: Four-sevenes.\nSpeaker B: Forty-six and five.\nSpeaker B: Six point.\nSpeaker D: Twenty-one.\nSpeaker C: Twenty-one.\nSpeaker C: Five.\nSpeaker C: That's six point five.\nSpeaker C: Six point five.\nSpeaker E: Close enough.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker C: That's pretty good, I think.\nSpeaker C: I'll wait until we get to finance the money.\nSpeaker C: If we can afford it.\nSpeaker C: That's all you've got at the moment.\nSpeaker C: Did you have anything there?\nSpeaker C: No, that's it.\nSpeaker C: Yeah.\nSpeaker C: All right.\nSpeaker C: So finance.\nSpeaker C: And we'll see if we can...\nSpeaker C: Unscrew this test.\nSpeaker E: Cool side.\nSpeaker E: So, I think we're going to just take...\nSpeaker E: There we go.\nSpeaker E: There we go.\nNone: Another mark.\nSpeaker C: Beautiful.\nSpeaker C: Yeah.\nSpeaker C: Not anymore.\nSpeaker C: Mm-hmm.\nSpeaker B: You have to wait for it.\nNone: There we go.\nSpeaker C: Okay.\nSpeaker C: And now we have to calculate the production costs.\nSpeaker C: So I've got an Excel spreadsheet to help us do that.\nSpeaker C: Can you read that?\nSpeaker C: Almost.\nSpeaker C: Oh, that's...\nSpeaker C: I started feeling it in.\nSpeaker C: But of course, these are provisional, so we have to go down.\nSpeaker C: No hand dynamo, right?\nSpeaker C: Yeah.\nSpeaker C: One simple battery.\nSpeaker C: No kinetic energy, no solar.\nSpeaker C: The chip we're going for an advanced chip on print.\nSpeaker C: We also said the sample sensor and sample speaker.\nSpeaker C: Yep.\nSpeaker C: Single curved surface, so that we can unfold with this case material.\nSpeaker C: This material was said rubber.\nSpeaker C: I don't know what special colour means.\nSpeaker B: Anything which is not one of those.\nSpeaker D: I think this is probably special colour.\nSpeaker E: Yeah.\nSpeaker E: Yeah, but rubber comes colour.\nSpeaker C: I would maybe it's like if you want titanium coloured or wood colour.\nSpeaker D: Or maybe if you want some kind of pattern thing.\nSpeaker E: Let's leave it as well.\nSpeaker E: Yeah, you might end up having to take off.\nSpeaker C: So we have screw wheel with push pattern.\nSpeaker C: No.\nSpeaker C: Yeah, we have for muting.\nSpeaker D: Yeah.\nSpeaker D: And we have LC display and buttons, supplements.\nSpeaker C: No.\nSpeaker C: No.\nNone: No.\nSpeaker D: We don't know.\nSpeaker B: We don't know.\nSpeaker D: We don't know.\nSpeaker D: We don't know.\nSpeaker D: We don't know.\nSpeaker D: We don't know.\nSpeaker C: We don't know.\nSpeaker C: We don't know.\nSpeaker C: I think maybe it's integrated with the LC display.\nSpeaker C: Okay, let's see.\nSpeaker C: Okay, well, we want a push button.\nSpeaker C: Yeah.\nSpeaker E: I think the push button...\nSpeaker C: So then you have...\nSpeaker E: I don't know if that's one.\nSpeaker E: That means you can only have 25 push buttons in total, in that not counting anything.\nSpeaker C: Yeah.\nSpeaker C: Still be in budget.\nSpeaker B: That seems unlucky.\nSpeaker C: What is the limit?\nSpeaker C: Whether push button means that we have to count all the colours.\nSpeaker C: I don't think so.\nSpeaker B: No, it says what is the kind of interface.\nSpeaker B: If it is push button, then you get a 0.5, it's a scroll wheel.\nSpeaker B: So we put its push button and scroll wheel and LCC display.\nSpeaker B: So that's the three kind of interfaces that we have.\nSpeaker C: So as we can see, that's way too expensive.\nSpeaker C: What's our strategy?\nSpeaker C: Some are budget, so 12.5.\nSpeaker E: The sample center will have to go because that's the most expensive thing.\nSpeaker C: So that has implications for the...\nSpeaker B: It does not have four voice recognition, but it does have for the feedback speaker.\nSpeaker B: When you press one, it says hello.\nSpeaker E: That's a bit of a gimmick anyway.\nSpeaker B: And the locator also goes away.\nSpeaker D: But it was very innovative.\nSpeaker D: So we're innovating this.\nSpeaker D: Well, this sample speaker is expensive, but we could just have some very easy device that just beeps.\nSpeaker C: Because the sample speaker was, I think, more complicated than it would be.\nSpeaker D: There you record your samples.\nSpeaker B: Also, in the case, I'm not sure that you will evaluate this as a curved surface, because it's just rubber.\nSpeaker B: So it's probably a flat surface rubber.\nSpeaker C: Okay, yeah.\nSpeaker E: So we need 150 off.\nSpeaker E: See, I was going to say the scroll wheel push button thing,\nSpeaker C: because take it down to just a scroll wheel.\nSpeaker B: So that would mean that we cannot press. Then we would be in level project.\nSpeaker B: If in the LCD we can scroll, right?\nSpeaker B: How do we make a selection if you cannot push it?\nSpeaker D: No, no, no, you can push this one, but we don't have a push button.\nSpeaker D: We don't have this muting mechanism for this scroll.\nSpeaker E: But that's where you just have to just put it down.\nSpeaker C: You can have to scroll it straight down.\nSpeaker C: So that's point three.\nSpeaker B: But hey, it's the scroll wheel.\nSpeaker B: I thought we were referring to this as a scroll spinning wheel and push button thing.\nSpeaker D: But I mean, we can put an additional mute button on the top as well.\nSpeaker E: Yeah, I mean, that wouldn't actually cost anymore.\nSpeaker E: That's the spin wheel, though, isn't it?\nSpeaker E: That can be the LCD.\nSpeaker C: That comes with the LCD?\nSpeaker C: Yeah.\nSpeaker C: We decided because it's not on our list.\nSpeaker B: So this is...\nSpeaker C: The scroll wheel is on the side.\nSpeaker B: So we are adding costs for both of them, right?\nSpeaker B: Okay, I mean, I think it's the good.\nSpeaker C: So we're point three over over at the moment.\nSpeaker E: And then we just take off the scroll wheel altogether and just have push button to the volume.\nSpeaker E: You can still put them on the side.\nSpeaker E: Yeah, I mean, the scroll wheel is pretty cool, but...\nSpeaker B: Instead of scrolling here, we have two buttons here, off or up and down.\nSpeaker B: Yeah.\nSpeaker B: Sounds good, actually.\nSpeaker C: Yeah, rather than having three different things that people have to do.\nSpeaker C: Okay.\nSpeaker C: There we go.\nSpeaker C: Oh, look, we're way under budget and we'll make huge profits and we'll get bonuses.\nSpeaker E: And we could admit to the single curve in the circle, could we?\nSpeaker C: Or that we have to have some sort of special colour.\nSpeaker E: Yeah.\nSpeaker C: So that's all right.\nSpeaker C: We'll leave it at that.\nSpeaker C: Yeah.\nSpeaker C: And then I'll take it back to management and say, we weren't quite sure about the colour of that cost extra, that we've still got some space for it.\nSpeaker C: We have, yeah.\nSpeaker C: Excellent.\nSpeaker C: All right.\nSpeaker C: So did we lose on our evaluation criteria?\nSpeaker C: Is that a little bit of a problem?\nSpeaker C: No, I'm not really, you know.\nSpeaker B: Because we keep all the features, we keep voice recognition, we keep LCD display.\nSpeaker B: Yeah.\nSpeaker B: Instead of having scrolling, we just put...\nSpeaker B: We just got rid of a gimmick that we never know anyway.\nSpeaker E: Yeah.\nSpeaker E: And the scroll wheel is, I mean, essentially the two buttons.\nSpeaker E: That is not very different.\nSpeaker E: Yeah.\nSpeaker E: All right.\nSpeaker B: But we lose the locator.\nSpeaker D: We don't have a beep.\nSpeaker D: We are going to have a beeping thing.\nSpeaker A: So instead of speaker.\nSpeaker D: Yeah, it's not like sample speaker, but it will just beep.\nSpeaker D: So we still have the locator.\nSpeaker D: Yeah.\nSpeaker D: Thank you.\nSpeaker E: That's not a very exciting colour.\nSpeaker E: I think you should make it more vegetable-like.\nSpeaker D: Much colour, the colour of the phone or the colour of the...\nSpeaker E: Oh, the beeper thing.\nSpeaker D: It can be yellow as well.\nSpeaker D: It can come in the same colour as the case.\nSpeaker C: We won't have run out of our pocket later.\nSpeaker E: Yeah, I think General wants it to vibrate.\nSpeaker E: You know, your pen vibrates.\nSpeaker E: Yeah, I know.\nSpeaker C: Yeah, but I don't think it's a very short time.\nSpeaker C: OK.\nSpeaker C: So it looks like we designed a banana.\nSpeaker C: We've just got about ten minutes or so left at the meeting.\nSpeaker C: So it would be good if we could just have a little talk about the project itself and how it went.\nSpeaker C: So that we can feed back to the management for the next time they designed the product.\nSpeaker C: Cool.\nSpeaker C: Feedback, I do.\nSpeaker D: Yeah, I'm as far as creativity is concerned.\nSpeaker D: I think that was wrong for creativity.\nSpeaker D: The only problem being that at the end we had to cut some things down because of the budget we had.\nSpeaker B: I think one thing that was lacking was that we did not know what the various things cost to begin with.\nSpeaker B: Yeah.\nSpeaker B: And we kept adding things randomly.\nSpeaker B: Yeah.\nSpeaker B: So, how do you know?\nSpeaker E: If we had the beginning, we should have been like, OK, we can have that a lot.\nSpeaker B: Yeah, we can see what we can.\nSpeaker B: So all the random decisions at the end could have been entered.\nSpeaker C: But in terms of the process of going and working individually and then coming back to the meeting, that works.\nSpeaker E: I think it would mean to say regular, you know, it wasn't like we were alone for very long.\nSpeaker E: So you didn't go off and think, wouldn't it be great to have a bomb rating?\nSpeaker A: I would have a big drop.\nSpeaker E: Shape like a banana.\nSpeaker E: And then, you know, you come back three days later and you're like, look, if I've been waiting, it's like a banana.\nSpeaker E: Yeah.\nSpeaker D: Yeah, the means were very, very good.\nSpeaker D: The means we used the whiteboard digital pen.\nSpeaker E: I like the pens.\nSpeaker E: And I think it's the only one that would just be so cool to do.\nSpeaker C: Yeah, he could take it to lecture and just write stuff down.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I wonder what one of these costs.\nSpeaker E: Do you think I've noticed if one may?\nSpeaker E: Maybe you should see that was the recording.\nSpeaker E: OK.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker C: But that worked well having a whiteboard that we could draw on as well as having the power point.\nSpeaker C: I find that the problem with power point often is that it's so static and you can't change it once.\nSpeaker C: It is a bit weird.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And this time also, the time limits that actually preparing the...\nSpeaker E: Nothing flew in.\nSpeaker E: You didn't have the whole swoosh-ing thing because it wasn't time for that.\nSpeaker E: Yeah.\nSpeaker E: That's all right.\nSpeaker E: But always it would.\nSpeaker E: Not that you can do that on the board either.\nSpeaker E: It would make some...\nSpeaker D: Yeah, but I mean, already just preparing the slides before the whole day.\nSpeaker E: Yeah, I mean, now it's fairly tight anyway.\nSpeaker E: I mean, especially with that last minute alteration.\nSpeaker E: Yeah.\nSpeaker E: You can see any brackets.\nSpeaker A: Yeah.\nSpeaker E: This is just hard to be changed.\nSpeaker E: And yeah.\nSpeaker E: It's a cool thing.\nSpeaker E: We're supposed to say nice things about Jen.\nSpeaker E: I'm pushing.\nSpeaker E: You know, moving's a nasty thing.\nSpeaker E: Yeah.\nSpeaker E: I was satisfied with my leadership.\nSpeaker D: Yeah.\nSpeaker E: You won't like it addictating you, that's always good.\nSpeaker C: You have to say that because I'm taking the next clip.\nSpeaker C: And then you've got the pen.\nSpeaker E: What time do you think?\nSpeaker D: And then the teamwork, I think it works quite nicely.\nSpeaker C: Did anyone feel that they were getting sort of covered up and not being able to express them?\nSpeaker B: I guess it was a fairly small group.\nSpeaker B: So all of us got to express our opinions.\nSpeaker D: Well, it's a pretty new, pretty novel solution for most control really.\nSpeaker D: All this flipping open thing.\nSpeaker E: I don't know.\nSpeaker E: I don't think you're shopping for my control.\nSpeaker D: But I've never seen anything.\nSpeaker D: And none of my examples were like this.\nSpeaker A: Yeah.\nSpeaker E: I'll be looking at how many people are next.\nSpeaker E: I'll be exploring.\nSpeaker D: Maybe we could have a patent on this one.\nSpeaker D: Yeah.\nSpeaker D: Patent patents.\nSpeaker E: I think we'd like to think the idea is we need, but we've got no way of finding it.\nSpeaker C: Yeah.\nSpeaker C: You can always go to Google and type in banana remote control.\nSpeaker C: I can't even worry.\nSpeaker E: I've just came up with like a thing to learn.\nSpeaker C: True.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: So costs are within budget.\nSpeaker C: Yes.\nSpeaker C: Yes.\nSpeaker C: Within budget.\nSpeaker C: Including a little, what have we got?\nSpeaker C: One euro left over.\nSpeaker C: It fits that we didn't possibly.\nSpeaker C: We've evaluated.\nSpeaker C: Yes.\nSpeaker C: A project.\nSpeaker C: And you've got the scores.\nSpeaker C: Can you put that in the project documents?\nSpeaker C: It's in the project.\nSpeaker C: And the process we all didn't really have any major problems with.\nSpeaker C: Was there anything that you found difficult or anything that didn't go as smoothly?\nSpeaker D: One way difficult was the time pressure.\nSpeaker E: Otherwise.\nSpeaker E: Yeah.\nSpeaker E: Sometimes it's like a little bit less.\nSpeaker B: I thought that was good though because if you're given too much time, then you've got nothing to do with your time.\nSpeaker B: Although we could have made the Rs better how we have time.\nSpeaker B: Yeah.\nNone: Okay.\nSpeaker C: I think we're still there within a...\nSpeaker C: Yeah.\nSpeaker C: We've got about five minutes left.\nSpeaker C: But if we're finished, then we're finished.\nSpeaker C: Which is too efficient.\nSpeaker C: We should never drag a meeting on just because you have extra time.\nSpeaker C: Yes.\nSpeaker C: So I would say that's the end of that meeting.\nSpeaker C: It's a pleasure working with you.\nSpeaker C: It was very productive day.\nSpeaker E: And four animals on the board again.\nNone: Oh no.\nSpeaker B: I don't think so.\nSpeaker D: It can make some animals.\nSpeaker D: Are you dead, why?\nSpeaker E: No.\nSpeaker E: It's just me and it smells so bad.\nSpeaker D: That's it?\nSpeaker D: It smells quite nice.\nSpeaker D: It smells very sweet.\nSpeaker C: So we have to complete the final question here, meeting summary when they...\nSpeaker C: Is there a question I've already sent?\nSpeaker C: I don't know if it's already sent or not.\nSpeaker C: No, it was a bit.\nSpeaker C: I'm pushing over there.\nSpeaker E: Do we have to go back into the other room or put this thing in?\nSpeaker C: I don't see why we can't stay here really.\nSpeaker C: Okay, it's moving.\nSpeaker C: I don't know if you can say this thing.\nSpeaker C: I'm glad.\nSpeaker D: I made your animal for you.\nSpeaker D: It was supposed to be pink.\nSpeaker D: Yeah, but it was blue.\nNone: It was blue.\n"}, {"meeting_id": "Buw001", "summary": "The team was setting up a new project in which they would record meetings and then generate summaries. The meeting began with introductions and a discussion of what kind of data the team could collect. They considered collecting visual data as well as notes. At the end of the meetings, the team wanted to ask participants to summarize what they took away as well as ask questions about the meeting. This would be a method for collecting more data to train a potential summarization model. One concern the team had was how they could reduce bias when collecting queries. Words like \"important\" could skew participant responses. The team also expressed some interest in collecting action items. Finally, the team discussed what each member should do to get the project up and running and the role of diversity in their data set.", "dialogue": "Speaker D: headphones that aren't so uncomfortable. I think this should be off the record, but I think\nSpeaker E: that okay. We're not recording yet are we?\nSpeaker I: No, that was some recording. I don't think that designed to be over your years.\nSpeaker D: Yeah, I know. It just, it really hurts. It gives you a headache. Temples, features. Yeah.\nSpeaker B: Yeah.\nSpeaker I: I thought I definitely haven't figured it out.\nSpeaker A: I guess I have to need a quarter of my contentment, you know, after\nSpeaker H: some of the things that Kevin Chiena has said. We know exactly how much you have left. What were we going to talk about again? We said data collection, which we're doing.\nSpeaker E: We're going to do it. Do we go around the room and do names or anything?\nSpeaker I: I think that's going to do. Usually we've done that. Also, we've done digits as well, but I forgot to print it out. So, besides with this big group.\nSpeaker I: No, I'd be better with this big. Yeah, but it takes too much time.\nSpeaker J: All right. What?\nSpeaker J: You'll think maybe point together in funny direction.\nSpeaker J: Sort of it.\nSpeaker J: The elephant would consider what it was.\nSpeaker E: Whoops.\nSpeaker J: Sort of it.\nSpeaker J: Would it?\nSpeaker J: Sort of thing.\nSpeaker J: Oh, this thing.\nSpeaker F: That's it.\nSpeaker F: Otherwise it just gets hard to read.\nSpeaker H: Oh, yeah, the element should be close to your mouth as possible.\nSpeaker E: All right. How's that working?\nSpeaker J: It's fine.\nSpeaker E: Okay.\nSpeaker H: All right. So, what we had was that we were going to talk about data collection.\nSpeaker H: And you put up their data format and other tasks during data collection.\nSpeaker E: So, I think the goal was, what can we do? How can you do the data collection differently to get?\nSpeaker E: What can you add to it to get some information that would be helpful for the user interface design?\nSpeaker E: Like especially for querying.\nSpeaker E: Especially for querying. So, getting people to do queries afterwards, getting people to do summaries afterwards.\nSpeaker F: One thing that came up in the morning was the, if he has, I remember Mr.\nSpeaker F: Andrew. So, he has these note-taking things that would sort of be a summary which he wouldn't have to solicit if he were able to.\nSpeaker E: Well, if you actually take notes as a summary as opposed to take notes in the sense of taking advantage of the timestamp.\nSpeaker E: So, action item or reminder to send this to so and so blah, blah, blah.\nSpeaker E: So, that wouldn't be a summary. That would just be, that would be related to the query side.\nSpeaker I: But if we had the cross paths, we could ask people, you know, if something comes up, write it down and market somehow.\nSpeaker J: Right. I mean, we, because you'd have several people with these paths, you could collect different things.\nNone: Right.\nSpeaker J: Because I tend to take notes which is summaries and so.\nSpeaker A: I mean, the downside to that is that he sort of indicated that the quality of the handwriting recognition was quite poor.\nSpeaker I: Well, that's all right. I don't think there'd be so many that you couldn't have someone clean it up pretty easily.\nSpeaker E: Yeah, we also could come up with some code for things that people want to do so that for frequent things.\nSpeaker E: And that other things people can write whatever they want.\nSpeaker E: I mean, it's to some extent for his benefit.\nSpeaker E: So, is that, you know, if we just keep it simple, then maybe it's still useful.\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker H: I just realized we skipped the part that we were saying we were going to do at the front or reach to do who we were.\nNone: The roll call.\nSpeaker E: Right. I thought you did that on purpose.\nSpeaker E: But anyways, we didn't roll call.\nSpeaker H: No, no, no, no. I just, I might go elsewhere.\nSpeaker H: So, yeah, I'm Morgan and I'm on channel three.\nSpeaker I: And I'm Adam Janon on channel eight.\nSpeaker F: I'm Jane Edwards. I think on channel B.\nSpeaker J: I'm Dan Ellis.\nSpeaker A: Eric on channel nine.\nSpeaker D: Liz on channel one.\nSpeaker E: Marie on channel zero.\nSpeaker C: Catherine on channel two.\nSpeaker F: Should we use pseudonyms?\nSpeaker F: Should we do it a second time with pseudonyms?\nSpeaker H: Rocky Rekun.\nSpeaker I: You want to hear your channels and the museums?\nSpeaker J: Uh, museums, tourists, tourists, next, next, next ones.\nSpeaker J: Barabist, PDM, right?\nSpeaker J: PZM, right?\nSpeaker J: PDA, right?\nSpeaker J: PDA left.\nSpeaker I: And eventually once this room gets a little more organized, the gym let will be mounted under the table.\nSpeaker I: And these guys will be permanently mounted somehow, you know, probably with double sided tape.\nSpeaker I: But so we won't have to go through that.\nSpeaker F: I have a question on protocol in these meetings, which is when you say, Jim, the person listening won't know what that is.\nSpeaker F: How do we get, is that important information?\nSpeaker F: No, the, Jim, I mean the box that contains it.\nSpeaker H: Well, I mean, supposed to be broadened out and go to a range of meetings besides just these internal ones.\nSpeaker H: There's going to be lots of things that any group of people know each other have in common, that we will not know.\nSpeaker E: Right. So there will be jargon that we, they'll be transcriptionaries.\nSpeaker H: Okay. I mean, we were originally going to do this with the illicit design.\nSpeaker H: And the reason we didn't go straight to that was because immediately 90% of what we heard would be jargon to us.\nSpeaker H: That's one of the reasons.\nSpeaker H: But yeah.\nSpeaker H: Yeah.\nSpeaker H: That's right.\nSpeaker H: We're on this, of course.\nSpeaker F: Okay, so we were on the data collection.\nSpeaker B: Right.\nSpeaker F: And the summary issue.\nSpeaker E: So, so actually there's kind of three issues.\nSpeaker E: There's the cross pad issue.\nSpeaker E: Should we do it?\nSpeaker E: And if so, what do we have them do?\nSpeaker E: Do we have people write summaries?\nSpeaker E: Everybody or one person?\nSpeaker E: And then do we ask people for how they would query things?\nSpeaker A: There's some problems in that.\nSpeaker A: And then when do you actually ask them about that?\nSpeaker A: I mean, that was one thing I was thinking about was is that Dan said earlier that maybe two weeks later, which is when you would want to query, you can say, wait, ask them that.\nSpeaker A: But there's a problem with that in that if you're not, if you don't have an interactive system, it's going to be hard to go beyond sort of the first level of question.\nSpeaker A: Right.\nSpeaker A: And for explore the data further.\nSpeaker A: Right.\nSpeaker A: So.\nSpeaker H: There's another problem, which is we certainly do want to branch out beyond recording meetings about meeting recorder.\nSpeaker H: And once we get out beyond our little group, people's motivation factor reduces enormously.\nSpeaker H: And if we start giving them a bunch of other things to do, we did another meeting here for another group.\nSpeaker H: And they were fine with it.\nSpeaker H: But if we'd said, okay, now all eight of you have to come up with a lesson.\nSpeaker I: Well, I asked them to.\nSpeaker I: And none of them did.\nSpeaker I: So I asked them to send me ideas for queries after the meeting.\nSpeaker I: And no one ever did.\nSpeaker I: I didn't follow up either.\nSpeaker I: So track them down and say, please do it now.\nSpeaker I: But no one spontaneously provided anything.\nSpeaker H: I'm worried that even if you did push them into it, it might be semi-random.\nSpeaker H: Right.\nSpeaker H: I suppose to what you'd really want to know if you were going to use this thing.\nSpeaker I: I just don't know how else to generate the query something getting an expert to actually listen to the meeting and say that's important.\nSpeaker I: That might be a query.\nSpeaker F: Well, there is this other thing, which we too are looking to early, which is there are certain key words, like action item, things like that, which could be used in.\nSpeaker J: So.\nSpeaker F: Yeah, and I think the structure around.\nSpeaker F: And I also was thinking with reference to the note-taking, the advantage there is that you get structure without the person having to do something artificial later.\nSpeaker F: One of the third thing I want to say is the summaries afterwards.\nSpeaker F: I think they should be recorded instead of written, because I think that it would take so long for people to write that I think you wouldn't get as good a summary.\nSpeaker E: How about this idea that normally, at most meetings, somebody is delegated to be a note taker?\nSpeaker E: I have a good point.\nSpeaker E: And so why don't we just use the notes that somebody takes?\nSpeaker E: I mean, that gives you a summary,\nSpeaker I: but it doesn't really, how do you generate queries from that?\nSpeaker J: Well, I mean, maybe a summary is one of the things we'd want from the app to the system, right?\nSpeaker E: Right.\nSpeaker J: And something actually.\nSpeaker I: Then James and I were talking about this during one of the breaks, and the problem with that is, I'm definitely going to do something with information retrieval, even if it's sort of full bore, what I'm going to do for my thesis, I'm going to do something.\nSpeaker I: I'm not going to do anything with summarization.\nSpeaker I: And so if someone wants to do that, that's fine, but it's not going to be me.\nSpeaker H: Well, I think that we, I mean, the core thing is that once we get some of these issues nailed down, we need to do a bunch of recordings, and send them off to IBM, and get a bunch of transcriptions, even if they're slightly flawed or need some other, and then we'll have some data there.\nSpeaker H: And then we can start looking at thinking, what do we want to know about these things, and the very least.\nSpeaker D: I actually want to say something about the note pad.\nSpeaker D: So if you could sense just when people are writing, and you tell them not to doodle or try not to be using that for other purposes, and each person has a note pad.\nSpeaker D: They just get it when they come in a room.\nSpeaker D: Then you can just have a plot of, you know, who's writing when?\nSpeaker D: That's all you, and you can also have notes of the meeting.\nSpeaker D: But I bet that's, that will allow you to go into the, sort of the hot places where people write things down.\nSpeaker A: I mean, you can tell when you're in a meeting,\nSpeaker D: when everyone stops and writes something down, that something was just said, it may not be kept in the later summary, but at that point in time, it was something that was important.\nSpeaker D: That wouldn't take any extra.\nSpeaker D: Absolutely.\nSpeaker D: Or someone could just, you could just put your hand on the pad and collect that if you want to.\nSpeaker H: That's a good idea, but maybe I'm missing something.\nSpeaker H: That doesn't get to the question of how we come up with queries, right?\nSpeaker H: Well, what it does,\nSpeaker D: you can go to the points where you can actually go to those points in time and find out what they were talking about.\nSpeaker D: And you, yeah.\nSpeaker E: Well, what it does is provide a different, I think it's an interesting thing.\nSpeaker E: I don't think it gets at the queries per se, but it does give us an information fusion sort of thing that you want to say, what were the hot points of the meeting?\nSpeaker H: That's what I mean is that I think it gets us something interesting.\nSpeaker H: But if we were asking a question, which I thought we were of, of how do we figure out what's the nature of the queries that people are going to want to ask of such a system, knowing what's important doesn't tell you what people are going to be asking.\nSpeaker D: But I bet it's a good superset of it.\nSpeaker E: Well, I think you could say they're going to ask about when did so-and-so talk about blah, and at least that gives you the word that they might run a query on.\nSpeaker D: These are the questions where there are maybe keywords.\nSpeaker I: This would tell you what the hit is, not what the query is.\nSpeaker I: Right, right.\nSpeaker E: It'll tell you the hit, but I think we're in a query for a moment.\nSpeaker D: It's a little bit dangerous right now.\nSpeaker D: We don't even know what, I mean, if you want to find out what an user will use, that might be true for one domain and one user, but I mean a different domain and a different user.\nSpeaker H: Yeah, but we're just looking for a place to start with that, because what James is going to be doing is looking at the user interface and he's looking at the query.\nSpeaker H: We have five hours of pilot data of the other stuff, but we have zero hours of queries.\nSpeaker H: So he's just sort of going where do I start?\nSpeaker E: Well, you could do, I think the summaries actually may help get us there for a couple reasons.\nSpeaker E: One, if you have a bunch of summaries, you can do a word frequency count and see what words come up in different types of meetings.\nSpeaker E: So action item is going to come up, whether it's a VLSI meeting or speech meeting or whatever.\nSpeaker E: So words that come up in different types of meetings, maybe something that you would want to query about.\nSpeaker E: The second thing you could possibly do with it is just run a little pilot experiment with somebody saying, here's a summary of a meeting, what questions might you want to ask about it to go back?\nSpeaker I: I think that's difficult because then they're not going to ask the questions that are in the summary.\nSpeaker E: Well, I think it would give...\nSpeaker E: That's one possible scenario, though, is you have the summary and you want to ask questions to get more detail.\nSpeaker I: Yeah, I think it has to be a participant.\nSpeaker B: Well, it doesn't have to be.\nSpeaker I: Okay, so that is another use of meeting recorder that we haven't really talked about, which is for someone else, as opposed to as a remembrance agent, which is what had been my primary thought in the information retrieval part of it, with the...\nSpeaker I: But I guess if you had a meeting participant, they could use the summary to refresh themselves about the meeting and then make up queries.\nSpeaker I: But it's not...\nSpeaker I: I don't know how to do it until you have a system.\nSpeaker D: The summary is actually going to drive the queries, then.\nSpeaker D: I mean, you're just going to be very circular.\nSpeaker J: But there is this class of queries, which are the things that you didn't realize were important at the time, that some in retrospect, you think, well, Hang on, didn't we talk about that?\nSpeaker J: And it's not like the dinner trend summary, but that's kind of what this complete data capture is kind of nice as for.\nSpeaker E: Right.\nSpeaker J: Because the thing is that you wouldn't have bothered to make an effort to record that they get recorded.\nSpeaker J: So I mean, and there's no way generally it knows until we just play at Shiaqa.\nSpeaker D: So you could always post-talk, label them.\nSpeaker J: Right, right, exactly.\nSpeaker J: But I mean, it's difficult to sort of say.\nSpeaker J: And if I was going to ask four questions about this, what would they be?\nSpeaker J: Those are the kind of things that come up.\nSpeaker I: But at least then we'd get us started.\nSpeaker J: Oh, yeah.\nSpeaker F: I have something that if you can use the summaries as an indication of the important points of the meeting, then you might get something like, so if the obscure item you want to know more about was the sum form of data collection, you know, maybe the summary would say, you know, we discuss types of data collection.\nSpeaker F: And, you know, and maybe you could get to it by that.\nSpeaker F: If you have the larger structure of the discourse, then if you can categorize what it is that you're looking for with reference to those larger headaches, then you can find that even if you don't have direct root to that.\nSpeaker I: Although it seems like that's a high burden on the note-taker.\nSpeaker I: That's a pretty fine grain that the note-taker will have to take.\nSpeaker D: Maybe Landay can put a student into the note-taker.\nSpeaker E: No, I think you've got to have somebody who knows the topic or, you know, whose job it is delegated to be the note-taker.\nSpeaker E: Somebody who's putting it in the...\nSpeaker D: No, I mean, but someone who can come sit in on the meetings and then takes the notes with them that the real note-taker.\nSpeaker D: And that way that one student has, you know, a rough idea of what was going on, and they can use it for their research.\nSpeaker D: I mean, this isn't really necessarily what you would do in a real system, because that's a lot of trouble, and maybe it's not the best way to do it.\nSpeaker D: But if he has some students that want to study that, then they should sort of get to know the people and attend those meetings and get the notes from the note-taker.\nSpeaker I: Well, I think that's a little bit of a problem.\nSpeaker I: They're sort of note-taking application stuff they've been doing for the last couple of years, and I don't think anyone is still working on it.\nSpeaker I: I think they're done.\nSpeaker I: So I'm not sure that they have anyone currently working on notes.\nSpeaker I: So we'd have to interest someone in is the combination of note and speech.\nSpeaker I: And so the question is, is there such a person, I think right now, the answer is no.\nSpeaker H: I've been thinking about a little bit here about this.\nSpeaker H: I think that the, now I'm thinking that the summary is actually a reasonable bootstrap into what we'd like to get at.\nSpeaker H: It's not ideal, but we have to get started someplace.\nSpeaker H: So I was just thinking about, suppose we wanted to get, we have this collection of meaning.\nSpeaker H: We have five hours of stuff.\nSpeaker H: We get that transcribes.\nSpeaker H: Now we have five hours of meetings, and you ask me, what kind of questions do you want to ask?\nSpeaker H: I wouldn't have any idea what kind of questions I want to ask.\nSpeaker H: I'd have to get started someplace.\nSpeaker H: So in fact, if I looked at a summary of it, I'd go, oh yeah, I was in that meeting.\nSpeaker H: I remember that.\nSpeaker H: What was the part that, and I think that might then help me to think of things, even things that aren't listed in the summary, but just as a refresh of what the general thing was going on in the meeting.\nSpeaker E: I think it serves two purposes.\nSpeaker E: One, as sort of a refresh to help bootstrap queries.\nSpeaker E: But also, I mean, maybe we do want to generate summaries.\nSpeaker B: Oh yeah, that's true.\nSpeaker E: And then it's, you know, it's kind of a key.\nSpeaker D: So how does the summary get generated?\nSpeaker D: I'm not a good idea of a summary, but I want to, to think carefully, suppose generating it, and how, because the summary will drive the queries.\nSpeaker D: So.\nSpeaker E: What I think, you know, in most meetings, this one being different, but in most meetings that I attend, there's somebody explicitly taking notes frequently on a laptop.\nSpeaker E: You can just make it be on a laptop.\nSpeaker E: So then you're dealing with ASCII and not somebody.\nSpeaker E: You don't have to go through handwriting recognition.\nSpeaker E: And then they post-edited into a summary, and they email it out for a minute.\nSpeaker E: I mean, that happens in most meetings.\nSpeaker F: I think that we're using summary in two different ways.\nSpeaker F: So what you just described, I would describe as minutes.\nSpeaker F: Right.\nSpeaker F: What I originally thought was, if you ask someone, what was the meeting about?\nSpeaker F: And then they would say, what we talked about this, and then we talked about that, and so on and so on, we talked about it.\nSpeaker F: And then you'd have like, my thought was to have multiple people summarize it on recording rather than writing, because writing takes time, and you get a relevant other thing to take time.\nSpeaker F: Whereas if you just say it immediately after the meeting, about two minutes summary of what the meeting was about, I think you get, see, I also worry about having a single note anchor, because that's just one person's perception.\nSpeaker F: And it's relevant to what your focus was on that meeting.\nSpeaker F: And people have different major topics that they're interested in.\nSpeaker F: So my proposal would be that you may be worth considering both of those types.\nSpeaker F: Okay.\nSpeaker F: Okay. Taking an spontaneous oral summary afterwards no longer than two minutes.\nSpeaker H: Adam, you can correct me on this, but my impression was that pretty much true that the meeting's here, nobody sits with a laptop.\nSpeaker I: And never seen it at the big scene.\nSpeaker I: Does anyone, I mean, Dan is the one who most frequently take notes.\nSpeaker J: Yeah.\nSpeaker J: And when we have other meetings, when I have meetings in the European projects, we have someone take that.\nSpeaker J: And we're like, yeah, but those are bigger, deal things, right?\nSpeaker H: Where you've got 15 people.\nSpeaker H: I mean, most, this is one of the larger meetings.\nSpeaker H: Most of the meetings we have here are four or five people.\nSpeaker H: And you're not, you don't have somebody sitting and taking minutes for it.\nSpeaker H: You just get together and talk about what you're doing.\nSpeaker E: I think it depends on whether it's a business meeting or a technical discussion.\nSpeaker E: And I agree technical discussions.\nSpeaker E: You don't usually have somebody taking notes.\nSpeaker I: But I ran a meeting.\nSpeaker I: I think notes that read.\nSpeaker I: Do they?\nSpeaker I: A person with a laptop.\nSpeaker J: Do you mean the only people that has any tips?\nSpeaker I: They're more, I mean, they're 10-ish.\nSpeaker I: Yeah.\nSpeaker D: You should also have a record of what's on the board.\nSpeaker D: I mean, I find it very hard to reconstruct what's going on.\nSpeaker D: I don't know.\nSpeaker D: And this is something early in the project we talked a lot about.\nSpeaker D: I don't know how, but for instance, I mean, the outline is sort of up here.\nSpeaker D: And that's what people are seeing.\nSpeaker D: And if you have a, or you could tell people not to use the board, but there's sort of this missing information.\nSpeaker I: We should, but you just, you could end up with video.\nSpeaker I: Well, and instrumented rooms.\nSpeaker I: And that's a different project, I think.\nSpeaker J: I think for this data capture, it would be nice to have a digital camera.\nSpeaker J: Just to take pictures of who's there, where the microphones are, and then we go, so if I didn't want to go, you'd have like three or four snaps.\nSpeaker D: I agree.\nSpeaker D: Yeah.\nSpeaker D: It's one of the things that we don't think that's...\nSpeaker D: I mean, it will have a very hard time understanding.\nSpeaker D: Don't you think that's the...\nSpeaker J: Well, no, I mean, I think, I mean, I think, I mean, I think the right now, we don't make a record of whether people are sitting on the tables.\nSpeaker J: Right.\nSpeaker J: And at some point, that might be off the use.\nSpeaker I: But I think adding...\nSpeaker I: I mean, photographs, at a whole other level of progress.\nSpeaker I: It's just a digital record.\nSpeaker J: Not as part of the data you have to recover.\nSpeaker B: Just in terms of...\nSpeaker J: Like archiving.\nSpeaker F: Yes, I agree.\nSpeaker F: I agree.\nSpeaker F: Because it's close enough about things, and then you have the things that are not later\nSpeaker D: might be able to take these and say, okay, they, you know, at least these are the people who are there.\nSpeaker D: And here's sort of what they started talking about.\nSpeaker F: And it's so simple, like, you said, three snapshots.\nSpeaker F: And...\nSpeaker H: Lizzy...\nSpeaker H: You said in on the subcommittee meeting or whatever, on the subcommittee meeting for at the workshop we were at that Mark Leverman was having.\nSpeaker H: So I wasn't there.\nSpeaker H: They must have had some discussion about video and visual aspect, and all that.\nSpeaker D: Big, big interest.\nSpeaker D: Yeah.\nSpeaker D: Huge, I mean, it personally, I don't, I would never want to deal with it.\nSpeaker D: I'm saying, first of all, there's a whole bunch of fusion issues that Dyerf was interested in, you know, fusing gesture and face recognition even live movement and things like that for this kind of task.\nSpeaker D: And there's also, I think, a personal interest on a part of Mark Leverman in this kind of, in storing these images in any data we collect for that later, we can do it, I think, with it.\nSpeaker H: So to address what Adam's saying, I mean, I think that the key thing there is that this is a description of database collection effort that they're talking about doing.\nSpeaker H: And if the database exists and includes some visual information that doesn't mean that an individual researcher is going to make any use of it, right?\nSpeaker H: So.\nSpeaker I: But then it's going to be a lot of effort on our part to create it and store it and.\nSpeaker H: Right, so we're going to.\nSpeaker I: Standards and to do anything with it.\nSpeaker H: So we're going to do what we're going to do, whatever is reasonable for us.\nSpeaker H: But having.\nSpeaker D: Very crude, like I know with Atis, we just had a tape recorder running all the time.\nSpeaker D: And later on, it turned out it was really good that you had a tape recorder of what was happening, even though you just got the speech from the machine.\nSpeaker D: So if you can find some really low, perplexity, yeah, way of doing that, I think it would be worthwhile.\nSpeaker F: I agree.\nSpeaker F: If it's simple, I mean, it's simple as to just the digital.\nSpeaker H: It's the digital.\nSpeaker H: Well, minimally, I mean, what Dan is referring to is, at least having some representation of the spatial position of the people, because we are interested in some spatial processing.\nSpeaker H: So.\nSpeaker I: Now what's the room a little more fixed?\nSpeaker I: That's a little easier to do.\nSpeaker I: Well, the wireless.\nSpeaker I: Yeah.\nSpeaker D: Also, CMU has been doing this.\nSpeaker D: They were the most vocal at this meeting, Alex Weibelskrup.\nSpeaker D: And they have said, I talked to the student who had done this, that with two fairly inexpensive cameras, they just recorded all the time, and were able to get all the information from, or maybe it was three, from all the pets.\nSpeaker D: So I think we might lose the chance to use this data for somebody later who wants to do some kind of processing on it if we don't.\nSpeaker I: I don't disagree.\nSpeaker I: I think that if you have that, then people who are interested in vision can use this database.\nSpeaker I: The problem with it is you have more people who don't want to be filmed than who don't want to be recorded.\nSpeaker I: So that there's going to be another group of people who are going to say, I won't participate.\nSpeaker D: Or you could put a paper bag over everybody's head and not look at each other and not look at boards, and just I'll be sitting talking that would be an interesting video.\nSpeaker D: Yeah.\nSpeaker F: But I think we're just proposing a minimal preservation.\nSpeaker D: I definitely won't participate as a camera.\nSpeaker D: It's a spatial organization, and you can\nSpeaker F: anonymize the faces for that matter.\nSpeaker I: This is a lot of infrastructure and work.\nSpeaker D: It's just set it up as long as anonymizing the movie. Not for CMU.\nSpeaker D: They have a pretty crude setup.\nSpeaker D: And they had, they just turned on these cameras.\nSpeaker D: They were not moving or anything.\nSpeaker D: Couldn't find it.\nSpeaker D: And stored it on analog media.\nSpeaker D: They didn't actually align it or anything.\nSpeaker D: They have it though.\nSpeaker F: That's worth considering.\nSpeaker F: Maybe you want to sit in that much more time.\nSpeaker D: Did they start digitally?\nSpeaker D: I think they just had the videotapes with a counter or something.\nSpeaker H: Well, I think for me, for our purposes, we probably might try that some.\nSpeaker H: And we certainly already have some recordings that don't have that.\nSpeaker H: Which we'll get other value out of, I think.\nSpeaker F: The thing is, if it's easy to collect, then I think it's a wise thing to do because once it's gone, it's gone.\nSpeaker D: And just the community, if LDC collects this data, I mean, if Mark Liverman is a strong proponent of how they collect it and what they collect, there will probably be some video data in there.\nSpeaker D: There we go.\nSpeaker D: And so that could argue for us.\nSpeaker D: Not doing it, or it could argue for us doing it.\nSpeaker D: The only place where it's available relapses when some of the summarization issues actually could be easier, maybe easier if you had the video.\nSpeaker H: I think at the moment, we should be determining this on the basis of our own interests and needs rather than hypothetical ones from a community thing.\nSpeaker H: As you say, if they decide it's really critical, then they will collect a lot more data than we can afford to.\nSpeaker H: And we'll include all that.\nSpeaker H: I'm not worried about the cost of setting it up.\nSpeaker H: I'm worried about the cost of people looking at it.\nSpeaker H: In other words, it'd be kind of silly to collect it all and not look at it at all.\nSpeaker H: And so I think that we do have to do some picking choosing of the stuff that we're doing.\nSpeaker H: But I do think that we minimally want something that we might want to look at some subsets of that.\nSpeaker H: For a meeting like this, at least take a polar ride of the board.\nSpeaker D: Or at least make sure that the note taker takes a snapshot of the board.\nSpeaker D: And know the position of the people.\nSpeaker D: That'll make it a lot easier for meetings that are structured.\nSpeaker D: I mean, otherwise, later on, if nobody wrote this stuff on the board down, we have a harder time summarizing it or agreeing on that.\nSpeaker F: I suppose this is comp-knowledge.\nSpeaker F: I mean, this is shared knowledge among all the participants.\nSpeaker F: And it's a shame to keep it off the recording.\nSpeaker I: If we weren't recording this, this would get lost, right?\nSpeaker F: Yeah.\nSpeaker F: I don't understand that point.\nSpeaker F: I just think that the point is that we're not saving it\nSpeaker I: anyway, right, in our real life setting.\nSpeaker E: What do you mean we're not saving it anyway? I've written all of this down and it's getting easier until you're able to.\nSpeaker I: Well, in that case, we don't need to take pictures of it.\nSpeaker D: Right.\nSpeaker D: That would be the other option.\nSpeaker D: I'm trying to make sure that anything that was on the board.\nSpeaker E: Well, that's where I'm saying that I think the note-taking would be, I think, for many meetings, there will be some sort of note-taking, in which case, that's a useful thing to have.\nSpeaker E: I mean, we don't need to require it.\nSpeaker E: I mean, I think it would be great if we try to get a picture with every meeting.\nSpeaker E: So we won't worry about requiring these things, but the more things that we can get it for, the more useful it will be for various applications.\nSpeaker H: So I mean, departing from the moment from the data collection question, but actually talking about this group and what we actually want to do.\nSpeaker H: So I guess that's the way you were figuring on doing was putting together some notes and sending them to everybody from today.\nSpeaker H: OK.\nSpeaker H: So the question that we started with was whether there was anything else we should do during the collection.\nSpeaker H: I guess the cross pads were certainly one idea, and we'll get them from him.\nSpeaker H: We'll just do that.\nSpeaker H: And then the next thing we talked about was the summaries.\nSpeaker H: And are we going to do anything about that?\nSpeaker E: Well, before we leave the cross pads and call it done, so if I'm collecting data, then there is this question of do I use cross pads.\nSpeaker E: So I think that if we really seriously have me collect data and I can't use cross pads, it's probably less useful for you guys to go to the trouble of using it, unless you think that the cross pads are going to.\nSpeaker E: I'm not sure what they're going to do, but having a small percentage of the data with it, I'm not sure whether that's useful or not.\nSpeaker E: Maybe it's something to do.\nSpeaker E: Maybe we just do it and see what happens.\nSpeaker H: I guess the point was to try again to try to collect more information that could be useful later for the UI stuff.\nSpeaker H: So it's sort of landy supplying it so that landy stuff can be easier to do.\nSpeaker H: So right now, he's operating from zero.\nSpeaker H: And so even if we didn't get it done from UW, it seems like that could still at least try it.\nSpeaker D: I think it would be useful to have a small amount of it just as a proof of concept.\nSpeaker D: Right.\nSpeaker I: And he's not able to give enough of them away so we could probably get more as well.\nSpeaker E: That's true.\nSpeaker E: So if it seems to be really useful to you guys, we could probably get a donation to me.\nSpeaker I: Yeah, I'm not sure.\nSpeaker I: I think it will again depend on landy.\nSpeaker I: And if he has a student who's interested in how much infrastructure we'll need, I mean, if it's easy, we can just do it.\nSpeaker I: But if it requires a lot of our time, we probably won't do it.\nSpeaker H: It's a lot of the stuff we're doing now really is pilot.\nSpeaker H: And once that's in other instance, we try it out.\nSpeaker H: We're going to see how it works.\nSpeaker D: Yeah.\nSpeaker D: I just wouldn't base any of the modeling on having those.\nSpeaker D: Right.\nSpeaker I: I agree with that.\nSpeaker E: Right.\nSpeaker E: OK.\nSpeaker I: I think the important smarting is a good idea, though, that if people want something in front of them.\nSpeaker D: Yeah.\nSpeaker D: That shouldn't be hard for them.\nSpeaker I: Yeah.\nSpeaker I: Doing pilots or laptops or something.\nSpeaker E: OK, if something's important, everyone clap.\nSpeaker E: OK, so cross paths, we're just going to try it and see what happens.\nSpeaker E: Yeah, that's right.\nSpeaker E: OK, note taking.\nSpeaker E: So I think that this is going to be useful.\nSpeaker E: So if we record data, I will definitely ask for it.\nSpeaker E: So I think we should just say, we don't want to put any extra burden on people.\nSpeaker E: But if they happen to generate minutes, could they send it to us?\nSpeaker E: Absolutely.\nSpeaker E: And then what it's going to say is that I don't want\nSpeaker I: to ask people to do something. They wouldn't normally do it in a meeting.\nSpeaker I: So I just want to keep doing it.\nSpeaker I: I'm just going to keep doing it.\nSpeaker I: I think definitely the thing is this.\nSpeaker I: And then James idea of summarization afterward, I think is not bad picking out, basically, to let you pick out keywords and construct queries.\nSpeaker I: So who does this?\nSpeaker I: Yeah, thank you.\nSpeaker I: People in the meeting.\nSpeaker I: Yeah, just at the end of the meeting before you go.\nSpeaker I: Without worrying each other.\nSpeaker A: Yeah.\nSpeaker A: Or even just one or two people stay behind.\nSpeaker J: Yeah.\nSpeaker J: Yeah.\nSpeaker J: People with radiant mics can go and separate rooms.\nSpeaker J: We continue recording.\nSpeaker J: But that is just nice thing.\nSpeaker D: Then you should try them a few weeks later.\nSpeaker D: And they have all these memory experiments about how we're going to actually retain it.\nSpeaker J: That's the thing.\nSpeaker J: We collect four different summaries.\nSpeaker J: You know, we're going to get all this weird data about how people perceive things differently.\nSpeaker J: It's like, this is not what we meant to reset.\nSpeaker D: Right.\nSpeaker J: OK.\nSpeaker J: That's right.\nSpeaker E: But again, like the cross paths, I don't think I would base a lot of stuff on it.\nSpeaker E: Because I think I know when I see the clock coming near the end of the meeting, I'm like inching towards the door.\nSpeaker E: So you're probably not going to get a lot of people wanting to do this.\nSpeaker I: I think maybe.\nSpeaker I: Is email easier?\nSpeaker I: I mean, when you first said, do it spoken.\nSpeaker I: What I was thinking is, oh, then, you want to come up and you want to hook them up to the recorder.\nSpeaker I: So if they're already here, I think that's good.\nSpeaker I: But if they're not already here, I'd rather do email.\nSpeaker F: Yeah, I just have to go over the least intrusive and quickest way is, and then closest to the meeting time to it, because people start to forget it is insane.\nSpeaker E: Yeah.\nSpeaker E: I think doing it orally at the end of the meeting is the best time.\nSpeaker E: Because they're kind of a captive audience once they leave.\nSpeaker E: Forget it.\nSpeaker E: But it's.\nSpeaker E: We did digits do the summary.\nSpeaker E: But I don't think that they'll necessarily, you'll get many people willing to stay.\nSpeaker E: But if you get even one.\nSpeaker H: Well, I think it's like the note-taking thing that you can't sort of be a provider or you don't even want to do this.\nSpeaker H: But if there's some cases where they will, then it would be helpful.\nSpeaker F: And I'm also wondering, couldn't that be included in the data sample so that you could increase the number of the boards that are recognized by a particular individual?\nSpeaker F: If you could include the person's meeting step and also the person's summary step, maybe that would be.\nSpeaker J: Yeah.\nSpeaker J: It's kind of nice.\nSpeaker J: Yeah.\nSpeaker F: Under the same acoustic circumstance, because they just want things to grow in their setup.\nSpeaker F: Nothing's changed.\nSpeaker A: So I have a question about queries.\nSpeaker A: I'm going to make.\nSpeaker I: Which is, can we turn that light off?\nSpeaker I: Can we turn that just?\nSpeaker H: The fluorescent light is flickering, yeah.\nSpeaker H: Oh, it is.\nSpeaker F: There it is.\nSpeaker F: OK.\nSpeaker E: How much was it?\nSpeaker E: For a little while, I thought it was just that I was really tired.\nSpeaker E: Too much caffeine.\nSpeaker E: Too much caffeine.\nSpeaker E: It really tired.\nSpeaker E: No, maybe that's real.\nSpeaker A: I thought it was the projector for a moment.\nSpeaker A: The question I had about queries was, so what we're planning to do is have people look at the sum raise and then generate queries?\nSpeaker I: Or are we going to try to generate queries?\nSpeaker I: Yeah.\nSpeaker A: And that was one suggestion.\nSpeaker A: So the question I had is, we've given any thought to how we would generate queries automatically given a summary.\nSpeaker A: I mean, I think there's a little research topic onto itself so that it may not be a feasible thing.\nSpeaker B: Yeah.\nSpeaker D: So did a land day in his group be in charge of figuring out how to do this?\nSpeaker D: I mean, this is an issue that goes a little bit beyond where we are right now.\nSpeaker D: Very, very expert.\nSpeaker J: Yeah?\nSpeaker E: It's a little too wide to be picked up.\nSpeaker E: What's our schedule?\nSpeaker H: Well, you still want to talk to the students.\nSpeaker E: If you and I need to just, no, we did the list talk.\nSpeaker E: Oh, well, we already did this talk.\nSpeaker E: OK.\nSpeaker E: So that was the positive thing.\nSpeaker E: We need to finish.\nSpeaker E: It's already 415.\nSpeaker E: After, we need to finish this discussion.\nSpeaker E: And you and I need a little time for wrap up in quad chart.\nSpeaker E: So.\nSpeaker G: I'm going to do the disposal.\nSpeaker E: What's the plan for this discussion?\nSpeaker H: We should.\nSpeaker H: I think we should be able to wind up another half hour or something at least.\nSpeaker H: Yeah, that's much left.\nSpeaker C: Yeah.\nSpeaker H: So I think we still haven't talked about the action items.\nSpeaker E: Action.\nSpeaker E: Yeah.\nSpeaker E: So why don't you say 530?\nSpeaker J: 530.\nSpeaker E: Is that OK?\nSpeaker E: We'll probably hit horrible traffic.\nSpeaker J: OK.\nSpeaker J: Thanks.\nSpeaker J: Bye.\nSpeaker E: That's a lot of time.\nSpeaker I: But.\nSpeaker I: Well, in that said to, is it land-based problem?\nSpeaker I: He doesn't have a student who's interested right now in doing anything.\nSpeaker I: So he has very little manpower.\nSpeaker I: There's very little allocated for him.\nSpeaker B: And also, he's pretty focused on user interface.\nSpeaker I: So I don't think he wants to do information retrieval, query generation.\nSpeaker I: That sort of stuff.\nSpeaker H: Yeah, well, there's going to be these student projects.\nSpeaker H: It can do some things, but it can't be yet.\nSpeaker H: Very deep.\nSpeaker H: I actually think that, again, just as a bootstrap, if we do have something like summaries, and having the people who are involved in the meetings themselves who are cooperative and willing to do yet more, come up with queries, could at least give land an idea of the kind of things that people might want to know.\nSpeaker H: Right, if he doesn't know anything about the area and the people we're talking about.\nSpeaker D: But the people will just look at the summaries.\nSpeaker D: They're the minute.\nSpeaker D: And sort of back generate the queries that I'm worried about.\nSpeaker D: So you might as well just give him the summaries.\nSpeaker D: And maybe.\nSpeaker A: Well, I'm not sure that's a solve problem.\nSpeaker A: Right.\nSpeaker A: How did generate queries from?\nSpeaker A: I'm from the summaries.\nSpeaker A: That was sort of what might.\nSpeaker D: So what he wants to do is people who are there who later see a minute since put in summary form, which is not going to be at the same time as the meeting, there's no way that it can happen.\nSpeaker D: Right, right.\nSpeaker D: Are going to later go over and make up some stuff to which these notes would be an answer or deeper.\nSpeaker D: Yeah.\nSpeaker D: Or just memory.\nSpeaker D: That's done.\nSpeaker D: I remember you were writing.\nSpeaker D: They have to do that offline.\nSpeaker D: Yeah.\nSpeaker F: I agree.\nSpeaker F: I'm also wondering if we could ask people a question, which would be, what was the most interesting thing you got out of this meeting?\nSpeaker F: Because in terms of like informantiness, it might be that the summary would not even include what the person thought was the most interesting.\nSpeaker D: Dan doesn't know what's actually happening.\nSpeaker E: But actually, I would say that's a better thing to ask than have them summarize the meeting.\nSpeaker F: I think you get two different types of things.\nSpeaker E: You get two?\nSpeaker E: Yeah.\nSpeaker E: That's true.\nSpeaker F: So you get like the general structure of important points of what the meeting was about.\nSpeaker F: You're still here?\nSpeaker F: So you get the general structure of important points of what the meeting was about with the summary.\nSpeaker F: But what's the most interesting thing you can learn?\nSpeaker F: So the fact that I know that transcriber usually is sad is something that I thought was interesting.\nSpeaker F: And then Dan worked on that.\nSpeaker F: So I thought that was really, yeah.\nSpeaker F: So I mean, you could get a pick up some of the micro items that wouldn't occur as major headaches, but it could be very informative.\nSpeaker E: Yeah, that's actually a really good idea.\nSpeaker F: I think it would be too cost intensive either.\nSpeaker F: It means like someone can do pretty easily on this further moment.\nSpeaker C: Are you thinking about just asking one participant or all of them?\nSpeaker I: If many are willing to do it.\nSpeaker J: Making the voluntary thing.\nSpeaker I: Because you get very different answers from everybody, right?\nSpeaker I: Yeah.\nSpeaker C: That's why it's so interesting.\nSpeaker I: Well, maybe one thing we could do is for the meetings we've already done.\nSpeaker I: We take minutes and we don't have summaries.\nSpeaker I: But people could like listen to them a little bit and generate some queries.\nSpeaker I: Of course, Jane doesn't need to.\nSpeaker I: I'm sure you have that meeting memorized right now.\nSpeaker E: But actually, it would be an easy thing to just go around the room and say what was the most interesting thing you learned for those people willing to stay.\nSpeaker F: And I think we pick up the microstructure.\nSpeaker F: Some of them.\nSpeaker E: And that might be something people are willing to stay for.\nSpeaker G: Well, I don't know how I want to end it.\nSpeaker C: If you could have them you could have them go around the room.\nSpeaker C: You might just get the effect that somebody says something and then you could go around the room and they say, yeah, me too, I agree.\nSpeaker C: That's fine.\nSpeaker J: Well, I might try and call the different ones.\nSpeaker J: I might say, well, I was going to say, well, I'm going to fix something else.\nSpeaker I: You have the other thing that they know why we're doing it.\nSpeaker I: Well, we'll be telling them the reason we're trying to do this is to generate queries in the future.\nSpeaker I: So try to pick things that other people didn't say.\nSpeaker H: It's going to take some time.\nSpeaker H: I mean, the kind of interest that I had in this thing initially was that basically the form that you're doing something else later.\nSpeaker H: And you want to pick up something from this meeting related to the something else.\nSpeaker H: So it's really the what's important is in the something else.\nSpeaker E: Right.\nSpeaker H: And it might be something minor of minor importance to the meeting.\nSpeaker H: In fact, if it was really major, it was the thing that really stuck in your head, then you might not need to go back and check on it.\nSpeaker H: So it's that you're trying to find your, now, you weren't interested.\nSpeaker H: Say, I said, well, I wasn't that much interested in dialogue again, or it was a person.\nSpeaker H: But three months from now, some reason I get really interested in dialogue.\nSpeaker G: What is, what was that part of that?\nSpeaker G: Yeah, Jim Bass says that.\nSpeaker I: A few lights on dialogue.\nSpeaker I: Yeah.\nSpeaker H: Yeah.\nSpeaker H: And then I'm trying to, I mean, that's when I look in general, when I look things up most, is when it's something that didn't really stick in my head the first time around.\nSpeaker H: But for some new reason, I'm interested in the old stuff.\nSpeaker I: But that's a very bad to generate.\nSpeaker E: That's hard to generate.\nSpeaker E: And I think that's half of what I would use it for.\nSpeaker E: But I also, a lot of times, think to myself, this is interesting.\nSpeaker E: I've got to come back and follow up on it.\nSpeaker E: So things that I think are interesting, I would be wanting to do a query about.\nSpeaker E: And also, I like the idea of going around the room, because if somebody else thought something was interesting, I'd kind of want to know about it.\nSpeaker E: And then I'd want to follow up on it.\nSpeaker H: Yeah, that might get it somewhat of what I was concerned about being interested in something later that I didn't consider to be important the first time, which for me is actually the dominant thing, because I thought it was really important that tends to stick more than if I didn't.\nSpeaker H: But some new task comes along, makes me want to know.\nSpeaker I: But what's interesting to me may not have been interesting.\nSpeaker H: Yeah, so having multiple people might get it some of that.\nSpeaker I: Yeah, I think you can't get it all of it.\nSpeaker I: Right?\nSpeaker I: We just need to start somewhere.\nSpeaker A: Yeah, this is a starting point.\nSpeaker A: Question men is, how much bias do we introduce by saying, this was important now, and maybe that is something else is important later?\nSpeaker A: I mean, does the bias matter?\nSpeaker A: I don't know.\nSpeaker A: I mean, I guess the question for you guys.\nSpeaker F: Well, and one thing, we were saying important, and we're saying interesting.\nSpeaker F: And I guess it was coming to a different thing.\nSpeaker A: Sure, sure.\nSpeaker A: But I guess that's the question, really, is that, I mean, does building queries based on what's important now introduce an irreversible bias on being able to do what more we want to do later?\nSpeaker A: Well, irreversible.\nSpeaker H: I mean, I guess what I keep coming back to in my own mind is that the soonest we can do it, we need to get some kind of system so that people who've been involved in the meeting can go back later, even if it's poor system in some ways, and ask the questions that they actually want to know.\nSpeaker H: As soon as we can get that going at any kind of level, then I think we'll have a much better handle on what kind of questions people want to ask than anything we do before that.\nSpeaker H: But obviously, we have to put some somehow.\nSpeaker E: Sure.\nNone: Right.\nSpeaker E: Right.\nSpeaker F: I will say that I chose interesting because I think it includes all some important in some cases.\nSpeaker F: But I feel like the summary gets a different time of information.\nSpeaker A: I think it's important to often be uninteresting.\nSpeaker A: And I also think it's a large thing that it puts.\nSpeaker F: It puts a lot of burden on the person to evaluate.\nSpeaker F: I think interest is not threatening you.\nSpeaker E: OK.\nSpeaker E: In the interest of generating an interesting summary, in the interest of generating some in this here, and also moving on to action items and other things, let me just go through the things that I wrote down as being important, that we at least decide on cross pads.\nSpeaker E: We're going to try if Landa can get them to you guys.\nSpeaker E: And see if they're interesting, and if they are, then we'll try to do it more.\nSpeaker E: Getting electronic summary from a note-taking person, if they happen to do it anyway, getting just individual pictures, a couple of digital pictures of the table and boards to set the context of the meeting.\nSpeaker E: And then going around the room at the end to just say, ask people to mention something interesting that they learned.\nSpeaker E: So rather than say the most interesting thing, something interesting in that way, you'll get more variety.\nSpeaker I: I wouldn't even say that they learned.\nSpeaker E: OK.\nSpeaker I: You might want to mention something that you brought up.\nSpeaker E: Being that was discussed.\nSpeaker E: And then the last thing would be for those people who are willing to stay afterwards and give an oral summary.\nSpeaker E: OK.\nSpeaker E: Does that pretty much cover everything we talked about?\nSpeaker E: Well, that we want to do.\nSpeaker F: I want to look at the patient on the oral summary.\nSpeaker F: They'd be separate.\nSpeaker F: They wouldn't be here.\nSpeaker E: OK.\nSpeaker I: That's what I think that's going to prognonately end up being whoever takes down the equipment.\nSpeaker F: And that would also be the thing we need to do.\nSpeaker F: We need to put in a database.\nSpeaker J: I mean, there is still this hope that people might actually think of real queries that really want to ask at some point.\nSpeaker J: And if that episode happened, then we should try to write down.\nSpeaker I: Give them a dollar or a query.\nSpeaker I: Yeah, really.\nSpeaker I: Yeah.\nSpeaker H: Real queries.\nSpeaker H: OK.\nSpeaker H: And again, if we can figure out a way to Jimmy, a very rough system, say, in a year.\nSpeaker H: Then so that in the second and third years, we actually have something to play with in generate real queries.\nSpeaker E: Asperger.\nSpeaker E: Asperger.\nSpeaker E: OK.\nSpeaker D: So I think I just wanted to say one thing about queries.\nSpeaker D: I mean, the level of the query can be very low level, or very high level.\nSpeaker D: And it gets fuzzy and fuzzy as you go.\nSpeaker D: Well, right.\nSpeaker D: So you need to have some sort of, if you start working with queries, some way of identifying with it.\nSpeaker D: This is something that requires a one word answer, or it's one place in the recording versus, was there a general agreement on this issue of all the people who had, you can ask queries that are meaningful for people.\nSpeaker D: In fact, they're very meaningful, because they're very high level.\nSpeaker D: But they won't exist anywhere in the app.\nSpeaker D: Absolutely.\nSpeaker I: So I think we're going to have to start with keywords.\nSpeaker I: And if someone comes more interested, we could work our way up.\nSpeaker I: But it may well.\nSpeaker H: But it may well.\nSpeaker H: It may well.\nSpeaker H: Because it depends on what a goal is.\nSpeaker H: If our goal is wizard or wasdish, we might want to know what is it that people would really like to know about this data.\nSpeaker H: And if it's something that we don't know how to do yet, that's great.\nSpeaker H: That's research project.\nSpeaker H: Mm-hmm.\nSpeaker H: Or something.\nSpeaker I: I was thinking about wizard of laws, but it requires wizard of the know all about the media.\nSpeaker H: Well, maybe not true wizard of it.\nSpeaker H: Oh, yes.\nSpeaker H: Because people are too.\nSpeaker H: Just imagining it.\nSpeaker H: Aware of what's going on.\nSpeaker J: But just get people to ask questions.\nSpeaker J: So they don't have the machine that they can't answer the mind.\nSpeaker I: Yeah.\nSpeaker H: But what would you like to do?\nSpeaker I: Neither could anyone else, though, is what my point is.\nSpeaker I: Yes.\nSpeaker F: I was wondering if there might be once more source of queries, which is indicator phrases, like action, I have.\nSpeaker E: OK.\nSpeaker F: So you can take the text.\nSpeaker F: Right.\nSpeaker F: So we have to transcript.\nSpeaker I: Dates, maybe, and I don't know.\nSpeaker I: That's something I always forget.\nSpeaker D: It's going to be determined.\nSpeaker D: Well, probably if you have to sit there at the end of a meeting and say one thing you remember, it's probably whatever action item was assigned to you.\nSpeaker D: That's all I remember.\nSpeaker D: That's all I wrote down.\nSpeaker D: So in general, I mean, that could be something you could say, right?\nSpeaker D: I'm supposed to do this.\nSpeaker D: Yeah.\nSpeaker F: It does.\nSpeaker F: You could be profan just saying not of your action item.\nSpeaker F: But the action item would be a way to get maybe an additional query.\nSpeaker D: I mean, that's realistically what people might well be remembering.\nSpeaker D: So we'll be not being able to do it.\nSpeaker E: Well, we're piloting.\nSpeaker E: We'll just do it and see what happens.\nSpeaker E: Yeah.\nSpeaker H: Actually, I don't remember my action items.\nSpeaker H: Yeah.\nSpeaker E: OK.\nSpeaker E: OK.\nSpeaker E: Speaking of action items, can we move on to action items?\nSpeaker E: Sure.\nSpeaker E: Are you hinting my?\nSpeaker E: Or maybe we should wait until this meeting is transcribed.\nSpeaker E: And then we will bet.\nSpeaker G: Yeah.\nSpeaker G: I mean, somewhere up there, we had milestones.\nSpeaker G: But I guess did you get enough milestones?\nSpeaker E: I got the description.\nSpeaker E: Yeah.\nSpeaker E: In fact, why don't you hand me those transparencies so that I remember to take them?\nSpeaker G: Yeah.\nSpeaker G: Yeah.\nSpeaker G: OK.\nSpeaker H: And then, obviously, you can't hide each of those as much as this need.\nSpeaker E: OK.\nSpeaker E: What I have down for action items is we're supposed to find out about our human subject requirements.\nSpeaker E: People are supposed to send me URLs for web pages.\nSpeaker E: They could not put together an overall cover.\nSpeaker J: And you're seeing.\nSpeaker J: We need to look at our web page and make a comment.\nSpeaker E: And you also need to look at your web page and clean it up by mid-July.\nSpeaker E: Let's see.\nSpeaker E: Do do do do.\nSpeaker E: Mailing lists.\nSpeaker E: Mailing list.\nSpeaker E: You need to put together a mailing list.\nSpeaker E: I need to email Adam or Jane about getting the data.\nSpeaker E: Who should I email?\nSpeaker I: How quickly do you want it?\nSpeaker I: My July is really very crowded.\nSpeaker E: And so how about if I just right now all I want, I personally only want text data.\nSpeaker E: I think the only thing Jeff would do anything with right now, but I'm just speaking based on a conversation with him two weeks ago in Turkey.\nSpeaker E: But I think all he would want is a digit.\nSpeaker E: But I'll just speak for myself.\nSpeaker E: I'm interested in getting the language model data.\nSpeaker E: So I'm just interested in getting transcriptions.\nSpeaker E: So then just email you.\nSpeaker E: Sure, sure, sure.\nSpeaker E: OK.\nSpeaker F: You could email the both of us.\nSpeaker F: Just if you want to, let me have a thing.\nSpeaker E: OK.\nSpeaker E: I'm going to write a line of each.\nSpeaker F: And your email is at.\nSpeaker F: Thanks.\nSpeaker E: OK.\nSpeaker H: In our phone call, before we turns out the way we're going to send the data is by CDROMs.\nSpeaker H: And then what they're going to do is take the CDROM and transfer it to analog tape and give it to a transcription service.\nSpeaker F: That was a side table.\nSpeaker F: Yeah.\nSpeaker H: Using foot pedals.\nSpeaker H: Yeah, foot pedals.\nSpeaker I: So how are they going to do the multi-chance?\nSpeaker H: See, that's a good question.\nSpeaker H: Yeah, they need all the way.\nSpeaker H: No, I mean, it'll be about like you did.\nSpeaker H: And then there'll be some things, you know, many things that don't work out well.\nSpeaker H: And that would go back to IBM.\nSpeaker H: And they'll run their aligner on it.\nSpeaker H: And it kicks out things don't work well, which, you know, the overlaps will certainly the examples of that.\nSpeaker H: And I mean, we will give them all of it, right?\nSpeaker H: I'll give them all of them.\nSpeaker I: So we'll give them all 16 channels.\nSpeaker I: And they'll do whatever they want with it.\nSpeaker I: Yeah.\nSpeaker D: But you also should probably give them the mixed, you know, equal.\nSpeaker D: I don't know.\nSpeaker D: I mean, they're not going to easily be able to do that.\nSpeaker D: It's not hard.\nSpeaker H: It also won't be adding much to the data data.\nSpeaker A: Yeah, right.\nSpeaker A: It is a difficult for us to do.\nSpeaker A: Yeah, you should.\nSpeaker D: That may be all that they want to send off to their transcript.\nSpeaker E: OK, related to the conversation with Puccini, I need to email him my shipping address.\nSpeaker E: And you need to email them something which you already did.\nSpeaker F: I mean, email them the trans cover URL, the online data that Adam set up, the URL so they can click on another instance here.\nSpeaker F: And I have emailed them the streamlined conventions, which you got a copy of today.\nSpeaker H: Right.\nSpeaker H: Now, I was going to email them a pointer to the web pages that we currently have, because particularly, they want to see the one with the way the recording room is set up.\nSpeaker H: So on here.\nSpeaker F: And it's possible.\nSpeaker F: I should have sent a message.\nSpeaker F: OK.\nSpeaker I: Not immediate action.\nSpeaker I: I have to put something I do have to worry about is data formats for higher level information.\nSpeaker E: OK.\nSpeaker I: Well, not even higher level, different level of processing and all that sort of stuff.\nSpeaker I: We're going to have to figure out how we're going to annotate that.\nSpeaker E: Yeah, we never had our data format discussion.\nSpeaker F: Why don't we do it?\nSpeaker F: We discussed these musicals for rotation.\nSpeaker F: So that's something that's displayed.\nSpeaker I: That's different than format.\nSpeaker E: Well, my feeling right now on format is you guys have been doing all the work and whatever you want.\nSpeaker E: We're happy to live with other people may not agree with that, because I'm not actually touching the data.\nSpeaker E: So I shouldn't be the one to talk.\nSpeaker H: But I think that's fine.\nSpeaker H: So key thing will be that we tell you.\nSpeaker H: Yeah.\nSpeaker H: Yeah.\nSpeaker H: But we also have the year speed that you're going to guess 800 number in the world.\nSpeaker E: Oh, yeah.\nSpeaker H: You're going to call your communicator thing and we're going to be good slash bad, depending on your users.\nSpeaker C: Something that I mentioned earlier to Murray and Liz is that it's probably important to get as many non-technical and non-speech people as possible in order to get some realistic users.\nSpeaker C: So if you could ask other people to call and use our system that would be good, because we don't want people who already know how to deal with dialogue systems who know that you shouldn't type or articulate, for instance, and things like that.\nSpeaker E: So like if you have somebody who makes your plane reservations for you, which is your parents to do it.\nSpeaker C: Yeah, for instance.\nSpeaker E: Yeah, seriously.\nSpeaker E: Your grandmother.\nSpeaker E: And you know, it could result in some good bloopers, which is always good for presentations.\nSpeaker E: So anyway.\nSpeaker I: Father would last through the second prompt pre-hang hung up.\nSpeaker I: I don't know where the memory is.\nSpeaker G: I don't know what that means.\nSpeaker E: OK.\nSpeaker E: Other, let's see, other action items.\nSpeaker E: So I have to.\nSpeaker G: I'm happy getting the recording equipment right in the W.\nSpeaker G: And so it depends.\nSpeaker H: There are if that comes together within the next month, there at least will be major communications between Dan\nSpeaker E: and W. I mean, I'm shooting to try to get it done, get it put together by the beginning of August.\nSpeaker E: So we have it.\nSpeaker H: It's pretty, we don't know.\nSpeaker H: I mean, he said that it was sitting in some room, collecting dust.\nSpeaker H: And so we don't know.\nSpeaker E: It's probably unlikely that we'll pull this off, but at least as we're trying.\nSpeaker H: What is it?\nSpeaker H: We don't know.\nSpeaker H: Recording equipment.\nSpeaker H: The paper recorder.\nSpeaker H: We know it's eight channels.\nSpeaker H: We know it's digital.\nSpeaker H: We don't even know if the microphones.\nSpeaker H: And so.\nSpeaker E: OK.\nSpeaker E: And I will email these notes.\nSpeaker E: I'm not sure what to do about action items for the data stuff, though.\nSpeaker E: Then I guess somebody needs to tell Andy that you want the pads.\nSpeaker G: Yeah, OK.\nSpeaker E: I'll do that.\nSpeaker E: OK.\nSpeaker G: And he also said something about outside the vacuum of the outside tech sources that he may have.\nSpeaker G: Oh.\nSpeaker G: Some tech sources that are close enough to the certain you can play with.\nSpeaker J: Yeah, that was, that was, what he was saying was this.\nSpeaker J: The thing that Jason have been working on finds web pages that are dramatically related to what you're talking about, that's the idea.\nSpeaker J: So that would be a source of text, which is supposedly got the right vocabulary.\nSpeaker J: Right.\nSpeaker J: But there's always a great difference between materials, not spike materials.\nSpeaker H: Yeah.\nSpeaker J: It might be.\nSpeaker E: But that's actually what I want to do.\nSpeaker E: That's what I want to work with.\nSpeaker E: This is being said, the wrong material, but the right source is the right.\nSpeaker I: Unfortunately, Andy told me that Jason is not going to be working on that anymore, switching to.\nSpeaker E: Yeah.\nSpeaker E: When I asked him if he could actually supply data, he seemed a little bit more reluctant.\nSpeaker E: So I'll send him email, put it in action on him that I sent him email about it.\nSpeaker E: And if I get something great, if I don't get something, the lander, Jason.\nSpeaker E: Lander.\nSpeaker E: And otherwise, if you guys have any papers, I could use your web pages.\nSpeaker E: That's what we can do.\nSpeaker E: You've got all the web pages on me.\nSpeaker J: So I search for them.\nSpeaker E: Oh, forget this.\nSpeaker E: One last action item.\nSpeaker E: I could use what web pages there are out there on meeting recorders.\nSpeaker I: I mean, that's basically what his software does is it picks out keywords and does a Google-like search.\nSpeaker H: You could do better than that.\nSpeaker H: There's some kind of ginal and stuff, right?\nSpeaker H: Yeah.\nSpeaker E: So you should look under like zero.\nSpeaker D: You could do intelligent environments, smart rooms.\nSpeaker I: The Georgia Tech classroom 2000.\nSpeaker D: See, I mean, right.\nSpeaker D: And then, right, that's where I thought you would want to eventually be able to have a board or a camera.\nSpeaker D: Because I think the last one.\nSpeaker I: No, Georgia Tech did a very elaborate instrumented room.\nSpeaker I: And I want to try to stay away from that.\nSpeaker E: OK, great.\nSpeaker E: That solves that problem.\nSpeaker E: One last action item.\nSpeaker E: OK, I think that's pretty much all I can think of.\nSpeaker E: I can ask one thing.\nSpeaker F: It relates to data collection.\nSpeaker F: And we mentioned earlier today this question of, so I know that with the near field likes, some of the problems that come with overlapping speech are lessened.\nSpeaker F: But I wonder if is that sufficient or should we consider maybe getting some data gathered in such a way that we could have a meeting with less overlap than what otherwise be the case.\nSpeaker F: So either by rules, participation, or whatever.\nSpeaker F: Now, I mean, it's true.\nSpeaker F: I mean, we were discussing this earlier that, depending on the task, so if you've got someone giving a report, you're not going to have as much overlap.\nSpeaker F: But so we're going to have non-overlapping samples anyway.\nSpeaker F: But in a meeting, which would otherwise be highly overlapping, is the near field like enough?\nSpeaker F: Or should we have some rules of participation for some of our samples to lessen the overlap?\nSpeaker E: I don't think we should have rules of participation, but I think we should try to get a variety of meetings.\nSpeaker E: That's something that if we get the meeting stuff going at UW, that I probably can do more than you guys, because you guys are probably mostly going to get exceed people here.\nSpeaker E: But we can get anybody in EE over and possibly also some CS people over at UW.\nSpeaker E: So I think that there's a good chance that we could get more variety.\nSpeaker D: There's still an overlap.\nSpeaker D: But Mark and others have said that there's quite a lot of found data from the discourse community that has this characteristic.\nSpeaker D: And also the political, anything that was televised for third party has the characteristic of not very much overlap.\nSpeaker H: But I think we're saying before I'll solve that the natural language group here had less overlap.\nSpeaker H: So it also depends on the style of the way\nSpeaker D: you talk about it. Dominance, relation, is the people is voting?\nSpeaker F: Because it is true.\nSpeaker F: People can modify the amount of overlap that they do if they're asked to, not entirely modify, but less than if it's desired.\nSpeaker F: But if that's sufficient data, I just want to be sure that we will not be having a lot of data which can't be processed.\nSpeaker E: OK, so I'm just writing here.\nSpeaker E: We're not going to try to specify rules of interaction, but we're going to try to get more variety by using different groups of people and different sizes.\nSpeaker F: And I know that you're a new bicycle to take care of also the problems of the history.\nSpeaker E: And then the other thing might be technical versus administrative.\nSpeaker E: Because if I recorded some administrative meetings, then that may have less overlap.\nSpeaker E: Because you might have more overlap when you're doing something technical and disagreeing or whatever.\nSpeaker F: Well, I just as a contributor, so I know that in legal depositions, people are prevented from overlapping, just saying, wait till each person is finished before you say something.\nSpeaker F: So it is possible to lessen if we wanted to.\nSpeaker F: But these other factors are fine.\nSpeaker F: I just wanted to reason.\nSpeaker E: Well, the reason why I didn't want to is, why I personally didn't want to is because I wanted it to be as unintrusive as possible.\nSpeaker E: You could be with these things hanging on you.\nSpeaker F: I think that's always the sort of, I just want to be sure that we're able to process as much data as we can.\nSpeaker H: Do they discuss any of that in the meeting they had with Lieberman?\nSpeaker D: What did they do?\nSpeaker D: There was a big division.\nSpeaker D: So Lieberman and others were interested in a lot of found data.\nSpeaker D: So there's lots of recordings that they're not close to off mic.\nSpeaker D: And there's lots of television stuff.\nSpeaker D: I'm political debates, things like that.\nSpeaker D: Congressional hearings, boring stuff like that.\nSpeaker D: And then the CMU folks and I were sort of on the other side.\nSpeaker D: And because they had collected a lot of meetings that were sort of like this and said that those are nothing like these meetings.\nSpeaker D: So they're really two different kinds of data.\nSpeaker D: I guess we just left it as that if there's found data that can be transformed for use and speech recognition easily, then of course we would do it.\nSpeaker D: But newly collected data would be natural meetings.\nSpeaker H: Actually, the CMU folks have collected a lot of data.\nSpeaker H: Is that going to be publicly available?\nSpeaker D: As far as I know, they have not.\nSpeaker I: So it's not near far, right?\nSpeaker D: I'm not sure.\nSpeaker D: If people were interested, they could talk to them.\nSpeaker D: But I got the feeling there was some politics on that.\nSpeaker I: That's one of my actions.\nSpeaker I: Just to check.\nSpeaker H: We should know what's out there.\nSpeaker I: I mean, I thought they'd only done farfew-wide intelligent room sort of things.\nSpeaker I: I hadn't known that they'd done anything.\nSpeaker H: Well, they only did the farfew.\nSpeaker D: Yeah.\nSpeaker D: I see.\nSpeaker D: But they had multiple mics and they did do recognition and they did do real conversations.\nSpeaker D: As far as I know, they didn't offer that data to the community at this meeting.\nSpeaker D: But that could change because Mark's really into this.\nSpeaker D: We should do that in touch with them.\nSpeaker H: Yeah.\nSpeaker H: I mean, I don't want to be sent out first note saying, hey, this list exists, but...\nSpeaker H: Is that an accident?\nSpeaker H: Yeah, I already had it.\nSpeaker H: That one.\nSpeaker H: I'm always going to do that.\nSpeaker H: So hopefully everybody here is on that list.\nSpeaker H: We should at least check that everybody here.\nSpeaker H: Thank you.\nSpeaker I: Everyone here is on the list.\nSpeaker I: Yeah.\nSpeaker I: I'm not.\nSpeaker I: I think you are.\nSpeaker H: We haven't sent anything to the list yet.\nSpeaker H: Oh.\nSpeaker H: We're just compiling.\nSpeaker I: I added a few people who I knew had to be on, even though they didn't tell me.\nSpeaker I: Like, change for your left boss.\nSpeaker I: You are on it, aren't you?\nSpeaker F: Yeah.\nSpeaker F: So I'm just just a clarification.\nSpeaker F: So, found data, they mean like established corporate linguistics and other fields, right?\nSpeaker F: Some things and stuff.\nSpeaker D: Some things and stuff.\nSpeaker D: They don't have to fund to collect.\nSpeaker D: And especially, well, I mean, found has also the meaning that it's very natural.\nSpeaker D: It's things that occurred without any...\nSpeaker D: These people weren't wearing clothes talking mics, but they were recorded anyway, like the congressional hearings and, you know, for legal purposes.\nSpeaker F: By any means, like standard corporate that had been used for years and linguistics and\nSpeaker D: marks aware of those. Okay.\nSpeaker D: That would be a found data because they found it.\nSpeaker D: And it didn't have to collect it.\nSpeaker D: Of course, it's not found in the sense that at the time it was collected.\nSpeaker D: But what he means is that, you know, Mark was really a fan of getting as much data as possible from, you know, reams and reams of stuff, broadcast stuff, web stuff, TV stuff, radio stuff.\nSpeaker D: But he well understands that that's very different than this type of meeting, but so what?\nSpeaker D: Still, it's interesting for other readers.\nSpeaker G: So, like we're winding down, right?\nSpeaker G: Mm-hmm.\nSpeaker G: Many ways.\nSpeaker G: Such a care of governance.\nSpeaker G: I'll buy the property.\nSpeaker J: Yeah.\nSpeaker J: I wish it could be on the same thing.\nSpeaker E: Oh, yes, we could do that.\nSpeaker I: Now, I was already thinking about it.\nSpeaker I: This is made for task.\nSpeaker I: I really liked the idea of what I thought was interesting was the combination of the cross-pad and the speech, especially the interaction of them rather than just note-taking.\nSpeaker I: So, can you determine the interesting points by whose writing can you do special gestures and so on that they have special meaning to the corporate?\nSpeaker I: I really like that.\nSpeaker F: Well, I just realized there's another category of interesting things, which is that I found this discussion very...\nSpeaker F: this question of how you get at queries, really interesting, and the fact that it's sort of nebulous, what kind of a query would be because it depends on what your purpose is.\nSpeaker F: So, I actually found that whole process of trying to think of what that would involve to be interesting, but that's not really a specific fact.\nSpeaker F: We went around a nice discussion of the factors involved there, which I think was worthwhile.\nSpeaker J: I had a real revelation about taking pictures that I know why I didn't do this before.\nSpeaker J: So, I was very interested in making pictures of the words.\nSpeaker J: The points on really relate to this meeting.\nSpeaker J: I mean, it will take pictures of them, but...\nSpeaker E: They're related to this morning's meeting.\nSpeaker J: So, the previous meeting, that's right.\nSpeaker J: That's about it for some of them.\nSpeaker A: I'm going to pass because I can't... I mean, Jane took my answer.\nSpeaker A: So, I'm going to pass for the moment, but come back to me.\nSpeaker E: Pass?\nSpeaker E: I think pass is socially acceptable, but I will say I will actually...\nSpeaker E: It's been on slightly different spin on what you said, this issue of realizing that we could take minutes, and that actually may be a goal.\nSpeaker E: So, that may be kind of the... in the sense test data, the template of what we want to test again.\nSpeaker E: It's generating a summary.\nSpeaker E: So, that was an interesting new twist on what we can do with this data.\nSpeaker C: I agree with Jane and Erica.\nSpeaker C: I think the question of how to generate queries automatically was the most interesting question that came up.\nSpeaker C: And it's something that, as you said, is her research topic in itself, so I don't think we'd be able to do anything on it, because we don't have funding on it in this project.\nSpeaker C: But so, it's definitely something I would want to do.\nSpeaker B: Do you want to work... we're already been done on that.\nSpeaker F: Okay, expert systems.\nSpeaker G: I suppose we're like the part where we were asked what questions you were...\nSpeaker G: Well, being more management lately than research, I think the thing that impressed me most was the people dynamics, and not any of the facts, and I really enjoyed hanging out with this group of people today.\nSpeaker G: So, that's what really impressed me most.\nSpeaker G: Can we get a fine bag?\nSpeaker I: Well, we had people wearing the wireless mics all the time.\nSpeaker A: Well, I mean, one thing you could search for is where people laughing a lot.\nSpeaker H: I'd probably search for something like that.\nSpeaker I: That actually has come up a couple times in queries.\nSpeaker I: I was starting to land in, that was one of his examples.\nSpeaker H: When did people laugh?\nSpeaker H: Find me the funny thing that Jeff said.\nSpeaker I: So, we need the laugh detector, yeah.\nSpeaker I: Because it seems to be pretty common.\nSpeaker I: I'm nervous now.\nSpeaker H: Quiet and solving.\nSpeaker E: So, I think we're done.\nNone: Okay, okay.\nNone: Great.\nNone: Can I turn it over?\nSpeaker H: I'm sorry, I can't turn it over.\nNone: Please, return off.\nNone: Do you have a job?\nNone: Do you have a job?\nSpeaker I: Do you have a job, Jim, with me in the box?\nSpeaker I: I don't like that.\n"}, {"meeting_id": "ES2007a", "summary": "The team got warmed-up through the opening drawing game which also works as the tool training process. The meeting was primarily composed of discussions among team members on potential functions and designs of remote controls based on their own life experiences. By then, feasibilities on budgets and demographic marketing strategies were also taken into consideration which influence greatly the potential functions and designs.", "dialogue": "Speaker B: The What's the name of the comment, or?\nNone: Just on the lines.\nNone: Okay, spin.\nNone: o that is a kil 10 earning rating.\nNone: Esqu locations like this.\nSpeaker B: HERO can now start the waiting time.\nSpeaker D: So okay.\nSpeaker E: I don't know\nSpeaker D: Can use\nSpeaker F: Is that it right? Nice and normal\nNone: Okay Okay I can just come here, it's fiddly.\nNone: What do we start?\nNone: Does anybody know?\nSpeaker B: Oh, another one.\nSpeaker F: So that's this.\nSpeaker C: Oh, okay.\nNone: Right.\nSpeaker C: Are we free to take notes?\nSpeaker B: Yeah.\nSpeaker B: I don't mean to say so.\nSpeaker B: Okay.\nSpeaker C: Just hang on a second, everybody.\nSpeaker C: I haven't actually looked at this yet.\nSpeaker C: I haven't looked at it, but let's just start it off.\nSpeaker C: If you're ready.\nSpeaker B: Okay.\nSpeaker C: It was the agenda.\nSpeaker C: Opening acquaintance, tool training, project plan discussion, and then closing.\nSpeaker C: Project A is a new remote control.\nSpeaker C: It's original, trendy, and it's user friendly.\nSpeaker C: Project method, functional design, individual work, and then a meeting, conceptual design, individual work, and a meeting, details design, individual work, and a meeting.\nSpeaker B: Tool training, try out the whiteboard.\nSpeaker C: Every participant should draw their favorite animal and sum up their favorite characteristics of that animal.\nSpeaker C: Okay.\nSpeaker C: Miss industrial designer, would you like to go first?\nSpeaker D: So are we supposed to bring the little things for these?\nSpeaker C: Yeah, I think just, do you have a belt?\nSpeaker C: Clip them back to you.\nSpeaker D: Oh, put them in your pocket.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: So my favorite animal.\nSpeaker C: Yeah, what's your favorite animal?\nSpeaker D: Okay.\nSpeaker F: Ah.\nSpeaker F: Is it rude?\nSpeaker C: That's a very good elephant.\nSpeaker C: Back into the elephant.\nSpeaker F: Oh my gosh.\nSpeaker F: I've never got a deal to draw that well.\nSpeaker C: Okay.\nSpeaker C: And you want to write up on there, so you've got to sum up your favorite characteristics of that animal.\nSpeaker E: Okay.\nSpeaker E: It's big.\nSpeaker E: It's got a great memory.\nSpeaker E: Is it?\nSpeaker E: Oh.\nSpeaker E: It's supposed to have a great memory.\nSpeaker E: You say an elephant never forgets.\nSpeaker E: And I don't know why, but it looks like nice to me.\nSpeaker C: Okay.\nSpeaker C: Wonderful.\nSpeaker C: Well done.\nSpeaker C: Do you want to use the wiper and wipe it off?\nSpeaker C: This is the aesthetics designer.\nSpeaker C: You want to go next?\nSpeaker F: I do want my favorite animal.\nSpeaker A: Okay.\nNone: My favorite animal.\nSpeaker D: Let's see.\nSpeaker A: Oh.\nSpeaker A: Enough of any of you have seen Napoleon Dynamite before.\nSpeaker A: It's a lager.\nSpeaker A: What?\nSpeaker A: A combination of a lion and tiger.\nSpeaker A: All right.\nSpeaker A: You're not seen Napoleon Dynamite.\nSpeaker A: No, it's a hilarious memory.\nSpeaker A: You have to see it.\nSpeaker A: And it's best characteristic is, and it's pretty much the awesomest animal.\nSpeaker A: But you have to see the movie and fully appreciate it.\nSpeaker A: Okay.\nSpeaker C: Well done.\nSpeaker F: Great.\nNone: Me?\nSpeaker F: Yep.\nSpeaker C: Miss Marfishing.\nSpeaker C: Okay.\nSpeaker F: I'll show you how this is going to work.\nNone: Cool.\nSpeaker F: Well, I'll try my best to draw.\nSpeaker F: Can I just draw the face?\nSpeaker C: Yeah, I think you can just draw the face, but then you'll have to describe in writing how the rest of it looks.\nSpeaker C: Ooh.\nSpeaker C: It's a cat.\nSpeaker F: That's a baby.\nSpeaker C: It also has a big fat body and a long tail.\nSpeaker F: Okay.\nSpeaker F: Do you want to do some?\nSpeaker C: Right.\nSpeaker F: Why?\nSpeaker F: Because cuddly and usually cats are very friendly usually.\nSpeaker F: And the healing well, the heal, and they can feel when a human's problem.\nSpeaker C: They're kind of spiritual.\nSpeaker F: So that's why I like cats.\nSpeaker C: Well done.\nSpeaker F: There we are.\nNone: It's me.\nSpeaker B: Do, do, do, do.\nSpeaker C: Okay.\nSpeaker C: I don't actually have a favorite animal, but for the, for the, for the thing else, draw a little.\nSpeaker B: I honestly can't draw for toffee.\nSpeaker C: In fact, no.\nSpeaker C: This is a prairie, Dad.\nSpeaker C: No.\nSpeaker C: That's exactly what it is.\nSpeaker C: Not a very good one.\nSpeaker E: Not bad, I would say.\nSpeaker B: Okay.\nSpeaker B: What do you got?\nSpeaker B: It's a squirrel.\nSpeaker B: And I like them because they're cute.\nSpeaker B: And stupid.\nSpeaker B: All right.\nSpeaker C: Okay.\nSpeaker C: I guess that was the test to see if this equipment was all working.\nSpeaker C: Let's move on to the next page.\nSpeaker C: Okay.\nSpeaker C: So, I'm going to go to the next page.\nSpeaker C: I'm going to go to the next page.\nSpeaker C: I'm going to go to the next page.\nSpeaker C: I'm going to go to the next page.\nSpeaker A: I'm going to go to the next page.\nSpeaker A: I'm going to go to the next page.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: I'm going to go to the next page.\nSpeaker C: Okay.\nSpeaker C: All right.\nSpeaker D: Who's got experiences with remote controls then?\nSpeaker D: Pretty much everybody.\nSpeaker F: Yeah.\nSpeaker C: Yeah.\nSpeaker F: It's TV remote control.\nSpeaker F: It's TV.\nSpeaker F: What would I like?\nSpeaker C: What?\nNone: Big.\nSpeaker C: Do you want it small?\nSpeaker C: Medium.\nSpeaker C: Are we going for like telephones or tiny things or we're going for something that's big?\nSpeaker A: It seems like there's sort of a tension between two ideas.\nSpeaker A: You want one remote that maybe can work all of the functions of TV and if you have whatever associated with the TV, the DVD players.\nSpeaker A: Did you know that?\nSpeaker A: At the same time you don't want a really busy remote with a thousand buttons on it or something.\nSpeaker C: Maybe you, yeah.\nSpeaker C: Now that's the other thing is just going to be cheap.\nSpeaker A: Yeah.\nSpeaker C: Because I mean I was thinking something that's got different, like maybe an LCD display on it that's got different patterns.\nSpeaker C: It's got different pages for different devices.\nSpeaker C: But that would probably be quite expensive.\nSpeaker D: But how do we know how much, I mean, how much do we have per...\nSpeaker D: How much?\nSpeaker C: It can't be more than 1250 per unit.\nSpeaker C: Per unit, yeah.\nSpeaker E: Cast.\nSpeaker E: So do we have to be realistic within the budget for it?\nSpeaker C: Well at the moment, we could, I mean, because we, this is what we're doing at the moment is just saying what we'd like.\nSpeaker C: And then after we, after we've found out what we can like, some different ideas, we can then go and do the research to find out if these, any of these ideas are feasible or not.\nSpeaker C: So it would be nice to have something that controls lots of different things.\nSpeaker A: Yeah.\nSpeaker F: Don't we have like one that comes out?\nSpeaker F: Like, so you have one in like, it doesn't have to be really thick.\nSpeaker F: I mean, rock controls can be thin bits. And then you have one for your DVD and you sort of slide it out.\nSpeaker F: You have another one you slide it out.\nSpeaker F: So it slides and then it all comes compact.\nSpeaker F: Okay, that's a light.\nSpeaker F: Into one. So it's not, you're actually just putting three or four different remotes together, but making them thinner and into one.\nSpeaker F: That's an idea.\nSpeaker F: So you just flip them out.\nSpeaker E: Okay.\nSpeaker E: Well, have one very complicated one on one side with all the DVD and BCR, access and stuff.\nSpeaker E: And then on the other side, one, a remote control, that would be very, very simple with just the channel program plus and minus and then just the mute button, for example, I can really keep it really, really simple on one side and the other side be very complicated.\nSpeaker C: One side to kids, one side for adults.\nSpeaker E: I'm not sure if it's a good idea to have a two-sided remote because it would be very hard to use in the way that you would slide it into something on the back, you wouldn't be able to press the buttons.\nSpeaker C: Like a flip telephone, something like that, maybe flip it up and then you've got all the buttons or you flip it closed and you just got their basic buttons on the outside maybe.\nSpeaker C: Oh, we've got five minutes left.\nSpeaker D: Start breaking up.\nSpeaker C: Okay.\nSpeaker C: Well, we've got a few ideas there.\nSpeaker A: Yeah, we should.\nSpeaker A: I guess by the end of this meeting, we should have at least a rough conceptual, stage one was technical functions design.\nSpeaker A: What effect the apparatus should have.\nSpeaker A: So we still have time in our next meeting to come up with actual concept for the user interface, but the functions that we're sure that we want are that it can control the TV, but also devices connected to the TV and be able to operate.\nSpeaker C: I think we should also have a backup plan of, because I just think that it might be expensive to make something that we haven't been told it has to be something that will control everything.\nSpeaker C: We should have a backup plan of just a really good television remote control that is just for a TV, but it's just a really good, nice one.\nSpeaker C: What do you reckon?\nSpeaker C: Yeah, because I'm just thinking, bear in mind, we have to have something that's cheap to make.\nSpeaker C: Yeah, that's true.\nSpeaker A: Maybe we should just concentrate on having a TV remote and have it be.\nSpeaker C: I think we'll be able to come up with ideas and stuff a lot quicker.\nSpeaker A: Have it be like ergonomic.\nSpeaker A: It's comfortable to use, simple to use, and looks decent.\nSpeaker F: What will make it interesting for people to buy, though?\nSpeaker C: Or maybe even something that's for disabled people or people that don't see very well or big buttons for touchy buttons for a market to a certain kind of people.\nSpeaker A: Oh, just one that looks really fucking cool.\nSpeaker E: Can you really light or something special?\nSpeaker A: Yeah, rather than focus on.\nSpeaker C: Otherwise, we'll be here all day talking about how to do this and do that.\nSpeaker C: I think we should...\nSpeaker E: At the end of the day, if it says just TV remote, it doesn't say a combination at all.\nSpeaker C: Obviously, everyone sounds like we're all a bit gadget heads, and we like things that do everything at once, but...\nSpeaker C: Yeah, that's right.\nSpeaker F: I think a flip up thing, because you always have this problem of if it's on the...\nSpeaker F: Well, I didn't anyway.\nSpeaker F: We had five or six remotes, and they'd be lying on the couch, and you'd come and sit down and...\nSpeaker F: Ooh, the television switch on or something.\nSpeaker F: Maybe something that does have a lid or closes, so you don't accidentally press a button or...\nSpeaker C: Okay, like a lot of fun.\nSpeaker C: Like a lot of funks on it like you have when you tell a phone.\nSpeaker F: Yeah.\nSpeaker F: But make it like really snazzy and cool with people who want it, so make it...\nSpeaker C: Yeah, it's gotta be sellable.\nSpeaker A: Yeah, that's true.\nSpeaker A: You were saying it doesn't have to have a flip function.\nSpeaker A: It can just have a lot of functions, so that it's not usable when you don't want it to be using it.\nSpeaker F: Yeah.\nSpeaker F: And even for kids as well, it's safer for them, I guess.\nSpeaker F: Like, they don't flick onto channels and not allow to flick onto.\nSpeaker C: No porn channel for children.\nSpeaker C: Okay.\nSpeaker C: All right, so we've got some ideas.\nSpeaker C: Anything else that's moving on?\nSpeaker C: Good for them.\nSpeaker B: Oh, just close that.\nSpeaker B: What does ID, VID and M-E stand for?\nSpeaker C: Industrial designer.\nSpeaker C: Oh, okay.\nSpeaker C: Which is...\nSpeaker F: Oh, these are kind of specific.\nSpeaker F: And I'm marketing.\nSpeaker B: Yeah, there you go.\nSpeaker C: So, let's...\nSpeaker C: User interface designer.\nSpeaker C: That's...\nSpeaker C: That's me.\nSpeaker C: That's...\nSpeaker C: Okay.\nSpeaker C: That's you, so you've got to...\nSpeaker C: There you go.\nSpeaker C: You're gonna be the one that's working out what buttons we need.\nSpeaker C: Industrial designer, you're the one that you're gonna be working out.\nSpeaker C: Kind of box it goes in, I guess.\nSpeaker C: Or...\nSpeaker C: Whether it's...\nNone: I...\nSpeaker C: What goes into the box somehow.\nSpeaker D: How it works.\nSpeaker D: I'm marketing.\nSpeaker C: User interface.\nSpeaker F: User interface.\nSpeaker F: User interface.\nSpeaker F: User interface.\nSpeaker F: So what the user requires.\nSpeaker A: Yeah, what...\nSpeaker A: Right, okay, yeah.\nSpeaker A: Yeah, it does seem like a responsibility.\nSpeaker C: You two are gonna...\nSpeaker C: I think you just dub up, you know, working together.\nSpeaker C: You're the one that's gonna go and find out, do the research.\nSpeaker C: See what people want in a remote.\nSpeaker C: What buttons are used more often and stuff like that.\nSpeaker C: Oh, we've been warned to finish the meeting now.\nSpeaker B: Okay.\nSpeaker C: Okay, everyone, well done.\nSpeaker C: Good meeting.\nSpeaker A: Alright, see you in 30 minutes.\nSpeaker B: Yeah.\nSpeaker F: So, do you take these off?\nSpeaker F: I don't...\n"}, {"meeting_id": "Bmr013", "summary": "The group discussed the collection status for a set of connected digits recordings that are nearly complete and ready to be trained on a recognizer. Anticipated results were discussed in reference to results obtained for other digits corpora, i.e. Aurora and TI-digits. The group also considered the prospect of performing fine-grained acoustic-phonetic analyses on a subset of Meeting Recorder digits or Switchboard data. Pre-segmentation manipulations that allow for the segmentation of channel-specific speech/non-speech portions of the signal and the distinction of foreground versus background speech were discussed. Finally, speaker fe008 and fe016 reported on new efforts to adapt transcriptions to the needs of the SRI recognizer, including conventions for encoding acronyms, numbers, ambient noise, and unidentified inbreaths.", "dialogue": "Speaker A: I'm not sure that that thing should move.\nSpeaker F: Yes, positive.\nSpeaker F: Ah, positive.\nSpeaker F: I'm positive.\nSpeaker F: It's the POS.\nSpeaker F: Yes.\nSpeaker F: You've soaked up all the positivity.\nSpeaker E: Yes, right.\nSpeaker E: It's sort of happened.\nSpeaker E: That would be good day, man.\nSpeaker E: That's nice.\nSpeaker A: It really is.\nSpeaker A: Oh, it's really bad.\nNone: It's so few.\nSpeaker G: Yeah, they're getting bigger.\nSpeaker G: Oh, so it's recording.\nSpeaker G: Thank you, man.\nSpeaker C: So we got this.\nSpeaker C: That's weird.\nSpeaker C: I remember when she was gathering her data for a dissertation.\nSpeaker C: Were you around?\nSpeaker C: I was doing the I-Ever.\nSpeaker C: All right.\nSpeaker G: Well, I don't know why the POS isn't moving.\nSpeaker E: Yeah.\nSpeaker E: That's something the trick cut.\nSpeaker E: Check out.\nSpeaker F: So, so, this is the first we are.\nSpeaker F: We have, our and have been recording.\nSpeaker H: Yeah, it is working.\nSpeaker H: I don't know why the numbers aren't going up, but the files are getting bigger.\nSpeaker H: And what is it?\nNone: Nothing.\nSpeaker H: Well, I didn't change anything.\nSpeaker H: It may have been working before.\nSpeaker H: It probably was working before.\nSpeaker H: And the only way I just checked is that the numbers are getting bigger and the files.\nSpeaker H: The files are getting bigger.\nSpeaker H: The file size is.\nSpeaker H: You want to put your mic on while we're talking?\nSpeaker D: Probably, stop.\nSpeaker D: So, in the future, if the POS is not moving.\nSpeaker H: Ignore it.\nSpeaker H: It's probably working.\nSpeaker H: I mean, we'll find out for sure in a moment.\nSpeaker H: So, that means we got the joke on that.\nSpeaker F: That's what we asked.\nSpeaker E: We have the joke.\nSpeaker E: Yeah.\nSpeaker H: So, I want to discuss digits briefly, but that won't take too long.\nSpeaker E: Okay, agenda items.\nSpeaker E: We have digits.\nSpeaker E: What else we got?\nSpeaker E: You were the presentation.\nSpeaker E: You were the presentation.\nSpeaker E: I'm going to ask you a question.\nSpeaker C: Do we want to say something?\nSpeaker C: Yeah, why don't you summarize an update of the transcript?\nSpeaker E: Update on transcripts.\nSpeaker D: And I guess that includes the filtering for the ASRFs.\nSpeaker F: And filtering for what?\nSpeaker D: For the references that we need to go from the fancy transcripts to the sort of...\nSpeaker C: Basically, it'll be recap on the meeting that we had in jointly this morning.\nSpeaker C: I think it's done as well.\nNone: Got it.\nSpeaker E: Anything else more pressing than those things?\nSpeaker E: So, finally, just do those.\nSpeaker H: As you can see from the numbers on the digits we're almost done.\nSpeaker H: The digits goes up to about 4,000.\nSpeaker H: And so, we probably will be done with the TI digits in another couple of weeks, depending on how many we read each time.\nSpeaker H: So, there were a bunch that we skipped.\nSpeaker H: Someone fills out the form and then they're not at the meeting.\nSpeaker H: So, it's blank.\nSpeaker H: But those are almost all filled in as well.\nSpeaker H: And so, once it's done, it would be very nice to train up a recognizer and actually start working with this data.\nSpeaker G: So, what's a corpus that's the size of the TI digits?\nSpeaker H: One particular test set of TI digits.\nSpeaker H: So, I extracted...\nSpeaker H: There was a file sitting around which people have used here as a test set.\nSpeaker H: It had been randomized.\nSpeaker H: That's just what I used to generate the order of these particular ones.\nSpeaker E: So, first of all, what we could do is take the standard training set for TI digits, train up with whatever great features we think we have.\nSpeaker E: For instance, and then test on this test set.\nSpeaker E: Presumably, you should do reasonably well on that.\nSpeaker E: And then, presumably, we should go to the distant mic and do poorly.\nSpeaker E: And then, you should get really smart over the next year or two.\nSpeaker H: And then, you should get a good reason by one or two percent.\nSpeaker H: But, in order to do that, we need to extract out the actual digits.\nSpeaker H: So, the reason it's not just a transcript is that their false starts and misreads and miscues and things like that.\nSpeaker H: And so, I have a set of scripts and next waves where you just select the portion, hit R, it tells you what the next one should be, and you just look for that.\nSpeaker H: So, it'll put on the screen.\nSpeaker H: The next set is 69922. And you find that and hit the key and it records it in a file in a particular format.\nSpeaker H: So, the question is, should we have the transcribers do that or should we just do it?\nSpeaker H: Or some of us? I've been doing, I've done eight meetings, something like that.\nSpeaker H: Just by hand. Just myself, rather.\nSpeaker H: So, it will not take long.\nSpeaker G: What do you think?\nSpeaker C: What's not going to be discussed is, we discussed this for coffee and I think it's a fine idea, partly because it's not unrelated to the present skill set.\nSpeaker C: But it will add, for them, an extra dimension might be an interesting break for them.\nSpeaker C: And also, it is contributing to the composition of the transcript, because we can incorporate this number strictly and it will be more complete transcripts.\nSpeaker C: So, I think it's fine.\nSpeaker E: So, you think it's fine to have the transcribers do it?\nSpeaker H: There's one other small bit which is just entering the information which is at the top of this form onto the computer to go along with where the digits are recorded automatically.\nSpeaker H: And so, it's just typing in name, time date and so on, which again, either they can do, but it is firing up an editor or again, I can do or someone else can do.\nSpeaker C: And that, you know, that one, I'm not so sure fits into the things that I want to use.\nSpeaker C: I want to use the hours for because in the time that they'd be spending doing that, they wouldn't be able to be putting more words on it.\nSpeaker C: But that's really your choice.\nSpeaker G: So, are these two separate tasks that can happen?\nSpeaker G: Or do they have to happen at the same time?\nSpeaker H: No, they don't have this.\nSpeaker H: You have to enter the data before you do the second task, but they don't have to happen at the same time.\nSpeaker H: So, it's just I have a file which has this information on it.\nSpeaker H: And then when you start using my scripts for extracting the times, it adds the times at the bottom of the file.\nSpeaker H: And so, I mean, it's easy to create the files and leave them blank.\nSpeaker H: And so, actually, we could do it in either order.\nSpeaker H: Okay.\nSpeaker H: It's sort of nice to have the same person do it just as a double check to make sure you're entering for the right person.\nSpeaker H: But either way.\nSpeaker E: Yeah. Yeah, just by way of a order of magnitude, we've been working with this Aurora data set.\nSpeaker E: And the best score on the nicest part of the data, that is where you've got training and test set that are basically the same kinds of noise and so forth, is about, I think, the best score was something like 5% error per digit.\nSpeaker E: So, that digit.\nSpeaker E: Right. So, if you were doing 10 digit recognition, it would really be in trouble.\nSpeaker E: So, the point there, and this is car noise, things but real situation, well, real.\nSpeaker E: There's one microphone that's close that they have.\nSpeaker E: It's this sort of thing, the close first is distant, but in a car instead of having a projector noise, it's car noise.\nSpeaker E: But it wasn't artificially added to get some artificial signal that was racially when it was just people driving around a car.\nSpeaker E: So, that's an indication.\nSpeaker E: That was with many sites competing, and this was the very best score and so forth.\nSpeaker G: Although, more typical models weren't that good, right?\nSpeaker G: I mean, the models are pretty crappy.\nSpeaker E: You're right. I think that we could have done better on the models.\nSpeaker E: But the thing is that this is a kind of typical number for all of the things in this task, all of the languages.\nSpeaker E: And so, I think we probably, the models, would be better in some than in others.\nSpeaker E: So, anyway, just an indication.\nSpeaker E: Once you get into this kind of realm, even if you're looking at connected digits, it can be pretty hard.\nSpeaker C: It's going to be fun to see how we compare it.\nSpeaker C: How do we do with the prosaudage?\nSpeaker H: There's so much difference.\nSpeaker H: It's going to be strange.\nSpeaker H: I mean, the prosaudics are not the same as TI digits, for example.\nSpeaker H: So, I'm not sure how much of effect that we'll have.\nSpeaker H: How do we connect the prosaudage?\nSpeaker H: Just what we were talking about with grouping.\nSpeaker H: That with these, the grouping, there's no grouping at all.\nSpeaker H: And so, it's just the only sort of discontinuity you have is at the beginning and the end.\nNone: So, what are they doing in Aurora?\nSpeaker H: Are they reading all number?\nSpeaker H: Aurora, I don't know.\nSpeaker D: I don't know what they're doing in Aurora.\nSpeaker E: I'm not sure.\nSpeaker E: No, no, I mean, it's connected.\nSpeaker E: It's connected to digits.\nSpeaker D: But it's also not just the prosaudage that cross the crossword model.\nSpeaker H: Right.\nSpeaker H: But in TI digits, they're reading things like zip codes and phone numbers and things like that.\nSpeaker H: How do we do on TI digits?\nSpeaker H: I don't remember.\nSpeaker H: I mean, very good, right?\nSpeaker H: Yeah, I mean, we were in one and a half percent, two percent.\nSpeaker E: Oh, I think we got under a percent.\nSpeaker E: Oh, really?\nSpeaker E: But I mean, the very best system that I saw in literature was 0.25 percent or something, and somebody had a Bell Labs.\nSpeaker E: Right.\nSpeaker E: But sort of pulling out all the stuff.\nSpeaker E: But I think a lot of systems sort of get half a percent.\nSpeaker E: Right.\nSpeaker E: We were in there somewhere.\nSpeaker H: But that means it's really, it's close talking mics, no noise, clean signal, just digits.\nSpeaker H: I mean, everything is good.\nSpeaker D: Yeah, exactly.\nSpeaker D: Yeah.\nSpeaker H: And we only recently got it to anywhere near a human.\nSpeaker H: Prehistory.\nSpeaker D: Beginning of life, yeah.\nSpeaker H: And it's still like an order of magnitude worse than humans, too.\nSpeaker E: Right.\nSpeaker E: When they're right away, yeah.\nSpeaker E: Yeah.\nSpeaker E: After coffee.\nSpeaker E: After coffee.\nSpeaker E: Not after lunch.\nSpeaker H: Okay, so what I'll do then is I'll go ahead and enter this data and then hand off to Jane and the transcribers to do the actual extraction of the digits.\nSpeaker E: Yeah.\nSpeaker E: One question I have that, I mean, we wouldn't know the answer to now, but do some guessing about it.\nSpeaker E: I was talking before about doing some modeling, marking of our territory features, with overlap, and so on.\nSpeaker E: And on some subset, one thought might be to do this on the digits or some of these digits.\nSpeaker E: It would be easier.\nSpeaker E: So if we're the only thing is I'm a little concerned that maybe the kind of phenomena.\nSpeaker E: And the reason for doing it is because the argument is that certainly with conversational speech, the stuff that we've looked at here before, just doing the simple mapping from the phone to the corresponding features that you could look up in a book, isn't right.\nSpeaker E: It isn't actually right.\nSpeaker E: In fact, there's these overlapping processes where some voicing comes up and then some, you know, some miziality comes in here and so forth.\nSpeaker E: And you do this gross thing and saying, well, I guess it's this phone starting there.\nSpeaker E: So that's the reasoning.\nSpeaker E: But it could be that when we're reading digits because it's for such a limited set that maybe that phenomenon doesn't encourage much.\nSpeaker E: Do you have any opinion about that?\nSpeaker C: It strikes me that there are more, each of them is more informative because it's so random and that people might articulate more and they might end up with more across the correspondence.\nSpeaker C: Yeah, I agree.\nSpeaker G: It's sort of less predictability.\nSpeaker D: Yeah.\nSpeaker D: Well, it's definitely true that when people are reading, even if they're re-reading what they had said spontaneously, that they have very different patterns, mid-show that and the desertations have shown that.\nSpeaker D: So the fact that they're reading, first of all, whether they're reading in a room of people or, you know, the fact that they're reading will make a difference.\nSpeaker D: Yeah.\nSpeaker D: Well, it depends what you're interested in.\nSpeaker H: Would this corpus really be the right one to even try that on?\nSpeaker E: See, I don't know.\nSpeaker E: So maybe the thing will be to do takes a very small subset.\nSpeaker E: I mean, I'd have a big program and take a small subset of the conversational speech and a small subset of the digits and look and just get a feeling for it.\nSpeaker E: Just take a look, really.\nSpeaker E: Because I don't think anybody is, at least I don't know of anybody.\nSpeaker E: Well, I don't know.\nSpeaker E: I can be interesting in design too.\nSpeaker C: Because then you have the comparison of the predictable speech versus the last predictability speech and maybe you'd find it at work too.\nSpeaker C: And the case of the beginning.\nSpeaker G: I have to think about the particular acoustic features to mark too because, I mean, the things they wouldn't be able to mark, like, you know, tens lacks.\nSpeaker G: And some things are really difficult, you know.\nSpeaker H: I think we can get a hollow end to give us some advice on that.\nSpeaker C: Also, I thought you were thinking of a much more restricted set of features.\nSpeaker E: Yeah, but I was like, he said, I was going to bring Todd in and then, but I mean, you want to be restrictive, but you also want to have coverage.\nSpeaker E: You know, you should, it should be such that if you, if you, if you had all of the features determined that you, that you have chosen, that that would tell you in the study's day case, the phone, so.\nSpeaker H: Even, I guess, with vowels, that would be pretty hard, wouldn't it?\nSpeaker H: Which identify, actually, you know, which one it is?\nSpeaker C: It seemed to me that the points of articulation would be more, I mean, that's, I think, about articulation, which means, rather than bells.\nSpeaker G: Points of articulation, what do you mean?\nSpeaker C: So is it bilabial or dental or is it, you know, paladine?\nSpeaker C: Which are all like, where are you talking about?\nSpeaker C: Place of articulation.\nSpeaker C: Yeah.\nSpeaker C: Just thank you, whatever I said.\nSpeaker E: Okay.\nSpeaker E: Yeah.\nSpeaker E: Okay, we got our jargon in.\nSpeaker E: Yeah.\nSpeaker D: Well, it's also, there's really a difference between the pronunciation, vowels, and the dictionary, and the pronunciations that people produce.\nSpeaker D: And so, you get some of that information from Steve's work on that labeling, and it really, I actually think that data should be used more, that maybe, although I think the meeting context is great, that he has transcriptions that give you the actual phone sequence, and you can go from, not from that to the articulatory features, but that would be a better starting point for marking the gestural features than data where you don't have that, because we, we want to know both about the way that they're producing a certain sound and what kinds of, what kinds of phonemic differences you get between the transcribed sequences and the dictionary ones.\nSpeaker E: Well, you might be right, that might be the way at getting it, what I was talking about, but the particular reason why I was interested in doing that was because I remember when that happened, and John O'Hall was over here, and he was looking at the spectrograms of the more difficult ones.\nSpeaker E: He didn't know what to say about what is the sequence of phones there.\nSpeaker E: They came up with some compromise, because that really wasn't what it looked like.\nSpeaker E: It didn't look like a sequence of phones.\nSpeaker E: It looked like this blending thing happening here.\nSpeaker G: Yeah, so you have this feature here.\nSpeaker D: There's no name for that.\nSpeaker D: But it still is, there's a, there are two steps.\nSpeaker D: One is going from a dictionary pronunciation of something, like, you're going to see you tomorrow.\nSpeaker D: It could be going to or gun to, you know, or gun to.\nSpeaker D: Yeah, you're going to see you tomorrow, you'll see you tomorrow.\nSpeaker D: And that would be nice to have these intermediate or these, some, these reduced pronunciations that those transcribed marked, or to have people mark those as well, because it's not, it's not that easy to go from the dictionary word pronunciation, the dictionary phone pronunciation to the gestural one without this intermediate sort of civil level kind of representations.\nSpeaker H: Well, I don't think more consistent thing that we do that, though.\nSpeaker E: Yeah, I mean, I'm just, the moment, of course, we're just talking about what to provide is a tool for people to do research.\nSpeaker E: You have different ideas about how to do it.\nSpeaker E: So for instance, you might have someone who just has a word, has words with states and has, comes from our particular gestures to that.\nSpeaker E: And someone else might actually want some phonetic, you know, intermediate thing.\nSpeaker E: So I think it would be best to have all of it if we could.\nSpeaker E: But, uh,\nSpeaker H: What I'm imagining is a score like notation, where each line is a particular feature.\nSpeaker H: Right? So you would say, you know, it's a voice through here, and you have label here, you have nasal here, and they could be overlapping in all sorts of bizarre ways that don't correspond to the timing on phones.\nSpeaker E: I mean, this is the kind of reason I remember when one of the switchboard workshops that when we talked about doing the transcription project, Dave Tolkien said can't be done.\nSpeaker E: Right.\nSpeaker E: And he was, what he meant was that this isn't, you know, a sequence of phones, and we actually look at switchboard, that's not what you see.\nSpeaker H: And in fact, the interannotary agreement was not that good, right?\nSpeaker H: On the harder ones?\nSpeaker H: Yeah, I mean...\nSpeaker D: I don't know how you look at it, and I understand what you're saying about this kind of transcription, exactly, because I've seen, you know, where does the voicing of our start and so forth.\nSpeaker D: All I'm saying is that it is useful to have that the transcription of what was really said and which syllables were reduced if you're going to add the features.\nSpeaker D: It's also useful to have some level of representation, which is a reduced...\nSpeaker D: It's a pronunciation variant that currently the dictionaries don't give you because if you add them to the dictionary and you run recognition, you add confusion.\nSpeaker D: So people purposely don't add them.\nSpeaker D: So it's useful to know which variant was produced at least at the phone.\nSpeaker G: So it would be great if we had either these kind of lablings on the same portion of switchboard that Steve marked or Steve's type markings on this data.\nSpeaker D: Exactly.\nSpeaker D: Exactly.\nSpeaker D: Yeah, no, I don't disagree with that.\nSpeaker D: It's not that slow, I don't know what that means.\nSpeaker E: Yeah, I don't disagree with it.\nSpeaker E: The only thing is that what you actually would end up with is something...\nSpeaker E: It's all compromise, right?\nSpeaker E: So the string that you end up with isn't actually what happened, but it's the best compromise that a group of people scratching their heads could come up with to describe what happened.\nSpeaker G: And it's more accurate than...\nSpeaker E: And it's more accurate than the dictionary, or if you've got a pronunciation lexicon that has three or four, this might have been the fifth one.\nSpeaker D: So it's like a catcher or whatever, go and start the way down.\nSpeaker D: That's what I meant.\nSpeaker D: And in some places it would fill in the kinds of gestural features are not everywhere.\nSpeaker D: So there are some things that you don't have access to either from your ear or the spectrogram, but you know what phone it was, and that's about all you can say.\nSpeaker D: And then there are other cases where...\nSpeaker G: It's basically to have multiple levels of information marking.\nSpeaker H: Well, the other difference is that the features are not synchronous, right? They overlap with each other in weird ways.\nSpeaker H: So it's not strictly one-dimensional signal.\nSpeaker H: So I think that's sort of qualitatively different.\nSpeaker D: You can add those features in, but it'll be under specified.\nSpeaker D: There'll be no way for you to actually mark what was said completely by features.\nSpeaker H: Well, not with our current system, but you could imagine designing a system that the states were features rather than phones.\nSpeaker D: Well, we probably have a separate discussion of that.\nSpeaker D: It's not that it was that.\nSpeaker C: But that was not the kind of direction I thought.\nSpeaker E: Yeah, so I mean, where this is...\nSpeaker E: I mean, I wanted...\nSpeaker E: I would like to have something that's useful to people other than those who are doing the specific kind of research.\nSpeaker E: I have in mind, so it should be something broader.\nSpeaker E: But what I'm coming from is we're coming off of stuff that Larry Saul did with...\nSpeaker E: with John Dallin and Muzzie Rahim, in which they have a multi-band system that is trained through combination and gradient learning and NDM to estimate the value for particular feature.\nSpeaker E: And this is part of a larger image that John Dallin has, but how he brain does it, which he's sort of imagining that individual frequency channels are coming up with their own estimate of these kinds of...\nSpeaker E: something like this might not be exact features that Jacobs and Thott over something.\nSpeaker E: But I mean, something like that's a kind of low-level features which are not fully phone classification.\nSpeaker E: And this particular image of how it's done is that then given all of these estimates at that level, there's a level above it, which is making some kind of sound unit classification such as phone.\nSpeaker E: And you could argue what those sound units should be.\nSpeaker E: That's sort of what it was imagining doing.\nSpeaker E: But it's still open within that, whether you would have an intermediate level in which it was actually phones or not, you wouldn't necessarily have to.\nSpeaker E: But again, I wouldn't want to...\nSpeaker E: I wouldn't want what we produced to be so local and perspective that it was matched what we were thinking of doing one week.\nSpeaker E: And what you're saying is absolutely right that if we can, we should put in another level of description there if we're going to get into some of this low-level stuff.\nSpeaker G: Well, you know, I mean, if we're talking about having the annotators annotate these kinds of features, it seems like, you know, the question is, do they do that on meeting data, or do they do that on switchboard?\nSpeaker H: That's what I was saying. Maybe meeting data isn't the right corpus.\nSpeaker C: Well, it seems like you could do both. I mean, I was thinking that it would be interesting to do it with respect to parts of switchboard anyway, in terms of, partly to see if you could generate first guesses at what the articulatory feature would be based on the phone representation of that lower level, that might be a time game, but also in terms of compatibility of...\nSpeaker G: Well, because then, yeah, and then also, if you did it on switchboard, you would have the full continuum of transcriptions.\nSpeaker G: You'd have it from the lowest level, the acoustic features, then you'd have the, you know, the fanatic level that Steve did.\nSpeaker G: It would be a complete set of things.\nSpeaker E: It could be an altered game.\nSpeaker E: It's so it's a little different.\nSpeaker E: So, I mean, we'll see how much we can get the people to do and how much more I have in all this stuff.\nSpeaker E: It might be good to do with your boat.\nSpeaker G: You know, seed it with guesses about what we think the features are based on, you know, the phone or Steve's transcriptions or something.\nSpeaker G: It's a basic...\nSpeaker H: Quicker.\nSpeaker H:...of the phone transcripts, so they would all be synchronous, but then you could imagine adjusting them here and there.\nSpeaker G: Exactly.\nSpeaker G: Scoot the voicing over a little bit.\nSpeaker E: Right. Well, I think what, I mean, I'm a little behind in what they're doing now and the stuff they're doing on switchboard now, but I think that Steve and getting are doing something with an automatic system first and then doing some adjustment as a recall.\nSpeaker E: So, I mean, that's probably the right way to go anyways.\nSpeaker E: It's to start off with an automatic system with a pretty rich pronunciation dictionary that, you know, tries to label it all.\nSpeaker E: And then people go through and fix it.\nSpeaker C: So, in our case, you'd think about us starting with regular dictionary.\nSpeaker E: Well, regular dictionary, I mean, it's a pretty rich dictionary. It's got a number of pronunciation.\nSpeaker G: You could start from the, if we were going to do the same set of sentences that Steve had done, we could start with those transcriptions.\nSpeaker G: That's actually what I was thinking about.\nSpeaker D: The problem is when you run, if you run a regular dictionary, even if you have variants in there, which most people don't, you don't always get out the actual pronunciation, so that's why the human transcribers giving you that pronunciation.\nSpeaker D: Actually, maybe they're using phone recognizers.\nSpeaker D: They, they, they, they were, I think they were, I think they would be good.\nSpeaker E: Yeah, so I think that we also don't have, I mean, we've got a good start on it, but we don't have a really good meeting recorder recognizer or transcribe or anything yet.\nSpeaker E: So, I mean, the other way to look at this is to, is to do some stuff on switchboard, which has all this other stuff to it.\nSpeaker E: And then, as we get further down the road, we can do more things ahead of time.\nSpeaker E: We can do some of the same things to the meeting data.\nSpeaker E: Yeah.\nSpeaker C: And these people might, they, they are, most of them are trained with IPA. They'd be able to do, than any level coaching.\nSpeaker G: Are they busy for the next couple of years?\nSpeaker C: You know, I mean, they, they're interested in continuing working with us.\nSpeaker C: So, I mean, I, and this would be up their alley, so we could, when, when you meet with, with John O'Hall and find what, just text on him, he want to apply them.\nSpeaker C: Maybe, you're good to train them too.\nSpeaker C: Yeah.\nSpeaker E: Anyway, this is not an urgent thing at all. It's just a team-mob team that have that data.\nSpeaker H: How would you do a forced alignment?\nSpeaker H: Interesting idea. You want to iterate somehow.\nSpeaker H: Interesting thing they think about.\nSpeaker H: I mean, you'd want models for spreading.\nSpeaker G: Of the acoustic teachers.\nSpeaker F: Yeah.\nSpeaker D: Well, it might be, you need to do some phonetic features.\nSpeaker D: These non-word words, or these kinds of words that people never, the, the, the, hums or the, these, no, I'm serious.\nSpeaker D: They're all these kinds of functional elements.\nSpeaker D: I don't know what you call them.\nSpeaker D: It's not just field-positive, but all kinds of ways of interrupting and some of them are, yeah, a hus and, hmm.\nSpeaker D: Okay. Grants. Now, that might be interesting.\nSpeaker C: He's got the Vibes eating.\nSpeaker F: We should move on.\nSpeaker F: New version of pre-signitation.\nSpeaker B: Oh, yeah. I worked a little bit on a pre-signitation to get the non-version, which does channel-specific speech-transpeak detection.\nSpeaker B: And what I did is I used some normalized features, which look into the, which is normalized energy, energy normalized by the mean over the channels and by the minimum over the, over within each channel.\nSpeaker B: And to, to, yet, to normalize also loudness and modified loudness and things, and those special features, which are in my feature actor.\nSpeaker B: And, and therefore, to be able to somewhat distinguish between foreground and background speech in, in the different, in each channel.\nSpeaker B: And, I tested it on, on three or four meetings, and it seems to work really well, I would say.\nSpeaker B: There are some problems with the lapel mic, of course.\nSpeaker H: That's great.\nSpeaker H: Yeah. So, I understand that's what you were saying about your problem with minimum.\nSpeaker B: Yeah, and I had, I had specific problems with the...\nSpeaker H: 90th quartile rather than minimum.\nSpeaker B: Wow.\nSpeaker B: Yeah, yeah. I did some, some, something like that.\nSpeaker B: And, I mean, there are, there are some, some problems in, when, in the channel there, there, there, there's a speaker dozen, doesn't work match or doesn't talk at all.\nSpeaker B: Then, the, yeah, there are, there are some problems with, with the normalization and, then, the system doesn't work at all.\nSpeaker B: So, I'm glad that there is the, the digit part where everybody's supposed to say something.\nSpeaker B: So, that's, that's great for, for my purpose.\nSpeaker B: And, the thing is, I, I, the evaluation of, of the system is a little bit hard as I don't have any references.\nSpeaker H: Well, we did the hand, the one by hand.\nSpeaker B: Yeah, that's the one, where, where I do the training on, so I come to the evaluation on that.\nSpeaker B: So, the thing is, can the transpray was perhaps do some, some, some meetings in, in terms of speech, non speech in, in the specific transpray?\nSpeaker B: Well, won't you have that from their transcriptions?\nSpeaker G: Well, okay.\nSpeaker C: So, I think I might have done what you're requesting, though I did it in a service of a different thing.\nSpeaker C: I have 30 minutes that I, more tightly transcribe with reference to individual channels.\nSpeaker B: Okay, that's great, that's great for me, okay.\nSpeaker H: So, hopefully that's not the same meeting that we did.\nSpeaker H: No, actually it's different meeting.\nSpeaker C: So, um, so, you know, we have the, they transcribe as if it's one channel with these, with the slashes to separate the overlapping parts.\nSpeaker C: And then we run it through, then I'm going to edit it, then I'm going to run it through channelize, which takes it into Dave Gilbert's format.\nSpeaker C: And then you have all these things split across according to channel.\nSpeaker C: And then that means that if a person contributed more than one synagogum overlap during that time bin that, that two parts of the utterance end up together, it's the same channel.\nSpeaker C: Okay.\nSpeaker C: And then I took his tool, and last night for the first 30 minutes of one of these transcripts, I tightened up the boundaries on individual speakers channels.\nSpeaker C: Because his, his interface allows me to have total flexibility in the time tags across the channels.\nSpeaker B: And, um, so, yeah, yeah, that's great, but would be nice to have some more meetings, not just one meeting to be sure that that we could get a couple meetings done with that level of precision.\nSpeaker H: I think that would be a good idea.\nSpeaker C: Oh, okay.\nSpeaker C: How much time, so the meeting is very in length.\nSpeaker C: What are we talking about in terms of the number of minutes you'd like to have as your, as your dream is at?\nSpeaker B: It seems to me I would be good to have a few minutes from, from different meetings.\nSpeaker B: So, but I'm not sure about how much.\nSpeaker C: Okay. Now, you're saying different meetings because of different speakers or because of different audio quality or both different, different, different number of speakers, different speakers, different.\nSpeaker E: Yeah, we don't have that much variety of meetings yet. I mean, we have this meeting and future meeting. We have a couple others that we have a couple examples of.\nSpeaker A: Even probably with the games differently will affect it.\nSpeaker B: Not really as, because of the normalization.\nSpeaker D: We can try running, we haven't done this yet because Andreas is going to move over the SRI recognizer.\nSpeaker D: Basically, I ran out of machines at SRI because we're running the e-vails and I just don't have machine time there.\nSpeaker D: But once that's moved over, hopefully in a couple days, then we can take what Jane just told us about as the pre-segmented, the segmentations that you did at level eight or some threshold that same, right.\nSpeaker D: And try doing forced alignment on the word strings. And if it's good, then that will, that may give you a good boundary. Of course, if it's good, we don't, then we're fine.\nSpeaker D: But I don't know yet whether these segments that contain a lot of pauses around the words will work or not.\nSpeaker B: I would quite like to have some manually transcribed references for a system as I'm not sure if it's really good to compare with some other automatic boundaries.\nSpeaker C: Well, now if we were to start with this and tweak it manually, they might be okay. It really depends on a lot of things.\nSpeaker D: But I would have maybe a transcriber look at the old forced alignment and then adjust those that might save some time.\nSpeaker D: If they're horrible, it won't help at all. But they might not be horrible.\nSpeaker D: So I'll let you know on the...\nSpeaker D: Okay, great.\nSpeaker C: How many minutes would you want from... I mean, we could easily get a section, you know, like say a minute or so, from every meeting that we have. So from the newer ones that we're working on.\nSpeaker C: And then...\nSpeaker B: If it's not the first minute of the meeting, that's okay with me, but in the first minute, there are often there are some strange things going on, which are really well for which are really good.\nSpeaker B: So what I'd quite like perhaps is to have some five minutes of different meetings.\nSpeaker C: Somewhere not in the very beginning, five minutes, okay.\nSpeaker C: And then I wanted to ask you just for my information then, would you be training...\nSpeaker C: So would you be training then the segment or so that it could on the basis of that segment the rest of the meeting?\nSpeaker C: So if I give you like five minutes, is the idea that this would then be applied to providing a higher...\nSpeaker B: I could do a retraining with that, yeah.\nSpeaker B: But I hope that I don't need to do it.\nSpeaker B: So it can be doing an unsupervised way.\nSpeaker C: Excellent.\nSpeaker C: Okay.\nSpeaker B: I'm not sure, but for those three meetings, which I did, it seems to be quite well.\nSpeaker B: But there are some, as I said, some of some problems with the lapel mic, but perhaps we can do something with cross correlations to get rid of those.\nSpeaker B: That's what I, that's my future work.\nSpeaker B: What I want to do is to look into cross correlations for removing those false overlaps.\nSpeaker D: Wonderful.\nSpeaker D: Are the wireless different than the wired mics at all?\nSpeaker B: I'm not sure. If there are any wired mics in those meetings, or I have to look at them.\nSpeaker B: But I think there's no difference between...\nSpeaker D: This is just the lapel versus everything else.\nSpeaker C: Okay, so then if that's five minutes per meeting, we got like 12 minutes, 12 meetings.\nSpeaker C: We're happy that I've been working with.\nSpeaker E: Of the meetings that you're working with, how many of them are different?\nSpeaker E: Are there any of them that are different than these two meetings?\nSpeaker C: Oh, in terms of the speakers or the...\nSpeaker C: Yes, speakers. Sorry.\nSpeaker C: We have different combinations of speakers.\nSpeaker C: I mean, just from what I've seen, there are some where your present are not present.\nSpeaker C: And then you have the difference between the network's group and this group.\nSpeaker C: Yeah, so I didn't know any of the group you had.\nSpeaker C: So you have the networks meeting?\nSpeaker E: Do you have any of Jerry's meetings in your time?\nSpeaker C: No.\nSpeaker C: No.\nSpeaker C: We could. I mean, you recorded one last week or so.\nSpeaker C: Yes, we.\nSpeaker D: We're going to be recording them every Monday.\nSpeaker E: Because I think he really needs a variety and having as much variety for speaker, we certainly would be a big part of that.\nSpeaker C: Okay, so if I, okay, include.\nSpeaker C: Okay, then if I were to include all together samples from 12 meetings, that would only take an hour.\nSpeaker C: And I could get the transcripts to do that, right?\nSpeaker C: I mean, what I mean is that would be an hour of sample, and then they transcribe those that hour, right?\nSpeaker C: That's what you should do.\nSpeaker E: Yeah.\nSpeaker E: And, right, like the yours.\nSpeaker C: I mean, I mean, adjust.\nSpeaker C: So they get it into the multi-channel format and then adjust the time bands.\nSpeaker E: So that should be faster than the 10 times.\nSpeaker C: Absolutely.\nSpeaker C: I did, I did.\nSpeaker C: So last night I did.\nSpeaker C: Well, last night I did about half an hour in three hours, which is not terrific, but anyway, it's an hour and a half per.\nSpeaker C: So I can't calculate on my.\nSpeaker B: The transcripts actually stop with transcribing new meetings or are they?\nSpeaker C: Well, they're still working. They still have enough to finish that I haven't assigned a new meeting.\nSpeaker C: But the next, I was about to need to assign a new meeting and I was going to take it from one of the new ones, and I can easily give them Jerry Feldman's meeting no problem.\nSpeaker C: Okay.\nSpeaker D: So they're really running out of data.\nSpeaker D: I mean, that's good.\nSpeaker D: At that first set, okay.\nSpeaker E: They're running out data unless we make the decision that we should go over and start transcribing the other set.\nSpeaker E: So the first set, the first set.\nSpeaker C: And so I was in the process of like adding this wonderful news.\nSpeaker C: We funded experiment, but also we were thinking maybe applying that to getting the final date very useful to getting the overlaps to be more precise all the time.\nSpeaker E: This blends nicely into the update on transcripts.\nSpeaker C: Yes, it does.\nSpeaker C: Well, Liz and Don and I met this morning in the Barco room and this afternoon.\nSpeaker C: It's afternoon.\nSpeaker C: It's afternoon.\nSpeaker C: It's afternoon to the afternoon.\nSpeaker C: Concerning this issue of the, well, there's basically the issue of the interplay between the transcript format and the processing that they need to do for the SRI recognizer.\nSpeaker C: And well, so I mentioned the process that I'm going through with the data.\nSpeaker C: So I get the data back from the transcript.\nSpeaker C: Well, I met it for like, get the data back from the transcriber and then I check for a simple things like spelling errors and things like that.\nSpeaker C: And I'm going to be doing a more thorough editing with respect to consistency of the conventions.\nSpeaker C: But they're generally very good.\nSpeaker C: And then I run it through the channelized program to get into the multi-channel format.\nSpeaker C: Okay.\nSpeaker C: And what we discussed this morning, I would summarize, is saying that these units that result in a particular channel and a particular time band at that level, very in length.\nSpeaker C: And their recognizer would prefer that the unit's not the overly long, but it's really an empirical question.\nSpeaker C: Whether the units we get at this point through just that process I described might be sufficient for them.\nSpeaker C: So as a first pass through, the first chance without having to do a lot of hand editing, what we're going to do is I'll run it through channelized, give them those data after I've done the editing process.\nSpeaker C: I'm sure it's great.\nSpeaker C: I can do that pretty quickly with just that minimal editing without having to hand break things.\nSpeaker C: And then we'll see if the units that we're getting at that level are sufficient and maybe don't need to be further broken down.\nSpeaker C: And if they do need to be further broken down, then maybe it just be piecewise.\nSpeaker C: Maybe it won't be the whole thing.\nSpeaker C: So that's what we were discussing.\nSpeaker C: Right?\nSpeaker C: Also, we discussed a lot of educational things.\nSpeaker C: So it's like, I hadn't incorporated a convention explicitly to handle acronyms, for example.\nSpeaker C: But if someone says PCM, it would be nice to have that be directly interpretable from the transcript, what they said, or tickle TCL.\nSpeaker C: And so I've incorporated also convention with that, but that's easy to handle with the post-editing phase.\nSpeaker C: And I'll mention it to transgarbers for the next phase.\nSpeaker C: And that's okay.\nSpeaker C: And then a similar convention for numbers.\nSpeaker C: So they say 183 versus 183.\nSpeaker C: And also I'll be encoding, as I do my post-editing, the things that are in curly brackets, which are clarificational material, to incorporate keyword at the beginning.\nSpeaker C: So it's going to be either a gloss, or it's going to be a vocal sound like a laugh or a cough, or a non-vocal sound like a dorsal, dorsal, and that can be easily done with a, you know, just a lemon-like additional thing in a general format.\nSpeaker D: Yeah, we just needed a way to strip, you know, all the comments, all the things that the linguist wants, but the recognizer can't do anything with.\nSpeaker D: But to keep things that we mapped to like reject models, or mouth noise, or cough.\nSpeaker D: And then there was this interesting issue, Jane, brought up, which I hadn't thought about before, but I was realizing as I went through the transcripts that there are some noises like, well, the good example was an in-breath where a transcriber working from the mixed signal doesn't know whose breath it is.\nSpeaker D: And they've been assigning it to someone that may or may not be correct. And what we do is, if it's a breath sound, you know, a sound from the speaker, we map it to a noise model, like a mouth noise model in the recognizer.\nSpeaker D: It probably doesn't hurt that much once in a while to have these, but if they're in the wrong channel, that's not a good idea.\nSpeaker D: And then there's also things like door slams that's really in no one's channel. They're like, it's in the room. And Jane had this nice idea of having like an extra...\nSpeaker D: An extra channel.\nSpeaker D: Yeah, I've been adding that.\nSpeaker D: And we were thinking that is useful also when there's uncertainty. So if they hear a breath and they don't know whose breath it is, better to put it in that channel than to put it in the speaker's channel because maybe it was someone else's breath.\nSpeaker D: So I think that's a good... you can always clean that up post-processing. There's a lot of little details, but I think we're coming to some kind of closure.\nSpeaker D: So the idea is that Don can take Jane's post-process channelized version and with some scripts convert that to a reference for the recognizer and we can run these.\nSpeaker D: So when that's ready, as soon as that's ready and as soon as the recognizer is here, we can get 12 hours of force aligned and recognize data and start working on it.\nSpeaker D: So I don't know, a couple of weeks or two away, I would say. If that process is automatic once we get your post-processed transcript.\nSpeaker C: And that doesn't mean that a better thing that it would require is not very much. They're just hoping that the units that are provided in that way will be sufficient because that would save a lot of time dividing things.\nSpeaker D: Yeah, some of them are quite long. I don't know how long you did one.\nSpeaker A: I saw a couple around 20 seconds and that was just without looking too hard for it. So I would imagine that there might be something that are longer.\nSpeaker A: One question, would that be a single speaker? Is that multiple speakers overlap? No, but if we're going to segment it, like if there's one speaker in there that says okay or something, right in the middle is going to happen around it.\nSpeaker D: It's not the fact that we can't process a 20 second segment. It's the fact that there's 20 seconds in which to place one word in the wrong point.\nSpeaker D: If someone has a very short utterance there and that's where we might want to have this individual, you know, have your pre-processed input.\nSpeaker B: And I just don't know how to do that. It's not that perhaps the transcripts could stop then from those multiple speakers detections.\nSpeaker D: Right. And enjoying the hand margin. Yeah, that's what I was thinking too. That's probably what will happen, but we'll try it this way and see.\nSpeaker D: I mean, it's probably good enough for a force alignment. If it's not then we're really, and we definitely, but for free recognition, it'll probably not be good enough.\nSpeaker D: Like it lots of errors because of the crosstalk and noise.\nSpeaker E: Good. I think that's probably agenda. Oh, I want to ask one thing. Yeah, then microphones and microphones. When do we get to?\nSpeaker D: They said to take about a week. You already. So what happens to our old microphones?\nSpeaker H: Well, the only thing we're going to have extra. Right. We don't have extra now is just the lapel, not the body pack, just lapel.\nSpeaker H: And then one of the one of those since what I decided to do on Morgan's suggestion was just get two new microphones and try them out.\nSpeaker H: And then if we like them, we'll get more. Since they're there like 200 bucks a piece. We want to at least try them out.\nSpeaker H: So it's a replacement for this headset and mic. Yeah. And they're going to do the wiring for us. What's the style of the headset?\nSpeaker H: It's it's by crown and it's one of these sort of mount around the ear thingies. And when I when I mentioned that we thought it was uncomfortable, he said it was a common problem with the Sony. And this is how apparently a lot of people are getting around it. And I checked on the web and every side I went to raved about this particular mic. It's apparently comfortable and stays on the head well. So we'll see if it's any good. But I think it's promising.\nSpeaker H: Yeah. Yeah. So it was.\nSpeaker H: It was accurate.\nSpeaker F: Yeah. It was a great employee. It was for the record Adam is not a patient. That's our self-employee. It's a listed out. Well we're using the crown. These are crown aren't they? The PCMs are crown aren't they?\nSpeaker H: Yeah. Yeah. They were you bet. And they worked very well. So if we go to a workshop about all this, it's going to be a meeting about meetings about meetings.\nSpeaker H: And then we have to go to the planning session for that workshop. Oh, that would be a meeting about the meeting.\nSpeaker H: Start saying M4. Yeah.\nSpeaker E: And we did it.\nSpeaker H: Yep. Go for it.\nSpeaker F: Okay.\nSpeaker E: Transcript.\nSpeaker E: 261-126-3-0. 3-20-5653-4-450-57566-662-789-0215263-2512-37-4706800-5681-93 04012\nSpeaker G: Transcript 2591-2610-2497980-3406-4656-7087621-93-0205-0413100-1754 280-2814-3405-620-73913-841-9720100-0106-7931\nSpeaker H: Transcript 3731-3750-7200-499-888-9800-2116-337-47-59304-69922-78-0104-19557-2990-34 509-714-1200\nSpeaker D: Transcript 3711-3730-607-307-890454-12439-263456095-81966-3810-057012-30708-308-1080 7 0 8 5 3 1 4 8 0 3 6 4 2\nSpeaker H: Paws between lines remember.\nSpeaker C: 2 1 7 1 dash 2 1 9 0 6 2 9 7 7 2 7 0 7 9 7 8 6 9 9 0 0 8 9 0 1 0 7 3 2 3 1 4 2 2 4 5 7 7 8 4 6 8 0 7 0 7 8 0 3 4 7 9 0 2 0 1 7 7 0 7 2 3 0 8 3 9 4 9 5\nSpeaker B: Transcript 2 1 5 1 dash 2 1 7 0 5 2 4 8 6 3 7 6 8 7 5 6 9 8 9 0 0 7 8 1 1 2 3 6 3 8 4 8 1 3 0 2 5 6 7 9 2 8 3 5 2 8 0 3 0 0 9 3 1 8 0 8 2 3 2 2 3\nSpeaker A: 0 4 8 6 0 9 6 5 Transcript 2 3 5 1 dash 2 3 7 0 2 3 5 4 0 4 2 6 4 1 7 7 4 8 7 2 1 8 9 0 0 2 3 3 0 4 8 2 5 8 9 6 0 9 0 4 8 2 7 8 0 9 8 0 4 0 0 0 1 7 0 4 3 0 2 8 1 1 7 8 0 5 7 2 9 6 2 0 excuse me 2 9 6 2 1 1 0 ok\n"}, {"meeting_id": "Bmr007", "summary": "Members of the team discussed options to encode aspects of conversation that cannot be captured through microphones. overlaps, as well as methods to deduce the length of pauses and the reasons behind them. This led to discussing how else to document the conversations, and archive each participant\u00e2\u0080\u0099s responses separately. ", "dialogue": "Speaker G: Yeah, yeah, okay.\nSpeaker G: We're on again, okay.\nSpeaker G: That is okay.\nSpeaker H: So if anyone hasn't signed the consent form, please do so.\nSpeaker H: The new consent form.\nSpeaker H: The new information form.\nSpeaker H: I mean, that's the one that we're going to bring to you back.\nSpeaker H: Okay.\nSpeaker H: And shall I go ahead and do some digits?\nSpeaker G: We're going to do that at the end, remember?\nSpeaker G: Okay, whatever you want.\nSpeaker G: Yeah, just to be consistent from you around and at least to the end.\nSpeaker G: It's a, yeah, it doesn't matter, okay.\nSpeaker G: Okay.\nSpeaker G: Well, it just, I mean, it might be that someone here has to go.\nSpeaker G: That was the other sort of the point.\nSpeaker G: So I had asked, actually, anybody who had any ideas for an agenda to send it to me and no one did.\nSpeaker G: So that was an over-guy.\nSpeaker E: From last time I wanted an issue.\nSpeaker E: Right.\nSpeaker E: From last time.\nSpeaker G: Okay, so one item for Denda is Jane has some, some research to talk about, research issues.\nSpeaker G: And I have some short research issues.\nSpeaker G: And I'm going to have some short research issues.\nSpeaker G: I have a list of things that I think were done over the last three months.\nSpeaker G: I was supposed to send off.\nSpeaker G: And I sent a note about it to Adam and Jane, but I think I'll just run through it also and see if someone thinks it's inaccurate or...\nSpeaker G: A list you have to send off to...\nSpeaker G: Oh.\nSpeaker G: Okay, you know.\nSpeaker G: So, so I'll go through that.\nSpeaker G: And...\nSpeaker G: Anything else?\nSpeaker G: Anything else to talk about?\nSpeaker I: No.\nSpeaker I: What about the new trip?\nSpeaker G: The next day.\nSpeaker G: Sort of off topic, I guess, because I was all about the...\nSpeaker G: I can chat with you about that.\nSpeaker G: Fine, that's another thing.\nSpeaker G: And...\nSpeaker G: Anything else?\nSpeaker G: Anything else?\nSpeaker G: There's a...\nSpeaker G: I mean, there is a...\nSpeaker G: A telephone call tomorrow, which would be a conference call.\nSpeaker G: Some of us are involved in...\nSpeaker G: For a possible proposal.\nSpeaker G: We'll talk about it next week.\nSpeaker H: Do you want me to give that?\nSpeaker H: I know you.\nSpeaker H: CC me, but I wasn't actually recipient.\nSpeaker H: I didn't quite know what to make of that.\nSpeaker H: Well, we'll talk about that after.\nSpeaker G: Okay.\nSpeaker G: Okay, so it sounds like the three main things that we have talked about are this list, Dean and Dean Adam have a series of items, and other than that, anything as usual, and if it goes beyond that.\nSpeaker G: Okay, Dean, since you were sort of cut off last time when we start with yours, make sure we get to it.\nSpeaker E: It's very brief.\nSpeaker E: I just want to just have these out.\nSpeaker H: This is the same as the email or different.\nSpeaker E: It's slightly just...\nSpeaker E: It's basically the same idea.\nSpeaker E: Same idea.\nSpeaker E: So I feel like that is using it before.\nSpeaker E: So basically, as you know, part of the encoding includes a mark that indicates that an overlap.\nSpeaker E: It's not indicated with a type precision.\nSpeaker E: It's just indicated that...\nSpeaker E: Okay, so...\nSpeaker E: It's indicated to some of the people know what parts of speech...\nSpeaker E: which stretches the speech were in the clear versus being overlap by others.\nSpeaker E: So I've used this mark and divided the enrol script, which divides things into individual minutes, which ended with 45 and a little bit.\nSpeaker E: And, you know, minute zero, of course, is the first minute up to 60 seconds.\nSpeaker E: And what you can see is the number of overlaps.\nSpeaker E: And then to the right, whether they involve two speakers, three speakers and one and three speakers.\nSpeaker E: And what I was looking for, specifically, was the question of...\nSpeaker E: whether they distributed evenly throughout or whether they're averse of them.\nSpeaker E: And it looked to me as though...\nSpeaker E: You know, this is just...\nSpeaker E: This is not specifically verified, but it did look to me as though they're averse throughout, rather than being unlocalized to a particular region.\nSpeaker E: So, part down there where the maximum number of overlaps is the area where we were discussing whether or not it was useful to code stress, set and stress as possible, indication of information and treatment.\nSpeaker E: So it's like, you know, rather lively discussion there.\nSpeaker G: What's the parenthesis stuff that says...\nSpeaker G: Like the first one says six overlaps and then 2.8?\nSpeaker E: That's the percent.\nSpeaker E: So, six is 2.8% of the total number of overlaps in that session.\nSpeaker E: So, here we end.\nSpeaker E: This is when people were, you know, packing up to go basically to this final stuff.\nSpeaker E: I don't remember where the digits felt.\nSpeaker E: I had to look at that.\nSpeaker E: But the final three there are no overlaps at all.\nSpeaker E: And a couple times there.\nSpeaker E: So it seems like it goes through verse.\nSpeaker E: That's going to be...\nSpeaker E: Now, another question is, are there individual differences in whether you're likely to be overlapped with or to overlap with others?\nSpeaker E: And again, I want to emphasize is just one particular meeting and also it's been a statistical testing involved.\nSpeaker E: But I took the coding of the...\nSpeaker E: You know, I had the script figure out who was the first speaker who was the second speaker involved in a two person overlap.\nSpeaker E: I didn't look at the ones involved in three or more.\nSpeaker E: And this is how it ranks down in the individual cells.\nSpeaker E: Who tended to be overlapping most often with who else?\nSpeaker E: And if you look at the marginal totals, which is the ones on the right side, across the bottom, you get the totals for an individual.\nSpeaker E: So, if you look at the bottom, those are the numbers overlaps in which Adam was involved as the person doing the overlap thing.\nSpeaker E: And if you look... I'm sorry, but you're alphabetical as my question is.\nSpeaker E: And then if you look across the right, then that's where he was the person who was the first speaker in the pair.\nSpeaker E: And he got overlapped with by somebody.\nSpeaker E: And then if you look down on that summary table, then you see that there are differences in whether a person overlap with or overlap.\nSpeaker H: Is this just raw counts or is it...?\nSpeaker H: Raw counts.\nSpeaker H: So it would be interesting to see how much each person spoke.\nSpeaker E: Yes, very true.\nSpeaker E: Very true.\nSpeaker E: We could normalize through that now in the table.\nSpeaker E: I did take one step toward away from the raw frequencies by putting percentages.\nSpeaker E: So the percentage of time, other times that the person spoke, what percentage...\nSpeaker E: Other times the person spoke and furthermore was involved in a two-person overlap.\nSpeaker E: What percentage of the time were they the overlap or what percent of the time were they the overlap be?\nSpeaker E: And there it looks like you see some differences.\nSpeaker E: That some people tend to be overlapped with more often than they're overlapped.\nSpeaker E: Of course, this is just one meeting. There's no statistical testing involved and that would be required for an in-line.\nSpeaker G: So it would be statistically incorrect to conclude from this that Adam talked too much?\nNone: No actually, it's still incorrect.\nSpeaker E: Yeah, that's right.\nSpeaker E: I don't see a point in signaling people out.\nSpeaker E: Now this is a kiss for obviously...\nSpeaker E: But the numbers for people themselves, that's right.\nSpeaker E: So I'm not saying on the tape who did better or worse because I think that...\nSpeaker E: And then here's a case where of course human subjects people would say, be sure that you anonymize results.\nSpeaker E: This is what...\nSpeaker H: This is actually when Jane sent this email first, it's what caused me to start thinking about anonymizing the data.\nSpeaker H: Well, fair enough.\nSpeaker E: And actually, the point is not about an individual. It's a point about tendencies toward different styles, different speakers.\nSpeaker E: Oh, sure.\nSpeaker E: It would be, you know, of course, there's also the question of what type of overlap was this and were they?\nSpeaker E: And I know that I can distinguish at least three types and probably more.\nSpeaker E: I mean, the general cultural idea, which the conversation analyst originally started with in the 70s, was that we had this strict model where politeness involves that you let person finish before you start talking.\nSpeaker E: And, you know, I mean, we know that...\nSpeaker E: And they've loosened up on that too, in the intermeaning time.\nSpeaker E: That that's viewed as being a culturally relative thing.\nSpeaker E: You have the high involvement style from the East Coast where people will overlap often as an indication of interest in what they are person saying.\nSpeaker E: And, you know, exactly, they're the same, fine?\nSpeaker E: And, you know, in contrast, and also, Divertannis, these, as she talked about differences in these types, that they're just different styles.\nSpeaker E: And you can impose a model of the ideal being, no overlaps in conversation.\nSpeaker E: And, you know, it's also bringing advances in the university, and I agree with that.\nSpeaker E: And, and also, I can't say university, but anyway, the people who used to say it was strict, now, don't.\nSpeaker E: I mean, they also acknowledge the influence of subcultural norms and cross cultural norms and things.\nSpeaker E: So, then it be kind of...\nSpeaker E: So, just superficially, you give a couple ideas of the types of overlaps involved.\nSpeaker E: I have the bottom several that I know. So, they're backchats, like, what I'm just now anticipating the end of a question.\nSpeaker E: And, simply answering it earlier. And, there are several of those in this in these days, where, because we're people who've talked to each other, we know basically what the topic is, what the possibilities are, and we've spoken with each other.\nSpeaker E: So, we know basically what the other person's style is likely to be.\nSpeaker E: And, so, and there are a number of places where someone just answered early.\nSpeaker E: And, places also, which I thought were interesting, where two or more people gave exactly the same answer in unison.\nSpeaker E: Different words, of course, but, you know, the basically, you know, everyone's saying yes, or, you know, or even more specific than that.\nSpeaker E: So, the point is that overlaps not necessarily bad thing, and that it would be useful to subdued by these further and see if their individual differences in styles, and respect the types involved.\nSpeaker E: And, that's all I want to say on that.\nSpeaker G: Thank you. Well, of course, the biggest result here, which is when we've talked about many times, and this is new to us, but I think would be interesting to show someone who isn't familiar with this.\nSpeaker G: It's just the sheer number of overlaps. See that, right? That, that, that, here's a relatively short meeting.\nSpeaker G: So, 40, 40 plus minute meeting, and not only were there 215 overlaps, but, I think there's one, one minute there.\nSpeaker G: Where, where, where there wasn't any overlap, I mean, it's throughout this thing.\nSpeaker E: Well, at the bottom, we have the bottom three. So, four, four minutes altogether.\nSpeaker G: Oh, so the bottom three did have stuff going on.\nSpeaker G: Yes, but just no overlaps.\nSpeaker I: Okay, so it's interesting to see what the total amount of time is in the overlap.\nSpeaker I: Yes, exactly. That's where Jose is.\nSpeaker B: I have this, that's where he comes in.\nSpeaker B: I have that information.\nSpeaker B: No. No, about how much is it?\nSpeaker G: Which is the duration of the, of each of the overlaps? What's the, what's the average?\nSpeaker B: I, I have in the average now, but I will, I will do the, the study of the, with the, with the, with the problem, with the, the, the different distribution of the duration of the overlap.\nSpeaker G: Okay, you, you don't have a feeling for roughly how much.\nSpeaker B: Because the budget is a good, the duration is the variation, the variation of the duration is the very, very...\nSpeaker E: I suspect that we'll also differ depending on the type of overlap.\nSpeaker B: Yeah, I'm sure.\nSpeaker B: Because are yourself a, a, a, a son of overlapping with duration, a, a, a, a overlap and another very, very short.\nSpeaker B: I probably is very difficult to, to speak, because the overlap is the only, the, the final S, the, the, the, the, the, the, the, the end, the end of the war, or the previous speaker, with the, the next word of the, the new speaker.\nSpeaker B: And I consider that son overlap, but it's very short, and X, with the, and the idea is probably, when, when we study the, that song, we have a, a, a, confusion with noise, with that, the effective sounds.\nSpeaker B: But, yeah, I have an information, but I have to,\nSpeaker C: use blip as by minute, sorry for the overlap, the strategy, the boundary between two minutes, that comes towards both of those.\nSpeaker E: Actually, actually not. So, it was, think about the case where A, start speaking, and then B, overlaps with A, and then the minute boundary happens. And let's say that, after that minute boundary, B is still speaking, and A overlaps with B, that would be a new overlap.\nSpeaker E: But otherwise, let's say B comes to the conclusion of that turn without anyone overlapping with him in which case, there would be no overlap counted in the second.\nSpeaker C: So most of them, talk simultaneously, both important moments, and one, and the other question to you.\nSpeaker E: Okay, in that case, my, the coding that I was using, since we haven't incorporated Adams coding on the lap, yes, the coding of, yes, not over.\nSpeaker E: So if we haven't incorporated Adams method of handling over, overlaps yet, then that would have fallen through the crack, cracks, it would be an underestimate of the number of overlaps, because I, I wouldn't be able to pick it up from the way it was encoding so far.\nSpeaker E: We just haven't done that. The precise second to second, you know, second to second coding.\nSpeaker G: I'm, I'm confused now. So, let me restate what I thought Andrew has been saying and, and see.\nSpeaker G: Let's say that in, in second, 57 of one minute, you start talking and I start talking, and we ignore each other and keep on talking for six seconds.\nSpeaker G: So we go over, so we were, we were talking over one another, and it's just, in each case, it's just sort of one interval, right?\nSpeaker G: So, we talked over the minute boundary. Is this considered, is one overlap in each of the minutes, the way you have done this?\nSpeaker E: No, it wouldn't. It would be considered as an overlap in the first one.\nSpeaker G: Okay, so that's good, I think, in the sense that I think Andrew has meant the rest of you.\nSpeaker E: I should also say I did a simplifying count in that, if A was speaking, B overlapped with A, and then A came back again, and overlapped with B again, I, I didn't count that as a three person overlap, I counted that as a two person overlap, and it was A being overlapped with by D.\nSpeaker E: Because the idea was the first speaker had the floor, and the second person started speaking, and then the first person re-ocerted the floor kind of thing.\nSpeaker E: These are simple, not an assumption. It didn't happen very often, there may be like three overlaps in that way in the long day.\nSpeaker H: I want to go back and listen to minute 41.\nSpeaker H: Yeah, yeah.\nSpeaker H: Because, yeah.\nSpeaker H: And I think that by interesting that there were a large number of overlaps and they were all two speaker. I mean, what I thought would have thought is that when there were a large number of overlaps, because everyone was talking at once, but apparently not, that's really neat.\nSpeaker H: Yeah, there's a lot of background.\nSpeaker F: Yeah, it is. I think it's really interesting though, before saying yes, meetings have a lot of overlaps, it's to actually find out how many more we have than two party.\nSpeaker F: And if you have a party conversation, like switchboard, does it offer a lot too? If you just look at the back channel, if you consider those overlaps, it's also very huge.\nSpeaker F: It's just that people haven't been looking at that because they've been doing single channel processing.\nSpeaker F: So the question is, how many more overlaps do you have of say the two person type by adding more people to a meeting?\nSpeaker F: And it may be a lot more, but it may not be.\nSpeaker G: Well, but see, I find it interesting, even if it wasn't anymore, because since we were dealing with this full duplex sort of thing in switchboard, where it was just all separated out, we just, everything was just nice.\nSpeaker G: So the issue is in a situation where that's...\nSpeaker F: It's really nice, it depends what you're doing.\nSpeaker F: So if you were actually having, depends what you're doing.\nSpeaker F: Right now, we're doing, we have individual mics on the people in this meeting.\nSpeaker F: And the question is, are there really more overlaps happening than there would be in a two person?\nSpeaker G: Let me rephrase what I'm saying, because I don't think I'm getting across.\nSpeaker G: But shouldn't these words like nice, because we're doing it precise.\nSpeaker G: What I mean is that in switchboard, despite the many other problems that we have, one problem that we're not considering is overlap.\nSpeaker G: And what we're doing now is, aside from the many other differences in the task, we are considering overlap.\nSpeaker G: And one of the reasons they were considering it, one of them, one of them is that, at least I'm very interested in the scenario in which both people talking here are pretty much equally audible in from a single microphone.\nSpeaker G: And so in that case, it does get mixed in, and it's pretty hard to just ignore it, to just do processing on one and on the other.\nSpeaker F: I agree that it's an issue here, but it's also an issue for switchboard. And if you think of meetings being recorded over the telephone, which I think, you know, this whole point of studying meetings isn't just at people in a room, but to also have meetings over different phone lines.\nSpeaker F: Maybe far, feel like people wouldn't be interested in that, but all the dialogue issues still have.\nSpeaker F: So each of us was calling and having a meeting that way, you know, a conference call.\nSpeaker F: And just the question is, you know, in switchboard, you would think that's the simplest case of a meeting, a more than one person.\nSpeaker F: I'm wondering how much more overlap of the types that, that you described happen with more people present. So it may be that having three people is very different from having two people or it may not be.\nSpeaker G: That's an important question to ask. I think what I'm certainly saying is that I don't think we were considering that switchboard. Not you, maybe.\nSpeaker F: Were you measuring it? I mean, there is actually to tell you the truth, the reason why it's hard to measure is because of, so from the point of view of studying dialogue, in which Dan Draskin and Dresden, I had some projects on.\nSpeaker F: I know the sequence of terms. So what happens is if you're talking and I have a back channel in the middle of your turn, and then you keep going, what it looks like in a dialogue model is your turn and then my back channel, even though my back channel occurred completely inside your turn.\nSpeaker F: Yeah.\nSpeaker F: So for things like language modeling or dialogue modeling, we know that that's wrong in real time. But because of the acoustic segmentations that were done and the fact that some of the acoustic data and switchboard were missing, people couldn't study it.\nSpeaker F: But that doesn't mean in the real world that people don't talk that way.\nSpeaker G: Yeah, I wasn't saying that. Right. I was just saying that now we're looking at it.\nSpeaker G: You maybe wanted to look at it before, but for these various technical reasons, in terms of how the data was, you weren't. So that's why it's coming to us as new, even though it may well be.\nSpeaker G: If you're hypothesis, you're offering, right, if it's the null hypothesis, and if actually you have as much overlap in the two person, we don't know the answer to that. The reason we don't know the answer to is because it wasn't studied and wasn't studied because it was a set of.\nSpeaker F: And if you're asking the question from the point of view of what's different from out of meeting, studying meetings of say more than two people versus what kinds of questions you can ask what the two person meeting.\nSpeaker F: It's important to distinguish that, you know, this project is getting a lot of overlap, but other projects for two, but we just couldn't study them.\nSpeaker F: May have been.\nSpeaker F: Well, there is a high rate, but I don't know how high. I have a question.\nSpeaker G: My point was just if you wanted to say to somebody, what do we learn about overlaps here? Just never mind comparison to something else.\nSpeaker G: What we've learned about is overlaps in this situation is that the first order thing I would say is that there's a lot of them.\nSpeaker G: In the sense that if you said, I guess what I'm comparing to is more the common sense notion of how much people overlap.\nSpeaker G: The fact that when Adam was looking for a stretch of speech before that didn't have any overlaps and he was having such a hard time.\nSpeaker G: And now I look at this and I go, well, I can see why he was having such a hard time.\nSpeaker F: That's also been a fish for.\nSpeaker G: I wasn't saying it wasn't. Right?\nSpeaker G: I was commenting about this.\nSpeaker G: I was saying if I have this complicated thing in front of me, which we're going to get much more sophisticated about when we get lots more data.\nSpeaker G: But then if I was going to describe somebody, what did you learn right here about the modest amount of data that was analyzed?\nSpeaker G: I'd say, well, the first order thing was there's a lot of overlaps. In fact, it's not just a bunch of overlap.\nSpeaker G: Second order thing is it's not just a bunch of overlaps in one particular point, but that there's overlaps throughout the thing.\nSpeaker F: And that's interesting. I agree with that. I'm just saying that the reason you get overlaps may or may not be due to the number of people in the meeting.\nSpeaker F: Yeah. Yeah. I wasn't making any statement about that.\nSpeaker F: And it would actually be interesting to find out because some of the data say switchboard, which is an exactly the same kind of contact.\nSpeaker F: And these are two people who don't know each other.\nSpeaker F: But we should still be able to somehow say what is the added contribution to the sort of overlap time of each additional person.\nSpeaker F: Yeah, that would be good to know, but we need the way.\nSpeaker E: Yeah, I agree with that. And the reason is because I think there's a limit.\nSpeaker E: There's an upper bound on how many you can have simply from the standpoint of audibility when we speak.\nSpeaker E: We do make a judgment as adults. I mean children don't adjust so well.\nSpeaker E: If a truck grows rolling past, adults will, depending, but mostly adults will hold off to what to finish the end of the sentence until the noise is passed.\nSpeaker E: And I think we generally do monitor things like that. Whether we, whether our utterance will be in the clear or not.\nSpeaker E: And partly it's related to rhythmic structure and conversation. So, you know, you, you, you, you, you, you, stop.\nSpeaker E: So people tend to time their, their, their, when they come into the conversation based on the overall rhythmic ambient thing.\nSpeaker E: So you don't want to be cross cutting and just to finish this that, that I think that there may be an upper bound on how many overlaps you can have simply from the standpoint of audibility and how loud the other people are who are already in the break.\nSpeaker E: But I, you know, I've certain types. Now, if it's just back channels, people may be doing that with less intention of being heard just sort of spontaneously doing back channels.\nSpeaker E: In which case that those might, there may be no upper bound.\nSpeaker C: I have this feeling that back channels which are the vast majority of overlaps in switchboard.\nSpeaker C: I didn't play it as big a role here because it's very unnatural.\nSpeaker C: I think to back channel in a multi audience, you know, in a multi person.\nSpeaker F: If you can see them actually. So if you watched people are going like, right, like this here, but they may not be the case.\nSpeaker C: But instead of audition, one person speaking and everybody's listening.\nSpeaker G: Actually, I think I've done it a fair number of times today.\nSpeaker G: I thought it had none.\nSpeaker H: Yeah, we need to put trackers on that.\nSpeaker I: In two person.\nSpeaker C: Yeah, so actually that's in part because the nodding, if you have very old contact, the nodding has the same function, but on the phone is switchboard.\nSpeaker C: You, that would work.\nSpeaker C: So you need to use the message here.\nSpeaker I: So in the two person conversations, when there's back channel, is there a great deal of overlap?\nSpeaker I: The speech or because I impression is sometimes it happens when there's a pause.\nSpeaker I: You know, like you get a lot of back channel when somebody's pausing.\nSpeaker E: She's doing that.\nSpeaker I: Sorry about what you're saying.\nSpeaker I: Talk to your both.\nSpeaker I: No, when there's back channel, I mean, just listening, when there's two people talking and there's back channel, it seems like the back channel happens when the pitch drops and the first person.\nSpeaker I: And a lot of times the first person actually stops talking and then there's a back channel and then they start up again.\nSpeaker I: So I'm wondering about, I just wonder how much overlap there is.\nSpeaker I: Is there a lot?\nSpeaker F: I think there's a lot of the kind that Jose was talking about where in the Scult position timing, conversation happens where they come in overlapping.\nSpeaker F: But at a point where the information is mostly complete.\nSpeaker F: So all you're missing is some last syllables or something, the last word, some highly predictable words.\nSpeaker F: So technically it's an overlap.\nSpeaker I: But maybe it's just a small overlap.\nSpeaker F: From the information flow point of view, it's not an overlapping.\nSpeaker I: I was just thinking more in terms of alignment overlap.\nSpeaker H: Language model prediction of overlap that would be really interesting.\nSpeaker F: Well, that's exactly why we wanted to study the pre-sized writing of overlapses in switchboard.\nSpeaker C: So here's a first interesting labeling task to distinguish between say back channels, precision timing, sort of, you know, benevolent overlaps and add, and pick it up and work for pants.\nSpeaker C: I don't know, I'll slide over to someone else's side over that before it's chaos.\nSpeaker C: That might be interesting from just to take over.\nSpeaker E: I think that in this meeting I really had the feeling that wasn't happening, that the hostile type.\nSpeaker E: These were benevolent types, as people finishing each other's sense.\nSpeaker C: I could imagine that there's a fair number of cases where this is not really hostile, but sort of competitive.\nSpeaker C: What person is finishing something and you have like two or three people trying to grab the next turn.\nSpeaker C: And so it's not against the person who talks first, because they're actually all waiting for that person to finish.\nSpeaker G: But they all are next.\nSpeaker G: I have a feeling most of these things that are not benevolent kind are competitive as opposed to really hostile.\nSpeaker I: I'm doing what determines who gets the floor.\nSpeaker G: I mean, vote in Florida.\nSpeaker G: What do you remember?\nSpeaker G: One thing I remember, or you can tell a good joke and then everybody's laughing and you get a chance to break in.\nSpeaker G: But the other thing I was thinking was that all these interesting questions are, of course, pretty hard to answer with a small amount of data.\nSpeaker G: So I wonder if what you're saying suggests that we should make a conscious attempt to have a fair number of meetings with a smaller number of people.\nSpeaker G: Most of our meetings are meetings currently with a 5, 6, 7, 8 people.\nSpeaker G: Should we really try to have some two person meetings or some three person meetings and record them just to beef up the statistics on that?\nSpeaker E: Well, it seems like there are two possibilities.\nSpeaker E: I mean, it seems like if you have just two people, it's not really like a meeting.\nSpeaker E: It's not as similar as the rest of the sample.\nSpeaker E: It depends on what you're after, of course. But it seems like that would be more in case of the control condition compared to an experimental condition with more than two.\nSpeaker G: Well, this was raising the question of whether it's the number.\nSpeaker G: There's a relationship between the number of people and the number of overlaps of type of overlaps.\nSpeaker G: And if you had two people meeting in this kind of circumstance, then you'd still have the visual.\nSpeaker G: So you wouldn't have that difference also that you have in the scenes which are poor data.\nSpeaker E: I'm just thinking that'd be more like a control condition.\nSpeaker H: But from the acoustic point of view, it's all good.\nSpeaker H: It's the same. The acoustic is fine.\nSpeaker C: If the goal were to just look at overlap view, what you could say of yourself saved yourself a lot of time, and not even transparent the words.\nSpeaker F: I was thinking you should be able to do this from the acoustic on the close track in my experience.\nSpeaker H: Well, that was my status for Florence.\nSpeaker H: I've never done with this type of discussion.\nSpeaker F: I don't think you would be able to have any kind of apology on the next one, but you get some rough statistics.\nSpeaker G: But what do you think about that? Do you think that would be useful?\nSpeaker G: I'm just thinking of as an action item, whether we should try to record some two words of meeting.\nSpeaker F: I guess my first comment was only that we should not attribute overlaps only to meetings, but maybe that's obvious.\nSpeaker F: Maybe anybody knew that.\nSpeaker F: In normal conversation with two people, there's an awful lot of the same kinds of overlap, and that it would be interesting to look at whether or there are these kinds of constraints that Jane mentioned, that maybe the additional people add to this competition that happens right after a turn.\nSpeaker F: You can have five people trying to grab a turn, but pretty quickly they back off and we go back to the sort of one person at a time with one person.\nSpeaker F: I don't know to answer your question.\nSpeaker F: I don't think it's crucial to have controls, but I think it's worth recording all the meetings we can.\nSpeaker F: I have a lot of that.\nSpeaker F: I would not record on the two person meeting because it only has a few.\nSpeaker C: Could we have the past and continue to have a fair number of phone calls?\nSpeaker C: We talked about this repeatedly.\nSpeaker C: We can see what happens in terms of role that way.\nSpeaker C: We have to set up for it.\nSpeaker G: We recorded this meeting so regularly.\nSpeaker H: We can look at each other.\nSpeaker H: Turn off the lights.\nSpeaker G: Let me take a picture.\nSpeaker G: That was the other thing.\nSpeaker G: Were we going to take a picture to the beginning of each of these meetings?\nSpeaker H: What I thought we were going to do was just take pictures of the whiteboards rather than take pictures of the meeting.\nSpeaker E: It's a head nodding here.\nSpeaker E: Why do we want to take pictures of the meeting?\nSpeaker E: Because you said the spatial relationship of the speech.\nSpeaker H: You could do that by just noting on the role.\nSpeaker H: I'll do that on the next set of forms.\nSpeaker B: I finally remembered to put the name of the speech.\nSpeaker H: We figured out from the mic.\nSpeaker H: No.\nSpeaker H: The wire was once.\nSpeaker H: I mean, I'm sitting here and the jack is over here.\nSpeaker F: Probably from the evening.\nSpeaker H: It would be another task.\nSpeaker H: Having ground to proof would be nice.\nSpeaker H: You could get it.\nSpeaker H: The game forming during the digit.\nSpeaker H: So I'm going to put the labels on all the chairs with the seat number.\nSpeaker H: That's a good idea.\nSpeaker C: But the chairs are in the same level.\nSpeaker C: Put them on the table.\nSpeaker E: But you know, the linguistic anthropologists would say it would be good to have a digital picture anyway.\nSpeaker E: Because you get a sense of a posture.\nSpeaker E: You could block out a person's face or whatever.\nSpeaker E: But these are important cues.\nSpeaker G: But from one picture, I don't know that you really get that.\nSpeaker G: You can be better than nothing.\nSpeaker E: You can just from a synch picture.\nSpeaker E: I think you can come up with some aspects.\nSpeaker E: I mean, I could tell you, if I'm in certain meetings, I really do the body language.\nSpeaker E: It's very interesting in terms of the dumbness.\nSpeaker C: That's funny.\nSpeaker C: Yeah.\nNone: I mean, you could block out the person's face.\nSpeaker H: You agree.\nSpeaker H: Of course, where we sit at the table, I find it's very interesting that we do tend to gravitate to the same place each time.\nSpeaker H: And it's somewhat coincidental.\nSpeaker H: I'm sitting here so I can run into the room if the hardware starts catching fire.\nSpeaker H: You'd like to be at charge.\nSpeaker G: I can start off at the head of the table.\nSpeaker G: Take control.\nSpeaker G: Speaking of taking control, you said you hid some research.\nSpeaker H: Yeah, I've been playing with using the close talking mic to try to figure out who's speaking.\nSpeaker H: So my first attempt was just using thresholding and filtering that we talked about about two weeks ago.\nSpeaker H: And so I play with that a little bit.\nSpeaker H: And it works okay, except that it's very sensitive to your choice of your filter width and your threshold.\nSpeaker H: So if you fiddle around with it a little bit and you get good numbers, you can actually do a pretty good job of segmenting when someone's talking and when they're not.\nSpeaker H: But if you try to use the same parameters on another speaker, it doesn't work anymore.\nSpeaker H: Even if you normalize it based on the absolute loudness.\nSpeaker F: It does work for that one's feature output meeting.\nSpeaker H: It does work for the one's feature output meeting.\nSpeaker H: Pretty well.\nSpeaker H: How did you do it then?\nSpeaker H: How did I do it?\nSpeaker I: What do you mean?\nSpeaker H: The algorithm was take every frame that's over the threshold and then median filter it and then look for runs.\nSpeaker H: So there was a minimum run length.\nSpeaker H: Every frame that's over what threshold?\nSpeaker I: A threshold that you pick.\nSpeaker I: In terms of energy?\nSpeaker I: Yeah.\nSpeaker I: Okay.\nSpeaker E: Say that again.\nSpeaker H: Okay.\nSpeaker H: So if you take each frame and you compute the energy and if it's over the threshold you set it to one.\nSpeaker H: And if it's under the threshold you set it to zero.\nSpeaker H: So now you have a bit stream.\nSpeaker H: That's zero's and once.\nSpeaker H: And then I median filtered that using a fairly long filter length.\nSpeaker H: Well, actually I guess it depends on what you mean by long.\nSpeaker H: You know, 10th of a second, source of numbers.\nSpeaker H: And that's to average out pitch, you know, the pitch converse and things like that.\nSpeaker H: And then look for long runs.\nSpeaker H: And that works okay if you filter if you tune the filter parameters, if you tune how long your median filter is and how high you're looking for your thresholds.\nSpeaker I: Did you ever try running the filter before you pick a threshold?\nSpeaker H: No.\nSpeaker H: Certainly could though.\nSpeaker H: But this was just I had the program mostly written already.\nSpeaker H: So it was easy to do.\nSpeaker H: Okay.\nSpeaker H: And then the other thing I did was I took, I took a few years speaker change detector, acoustic change detector and I implemented that with the closed talking mics.\nSpeaker H: And unfortunately that's not working real well and it looks like it's the problem is he does it in two passes.\nSpeaker H: The first pass is to find candidate places to do a break.\nSpeaker H: And he does that using a neural net doing broad phone classification and he has the one of the phone classes is silence.\nSpeaker H: And so the possible breaks are where silence starts and ends.\nSpeaker H: And then he has the second pass which is a modeling a Gaussian mixture model looking for whether it improves or degrades to split at one of those particular places.\nSpeaker H: And what looks like it's happening is that the even on the closed talking mic, the broad phone class classifiers doing a really bad job.\nSpeaker H: What was it trained on?\nSpeaker H: I have no idea. I don't remember. Does anyone do you remember Morgan? Was it broadcast news?\nSpeaker H: I think so, yeah.\nSpeaker H: So I didn't rate my next attempt which I am in the midst of and having quite finished yet was actually using the thresholding as the way of generating the candidates.\nSpeaker H: Because one of the things that definitely happens is if you put the threshold low, you get lots of breaks, all of which are definitely acoustic events.\nSpeaker H: They're definitely someone talking. Like it could be someone who isn't the person here, but the person over there or it could be the person breathing.\nSpeaker H: And then feeding that into the acoustic change detector.\nSpeaker H: And so I think that might work. I haven't gotten very far on that.\nSpeaker H: But all of this is closed talking mic.\nSpeaker H: Just trying to get some ground truth.\nSpeaker B: I think there is a good difference in the same thing.\nSpeaker H: Oh, absolutely. So my attention to this is as an aid for ground truth.\nSpeaker B: I mean that in the Mr. File you can find some with a great different level of energy.\nSpeaker B: I think for a minimum basic energy, more or less like a fair sign and detector.\nSpeaker B: Say it again. When you detect the fairs at the end of the detector of the name and the English, the detector of a war in the isolated war.\nSpeaker B: I'm not a launcher, but you're saying contract.\nSpeaker H: And you use an outside detector.\nSpeaker B: I think it's probably to war well because you have in the Mr. File it's a level of energy and a great difference between the speaker.\nSpeaker B: And probably it's not so easy when you use the PDA because the signal is the energy level in that speech file is more similar between the different speakers.\nSpeaker B: I think it will be in my opinion. It will be more difficult to detect basic energy, the chain.\nSpeaker H: In the PDA. Absolutely. No question. It will be much harder.\nSpeaker B: Another question. When I review the work of Javier, I think the idea of using neural network to get a broad class phonetic from the speaker.\nSpeaker B: If you have a, I'm considering only because Javier only consider like candidate the silence because this is the only model he used to detect the possibility of a chain between the speaker.\nSpeaker B: And another research in different groups working on broadcast news prefer to consider hypothesis between each 40 when the phone changes.\nSpeaker B: Because I think it's more realistic that only consider the silent between speaker. There is a seat silent between speaker is acoustic event important to consider.\nSpeaker B: I found that silent in many occasions in the speech file. But when you have to speak together with enough silent between the end, I think it's better to use the acoustic chain detector.\nSpeaker B: I, I, I, X or bike, the criteria for consider all the friends in my opinion.\nSpeaker G: The reason that he just used silence was not because he thought it was better. It was, it was the place he was starting. So I was trying to get something going. And as, as in your case, if you're here for on the amount of number of months, you try to get the holistic goal.\nSpeaker B: But his, his goal was always to proceed from there to then allow broad category change also. But do you think that if you consider all the friends to apply the, the bike, to the date, the different acoustic chain between speaker without and with silence or with overlapping.\nSpeaker B: I think like a general way of process the acoustic chain in the first step, I mean. And then with that, considering the, you, you can consider the energy like another parameter in the, in the future.\nSpeaker B: Absolutely. This is the idea. And if you do that with a bike, to tell you for example, or with another kind of a, of this time in the first step. And then you, you get the hypothesis to the, this chain acoustic.\nSpeaker B: Process, right. Because probably you, you can find the small gap of silence between speaker with a small duration less than 200 seconds. For example, and apply another, another algorithm, another approach like the texture of an, the texture, the basic energy to, to consider that, that, the sound of small silence between speaker or another algorithm to, to process the, the segment between marks, founded by the, the bike, the unit applied for, for each thing.\nSpeaker B: And it will be, and, and, and more general approach. That if we compare with use a neural net or another, a, a, a, a, a, a, a, a, a, with a broad class or, or not broad class because, in my opinion, it's, in my opinion.\nSpeaker B: If you, if you change the condition of the speech, I mean, if you adjust to your algorithm with the mr speech file and to, to, to, a, that, the neural net, you said by heavier with the mr file, with the, with the mix, mix, mix, sorry. And, and then you, you, you tried to, to apply that, a, a, a, a, a, speaker recognizer to that signal, to the pdA, a, a, a, a, a, a, I think you will have problems with the, a, a, the, a, a, a, you will need to, a, you will need to, to, to, to, retrain.\nSpeaker G: Look, I think this is a, once, I used to work like a voice, not voice silence detection young, this is this kind of thing.\nSpeaker G: If you have somebody who has some experience with this sort of thing and they work on it for a couple of months, they can come up with something that gets most of the cases fairly easily.\nSpeaker G: Then you say, okay, I don't want to just get most of the cases, I want it to be really accurate.\nSpeaker G: Then it gets really hard to measure what you do.\nSpeaker G: So the problem is, is that if you say, well, I have to use other data over here that I learn things from either explicit training of neural nets or Gaussian mixture models or whatever.\nSpeaker G: So, as you don't use any of those things, you say you have looked for acoustic change.\nSpeaker G: What does that mean?\nSpeaker G: That means you set some threshold somewhere or something, right?\nSpeaker G: And so, where do you get your thresholds from?\nSpeaker G: From something that you looked at.\nSpeaker G: So you always have this problem, you're going to new data.\nSpeaker G: How are you going to adapt to whatever you can very quickly learn about the new data if it's going to be different from old data that you have?\nSpeaker G: I think that's a problem.\nSpeaker H: Well, also, what I'm doing right now is not intended to be an acoustic change detector for far-field mics.\nSpeaker H: What I'm doing is trying to use the close talking mic and just use candidates.\nSpeaker H: Can't just get in and just try to get a first pass at some of the smart things.\nSpeaker H: And I haven't spent a lot of time on it.\nSpeaker H: And I'm not intending to spend a lot of time on it.\nSpeaker C: I'm unfortunately out to run, but I can imagine building a model of speaker change detection that takes into account both the far-field and actually not just the close talk mic for that speaker, but actually for all of the speakers.\nSpeaker C: If you model the effect that me speaking has on your microphone and everybody else's microphone as well as on that.\nSpeaker C: And you build basically, I think you would build an HMM that has a state space all of the possible speaker combinations.\nSpeaker C: And you can control it.\nSpeaker H: It's not that big actually.\nSpeaker H: Two to the end.\nSpeaker H: Two to the number of people in the meeting.\nSpeaker G: But actually, maybe just something simpler, but the long lines of what you're saying, I was just realizing, you know, the sky used to build mic mixers, automatic mic mixers where, you know, in order to be able to turn up the gain, you know, as much as you can, you lower the gain on the mics of people who aren't talking, right?\nSpeaker G: And you have some sort of reasonable way of doing that.\nSpeaker G: But what if you were just looking at very simple measures like energy measures, but you don't just compare it to some threshold overall, but you compare it to the energy in the other\nSpeaker H: microphones. I was thinking about doing that originally to find out who's the loudest in that person certainly talking, but I also wanted to find threshold, excuse me, overlap.\nSpeaker H: So not just just loudest.\nSpeaker B: Sorry.\nSpeaker B: I found it when I analyzed the speech file from the mic across microphone.\nSpeaker B: I found sound with a different level of energy, including overlap sound, including because the, depending on the position of the mic, or the feature speaker to get more or less energy in the signal.\nSpeaker B: And then if you consider energy to detect overlapping, and you process the speech file from the reset signals, the mic signals, I think it's difficult only to end with the energy to consider that in that zone we have a overlapping sound if you process only the energy of each frame.\nSpeaker G: Well, it's probably harder, but I think what I was noting just when he raised that was that there's other information to be gained from looking at all of the microphones and you may not need to look at very sophisticated things.\nSpeaker G: Because if most of the overlaps, this doesn't cover C3, but if most of the overlaps say are two, if the distribution looks like there's a couple high ones and an address down below, you know, I mean, there's some information there about the distribution even with very simple measures.\nSpeaker G: By the way, I had an idea that what I was watching Chuck nodding and a lot of these things is that we could all wear little bells in our heads.\nSpeaker H: You know that.\nSpeaker H: Sorry, I'm just going to sleep.\nSpeaker I: Yeah.\nSpeaker I: Actually, I saw a woman at the bus stop the other day who was talking on her cell phone speaking Japanese and was bowing, you know, profusely, just kept.\nSpeaker F: It's very difficult if you try while you're trying to say to convince somebody on the\nSpeaker D: phone, it's difficult not to move your hands.\nSpeaker F: If you watch the whole thing, you do things. I still think we should try a meeting or two with the blindfolds.\nSpeaker F: At least of this meeting that we have lots of recording though.\nSpeaker F: Yeah, I think.\nSpeaker F: Maybe for a part of the meeting, we're going to have to do it the whole evening.\nSpeaker F: I think it's a great idea.\nSpeaker F: It'll be too hard to make barriers that I was thinking because they have to go all the way.\nSpeaker F: I can see Chuck even if they're going to go to bed.\nSpeaker H: I can see that they're just turning out of lights.\nSpeaker E: I can say I made barriers so that the stuff I was doing was calling.\nSpeaker E: Which just used this kind of foam board, really inexpensive.\nSpeaker E: You can mask and tape it together.\nSpeaker E: These are very large.\nSpeaker E: We also have tissues.\nSpeaker F: These are the other things I think.\nSpeaker F: So we need a barrier that doesn't disturb the long range of life.\nSpeaker G: That would be good.\nSpeaker G: Probably we should wait until Adam set up the mics.\nSpeaker F: The interesting of the camera going.\nSpeaker F: I think we're going to have to work on the human subjects.\nSpeaker D: Yeah, that's right.\nSpeaker D: We did that one with people.\nSpeaker E: Do you mind being blindfolded while you're doing it?\nSpeaker G: That's the one that we videotape.\nSpeaker G: So I want to move this along.\nSpeaker G: I did have this other agenda item, which is a list which I sent to a couple folks but I wanted to get broader input on it.\nSpeaker G: So this is the things that I think we did in the last three months.\nSpeaker G: Obviously, not everything did, but sort of highlights that I can tell some outside person.\nSpeaker G: What were you actually working on?\nSpeaker G: In no particular order.\nSpeaker G: One, ten more hours of meetings recorded.\nSpeaker G: Something like that from three months ago.\nSpeaker G: XML formats and other transcription aspects sorted out.\nSpeaker G: Sent to IBM.\nSpeaker G: Pilot data put together.\nSpeaker G: Sent to IBM for transcription.\nSpeaker G: Next batch of recorded data put together on the CD-ROMs for shipment to IBM.\nSpeaker G: Hasn't been sent yet.\nSpeaker G: But it's getting ready.\nSpeaker G: Human subjects approval on campus and release forms worked out so the meeting participants have a chance to request audio pixelization of selected parts of the speech.\nSpeaker G: Audio pixelization software written and tested.\nSpeaker G: Eliminary analysis of overlaps in the pilot data we have transcribed.\nSpeaker G: And exploratory analysis of long distance inferences for topic coherence.\nSpeaker G: I was just sure if those were the right way.\nSpeaker G: That was the right way to describe that.\nSpeaker G: That was the lecture size that you and look under did.\nSpeaker G: What was that called?\nSpeaker G: Well, I'm probably saying this wrong.\nSpeaker G: But what I said was exploratory analysis of long distance inferences for topic coherence.\nSpeaker G: Something like that.\nSpeaker G: So a lot of that was from what you two were doing.\nSpeaker G: So I sent to you and please mail me corrections or suggestions for changing.\nSpeaker G: I don't want to make this twice its length, but just improve it.\nSpeaker H: Is there anything I need to do?\nSpeaker H: I did a bunch of stuff for supporting of digits.\nSpeaker G: A bunch of stuff for, okay, maybe send me a sentence.\nSpeaker G: That's a little bit of a sentence.\nSpeaker G: That doesn't just say a bunch of stuff.\nSpeaker H: Stuff is probably bad too.\nSpeaker H: It's not very technical.\nSpeaker H: I'll try to phrase it in passive voice.\nSpeaker H: Yeah, technical stuff.\nSpeaker G: A range of things.\nSpeaker G: Yeah.\nSpeaker G: And I sort of threw in what you did with what Jane did under the preliminary analysis for relapse.\nSpeaker G: Do you all can you tell us about all the work you've done on this slide?\nSpeaker G: Last three months.\nSpeaker G: It's really too complicated.\nSpeaker A: I didn't get what is audio pixelization.\nSpeaker G: Audio pixelization.\nSpeaker G: He did it.\nSpeaker H: So why don't you explain it quickly?\nSpeaker H: It's just beeping out parts that you don't want to include it in the meeting.\nSpeaker H: So you can say things like, well, they should probably not be on the record.\nSpeaker G: But we spent a fair amount of time early on just talking dealing with this issue about, we realized what people are speaking in an impromptu way.\nSpeaker G: And they might say something that would embarrass them or others later.\nSpeaker G: And how did you get around that?\nSpeaker G: So in the consent form it says, well, we'll look at the transcripts later and if there's something that you're unhappy with.\nSpeaker G: But you don't want to just totally excise it because, well, you have to be careful about excising how you excise it, keeping the timing right and so forth.\nSpeaker G: So the moment the idea we're running with is putting the beep.\nSpeaker H: You can either beep or it can be silence.\nSpeaker H: I couldn't decide which was the right way to do it.\nSpeaker H: It's good auditorily. If someone is listening to it, there's no mistake that it's been beeped out.\nSpeaker H: But for software, it's probably better for it to be silence.\nSpeaker I: You could make a model of that beep.\nSpeaker H: I like that.\nSpeaker H: It's an A low-poss.\nSpeaker I: It's more obvious that there was something there than it's just silence.\nSpeaker I: I mean, he's removing the old thing.\nSpeaker I: Right.\nSpeaker I: I mean, if you just replace it with silence, it's not clear whether that's really silence.\nSpeaker E: What question do you do it on all channels?\nSpeaker E: Of course.\nSpeaker H: Interesting. I like that.\nSpeaker H: You have to do it on all channels because it's audible.\nSpeaker H: It's potentially audible. You could potentially recover it.\nSpeaker H: You could keep it back door.\nSpeaker H: I haven't thrown away any of the meetings that I beeped.\nSpeaker H: Actually, yours is the only one that I beeped and then the ARP, the DARPA meeting.\nSpeaker H: Sorry.\nSpeaker H: It's the DARPA meeting I just excised completely.\nSpeaker F: It's some people who only have beeped their speed.\nSpeaker G: Very easy to find.\nSpeaker G: All right.\nSpeaker G: I think we should go on to the digits.\nSpeaker E: I have one concept.\nSpeaker E: I want to say, which is that I think it's nice that you're preserving the time relations.\nSpeaker E: So you're not just cutting, you're not doing scissors.\nSpeaker E: Right.\nSpeaker E: You're keeping the time duration that I've been in.\nSpeaker E: It's a deleted part.\nSpeaker H: Okay.\nSpeaker F: I think the other thing is that I'm saying something unbelievable.\nSpeaker F: You'll lose it.\nSpeaker H: There's no way around that.\nSpeaker G: Before we do the digit I did also want to remind people.\nSpeaker G: Please do send me thoughts for an agenda.\nSpeaker G: That would be good.\nSpeaker G: So that it's kind of stuck on me this week.\nSpeaker G: Yeah, well the screen.\nSpeaker F: Okay, and I wanted to say I think this is really interesting.\nSpeaker F: It's cool stuff.\nSpeaker F: I was going to say can you do that for the other meetings?\nSpeaker H: Can you do it for them?\nSpeaker I: No, actually I thought that's what you were given.\nSpeaker I: So it was another meeting.\nSpeaker I: I was like, oh cool.\nSpeaker E: How long did it take just briefly?\nSpeaker E: I have the script now.\nSpeaker E: So I mean it can work off the label.\nSpeaker E: Okay, so it has to be handled label first.\nSpeaker E: Well, yeah, because, well, I mean, once his is algorithm is that it works well enough.\nSpeaker E: Then we could do that way.\nSpeaker E: Okay, but it's really worked out with my things.\nSpeaker E: I'm not quite to the point where it was.\nSpeaker E: What this is caused me.\nSpeaker E: So this discussion caused me to want to subdivide these further.\nSpeaker E: I'm going to take a look at the back channel.\nSpeaker E: So much we have the panel.\nSpeaker E: I hope to have that.\nSpeaker H: And my algorithm worked great actually on these.\nSpeaker H: But when you wear it like that or with the lip hell, or if you have it very far from your face, that's when it starts failing.\nSpeaker F: Well, I can wear it.\nSpeaker F: It doesn't matter.\nSpeaker H: I mean, we wanted to wear it, right?\nSpeaker H: It's too late.\nSpeaker H: I don't want to change the way we do to be in it.\nSpeaker H: It's just a comment on the software, not a comment on the prescriptions on how you wear microphones.\nSpeaker H: Okay, let's get the bolts.\nSpeaker H: Let's do digits.\nSpeaker H: Transcript 16111630.\nSpeaker H: 5306185750.\nSpeaker H: Strike that.\nSpeaker H: 75108.\nSpeaker H: 8770980091392133442890597809869023857805094650207117529266007234.\nSpeaker I: Transcript 1851-187046055987681478054064016007000400400416004651222716634.\nSpeaker I: 5074-725448331-93421-00991241.\nSpeaker F: Transcript 18311850.\nSpeaker F: 3608456-83792-948150880301905625353527096520781298893623390060399024, I'm sorry, 2244340.\nSpeaker G: Transcript 1791-181012005843051665540890035840143692243724756.\nSpeaker G: 7056-911-0578703811729102.\nSpeaker A: Transcript 1811-1830230654066547067737890089529709649101032445606864706.\nSpeaker B: Transcript 1771-1790-005782021348358645205550678041060780410607706.\nSpeaker B: 41039-0230-C492-01543279340540403271345308258146955708909.\nSpeaker E: Transcript 1751-1770-011492887897354500604018120276.\nSpeaker E: 812-0276-923-1518-030-205-0123-0303184-524-64522-851-8900.\nSpeaker E: The only one that wasn't red is known, so we don't do it properly.\n"}, {"meeting_id": "TS3003a", "summary": "The whole meeting was the initial meeting of a new remote control project. Firstly, Project Manager introduced the scope and agenda of the project, and the team got acquainted with each other and technical devices. Then Project Manager made clear that remote control would be priced at 25 Euros and a production cost of 12.5 Euros, in order to achieve a profit aim of 50 million Euros. After an accident, Marketing suggested that remote control should be made multipurpose and the consensus was reached on this point. Finally, the group brainstormed some other functions for the remote control despite a potential increase in production cost.", "dialogue": "Speaker A: Good morning.\nSpeaker A: Good morning.\nSpeaker A: I see you all find your places.\nSpeaker A: Everybody sitting on the right place.\nSpeaker A: I guess so.\nSpeaker A: Let's see.\nSpeaker A: First I'll introduce myself.\nSpeaker A: I don't know if everybody knows me.\nSpeaker A: I'm Bart.\nSpeaker A: Hello.\nSpeaker A: Hello.\nSpeaker A: Bart.\nSpeaker A: Hello.\nSpeaker A: Hey, Bart.\nSpeaker A: Welcome.\nSpeaker A: Thank you.\nSpeaker A: Let's see.\nSpeaker A: Let's start off with a little presentation.\nSpeaker A: First I'll tell you a little bit about the setting.\nSpeaker A: You can see a few cameras here.\nSpeaker A: They'll record our actions.\nSpeaker A: And you'll have wires and microphones.\nSpeaker A: They will record your voice.\nSpeaker A: There are also some microphones there.\nSpeaker A: But you don't have to pay a lot of attention on those because it will disappear when you don't intend to do it.\nSpeaker A: So this is our project documents folder.\nSpeaker A: There are some notes in it already.\nSpeaker A: I see some documents.\nSpeaker A: I'll start with the presentation kick off.\nSpeaker A: I'm being modified by the administrator.\nSpeaker A: Let's do it redonely.\nSpeaker A: Well, I don't know if you've noticed, but we're working for real reaction.\nSpeaker A: It's a company in electronics.\nSpeaker A: We put fashion in electronics.\nSpeaker A: We make it work.\nSpeaker A: We put a lot of effort in design and in the product itself.\nSpeaker A: We put it in the project manager.\nSpeaker A: So I'll direct you through the project.\nSpeaker A: This is our agenda.\nSpeaker A: We have our opening of Queensland's tool training project plan discussion closing.\nSpeaker A: Maybe I can sit down.\nSpeaker A: Then I can take some notes.\nSpeaker A: Let's see.\nSpeaker A: Maybe you can take the minutes once in a while.\nSpeaker A: I don't know if it's not a lot of work.\nSpeaker A: If you hear something, you can write it down.\nSpeaker A: You can see it's the opening of Queensland's tools.\nSpeaker A: Queensland's is the point we've done a bit.\nSpeaker A: Have you all seen the corporate website already?\nSpeaker A: Have you seen any flaws in it?\nSpeaker A: I think I found one.\nSpeaker B: I paid much attention to it.\nSpeaker A: I see if it works.\nSpeaker A: No problem.\nSpeaker A: On the corporate information side, there was real remote instead of real reaction.\nSpeaker A: Oh, yeah.\nSpeaker A: Real remote is not really the company we are, but it's just a little.\nSpeaker A: Okay.\nSpeaker A: What are we going to do?\nSpeaker A: Our project aim is, as you can see, a new remote control.\nSpeaker A: It has to be original, trendy, and user-friendly.\nSpeaker A: So these are the points we also hired you.\nSpeaker A: We've got the marketing expert for the trendy and user-friendly look.\nSpeaker A: And in this real designer, there's also user-friendly and a bit original.\nSpeaker A: And we've got our user interface designer.\nSpeaker A: That's about the new remote control.\nSpeaker A: Our project method is, there are three phases we're going through.\nSpeaker A: The first is functional design, individual work meetings.\nSpeaker A: After the functional design, then the conceptual design and the detailed design, I had some role indications on here.\nSpeaker A: But I think you know it already by yourself.\nSpeaker A: The industrial designer is going to work on the working design, the components design, and a bit of the look and feel design.\nSpeaker A: The user interface designer is going to do the technical function design, user interface concept, and user interface design.\nSpeaker A: And the marketing expert is doing a little bit of user requirements, specification, trend watching, and product evaluation.\nSpeaker A: So that's a bit what you're going to do.\nSpeaker A: That will be all worked out in other meetings.\nSpeaker A: Then we've got our first tool training.\nSpeaker A: We're going to work with a lot of high-tech tools here.\nSpeaker A: So it's handy if we have a little bit of training first.\nSpeaker A: As you can see, we've got the smart boards here and here, and the white boards.\nSpeaker A: In the whiteboard here, there's a little toolbar on this side.\nSpeaker A: Here are some functions.\nSpeaker A: You can save these functions.\nSpeaker A: We don't have anything to do with.\nSpeaker A: Only undo.\nSpeaker A: You can undo a little piece of drawing, a blank new document for each person, select a pen, razor, capture redone, they have to do anything with.\nSpeaker A: Then we've got our pen, this pen.\nSpeaker A: It's really funny because you can draw with it on this page.\nSpeaker A: In the thing, it is formal.\nSpeaker A: You can also select the current color and the line width.\nSpeaker A: With that first, you have to select the pen function.\nSpeaker A: But we're going to work with it in a minute.\nSpeaker A: That's very simple and it's easy to draw your findings and drawings in there.\nSpeaker A: Then a short thing about documents.\nSpeaker A: We've got our shared folder, project document, I think.\nSpeaker A: But all of you have all found it already because there are a lot of documents in it already, and these are available on the smartboards as well.\nSpeaker A: If you have a document you want to show, just open it from the folder.\nSpeaker A: Here is the simple toolbar.\nSpeaker A: It's what I just said, it's safe, print, move back or forward one page.\nSpeaker A: You can switch between the different drawings.\nSpeaker A: Then we're going to try out the whiteboard.\nSpeaker A: As you can see, we're going to draw an animal.\nSpeaker A: Just to get a bit familiar with it.\nSpeaker A: I'm not supposed to run away.\nSpeaker A: Is anybody playing with the mouse?\nSpeaker A: We're going to draw an animal.\nSpeaker A: Just sum up a few of its favorite characteristics.\nSpeaker A: The only thing we have to look after is that we use different colors and different line with this.\nSpeaker A: I can start from there.\nSpeaker A: You can use this pen by holding it like a little child because if you hold it like this, the sensors will get blocked and the drawing won't get good.\nSpeaker A: Another thing is you have to be a bit slow because if you're going to draw it really fast, then the pen won't hold up.\nSpeaker A: We choose four month.\nSpeaker A: Current color, I think gray is appropriate.\nSpeaker A: Then the line width, I think seven will be nice.\nSpeaker A: Now you'll see my drawing capabilities.\nSpeaker A: It means another very much.\nSpeaker A: You have to do it real slow.\nSpeaker A: I'm trying to draw a dolphin.\nSpeaker A: I'm thinking about a short fish.\nSpeaker A: We've got another function.\nSpeaker A: You can do this easily.\nSpeaker A: I've got to write down a few of its characteristics.\nSpeaker A: You've got no text tool?\nSpeaker A: This is typically a new action.\nSpeaker A: Maybe I have to hold it a bit upside down.\nSpeaker A: I think that's it because if you want to do it like this, it will be a stripe.\nSpeaker A: I don't know.\nSpeaker A: This is not my work.\nSpeaker A: I think it wants to draw another animal.\nSpeaker A: It lives for the fin.\nSpeaker A: It's my characteristic about the dolphin.\nSpeaker A: Now I'm going to hand over the pen on a new blank sheet to you.\nSpeaker B: I'm going to draw a black.\nSpeaker B: I'm not an artist, but here we go.\nSpeaker A: It's easier to draw with a smaller line.\nSpeaker A: I did.\nSpeaker B: I'm not an artist.\nSpeaker B: Okay, this is my, um, cheap.\nSpeaker A: It's nice with, of course.\nSpeaker B: Blue dot they always get sprayed on their butts.\nSpeaker A: It's a real dead sheep.\nSpeaker A: Yeah, recognition.\nSpeaker A: Maybe you can also write your name somewhere.\nSpeaker B: They are.\nNone: Come on.\nSpeaker B: Let's go really slow when you're writing.\nSpeaker A: Yeah.\nSpeaker B: They're brilliant animals.\nSpeaker B: And that's just a little me thing.\nSpeaker B: So, I guess I'll pass the pen to her.\nSpeaker B: Okay, let's use her interface.\nSpeaker B: I'm just going to add her to her head.\nSpeaker B: Let's see.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker A: Sweet.\nSpeaker B: You know what other it is?\nSpeaker B: Garfule.\nSpeaker A: Yeah.\nSpeaker B: So, I'll just try something else.\nSpeaker B: Something different than Garfule.\nSpeaker B: I'm a little bit more skinny.\nSpeaker A: This is your kid.\nSpeaker A: Did you find him on the street?\nSpeaker B: Well, it's supposed to be a cat.\nSpeaker B: I like cats because they are, uh, independent.\nSpeaker B: So, okay.\nSpeaker A: That's pretty clear.\nSpeaker A: So, everybody knows how to work with the whiteboard now.\nSpeaker A: So, if you have any ideas or if you want to draw anything on the whiteboard, you can go ahead.\nSpeaker A: We're being home.\nSpeaker A: So, we've got the tool introduction.\nSpeaker A: Move along to the project finance.\nSpeaker A: As you can see, we, for a remote control, selling price is 25 euros, our selling price.\nSpeaker A: Our profit aim is 50 million euros.\nSpeaker A: That's the least we have to get from our remote controls.\nSpeaker A: So, we have to work together to reach our aims.\nSpeaker A: We can do it international.\nSpeaker A: So, we have to focus on different kind of users, different kind of cultures and different kind of trends as well.\nSpeaker A: But that's all in the later stadium.\nSpeaker A: Our production cost, um, can be maximal 12 and a half euros.\nSpeaker A: So, that's also a point we have to keep in mind that we won't make remote controls with small televisions inside and stuff like that.\nSpeaker A: So, we're going to do a little bit of work.\nSpeaker A: So, just try to remember these points.\nSpeaker A: Selling price is 25 profit aims, 50 million, but more important is the international market we're trying to focus on.\nSpeaker A: And production costs maximal 12 and a half euros.\nSpeaker A: So, that leads us to our little discussion.\nSpeaker A: So, I'm going to sit down and think.\nSpeaker A: It's easier.\nSpeaker A: I've got a message.\nSpeaker A: Five minutes.\nSpeaker A: Okay, that's a good timing.\nSpeaker B: So, just on the side now, why is it my laptop is only giving me a black screen?\nSpeaker A: Maybe you have to say the magic word.\nSpeaker A: Yeah.\nSpeaker A: Right.\nSpeaker A: Does it do anything?\nSpeaker A: No.\nSpeaker A: Maybe you have to just clap it down.\nSpeaker A: Back up again.\nSpeaker B: It's off.\nSpeaker A: It's off.\nSpeaker A: I have to put it back.\nSpeaker A: Yeah.\nSpeaker B: It'll be okay.\nSpeaker B: Well, it was on.\nSpeaker A: Yeah.\nSpeaker A: It's those laptops.\nSpeaker B: Oh, there we are.\nSpeaker A: Nice.\nSpeaker A: Okay.\nSpeaker A: But does everybody else experience this with remote controls?\nSpeaker A: I mean, not the ordinary remote controls, but also a little bit different ones.\nSpeaker A: Like you can use for other.\nSpeaker A: No, really.\nSpeaker B: Yeah.\nSpeaker B: Well, we're not here.\nSpeaker B: Kind of brought TV at home and a DVD player.\nSpeaker B: So, we've got a lot of remote controls.\nSpeaker B: One for the TV, one for the video recorder, one for the DVD player.\nSpeaker B: Yeah.\nSpeaker B: And I think it would be best to just make one remote control that can operate the mall.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: I've got one at home.\nSpeaker A: You can program, I think, eight different devices in it.\nSpeaker A: And you can use it for your television.\nSpeaker A: Okay.\nSpeaker A: Anything else.\nSpeaker A: And it also operates on infrared.\nSpeaker A: So, you've got a little device inside your room.\nSpeaker A: And then you can operate it from the third or the second floor.\nSpeaker A: Oh, really?\nSpeaker A: So, that's pretty handy when you have a video recorder or DVD player downstairs.\nSpeaker A: And you've got a link to your TV on the second floor.\nSpeaker A: Yeah.\nSpeaker A: So, that's a pretty handy thing.\nSpeaker A: But only the, I think, if you can put different kind of devices in one remote control, it makes it a lot easier as well.\nSpeaker A: It's good to remember.\nSpeaker B: So, I think you can take minutes again.\nSpeaker A: Yeah.\nSpeaker A: It's nice.\nSpeaker A: Thank you.\nSpeaker A: So, we want different functions.\nSpeaker A: Yeah.\nSpeaker A: And we can maybe see if we can do something with infrared.\nSpeaker A: But I don't know if that will exceed the production costs.\nSpeaker A: So, that's something we have to find out, I think.\nSpeaker B: Yeah.\nSpeaker B: But that would be really good if we could do that.\nSpeaker A: And other functions for remote control.\nSpeaker A: Maybe we can make it, I think it has to be shockproof.\nSpeaker B: Sure.\nSpeaker B: It has shockproof.\nSpeaker B: My remote control tends to fall a lot.\nSpeaker B: So, I would, I mean, yeah.\nSpeaker A: What a proof.\nSpeaker A: Okay.\nSpeaker A: So, a few things we can think of.\nSpeaker A: I will put the minutes from this meeting in our project folder.\nSpeaker B: Yeah.\nSpeaker B: One other little thing.\nSpeaker B: I thought it might be handy to put a battery status display on it.\nSpeaker B: So, you can see that it stays.\nSpeaker B: Okay.\nSpeaker B: How much is left in the battery?\nSpeaker B: But they'll also really drag up the production costs.\nSpeaker B: Yeah.\nSpeaker B: But maybe just a little led.\nSpeaker A: That's an idea as well.\nSpeaker A: Other ideas?\nSpeaker A: Quick ideas.\nSpeaker B: Nope.\nSpeaker B: They were all mentioned.\nSpeaker A: Okay.\nSpeaker A: Any questions about this presentation?\nSpeaker A: Kick off presentation.\nSpeaker B: Nope.\nSpeaker B: I don't think so.\nSpeaker A: Okay.\nSpeaker A: Then I'll put the minutes from this meeting in the project folder.\nSpeaker A: And then we can all work.\nSpeaker A: Finish meeting now.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: We can all work on our own projects.\nSpeaker A: Okay.\nSpeaker A: Then I'll meet you in about half an hour, I think.\nSpeaker A: Okay.\nSpeaker A: So, good luck.\nSpeaker A: Yep.\nSpeaker A: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "Bed005", "summary": "The meeting discussed various topics such as data collection and updates on the German parser. They discussed XML modifications and future thoughts on the ontology. Additionally, object representations will include an EVA vector. This can be incorporated in the database entry for a particular building or inherited from the ontology of the building type. The actual number of the inputs can create a combinatorial explosion when setting the probabilities. In any case, further to fulfilling the basic requirements (translating the parser and the generator into english), the project is entirely open-ended in terms of focus of research.", "dialogue": "None: I think I got my mic on.\nNone: Okay.\nSpeaker E: See, 38, 871, 4627, 692, 640, 979, 271, 4627, 692, 640, 979, 2116, 462, 3096, 861, 6489, 36408, 569, 880, 0742, 824491577, 1197, 487, 93, 4240.\nSpeaker F: Okay, I may do yours then.\nSpeaker F: Is she a son?\nSpeaker F: Yeah.\nSpeaker E: Yeah, it's a little thing on top.\nSpeaker E: It doesn't, it should be the other way.\nSpeaker B: 156, 741, 564, 051, 453, 9205, 1, 3, 3, 8, 0, 250, 8, 0369, 289, 671, 19, 819, 8508, 98455, 8307, 7884817, 0839, 8383.\nSpeaker F: Okay, so we all switched on?\nSpeaker E: We all switched on, yeah.\nSpeaker F: It's all right.\nSpeaker F: So before we get started with the technical part, I just want to review what I think is happening with the R-Data collection.\nSpeaker F: So probably after today, that shouldn't come up in this meeting.\nSpeaker F: This should be, there's another thing going on of gathering data, and that's pretty much independent of this.\nSpeaker F: But I just want to make sure we're all together on this.\nSpeaker F: What we think is going to happen is that, in parallel, starting about now, we're going to get say to, were you working with me in Robert Draft, a note that we're going to send out to various cogs like another class is saying, here's an opportunity to be a subject, contact, say, and then there'll be a certain number of hours during the week, which she will be available, and we'll bring in people, roughly how many Robert?\nSpeaker F: Do we know?\nSpeaker D: 50 was sort of a first.\nSpeaker F: So we're looking for a total of 50 people, not necessarily by any means, all students, but we'll start with that.\nSpeaker F: In parallel with that, we're going to need to actually do the script.\nSpeaker F: And so, I guess there's a plan to have a meeting Friday afternoon, with Jane, and maybe Liz, and whoever, on actually getting the script worked out.\nSpeaker F: But what I'd like to do if it's okay, is to, I say start recruiting in parallel and possibly start running subjects next week.\nSpeaker F: Week after that spring break, and maybe we'll look for some subjects next door,\nSpeaker D: or... Also, Fay will not be here during spring break,\nSpeaker F: but we won't do it. Okay, so that's easy.\nSpeaker F: So, is that...\nSpeaker F: makes sense to everybody?\nSpeaker D: Yeah.\nSpeaker D: Also, both Fay and I will do something of which I may kindly ask you to do the same thing, which is we're going to check out our social infrastructures for possible subjects, meaning, kid children's, gymnastics classes, preschool parents, and so forth.\nSpeaker D: They also sometimes have flexible schedules.\nSpeaker D: So if you happen to be sort of in a non-student social setting, and you know people who may be interested in being subjects, we also consider using the Berkeley High School and their teachers, maybe, and get them interested in...\nSpeaker D: That's a good idea.\nSpeaker D:...stuff.\nSpeaker D: And so, as far as our brainstorming was concerned.\nSpeaker F: Oh, yeah, the high school is a great idea.\nSpeaker D: So...\nSpeaker D: But I will just make a first draft of the note, the write-up notes, and it'll you and Fay, and then...\nSpeaker F: And one else, it'll copy Jane on it.\nSpeaker D: And are we, have we conquered that these forms are sufficient for us, and necessary?\nSpeaker D: I think they're necessary.\nSpeaker F: This is the permission form.\nSpeaker F: There has to be one, and I think we're just going to use it as it is.\nSpeaker F: And...\nSpeaker F: Yeah, everybody with that.\nSpeaker F: Yeah.\nSpeaker F: There's one tricky part about...\nSpeaker F: They have the right...\nSpeaker F: The last paragraph, if you agree to participate, do other opportunities to have anything excised, which you would prefer not to have included in the data set.\nSpeaker F: Okay?\nSpeaker F: Now, that we had to be included for this other one, which might have meetings, you know, about something.\nSpeaker F: In this case, it doesn't really make sense.\nSpeaker F: So, what I'd like to do is also have our subject sign a waiver, saying, I don't want to see the final transcript.\nSpeaker F: And if they don't, if they say, no, I'm not willing to sign that, then we'll show them the final transcript.\nSpeaker F: But...\nSpeaker F: That makes sense.\nSpeaker F: That...\nSpeaker F: So, we might actually...\nSpeaker F: Jane may say that, you know, you can't do this on the same form.\nSpeaker F: We need a separate form.\nSpeaker F: But anyway, I'd like to add a little thing.\nSpeaker F: I think for them to initial saying that, I don't want to see the final transcript.\nSpeaker F: But other than that, that one's been approved.\nSpeaker F: And this really is the same project.\nSpeaker F: You know, and so forth.\nSpeaker F: I think we just go ahead.\nSpeaker D: Okay.\nSpeaker D: So much for the data.\nSpeaker D: Except that with Munich, everything is fine now.\nSpeaker D: I'm going to transcribe.\nSpeaker D: They also got a translate, the German data from the GV in cinema stuff for Andreas, so they're awesome to be happy now.\nSpeaker D: With that.\nSpeaker D: So, we move on to the technical side.\nSpeaker D: Well, I guess the good news of last week was the parser.\nSpeaker D: So, Baskar and I started working on the parser.\nSpeaker D: Then Baskar went to class and once he came back, it was finished.\nSpeaker D: So, I didn't measure it, but it was about an hour and ten minutes.\nSpeaker D: Yep.\nSpeaker D: And now we have a complete English parser that does everything determined parser does.\nSpeaker F: Which is not a lot.\nSpeaker D: That's not a lot.\nSpeaker D: Right.\nSpeaker D: And...\nSpeaker A: What did you end up having to do?\nSpeaker A: I mean, was there anything...\nSpeaker D: Well, we can show it then.\nSpeaker D: Well, the first thing we did is we tried to do change the the lauf into run or running or runs.\nSpeaker D: And we noticed that whatever we tried to do, it had no effect.\nSpeaker D: And we were puzzled.\nSpeaker D: And the reason was that the parser completely ignores the verb.\nSpeaker D: So, this sentence...\nSpeaker D: Just the parser property.\nSpeaker D: Sentence is parses, the same output, even if you leave out all of this, so it's basically feature film and TV.\nSpeaker D: That's what you need.\nSpeaker D: And the time.\nSpeaker D: If you add today and evening, it'll add time or not.\nSpeaker D: So, it does look at that.\nSpeaker D: But all the rest is simply frosting on the cake and it's optional for that parser.\nSpeaker F: So, are you going to show us the little templates?\nSpeaker D: Yeah.\nSpeaker D: I can show you the templates.\nSpeaker D: I also have it running here.\nSpeaker D: So, if I do this now, you can see that it parsed the wonderful English sentence which films are on the cinema today, evening.\nSpeaker D: Don't worry about it.\nSpeaker D: It could be this evening which films are on the cinema or running in the cinema.\nSpeaker D: Which today evening is anything happening in the cinema this evening?\nSpeaker A: Okay, keywords, basically.\nSpeaker D: Will elaborate or more or less...\nSpeaker F: Actually, it's a little tricky in that there are some allowable German orders which aren't allowable English orders and so forth.\nSpeaker F: And it is order-based.\nSpeaker F: So, isn't it?\nSpeaker F: No.\nSpeaker F: Oh, so it's actually a set, not a sequence?\nSpeaker D: Yeah, we were...\nSpeaker D: I was afraid that...\nSpeaker D: It really is keyword.\nSpeaker A: Really is, basically.\nSpeaker F: Oh, wow.\nSpeaker D: I mean, these sentences are just silly.\nSpeaker D: I mean, these were not the ones we actually did it.\nSpeaker D: What's an idiomatic way of phrasing this?\nSpeaker D: Which films are showing?\nSpeaker D: Playing?\nSpeaker D: Playing?\nSpeaker A: Tonight.\nSpeaker C: I changed this file actually, but it's on my card.\nSpeaker C: Do you want to get it?\nSpeaker D: Or is it easy to get it?\nSpeaker D: I have no net here.\nNone: Okay.\nSpeaker D: So, wonderful parts, same thing.\nSpeaker D: Except that we don't have this time information here now, which is...\nSpeaker D: Oh, the reserve.\nSpeaker D: Anyways.\nSpeaker D: So, these are sort of the ten different sentence types that the parts that was able to do, and still is now in English.\nSpeaker D: And...\nSpeaker D: Sorry.\nSpeaker D: And you have already started to make it a little bit more elaborate, right?\nSpeaker C: Yeah, I mean, I changed those sentences to make it more idiomatic.\nSpeaker C: And of course, you can have it.\nSpeaker C: Many variations of those sentences, they will still be fine.\nSpeaker C: So, in a sense, it's pretty broad.\nSpeaker D: Okay.\nSpeaker D: So, if you want to look at the templates, they're conveniently located in a file template.\nSpeaker D: And this is what I had to do.\nSpeaker D: I had to change...\nSpeaker D: Spiels Film to Film, Film to Movie, Kino to Cinema.\nSpeaker D: Today, or today, evening, evening.\nSpeaker D: Capitalized as well.\nSpeaker C: And...\nSpeaker C: One thing I was wondering was, those functions,\nSpeaker D: are those things that modify the M3L, basically? Yeah.\nSpeaker D: Okay.\nSpeaker D: And that's the next step, but we'll get to that in a second.\nSpeaker D: And so, this means...\nSpeaker D: This and C are not optional.\nSpeaker D: What I like is all maybe in there, but may also not be in there.\nSpeaker F: So, the point is, if it says this and C, in the other order, it says 2 keywords.\nSpeaker F: This is the one I want to see.\nSpeaker F: Or whatever.\nSpeaker D: Okay.\nSpeaker D: Action Watch.\nSpeaker D: Whatever.\nSpeaker D: Nothing was specified.\nSpeaker D: Except that it has some references to audio-visual media here, where it gets set from...\nSpeaker D: It's correct, but I don't know where it gets the phone.\nSpeaker D: Oh, C.\nSpeaker D: Yeah.\nSpeaker D: Oh, it's order.\nSpeaker D: Yeah.\nSpeaker D: And see this...\nSpeaker D: It's exactly the same thing.\nSpeaker D: Okay, so it is set-based.\nSpeaker C: All right.\nSpeaker C: One thing I was wondering was, those percentage signs, right?\nSpeaker C: So, I mean, why do you have them?\nSpeaker C: Because if you didn't have them...\nSpeaker D: I tell you why, because it gives you a score.\nSpeaker D: And the value of the score is, I assume, I guess, the more of these optional things are actually in there, the higher the score...\nSpeaker D: That's the arrangement purpose, all right.\nSpeaker D: So, we shouldn't be a little bit too much.\nNone: It's doing something.\nNone: Somethings.\nSpeaker D: And it's very flexible.\nSpeaker D: I've just tried to be nice.\nSpeaker D: I know, fine.\nSpeaker D: Yeah, flexible.\nSpeaker D: Okay.\nSpeaker D: Let's hope that the generation will not be more difficult,\nSpeaker F: even though the generator is a little bit more complex.\nSpeaker D: And that's the way it is. And the next thing I would like to be able to do, and it seems like this should not be too difficult either, is to say, okay, let's now pretend we actually wanted to...\nSpeaker D: Not only change the mapping of words to the M3, but we're not going to change the map.\nSpeaker D: And we're going to change the map.\nSpeaker D: And we're going to change the map.\nSpeaker D: And we're going to change the map.\nSpeaker D: And the mapping of words to the M3, but we also wanted to change, add a new sentence type\nSpeaker F: and make up some new M3. A big, great.\nSpeaker F: A big, good exercise to just see whether one can get that run.\nSpeaker F: Yeah.\nSpeaker D: And...\nSpeaker D: That shouldn't be too tough.\nSpeaker C: Yeah, so where are those functions actually in good buy and so on, right?\nSpeaker C: Are they actually...\nSpeaker C: Are they going to be called?\nSpeaker C: Are they present in the code for the parser?\nSpeaker D: Yeah, I think what it does...\nSpeaker D: It does something sort of fancy it loads.\nSpeaker D: It has these style sheets and also the...\nSpeaker D: Schimata.\nSpeaker D: So what it probably does is it takes the...\nSpeaker D: This is where it is.\nSpeaker D: This is already the XML stuff.\nSpeaker D: This is where it takes its own syntax.\nSpeaker D: And converts it somehow.\nSpeaker D: Where is the...\nSpeaker D: We didn't for.\nSpeaker D: Where it actually produces the XML out of the parsed stuff.\nSpeaker D: Now this...\nSpeaker D: I can't find it now.\nSpeaker D: I mean, where the action...\nSpeaker D: How the action could buy maps into something...\nSpeaker E: Yeah, where are those constructors defining?\nSpeaker D: No.\nSpeaker D: Yeah, this is sort of what happens.\nSpeaker D: This is what you would need to change to get the XML changed.\nSpeaker D: So when it encounters and contrasts day, it will...\nSpeaker D: Activate those classes in the XML stuff.\nSpeaker D: But I saw these actions.\nSpeaker D: They're good buy steps and where.\nSpeaker E: Can you grab for it?\nSpeaker D: Yeah, let's do that.\nSpeaker A: M3L.TTD.\nSpeaker C: Yeah.\nSpeaker C: That should start specification for the XML.\nSpeaker E: Yeah.\nSpeaker D: Well, we'll find that out.\nSpeaker D: So whatever this does, I mean, this is basically looks to me like a function call, right?\nSpeaker D: Oh, yeah.\nSpeaker D: So whenever it encounters good buy, which we can make it do in a second here?\nSpeaker C: That function auto might be generates an initialized XML structure.\nSpeaker C: I think these are those functions act in the current XML structure and change it in some way.\nSpeaker F: For example, by adding a field to it or something.\nSpeaker F: Yeah, they also seem to affect state because some of them...\nSpeaker F:...stake variable somewhere like the discourse status confirm.\nSpeaker F: Yeah.\nSpeaker F: Okay, so that's going to be a call on the discourse and confirm that it's...\nSpeaker C: Oh, you mean that's not going to actually modify the tree, but it's...\nSpeaker F: I think that's right.\nSpeaker F: I think it's actually...\nSpeaker F: That looks like it's state modification.\nSpeaker D: Well, there is a feature called discourse status.\nSpeaker D: Yeah.\nSpeaker D: And so whenever I just say write, it will put this in here.\nSpeaker F: Oh, so it always just...\nSpeaker F: So let's go back then because maybe that all those things...\nSpeaker F:...while they look like function calls are just a way of adding exactly that to the XML.\nSpeaker F: Yeah.\nSpeaker F: Ah-ha.\nSpeaker F: So this...\nSpeaker D: I'm not sure.\nSpeaker D: I'm not sure that...\nSpeaker D: Well, we'll see.\nSpeaker D: When we say let's test something good buy.\nSpeaker D: It causes it to create basically an action, good buy and action.\nSpeaker D: Which is a means of telling the system to shut down.\nSpeaker D: Right.\nSpeaker D: So if we know that write produces a feature discourse status, confirm discourse status.\nSpeaker D: So if I now say write, good buy, it should do that.\nSpeaker D: It creates this, confirm, good buy.\nSpeaker C: Yeah.\nSpeaker C: But there is some kind of function called because how does it know to put good buy and content, but...\nSpeaker C:...confirming features?\nSpeaker C: Oh, it did.\nSpeaker C: That's because...\nSpeaker C: So it's not just that it's adding that field.\nSpeaker F: Right.\nSpeaker F: It's a good point.\nSpeaker F: It's under what subtype you're doing it.\nSpeaker E: It's mystery functions.\nSpeaker D: Well, sometimes...\nSpeaker D: It's a bit of a bit of a presentation.\nSpeaker F: Sometimes.\nSpeaker F: Yeah.\nSpeaker F: So that's funny.\nSpeaker F: You bury this state in the function.\nSpeaker E: All right.\nSpeaker E: So you just automatically initialize this thing, stir comment, right?\nSpeaker E: So it's just a story.\nSpeaker D: For example, oh, this is German, sorry.\nSpeaker D: So now this, it cannot do anymore.\nSpeaker D: Nothing comes out of here.\nSpeaker E: Not a number is value awesome.\nSpeaker D: Sorry.\nSpeaker D: It doesn't speak German anymore, but it does speak English.\nSpeaker D: There is here a reference.\nSpeaker D: So this tells us that whatever has the ID zero is referenced here by the restriction seat.\nSpeaker D: And this is actually...\nSpeaker D: I want...\nSpeaker D: What was the sentence?\nSpeaker D: I want to see it here.\nSpeaker D: Here to see it here.\nSpeaker D: And where is it playing?\nSpeaker D: There should also be a reference or something.\nSpeaker D: Maybe...\nSpeaker D: Or this is...\nSpeaker D: Here we changed...\nSpeaker D: So we here we add something to the discourse status that the user wants to change something that was sort of done before.\nSpeaker D: And that whatever is being changed has something to do with the cinema.\nSpeaker E: And then whatever takes us in three others that actually changes the state.\nSpeaker F: Yeah.\nSpeaker F: Under the discourse maintainer.\nSpeaker F: And it runs around looking for discourse status tags and doing whatever it does with them.\nSpeaker F: And other people ignore those tags.\nNone: All right.\nSpeaker F: So yeah, I definitely think it's worth the exercise of trying to actually add something that isn't there.\nSpeaker F: So I'm just going to complete it on this thing.\nSpeaker F: Yeah, I can understand it much more.\nSpeaker F: But then the next thing we talked about is actually thinking on how to add our own tags and stuff like that.\nSpeaker D: Okay.\nSpeaker D: Point number two.\nSpeaker D: I got the M3L for the routes today.\nSpeaker D: So I got some more.\nSpeaker D: This is sort of the interesting.\nSpeaker D: It's just going up.\nSpeaker D: It's not going back down.\nSpeaker D: So this is what I got today is the new M3L for the maps.\nSpeaker D: With some examples.\nSpeaker D: So this is the XML.\nSpeaker D: And this is sort of what it will look like later on.\nSpeaker D: You can't see it on this resolution.\nSpeaker D: And this is what it sort of is the structure of map requests.\nSpeaker D: Also not very interesting.\nSpeaker D: And here is the more interesting stuff for us is the routes.\nSpeaker D: Route elements.\nSpeaker D: And again, as we thought, it's really simple.\nSpeaker D: This is sort of the parameters.\nSpeaker D: We have simple from objects and two objects and so forth.\nSpeaker D: Points of interest along the way.\nSpeaker D: And I asked them whether or not we could, first of all, I was a little bit.\nSpeaker D: It seemed to me that this way of doing it is sort of a stack step backwards from the way we've done it before.\nSpeaker D: It seems to me that some notions were missing.\nSpeaker D: So these are these are your friends back at eml.\nSpeaker F: So this is not a complicated negotiation.\nSpeaker F: No.\nSpeaker D: It's not a seven committee user anything, right?\nSpeaker D: No, this is very straightforward.\nSpeaker F: So this is just trying to, it's a design thing, not a political thing.\nSpeaker F: Yeah, we can sort of agree on what ought to be done.\nSpeaker F: Exactly.\nSpeaker D: And however the, so that you understand it, it's really simple.\nSpeaker D: You have a route and you cut it up in different pieces.\nSpeaker D: And every element of that, of that, every segment we call a route element.\nSpeaker D: And so from A to B we cut up in three different steps.\nSpeaker D: And every step has a firm object where you start a two object way, where you sort of end in some points of interest along the way.\nSpeaker D: But I was sort of missing here and maybe it was just me being too stupid as I didn't sort of get the notion of the global goal of the whole route.\nSpeaker D: It was not straightforward, visibly for me.\nSpeaker D: And some other stuff in I suggested that they should be kind enough to do two things for us.\nSpeaker D: One, also allocating some tags for our action schema, enter Vista approach.\nSpeaker D: And also, since you had suggested that we figure out if we ever for a demo reason wanted to shortcut directly to the GIS and the planner of how we can do it.\nSpeaker D: Now what's the state of the art of getting to entrances? What's the syntax for that? Getting to Vista points and calculating those on the spot.\nSpeaker D: And the approach mode anyhow is the default. That's all they do with these days.\nSpeaker D: Wherever you find a route planner, it does nothing but get to the closest point where the street network is at minimal distance to the geometric center.\nSpeaker F: So, well now this is important. Again, it's kind of almost managerial point.\nSpeaker F: You're in the midst of this, you know better, but it seems to me it's probably a good idea to minimize the number of change requests we make of them.\nSpeaker F: So, it seems to me what we ought to do is get our story together. Okay, and think about it some internally before asking them to make changes.\nSpeaker F: Is this makes sense to you guys? I mean, you're doing the interaction, but it seemed to me that what we ought to do is come up with something where you...\nSpeaker F: And I don't know who's working most closely on it, probably, Chano. Okay, take what they have.\nSpeaker F: Send it to everybody saying this is what they have. This is what we think we should add. Okay, and then have an iteration within our group saying, well, okay, and get our best idea of what we should add.\nSpeaker D: And then go back to them. I don't know, just make sense to you or... Yeah, especially if we want sort of what I... my feeling was we sort of reserve something that has an okay label.\nSpeaker D: That was my first sort of step. No matter how we want to call it, this is sort of our playground.\nSpeaker D: And if we get something in there that is a structure elaborate and complex enough to maybe enable a whole simulation. Right, one of these days. That would be...\nSpeaker F: That's right, so it's a good goal. Yeah, the problem isn't the short range optimization. It's the sort of one or two-year kind of thing.\nSpeaker F: One of the classic things we think we might try to do in a year or two, how would we try to characterize those and what do we want to request now that's going to leave enough space to do all of this stuff? Right, and that requires some thought.\nSpeaker F: And so that's something like a great thing to do as a priority item as soon as we can do it.\nSpeaker F: So you guys will send to the rest of us a version of this and the description.\nSpeaker E: Yes, suggested improvements.\nSpeaker F: Yeah, so not everyone reads German, so if you change the description to English. And then, yeah, then with some suggestions about where do we go from here.\nSpeaker F: And this, of course, is just the action end. At some point, we're going to have to worry about the language end, but for the moment, just for this class of things we might want to try to encompass.\nSpeaker E: And this goes with this is beyond approach and...\nSpeaker F: Oh, yeah, yeah, yeah, yeah, yeah. This is everything that...\nSpeaker F: We might want to do in the next couple of years. So what would... Okay.\nSpeaker F: I mean, that's an issue. We don't know one entirely.\nSpeaker E: Yeah, but I'm just...\nSpeaker E: But the... Yeah, okay. So this XML stuff here just has to do with source path goal type stuff in terms of traveling through Hadelberg or travel specifically.\nSpeaker E: So is the domain greater than that?\nSpeaker F: No, I think the idea is that...\nSpeaker F: Well, it's beyond source path goal, but I think we don't need to get beyond the tourists in Hadelberg.\nSpeaker F: It seems to me we can get all the complexity we want in actions and in language without going outside of tourists in Hadelberg.\nSpeaker F: But depending on people are interested in, one could have tours, one could have explanations of why something is...\nSpeaker F: Why was this done? No, there's no end to the complexity you can build into the...\nSpeaker F: What a tourist in Hadelberg might ask.\nSpeaker F: So at least unless somebody else wants to suggest otherwise, I think the general domain...\nSpeaker F: We don't have to broaden that is tourists in Hadelberg.\nSpeaker F: And if there's something somebody comes up with that can't be done that way, then sure.\nSpeaker F: We'll look at that. But I'd be surprised that if there's any important issue then.\nSpeaker F: And I may have given you one to push us into reference problems, that would be great.\nSpeaker F: Okay, so this is his specialty is reference.\nSpeaker F: And what are these things referring to? Not only an Afro, but more generally, this whole issue of referring expressions and what is it that they're actually dealing with in the world.\nSpeaker F: And again, this is in the database. This is also pretty well formed because there is an ontology and the database and stuff.\nSpeaker F: So it isn't like the evening star or stuff like that. All the entities do have concrete reference.\nSpeaker F: Although to get at them from the language may not be trivial. There aren't really deep mysteries about what things the system knows about.\nSpeaker B: We have both proper names and descriptions.\nSpeaker F: All those things. You have proper names and descriptions. And the offer and pronouns and all those things.\nSpeaker D: And unfortunately the whole database is in German. We have just commissioned someone to translate some bits of it, i.e. the short, the more general descriptions of all the objects and persons and events.\nSpeaker D: It's a relational database with persons events and objects. And it's quite, I think that would be great because the reference problem really is not trivial.\nSpeaker D: Even if you have such a well-defined world.\nSpeaker D: Throwing a, carrying hours to Athens.\nSpeaker D: How do I get to the powder tower? We sort of think that our bit in this problem is interesting. But just to get from powder tower to an object ID in a database is also not really trivial.\nSpeaker B: Or if you take something even more scary. How do I get to the self-building after the tower? I don't know. It's something that I'm using.\nSpeaker E: Or the church across the city hall. Or the restaurant where they were later. That would be fine.\nSpeaker D: Or this tower or that building.\nSpeaker F: Or you can say how do I get back?\nSpeaker F: And again, it's just a question of which of these things people want to dive into. What I think I'm going to try to do. And I guess, let's say that by the end of spring break, I'll try to come up with some general story about construction grammar and what constructions we'd use and how all this might fit together.\nSpeaker F: There's this whole framework problem that I'm feeling really uncomfortable about. And I haven't had a chance to think about it seriously.\nSpeaker F: But I want to do that early rather than late. You and I will try to talk about this.\nSpeaker D: It also strikes me that we did a small, something maybe we should address one of these days. That most of the work people actually always do is look at some statements and analyze those, whether it's abstracts or newspapers and so on.\nSpeaker D: But the whole is it really relevant that we are dealing mostly with questions?\nSpeaker D: Well, I mean, yeah, and this seems to be that we shouldn't maybe at least spend a session of brainstorm a little bit about whether that is a special case in that sense.\nSpeaker D: Do we ever find metaphorical use and questions in that sense really? Yeah. You will in house or not. Oh, yeah.\nSpeaker F: Who got kicked out of friends? Yeah, or you know, why is he pushing for promotion or who's pushing? Just pick any of them and just do the, I don't think it's at all difficult to convert them to question forms that really exist and people say all the time.\nSpeaker F: We don't know how to handle them too. We don't know how to handle the declarative forms really and then the interrogative forms.\nSpeaker A: So, you know, I think it's a very different case. So, we had this problem last year and we first thought about this domain actually was that most of the things we talked about our story understanding, we're going to have a short discourse and the person talking is trying to, I don't know, give you a statement and tell you something and here.\nSpeaker A: Is it a mental problem? Yeah, I guess so. And then here, the person is getting information and they may or may not be following some larger plan that we have to recognize or infer and their discourse pattern probably don't follow quite as many.\nSpeaker F: Right, now I think that's one of the things that's interesting is in this sort of overarching story, we worked it out for, as you say, the storytelling scenario and I think it's really worth thinking through what it looks like, what does the SIM spec mean, etc.\nSpeaker A: Right, because for a while we're thinking, well, how can we change the data to sort of illicit, illicit actions out of more like what we are used to, but obviously we'd rather try to figure out what.\nSpeaker F: Well, I don't know, I mean maybe that's what we'll do is we can do anything we want with it.\nSpeaker F: I mean, once we have fulfilled these requirements, okay, and the one for next summer is just half done and then the other half is this generation thing which we think isn't much different.\nSpeaker F: So once that's done, then all the rest of it is sort of what we want to do for the research and we can do all sorts of things that don't fit into their framework at all.\nSpeaker F: There's no reason why we're constrained to do that.\nSpeaker F: If we can use all the execution engines, then we can really try things that would be too much pain to do ourselves, but there's no obligation on it.\nSpeaker F: So if we want to turn it into understanding stories about Heidelberg, we can do that.\nSpeaker D: And that would just be a, always matter of fact, we need, I mean, if we take a 10 year perspective, we need to do that because assuming we have this, in that case we actually do have these wonderful stories and historical anecdotes and nights jumping out of windows and tons of stuff.\nSpeaker D: So the database is huge and if we want to answer a question on that, we actually have to go one step before that and understand that.\nSpeaker D: No, do sensible information extraction.\nSpeaker D: And so this has been a deep map research issue that was part of the unresolved and to do something for the future is how can we sort of run our text, our content through a machine that will enable us later to.\nSpeaker D: Retrieve or answer questions more sensibly.\nSpeaker B: Anyway, so what is the basic thing that you are obligated to do?\nSpeaker F: So what happened is there's this, Robert was describing the, there's two packages, there's a quote parser, there's a particular piece of this big system which in German takes these to sentence templates and produces XML structures.\nSpeaker F: And one of our jobs was to make the English equivalent of that, that these guys did in a day. The other thing is that at the other end, roughly at the same level, there's something that takes XML structures, produces an output XML structure which is instructions for the generator.\nSpeaker F: Okay, and then there's a language generator and then after that, a synthesizer that goes from an XML structure to a language generation to actual specifications for a synthesizer.\nSpeaker F: But again, there's one module in which there's one piece that we have to convert to English.\nSpeaker F: Is that, okay, and this is all along with viewed as a kind of a minor thing, necessary but not, okay.\nSpeaker F: And much more interesting is the fact that as part of doing this, we are inheriting this system that does all these other things, not precisely what we want.\nSpeaker F: That's where it gets difficult and I don't pretend to understand yet what I think we really ought to do.\nSpeaker D: Okay. So, you know how that, but the sort of journal and I will take up that responsibility and get a first draft of that.\nSpeaker D: Now we have just I think two more short things. You guys sort of started fighting on the basenet, noisy or front?\nSpeaker C: Yeah, I thought I should talk a little bit about that because it might be a good sort of architecture to have in general for problems with multiple inputs to anode.\nSpeaker F: And what's the other one so that just we know what the agenda is? The wallpaper, I think.\nSpeaker F: Oh, yeah, I've got a couple new whoo papers as well. So I've been in contact with who.\nSpeaker F: So probably let's put that off till I to I understand better what he's doing.\nSpeaker F: It's a little embarrassing because all this was in his thesis and I was on his thesis committee.\nSpeaker F: And so I really knew this at one time. But I, it's part of what I haven't figured out yet is how all this goes together.\nSpeaker F: So they got some more stuff from DeKai. And so why don't we just do the.\nSpeaker C: Okay. So should I, is there a whiteboard here that I can use?\nSpeaker F: Yeah, you could, or should I just use this? Probably just as easy.\nSpeaker E: Yeah, yeah, you can put the microphone in your pocket.\nSpeaker E: Okay.\nSpeaker E: I was in being you in your pocket because I don't have one.\nSpeaker C: That's why they invented pocket. Yeah. So we call it.\nSpeaker C: We want to have this kind of structure in our basements, namely that.\nSpeaker C: You have these nodes that have several parents, right? So the, I mean, this sort of.\nSpeaker C: The typical example is that these are all a bunch of cues for something.\nSpeaker C: And this is a certain effect that we'd like to conclude. So like, let's just look at the case where.\nSpeaker C: And this is actually the final action, right? So this is like.\nSpeaker C: You know, touch or sorry.\nSpeaker C: Eva.\nSpeaker C: Yeah. Eva, right.\nSpeaker C: Enter view approach.\nSpeaker C: Right.\nSpeaker C: So it's decided out for.\nSpeaker C: Yeah. Enter view approach.\nSpeaker C: Right.\nSpeaker C: So I mean, we'd like to take all these various cues, right? So this one might be say.\nSpeaker C: Well, I'm going to say.\nSpeaker C: I don't know.\nSpeaker C: I could be like.\nSpeaker C: This isn't the way it really is, but let me say that's supposed someone mentioned.\nSpeaker C: Admission fees.\nSpeaker C: It takes too long, right? Let me just say landmark.\nSpeaker C: If the thing is a landmark, you know, there's another thing that says if.\nSpeaker C: If it's closed or not at the moment.\nSpeaker C: Right. So you have nodes, right?\nSpeaker C: And the problem that we were having was that, you know, given end nodes, there's two to the end.\nSpeaker C: Given end nodes.\nSpeaker C: And for the more the fact of these three things here, we need to specify three times two to the end probabilities.\nSpeaker C: Right.\nSpeaker C: That's assuming these are all binary, which they may not be. For example, they could be time of day in which case we could say, you know, morning, afternoon, evening night.\nSpeaker C: So this could be more.\nSpeaker C: So it's a lot anyway.\nSpeaker C: And that's a lot of probabilities to put here, which is kind of a pain.\nSpeaker C: So no, as you are, is our way to sort of deal with this.\nSpeaker C: So the idea is that.\nSpeaker C: Let's call these c1, c2, c3, and c4, and e for a cause and effect.\nSpeaker C: I guess.\nSpeaker C: The idea is to have these intermediate nodes.\nSpeaker C: Well, actually, the idea first of all is that each of these things has a quote-unquote distinguished state, which means that this is the state in which we don't really know anything about it.\nSpeaker C: So for example, if we don't really know if the thing is a landmark or not, or if that distance isn't relevant, then that would be sort of the distinguished state.\nSpeaker C: And similarly, you know, if there's something for the person talking about the admission fee, you know, if they didn't talk about that would be the distinguished state.\nSpeaker D: So this is a principle way of saying default.\nSpeaker C: Yeah.\nSpeaker C: That's just what they were used in that paper.\nSpeaker C: So the idea is that you have these intermediate nodes, e1, e2, e3, and e4.\nSpeaker C: So this is the heck of a paper you're working with.\nSpeaker C: So the idea is that each of these ei represents what this would be if all the other ones were in the distinguished state.\nSpeaker C: So for example, suppose the thing that they talked about is a landmark, but none of the other sort of queues really apply.\nSpeaker C: Then this would be the, this would just represent the probability distribution of this assuming that this queue was turned on and the other one just didn't apply.\nSpeaker C: So you know, it was a landmark and none of the other things really applicable, then this would represent the probability distribution.\nSpeaker C: So maybe in this case, maybe we just, maybe we decide that if the thing is a landmark and we don't know anything else, then we're going to conclude that they want to view it with probability, you know, 0.4.\nSpeaker C: They want to enter it with probability, with probability 0.5 and they want to approach it, probably 0.1 say.\nSpeaker C: So we come up with these little tables for each of those.\nSpeaker C: Okay.\nSpeaker C: And the final thing is that this is a deterministic function of these.\nSpeaker C: So we don't specify any probabilities.\nSpeaker C: We just have to say what function this is.\nSpeaker C: So we can let this be g of e1, e2, e3, e4.\nSpeaker C: And our example g would be majority vote.\nSpeaker F: Well, okay. So the important point is not what the g function is. The important point is that there's a general kind of idea of short cutting the full CPT, the full conditional probability table, with some function, okay.\nSpeaker F: Which you choose appropriately for each case. So depending on what your situation is, they're different functions which are most appropriate.\nSpeaker F: And so I gave Baskar a copy of this sort of 92 paper. And you got one rubber. I don't know who else has seen it.\nSpeaker C: I mean, yeah. It's hecka min, short. It's short.\nSpeaker F: So, if you read it yet.\nSpeaker F: Now you should take a look at it.\nSpeaker F: Okay. So you should take a look. Nancy, I'm sure you read it at some point in life.\nSpeaker F: Okay. And so you guys can decide how interesting.\nSpeaker F: No, wait. So the paper isn't real hard. And one of the questions is, how much of this does Java-based support?\nSpeaker C: Yeah. It's a good question.\nSpeaker C: So what we want is basically Java-based support deterministic functions. And in a sense, we can make it supported by manually entering probabilities that are 1 and 0's.\nSpeaker F: Right. So the little thing that I sent, I sent a message saying, here's a way to take one thing you could do, which is kind of in a way stupid.\nSpeaker F: So let's take this deterministic function and use it to build the CPT. So the Java-based won't do it for you.\nSpeaker F: That you can convert all that into what the CPT would be.\nSpeaker F: And what I sent out about a week ago was an idea of how to do that for evidence combination.\nSpeaker F: So one function that you could use as your G function is evidence combining. So you just take the, if each of the ones has its own little table like that, then you could take the strength of each of those times its little table when you'd add up the total evidence for the E and A.\nSpeaker C: I don't think you can do this because if G is a function from that to that. Right. So there's no numbers. There's just quadruplets of, well, n-tuplets of, you know what I'm saying is, there is, even if you decide what is appropriate is probabilistic evidence combination.\nSpeaker F: You can write a function that does it. But it's actually one of the examples he's got in there. But anyways, skipping, skipping the question of exactly which functions.\nSpeaker F: Is it clear that you might like to be able to shortcut the whole conditional probability table?\nSpeaker D: I mean, it seems very plausible in some sense where we will be likely to not be observed some of the stuff because we don't have the access to the information.\nSpeaker F: Right. That's one of the problems is where we're at all.\nSpeaker C: So it's all right. Would not be able to observe what?\nSpeaker D: If it's a discourse initial phrase, we will have nothing in the discourse history.\nSpeaker C: So if we ever want to wonder what? If you're not able to observe certain, that's fine. That is not the only thing.\nSpeaker F: There's two separate things, Robert. The basenets in general are quite good at saying if you have no current information about this variable, just take the prior for that.\nSpeaker F: That's what they're real good at. So if you don't have any information about the discourse, you just use your priors of whatever the discourse, and it's basically whatever, it's probabilistically whatever would be. It's not a great estimate, but it's the best one you have.\nSpeaker F: And so forth. So that they're good at. The other problem is how do you fill in all these numbers? And I think that's the one he was getting.\nSpeaker C: So specifically, in this case, you have to have this many numbers. Whereas in this case, you have to have three for this, three for this, three for this, three for this.\nSpeaker C: So you have to have just three n. So this is much smaller than that.\nSpeaker E: That's not a good estimate, I think. Well, pretty quickly. You don't need data enough to cover.\nSpeaker E: Really, what a nice thing to kind of neural net size these basenets.\nSpeaker F: No, no, so noise use is a funny way of referring this because the noise error is only one instance.\nSpeaker C: This isn't a noise or that one actually isn't the noise or so what to think of of a way to.\nSpeaker E: It's a noisy arg max or noisy whatever. Yeah, whatever. Yeah. So my point was more that we just, with the neural net right, things come in, you have a function that combines them.\nSpeaker F: Yeah, that's true. It's also more neural net like, although it isn't necessarily some, you know, some of weights or anything like that.\nSpeaker F: I mean, you could have, like the noisy or function really is one that's essentially says, take the max. Well, the or.\nSpeaker F: Same. Right. I guess you're right. Yeah. But anyway, so.\nSpeaker F: And I think that's the standard way people get around the. There are a couple other ones. There are ways of breaking this up into some nets and stuff like that. But.\nSpeaker F: I think we definitely, I think it's a great idea to pursue that.\nSpeaker D: Well, still sort of leaves one question. I mean, you can always see easily that I'm not grasping everything correctly. But what seems attractive to me in the last discussion we had was that we find a means of.\nSpeaker D: Of getting these point 4.5.1 of C4. Not because you know, A is then mug or not, but we we we we label this whatever object type. And if it's a garden, it's 0.3.4.2. If it's a castle, it's 0.8.1.1. If it's.\nSpeaker D: Town hall, it's 0.2.3.5. And so forth. And we don't want to write this down necessarily every time for something. But.\nSpeaker D: It would be students will be stored. That's the question. Well, in the beginning, we'll write up a flat file. We know we have 20 objects and we'll write it down.\nSpeaker F: Let me say something because there's a pretty point about this. We're not so good in right now, which is the hierarchy that comes with the ontology is just what you want for this.\nSpeaker F: So that. If you know about it, let's say a particular town hall. That it's one that is a monument, then that would be stored there. If you don't, you look up the hierarchy.\nSpeaker F: So you may or so that you'd have this little vector of. You know, approach mode or even mode. Let's OK. So we have the eve of vector for various kinds of landmarks. If you know it for a specific landmark, you put it there.\nSpeaker F: If you don't, you just grab the hierarchy to the first place you find one. So is the idea to put it in the ontology? Absolutely. OK.\nSpeaker F: Or link to or but in any case, view it logically as being in the ontology. It's part of what you know about an object is its eve vector. OK.\nSpeaker F: And if you say if you know about a specific object, you put it there. This is part of what the guy was doing. So when you get the row, we'll see what he says about that. And then if you if it isn't there, it's higher.\nSpeaker F: And if you don't know anything except that it's it's a building, then I put the highest thing you have the what amounts to a prior. You don't know anything else about a building.\nSpeaker F: You just take whatever your crew approximation is up at that level, which might be equal or whatever it is. So that's a very pretty relationship between these local vectors and the ontology.\nSpeaker F: And it seems to me the obvious thing to do unless we find a reason to do something different.\nSpeaker F: Does it make sense to you? Yeah. So we are doing the ontology. So we have to get to over is doing the ultimately. And that's another thing we're going to need to do is to either we're going to need some way to either get a tag in the ontology or add fields or some way to associate or it may be that all we can do is some our own hash table that.\nSpeaker F: There's always a way to do that just a question. Yeah, hash on object name to. Yeah, the probability. Right. And so.\nSpeaker D: It's most strikes me as a word. If we get the mechanism. That would be sort of the wonderful part and then how to make it work is the second part in the sense that. I mean, the guy who's doing the ontology.\nSpeaker D: The apologize that it will take him another through two to three days because they're having really trouble getting the upper level straight. Right now the reason is given the crop, but the projects that all carry their own taxonomy and on all history.\nSpeaker D: They're really trying to build one top level ontology that covers all the email projects and that's sort of a tough cookie a little bit tougher than they figured I could have told them. So. Right. Yeah. But nevertheless, it's going to be there by next Monday. And I will show you what's what some examples from that for towers and stuff. And what I don't think is ever going to be in the ontology is sort of, you know, the likelihood of people.\nSpeaker D: People entering town halls and looking at town halls and approaching town halls, especially since we're dealing with a case based on an instant based ontology.\nSpeaker D: So there will be nothing on on that town hall or on the Berkeley town hall or on the high work town hall. Just be information on town halls.\nSpeaker D: But how what are they going to do with instances? Well, that's that's a different question. I mean, the first they had to make a design question.\nSpeaker D: Do we take ontologies that have instances or just one that does not that just says the types. Okay. And so since the decision was on types on a simply type based.\nSpeaker D: We now have to hook it up to instances.\nSpeaker F: What is smart come going to do about that?\nSpeaker D: Because they have instances all the time. Yeah, the ontology is really not a smart come thing in enough itself. That's more something that I kick loose in email.\nSpeaker D: So it's a complete email thing. But smart comes going to need an ontology. Yes. A lot of people are aware of that.\nSpeaker F: I understand. But we do anything about it. Okay. Political problem. We won't worry about it. No, but the I still think that there is enough information in there.\nSpeaker D: For example, whether. Okay. So it will know about the 20 object types there are in the world. Let's assume there are only 20 object types in this world.\nSpeaker D: And it will know if any of those have institutional meanings. So in a sense, I used as institutions for some some sense or the other.\nSpeaker D: Which makes them. Entrubble right in a sense. Anyway, so we may have to this is the whole thing. We may have to build another data.\nSpeaker F: Conceptually, we know what should be done. When we see what people have done, it may turn out that the easiest thing to do is to build a separate thing that that just pulls.\nSpeaker F: Like it may be that the instance, we have to build our own instance. Yes.\nSpeaker F: Things that. Right. We can just. Those are types and then it goes off the oncology once you have its type. So we build a little data structure.\nSpeaker F: And so what we would do in that case is in our instance gadget have our EVAs. And if there isn't one, we'd get the type and then have the EVAs for the type.\nSpeaker F: So we would have our own little. EVA tree. And then for other. Vectors that we need. So we'd have our own little things so that whenever we needed one, we'd just use the oncology to get the type.\nSpeaker F: And then would hash or whatever we do to say off it's that type of thing and we want its EVA vector. It's that.\nSpeaker F: So I think we can handle that and then the combination functions and whether we can put those in Java Bayes and all that sort of stuff is is the bigger deal.\nSpeaker F: Yeah. I think that's where we have to get technically clever.\nSpeaker E: We could just steal the classes in Java Bayes and then interface to them with our own code. Well, you quite understanding the classes in Java Bayes.\nSpeaker F: Yeah, I mean, that's cute. I mean, you've been around enough to just. Well, I mean, this is huge package, which may or may not be consistent.\nSpeaker E: But yeah, we could look at it. Well, just. Okay. Yeah.\nSpeaker F: It's sort of a kind of a thing is it's kind of an interpreter and it expects its data structures to be in a given form. And if you say, hey, we're going to make a different kind of data structure to stick in there.\nSpeaker F: Well, no, but that just means there's a protocol right that you could. It may or may not. I don't know. That's a question is to what extent is it allows us to put in these G functions.\nSpeaker E: And I don't know. No, but I mean, what I would say you could have four different basements as you're running and then run your own right your own function that would take the output of those four and make your own G function.\nSpeaker F: Yeah, that's fine. If it comes only at the end, but suppose you wanted embedded.\nSpeaker E: Well, then you'd have to break all of your basements up in the smaller basements with all the.\nSpeaker F: Oh, yeah, that's a truly horrible way to do it. One would hope. Yeah, but I'm just. Yeah, yeah, yeah, yeah, yeah. But at that point, you may say, hey, Java base isn't the only package in town. Let's see if there's another package that's more civilized about this.\nSpeaker F: Now, Streney is worth talking to on this because he said that he actually did.\nSpeaker F: He had some combining functions in the Java base, but he doesn't remember unless when I talked to him, he didn't remember whether it was an easy thing in natural thing or whether he had to do some violence to it to make it work.\nSpeaker C: But he did do it. I don't see why the combining function have to be directly hacked into Java base. I mean, they're used to create tables so we can just make our own functions that create tables.\nSpeaker F: So what I did do it is to just convert it into into a CPD that uses it's blown up. It's huge, but it doesn't require any data fitting or complication.\nSpeaker C: I don't think I mean, the fact that it blows up is a huge issue in the sense that I mean, okay, so say it blows up, right? So say this like.\nSpeaker C: 10, 10, 15 things. It's going to be like two to the bat, which isn't so bad. I understand. I'm just saying that that was my note. The little note I said said that.\nSpeaker F: I said, here's the way you take the logical G function and turn it into a CPD. I mean, the the evidence combining function.\nSpeaker F: So we could do that. And maybe that's what we'll do. But don't know.\nSpeaker F: So I will be for next week. Push somewhere on the stuff that Decker I would did and try to understand it.\nSpeaker F: You'll make a couple more copies of the document paper to give people. Okay. Okay. And.\nSpeaker D: I think through this getting even vacatars dynamically out of autologes, what would I because I'm not quite sure whether we all think of the same thing or not here. You and I should talk about it.\nSpeaker F: Okay. All right. Great. And Robert, thank you for coming in under. He's been sick. Robert.\nSpeaker E: I was thinking maybe you should just cough into the microphone and see if they can handle it.\nSpeaker A: Okay. So Nancy, about. Yep. Sure.\nSpeaker D: And. So Robert, we have the money to hire. This is.\nSpeaker D: The question.\nSpeaker D: Also.\nNone: You You You You\n"}, {"meeting_id": "Bed011", "summary": "The main focus of the meeting was firstly on the structure of the belief-net, its decision nodes and the parameters that influence them, and secondly, on the design of the data collection tasks. For the latter, there are already 30 subjects lined up and more are expected to be recruited off campus. Finally, as to the semantic and syntactic constructions, work will start with more general and brief descriptions, before moving to exhaustive analysis of at least a subset. Similarly, the construction parser that is to be built within a year is expected to be relatively basic, yet robust.  ", "dialogue": "Speaker E: No, can you give me the\nSpeaker B: three, two, five, four, one, six, zero, zero, three, nine, six, seven, eight, five, three, two, seven, three, three, one, seven, nine, three, nine, five, seven, six, two, one, two, one, six, four, seven, zero, nine, three, four, three, six, one, four, nine, zero, one, six, zero, six, four,\nSpeaker A: OK, so, can you read your numbers? Zero, one, nine, four, five, zero, five, eight, five eight, nine, four, two, eight, eight, seven, three, seven, four, seven, six, five, one, seven, eight, eight, three, nine, four, one, three, zero, two, 9, 2, 8, 1, 3, 4, 5, 6, 5, 6, 6, 7, 8, 6, 8, 8, 7, 5, 6, 8, 9, 5, 6, 8, 9, 5, 4, 0, 8, 8, 7, 7, 9, 9, 4, 9, 3, 9, 4, 7, 0, 1, 7, 8, 0, 1, 4, 3.\nSpeaker C: Yeah, let's get started. Hopefully Nassio come as not, she won't.\nSpeaker G: Robert, do you have any way to turn off your screen saver on there so that it's not going off every, it seems to have about a two minute?\nSpeaker E: Yeah, it's not that I didn't try.\nSpeaker E: And I told it to stay on forever and ever but if it's not plugged in it just doesn't obey my commands.\nSpeaker E: Okay.\nNone: Has a mind.\nSpeaker E: Got it, but I just sort of keep on wailing.\nSpeaker E: We just be working on it as intensely so it doesn't happen.\nSpeaker E: We'll see.\nSpeaker E: Should we plunge right into it?\nSpeaker C: Yeah, I think so.\nSpeaker E: So what I've tried to do here is list all the decision notes that we have identified on this side, commented and what they're about in sort of the properties we may give them.\nSpeaker E: And here are the tasks to be implemented via our data collection.\nSpeaker E: So all of these tasks, the reading is all of these tasks more or less imply that the user wants to go there, sometimes for the other.\nSpeaker E: And inadequately for example here we have our Eva intention and these are the data tasks where we can assume the person would like to enter view or just approach the thing analogously the same on the object information.\nSpeaker E: We can see that we have sort of created these tasks before we came up with our decision notes.\nSpeaker E: So there's a lot of things where we have no analogous tasks and that may or may not be a problem.\nSpeaker E: So we've changed the tasks slightly if we feel that we should have data for sort of for every decision notes trying to implant the intention of going to a place now, going to place later on the same tool, trying to plan the intention of going sometime on the next tour or the next day or whenever.\nSpeaker E: I think that might be overdoing it.\nSpeaker C: So let me pop up a level and make sure that we're all oriented the same.\nSpeaker C: So what we're going to do today is two related things.\nSpeaker C: One of them is to work on the semantics of the belief net which is going to be the main inference engine for the system making decisions and decisions are going to turn out the B parameter choices for calls on other modules.\nSpeaker C: So the natural language understanding thing is we think we only have to choose parameters but a fairly large set of parameters.\nSpeaker C: So to do that we need to do two things, one of which is figure out what all the choices are which we've done a fair amount.\nSpeaker C: Then we need to figure out what influences choices and finally we have to do some technical work on the actual belief relations and presumably estimates of the probabilities and stuff.\nSpeaker C: We are going to do the probabilities stuff today.\nSpeaker C: Technical stuff will do another day probably next week.\nSpeaker C: We are going to worry about all the decisions and the things that contribute to them.\nSpeaker C: And we're also sort of in the same process going to work with FAY on what there should be in the dialogues.\nSpeaker C: So one of the steps that's coming up real soon is to actually get subjects in here and have them actually record like this, record dialogues more or less.\nSpeaker C: And depending on what FAY sort of provokes them to say we'll get information on different things.\nSpeaker C: So if you get people who defraise them and so for Keith and people worrying about what constructions people use we have some ways to affect that by the way the dialogues go.\nSpeaker C: So what Robert kindly did is to lay out a table of the kinds of things that that might come up and the kinds of decisions.\nSpeaker C: So on the left are decision nodes and discrete values.\nSpeaker C: So if we're right you can get by with just this middle column worth of decisions and it's not all that many and it's perfectly feasible technically to build belief nets that we'll do that.\nSpeaker C: Andy has a handout.\nSpeaker E: Yeah maybe it was too fast plunging in there because we have two updates.\nSpeaker E: You can look at this if you want.\nSpeaker E: These are what are subjects going to have to fill out.\nSpeaker E: Any comments are thinking can still be made and the changes will be put in correspondingly.\nSpeaker E: Let me summarize in two sentences mainly for EWAS benefit who probably has not heard about the data collection.\nSpeaker E: Or have you heard about it?\nSpeaker E: No.\nSpeaker E: Okay we're going to put this in front of people.\nSpeaker E: They give us some information on themselves.\nSpeaker E: Then they will read a task where lots of German words are sort of thrown in between.\nSpeaker E: And they have to read isolated proper names.\nSpeaker E: And these are.\nSpeaker E: I don't see a release.\nSpeaker C: No this is not a release form.\nSpeaker C: This is a speaker.\nSpeaker C: Okay fine.\nSpeaker E: The release form is over there in that box.\nSpeaker E: Alright, fair enough.\nSpeaker E: And then they're going to have to choose from one of these tasks which are listed here.\nSpeaker E: They pick a couple say three six is what effect.\nSpeaker E: Six different things they sort of think they would do if they were in Hydeburg or traveling someplace.\nSpeaker E: Then they have a map like this very sketchy simplified map.\nSpeaker E: They can take notes on that map.\nSpeaker E: And then they call this computer system that works perfectly on in a sense everything.\nSpeaker E: And this is a fictional system obviously.\nSpeaker E: Yeah the computer system sits right in front of you that's.\nSpeaker E: And she does know everything.\nSpeaker E: And she has a way of making this machine talk.\nSpeaker E: So she can copy sentences into a window or type really fast and this machine will use speech synthesis to produce that.\nSpeaker E: So if you ask how do I get to the castle then several seconds later it will come out of here in order to get to the castle you do okay.\nSpeaker E: And then after three tests the system breaks down.\nSpeaker E: And Fay comes on the phone as a human operator.\nSpeaker E: And says sorry the system broke down but let's continue.\nSpeaker E: And we sort of get the idea what people do when they think they speak to a machine and what people say when they think they speak to a human or no or assume they speak to a human.\nSpeaker E: That's the data collection.\nSpeaker E: And Fay has some 30 subjects lined up something.\nSpeaker E: And they're ready to roll and we're going to start tomorrow at 3, 4, 1.\nSpeaker B: Tomorrow?\nSpeaker B: Well we don't know for sure because we don't know whether that person is coming up.\nSpeaker E: Okay run for it.\nSpeaker E: And we're still looking for a room on the sixth floor because they stole away that conference room behind our backs.\nSpeaker C: There are these.\nSpeaker C: Oh I see we have the yeah it's tricky.\nSpeaker C: Let's let's get it offline.\nSpeaker E: It's happening.\nSpeaker E: David and Jane and Laila are working on that as we speak.\nSpeaker E: Okay.\nSpeaker E: There was the data collection in a nutshell.\nSpeaker E: And I can report it.\nSpeaker E: So I did this but I also tried to do this.\nSpeaker E: So if I click on here is this wonderful.\nSpeaker E: We get to the belief net just focusing on the go their note.\nSpeaker E: And elegantly this would be sort of the reason note and the timing note and so forth.\nSpeaker E: And what happened is that design wise I'd sort of notice that we can we still get a lot of errors from a lot of points to one of the sub go there user go there situation notes.\nSpeaker E: So I came up with a couple of additional notes here where whether the user is thrifty or not and what is budget is currently like is going to result in some financial state of the user.\nSpeaker E: How much willing is he willing to spend or can spend being the same with this just the money available which may influence us whether he wants to go there if it's you're charging tons of dollars for admission or is going to cost a lot of whatever 22 million to fly to international space station.\nSpeaker E: Not all of you can do that.\nSpeaker E: So and this actually turned out to be pretty key because it's having specified sort of these this this intermediate level and sort of noticing that everything that happens here let's go to our favorite end point one is again more or less we have than the situation notes contributing to the end point situation note which contributes to the end point and so forth.\nSpeaker E: I can now sort of draw straight lines from these to here meaning it of course goes with the sub S everything that comes from situation everything that comes from user goes with the sub U and whatever was specified for the so called key note or the discourse would comes from the parser construction parser will contribute to the D and the ontology to the sub O note and one just sort of has to watch which also final decision notes so it doesn't make sense to figure out whether he wants to interview or approach an object if he never wants to go there in the first place.\nSpeaker E: This makes the design thing fairly simple and now all that's left to do then is the CPT's the conditional probabilities for the likelihood of a person having enough money actually wanting to go a place if it costs you know and and once Baskara has finished his classwork that's what we're going to end up doing you get involved in that process too and and for now the question is how much of these decisions do we want to build and explicitly into our data collection so one could sort of think of you know we could call this C or you know people who visit the zoo we could call it visit the zoo tomorrow so we have a tension of seeing something but not now but later.\nSpeaker C: So let's see I think that from one point of view all these places are the same so that that in terms of the linguistics and stuff there may be a few different kinds of places so it seems to me that we ought to decide you know what things are actually going to matter to us and so the zoo in the university in the castle etc are all big-ish things that you know have different parts to them and one of them might be fine.\nSpeaker E: The reason why we did it that way as a reminder is no person is going to do all of them they're just going to select to their preferences.\nSpeaker C: Oh yeah I usually visit zoos or I usually visit castles or I usually the point is to build a system that's got everything in it that might happen you do one thing to build a system that had the most data on a relatively confined set of things should do something else and the speech people for example are going to do better if they if things come up repeatedly.\nSpeaker C: Now of course if everybody says exactly the same thing then it's not interesting so all of them say there's a kind of question of what we're trying to accomplish and I think my temptation for the data gathering would be to you know these persons only going to do it once you have to worry about them being bored so if it's one service one luxury item one bigish place and so forth and so on then my guess is that the data is going to be easier to handle.\nSpeaker C: Of course you have this I get a possible danger that somehow there are certain constructions that people use when talking about a museum that they wouldn't talk about with the university and stuff but I guess my temptation is to go for simpler you know less variation but\nSpeaker G: I don't know what other people think about this in terms of so I don't exactly understand like I guess we're trying to limit the tale of our ontology of types of places that someone could go right but who is it that has to care about this or what component of the system\nSpeaker C: oh well I think there are two places where it comes up one is in the people who are going to take this and try to do speech with it lots of pronunciations of the same thing are going to give you better data than you know a few pronunciations of lots more things that's one.\nSpeaker G: So we would rather just ask and have a bunch of people talk about the zoo and assume that the constructions they use there will give us everything we need to know about these sort of big zoo castle whatever type things these bigger places.\nSpeaker G: Yeah this is an action for speech data people saying zoo over and over again and over\nSpeaker C: two. So this is a question for you and you know if we do we probably will actually try to build a prototype probably we could get by with the prototype only handling a few of them anyway.\nSpeaker F: So yeah this was sort of these are all different sort of activities but I think I got the\nSpeaker E: point and I think I like it we can do put them in a more hierarchical fashion so go to place and then give them a choice either the symphony type or opera type or the tourist side type or the nightclub disco type person and they say yeah this is on that go to biggest place this is what I would do then we have the fixed thing and then maybe do something the other day thing.\nSpeaker E: So my question is I guess to some extent we should you just have to try it out see if it works it would be challenging in a sense try to make it so so complex that they even really should schedule or plan it more complex thing in terms of okay you know they should get the feeling that there are these six things they have to do and they can be done maybe in two days.\nSpeaker E: Well I think they make these decisions can I go there tomorrow or influences.\nSpeaker C: Well I think it's easy enough to set that up if that's your expectation so the system could say well I would like to set up your program for two days in a Heidelberg you know let's first think about all the things you might like to do.\nSpeaker C: So there I mean I'm sure that if that's what you did then they would start telling you about that and then you could get into various things about ordering if you wanted.\nSpeaker E: But I think this is part of the instructors job and that can be done sort of to say okay now we've picked these six tasks now you have you can call the system and you have two days.\nSpeaker C: I'm sorry but we have to help we have to decide fate we'll carry out whatever we decide but we have to decide you know what is the appropriate scenario that's what we're going to talk about.\nNone: Yeah.\nSpeaker F: So we have two different scenarios in the early one is a plan of the only kind of human functions on the spot.\nSpeaker E: Yeah but I don't I'm not really interested in sort of face planning capabilities but it's more the how do people phrase these planning requests.\nSpeaker E: So are we going to masquerade the system as this as you said simple response system I have one question I get one response or should we allow for a certain level of complexity and I think the data would be nicer if we get temporal references.\nSpeaker C: So Keith what do you think?\nSpeaker G: Well it seems yeah I mean off the top of my head it kind of seems like you would probably just want you know richer data more complex stuff going on people trying to do more complex things I mean you know if our goal is to really sort of be able to handle a whole bunch of different stuff than throwing harder situations that people will get them to do more linguistic more interesting linguistic stuff but I mean I'm not really sure because I don't fully understand like what our choices are ways to do this here.\nSpeaker E: We have tested this and have you heard listen to the first two or the second person is was faced with exactly this kind of setup.\nSpeaker G: I started to listen to one and it was just like sort of depressing sort of listening to the beginning part and the person was just sort of reading off a script or something.\nSpeaker E: Oh okay that was the first person.\nSpeaker G: First of all it wasn't very good.\nSpeaker E: So it is already with this setup and that particular subject it got pretty complex.\nSpeaker E: So I suggest we make some fine tuning of these get sort of run through ten or so subjects and then take a breather and see whether we want to make it more complex or not depending on what sort of results we're getting.\nSpeaker G: Yeah in fact I am just you know today next couple days going to start really diving into this data I basically looked at one of the files you know one of these you gave me those dozens of files and I looked at one of them which is about ten sentences found fifteen twenty different construction types that we would have to look for and so on and like all right well let's start here.\nSpeaker G: So I haven't really gone into the you know looked at all of the stuff that's going on so I don't really right I mean once I start doing that I'll have more to say about this kind of thing.\nSpeaker C: Okay well you did say something important which is that you can probably keep yourself fairly well occupied with the simple cases for quite a while.\nSpeaker C: Yeah although obviously so that's not so does suggest that now I have looked at all the data and it's actually at least to an amateur quite redundant.\nSpeaker C: Yeah it was very stylized and quite a lot of people said more or less the same thing.\nSpeaker G: I did sort of scan it at first and notice that and then looked into tail at one of them but yeah I noticed that too.\nSpeaker C: So we want to do more than that.\nSpeaker E: And with this we're getting more no question.\nSpeaker E: So do we want to get going beyond more?\nSpeaker C: Well okay so let's take let's I think your suggestion is good which is we'll do a batch okay and say how long is it going to be till you have ten subjects a couple days or the week or I don't have a few friends.\nSpeaker B: I can yeah I mean I think I probably scheduled ten people whenever.\nSpeaker C: Well it's up to you I mean I we don't have any huge time pressures just when you have\nSpeaker B: the yeah. I would say maybe two weeks.\nSpeaker C: Oh okay so let's do this let's plan next Monday okay to have a review of what we have so far.\nSpeaker E: And no transcriptions of course.\nSpeaker C: No I won't have the transcriptions but what we should be able to do and I don't know if you'll have time to do this but it would be great if you could not transcribe it all but pick out some stuff.\nSpeaker C: I mean we could listen just sit here and listen to it all.\nSpeaker C: Are you going to have the audio on the website?\nSpeaker C: Okay.\nSpeaker E: Until we reach the gig up by a thing and David Johnson kills me and we're going to put it on the website yeah.\nSpeaker C: Oh we can get I mean you can buy another disc for $200 right I mean it's okay so I\nSpeaker E: won't take care of David Johnson. Okay.\nSpeaker E: He's been solving all our problems so it's wonderful.\nSpeaker C: So anyway so if you can think of a way to point us to interesting things sort of as you're doing this or something make notes or something that this is you know something worth looking at and other than that yeah I guess we'll just have to listen although I guess it's only 10 minutes each right roughly.\nSpeaker B: I guess I'm not sure how long it's actually.\nSpeaker E: The reading test is a lot shorter therefore it's cut by 50% and the reading nobody's interested in that.\nSpeaker E: Right we don't care about that at all.\nSpeaker E: So it's actually like five minutes dialogue.\nSpeaker C: My guess is it's going to be 10.\nSpeaker E: 10 minutes as long.\nSpeaker C: I understand but people it feels like a long time.\nSpeaker E: It feels like forever when you're doing it but then it turns out to be three minutes and 45 seconds.\nSpeaker C: Could be.\nSpeaker C: Okay.\nSpeaker C: I was thinking people would you know hesitate whatever whatever it is we'll do with it.\nSpeaker C: Yeah it's not.\nSpeaker C: Okay so that'll be on the webpage.\nSpeaker C: That's great.\nSpeaker C: But anyway yeah so I think it's a good idea to start with the sort of relatively straightforward response system and then if we want to get them to start doing multiple step planning with a whole bunch of things and then organize them and tell them which things are near each other and you know any of that stuff.\nSpeaker C: Which things would you like to do Tuesday morning?\nSpeaker C: So yeah that seems pretty straightforward.\nSpeaker B: But were you saying that?\nSpeaker E: I need those back by the way.\nNone: Okay.\nSpeaker C: That's for if anyone.\nSpeaker B: Maybe one thing we should do is go through this list and sort of select things that are categories and then offer only one member of that category.\nSpeaker C: That's what I was suggesting for the first round.\nSpeaker C: Yeah.\nSpeaker B: Okay.\nSpeaker B: So rather than having a zoom in.\nSpeaker B: And then I mean they could be alternate versions of the same.\nSpeaker B: They could but it would be one of data on difference.\nSpeaker C: Yeah.\nSpeaker C: But.\nSpeaker B: Like one person gets the version with the zoo as a choice.\nSpeaker E: The person don't get it.\nSpeaker E: I mean this is what we did it because when we gave them just three tasks for part A and\nSpeaker B: three tasks for part B. They just want to be able to choose both zoo and say touring the castle.\nSpeaker E: Exactly.\nSpeaker E: This is limiting the choices but yeah right okay sorry.\nSpeaker E: But I think this approach worked very well but the person was able to look at it and say okay this is what I would actually do.\nSpeaker E: Okay.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: He was vicious.\nSpeaker B: Okay.\nSpeaker E: We got it.\nSpeaker E: We got to disallow traveling to zoos and castles at the same time sort of.\nSpeaker B: I mean they are significantly different but.\nSpeaker E: But no they're sort of this is where tour becomes you know tours maybe a bit different.\nSpeaker E: And these are just places where you enter much like here but we can.\nSpeaker C: Yeah.\nSpeaker C: In fact if you use the right verb for each in common like it you know attend a theater symphony or opera as a group and tour the university castle or zoo all of these do have this kind of tour aspect about the way you would go to them.\nSpeaker C: And the movie theater is probably also a tent etc.\nSpeaker C: So it may turn out to be not so many different kinds of things and then what we would expect is that the sentence types would their responses would tend to be grouped according to the kind of activity you would expect.\nSpeaker F: It seems that the different between going to see something and change my.\nSpeaker C: Oh absolutely.\nSpeaker E: Yeah.\nSpeaker E: Yeah this is where the function itself is definitely different in the getting information or stuff.\nSpeaker E: Okay.\nSpeaker E: But this is open.\nSpeaker E: So since people are going to still pick something we're not going to get any significant amount of redundancy and for reasons we don't want it really in that sense.\nSpeaker E: And we would be ultimately more interested in getting all the possible ways of people asking or for different things with a computer.\nSpeaker E: And so if you can think of any other sort of high level tasks a tourist may do just always just mail them to us and we'll sneak them into the collection.\nSpeaker E: We're not going to do much statistical stuff with it.\nSpeaker E: We don't have enough.\nSpeaker E: But it seems like since we are getting towards subject 50 subjects and if we can keep it up to a sort of five fourish per week rate we may even reach the 100 before Faye takes herself to Chicago.\nSpeaker B: That means that 100 people have to be interested.\nSpeaker C: Good luck.\nSpeaker C: These are all people from campus so far.\nSpeaker C: So we don't know how many we can get next door at the shelter for example.\nSpeaker C: For ten bucks probably quite a few.\nSpeaker C: That's right.\nSpeaker C: All right, so let's go back then to the chart with all the decisions and stuff and see how we're doing.\nSpeaker C: People think that this is going to cover what we need or should we be thinking about more.\nSpeaker E: Okay, in terms of decision notes.\nSpeaker E: Yep.\nSpeaker E: I mean go there is a yes or no.\nSpeaker E: Right?\nSpeaker E: Yep.\nSpeaker E: I'm also interested in this property line here.\nSpeaker E: So if you look at, sorry, look at that.\nSpeaker E: Timing was, I have these three.\nSpeaker E: Do we need a final differentiation there?\nSpeaker E: Now later on the same tour, sometimes on the next tour.\nSpeaker G: What's this idea of next tour?\nSpeaker E: I mean, it's sort of the next day.\nSpeaker E: So you're doing something now and you're planning to do these three four things and you can do something immediately you can sort of tag it on to that tour or you can say this is something I want to do sometime in my life.\nSpeaker G: Okay.\nSpeaker G: So this tour is sort of just like the idea of current round of touristness.\nSpeaker C: Yeah, probably between stuff's back at the hotel.\nSpeaker C: Okay.\nSpeaker C: I mean, if you want to be precise about it, you know, and I think that's the way tours do organize their lives.\nSpeaker C: Sure.\nSpeaker C: Sure.\nSpeaker C: So you can go on then we'll go off and okay.\nSpeaker F: So all tours of eight tour happens only within one day.\nSpeaker C: Yes.\nSpeaker F: It's too early to move.\nSpeaker G: Right.\nSpeaker G: For this.\nSpeaker G: Okay.\nSpeaker G: Just to be totally clear.\nSpeaker E: Well, my visit to Prague, there were some nights where I had never went back to the hotel.\nSpeaker G: So whether that's a two day tour or not, I think you just spend the whole time at Uflekland.\nSpeaker C: We will not ask you more.\nSpeaker E: That's enough.\nSpeaker E: I don't know.\nSpeaker E: What does the English cognate if you want for Zank Nimalans talk?\nSpeaker E: Kind of ano.\nSpeaker E: Sort of.\nSpeaker E: We'll do it.\nSpeaker E: When you say on that date, it means it'll never happen.\nSpeaker E: Right.\nSpeaker E: Okay.\nSpeaker E: Do you have an expression?\nSpeaker E: Probably.\nSpeaker G: Not that I know, I've actually.\nSpeaker E: Yeah, I went out.\nSpeaker E: Yeah.\nSpeaker E: We'll do it when he'll pres us over.\nSpeaker E: So maybe that should be another.\nSpeaker E: Yeah.\nSpeaker E: Property in there.\nSpeaker E: Okay.\nSpeaker E: The reason why do we go there in the first place, I see the foreset scene for meeting people for running errands or doing business.\nSpeaker E: Entertainment is a good one in there, I think.\nSpeaker E: I agree.\nSpeaker G: So business is supposed to be sort of like professional type stuff.\nSpeaker G: Right?\nSpeaker G: Yeah.\nNone: Okay.\nSpeaker E: I mean, this is an old, a general thing he sort of had it in there.\nSpeaker E: Who is the, is the person?\nSpeaker E: So it might be a tourist.\nSpeaker E: It might be a businessman.\nSpeaker E: Yeah.\nSpeaker E: So it's a good system, what's the sort of go-toism?\nSpeaker G: Yeah, or both.\nSpeaker G: Yeah.\nSpeaker G: I mean, like, for example, my father is about to travel to Prague.\nSpeaker G: He'll be there for two weeks.\nSpeaker G: He's going to, he's there to teach a course at the business school, but he also is touring around, so I may have some.\nSpeaker G: Sure.\nNone: He's doing things.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker F: What do you have in mind, the number of socializing people?\nSpeaker E: Just meeting people, basically.\nSpeaker E: I want to meet someone somewhere, which puts a very heavy constraint on the Eva, you know, because then if you're meeting somebody at the town hall, you're not entering it.\nSpeaker E: Usually you just want to approach it.\nSpeaker G: So, I mean, does this capture, like, where do you put exchange money as an errand, right?\nSpeaker G: So I go to a movie, where is now entertainment, Dineout is?\nSpeaker C: No, I, well, I didn't.\nSpeaker C: Well, I, let, let, well, we'll put it somewhere, but, but, I would say that if Dineout is a special, if you're doing it for that purpose, then it's entertainment.\nSpeaker C: Yeah.\nSpeaker C: And we'll also, as, as you, further along, we'll get into business about, well, you're, you know, this is going over a meal time, do you want to stop for a meal or pick up food or something, and that's different.\nSpeaker C: That's, that's sort of part of the, that's not a destination reason, that's sort of, um, a song.\nSpeaker E: That goes with the energy depletion function.\nSpeaker E: Right.\nSpeaker E: Yes.\nSpeaker E: Okay.\nSpeaker G: End point.\nSpeaker G: Bad laugh.\nSpeaker E: End point is pretty clear.\nSpeaker E: Um, mode, uh, I have phone three drive their walk there or, or be driven, which means bus taxi art.\nSpeaker E: Okay.\nSpeaker A: The public transport in general.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker C: Obviously, taxis are very different than buses, but on the other hand, the system doesn't have any public transport.\nSpeaker C: The planar system doesn't have any public transport in it yet.\nSpeaker E: So this granularity would suffice, I think, if we say the person, probably based on the adherence, in this situation, we can conclude once drive their walk there or use some other form of transportation.\nSpeaker G: How much of hydrogen bill can you get around by public transport?\nSpeaker G: I mean, in terms of the interesting bits, there's lots of bits where you don't really, I've only, I was there 10 years ago for a day, so I don't remember, but, you're good.\nSpeaker G: I mean, like, sort of the touristy bits.\nSpeaker C: You can't get to the philosophers way very well, but, I mean, there are hikes that you can't get to.\nSpeaker C: But, yeah.\nSpeaker C: But I think other things, you can't, if I remember right.\nSpeaker A: So it's like biking there, part of the timing?\nSpeaker E: Yeah.\nSpeaker E: And we actually biking should be, should be a separate point because we have a very strong bicycle planning component.\nSpeaker E: Oh.\nSpeaker E: So, uh, bicycle should be in there.\nSpeaker E: But, will we have, I mean, is this realistic?\nSpeaker E: Yeah.\nSpeaker C: Okay, we can leave that.\nSpeaker E: Yeah.\nSpeaker E: We can sort of drive.\nSpeaker G: I would lump it with walk because it will matter, right?\nSpeaker G: Yeah.\nSpeaker G: Like that.\nSpeaker C: Skateboards, I'm going to write it right anyway.\nSpeaker C: Scooters, right?\nSpeaker E: Right.\nSpeaker E: All right.\nSpeaker E: Okay, length is, you want to get this over with as fast as possible.\nSpeaker E: You want to use some part of what, of the time you have.\nSpeaker E: They can, but we should just make a decision whether we feel that they want to use some substantial or some fraction of their time.\nSpeaker E: You know, they want to do it so badly that they are willing to spend, you know, the necessary and plus time.\nSpeaker E: And, um, and, you know, if we feel that they want to do nothing but that thing, then, you know, we should point out that to the planner that they probably want to use all the time they have.\nSpeaker E: So stretch out that visit.\nSpeaker G: Wow, it seems like this would be really hard to guess. I mean, on the part of the system, it seems like it.\nSpeaker G: I mean, you're talking about rather than having the user decide this, you're supposed to figure it out.\nSpeaker E: The user can always say it, right?\nSpeaker E: It's just sort of, we hand over these parameters.\nSpeaker E: If we make, if we have a feeling that they are important and that we can actually infer them to a significant degree, all we ask.\nSpeaker C: Yeah, and part of the system design is that if it looks to be important and you can't figure it out, then you ask.\nSpeaker C: But hopefully you don't ask, you know, all these things all the time.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: But there's definitely a back-off position to asking.\nSpeaker E: And if no part of the system ever comes up with the idea that this could be important, no planner is ever going to ask for it.\nSpeaker E: And I like the idea that, you know, sort of, Jerry pushed this idea from the very beginning that it's part of the understanding business to sort of make a good question of what's sort of important in this general picture, what you need.\nSpeaker E: If you want a simulated, for example, what parameters would you need for the simulation?\nSpeaker E: And timing, a length would definitely be part of it.\nSpeaker E: Costs, little money, some money, lots of money.\nSpeaker E: Actually, maybe, yeah, okay.\nSpeaker G: You could say some there.\nNone: I must say that this one looks a bit strange to me.\nNone: It's like a problem here.\nNone: If I go to a sped us.\nNone: I'm willing to lose.\nSpeaker C: I think there are different things where you have a choice, for example, this interacts with, you're willing to take a taxi.\nSpeaker C: You know, if you're going to the opera, are you going to look for the best seats or the peanut gallery or whatever?\nSpeaker C: So I think there are a variety of things in which tourists really do have different styles eating another one.\nSpeaker E: What my sort of sentiment is there.\nSpeaker E: I once had to write a charter for a student organization.\nSpeaker E: And they had wanted me to define what the quorum is going to be.\nSpeaker E: I looked at the other ones and they were 10% of the student body as we present at the general meeting.\nSpeaker E: Otherwise, it's not.\nSpeaker E: And I wrote in there enough people have to be there.\nSpeaker E: And it was hardly debated, but people agreed with me that everybody probably has a good feeling whether it was a farce joke, whether there were enough people.\nSpeaker E: And if you go to Turkey, you will find when people go shopping, they will say, how much she's you want and they say, enough.\nSpeaker E: This used all over the place because the person is adding the cheese nose.\nSpeaker E: That person has two kids and a husband that dislikes cheese.\nSpeaker E: And so the middle part is always sort of the golden way, right?\nSpeaker E: You can really make it as cheap as possible or you can say, I don't care.\nSpeaker E: Money is no object.\nSpeaker E: Money is no object.\nSpeaker E: Or you say, this one is spent enough or the sufficient or the appropriate amount.\nSpeaker E: But then again, this may turn out to be insufficient for all purposes.\nSpeaker E: But this is my first guess.\nSpeaker E: In much of the same way as how should the route be?\nSpeaker E: Should it be the easiest route, even if it's a little bit longer?\nSpeaker E: No steep incline nations go the normal way.\nSpeaker E: Whatever that again means, or does the person want to rough it?\nSpeaker G: There's a couple of different ways you can interpret these things.\nSpeaker G: I want to go there and I don't care if it's really hard or if you're an extreme sport person.\nSpeaker G: I want to go there and I insist on it being the hard way.\nSpeaker G: I assume we're going for the first interpretation.\nSpeaker G: Something like, I'll go with, I don't know.\nSpeaker C: He was going for the second one, actually.\nSpeaker F: We'll sort that out.\nSpeaker D: What's the answer sign?\nSpeaker E: Absolutely.\nSpeaker E: Well, this is all sort of trouble of my head.\nSpeaker E: No research behind that.\nSpeaker E: Object information, do I want to know anything about that object?\nSpeaker E: Is it the true or false?\nSpeaker E: If I care about it being open accessible or not, I don't think there's any middle ground there.\nSpeaker E: Either I want to know where it is or not.\nSpeaker E: I want to know if it's history or not.\nSpeaker E: Or I want to know what it's good for or not.\nSpeaker E: Maybe one could put scales in there too.\nSpeaker E: What you were saying?\nSpeaker E: One could put scales in there.\nSpeaker C: Yeah, right.\nSpeaker C: Object becomes entity, right?\nSpeaker E: Yep, that's true.\nSpeaker E: I have to do it now.\nSpeaker E: That's the wrong shortcut anyhow.\nSpeaker C: And we think that's it, interestingly enough, that, you know, or something very close to what is going to be enough.\nSpeaker C: Going to be enough.\nSpeaker C: And...\nNone: Don't run.\nSpeaker E: Yep.\nSpeaker C: Okay.\nSpeaker C: So I think the order of things is that Robert will clean this up a little bit, although it looks pretty good.\nSpeaker E: Well, this is the part that needs to work.\nSpeaker C: Right, so in parallel, three things are going to happen.\nSpeaker C: Robert and even Busker are going to actually build a belief net that has CPTs and tries to infer this from various kinds of information.\nSpeaker C: And Faye is going to start collecting data.\nSpeaker C: And we're going to start thinking about what constructions we want to do to make a data collection.\nSpeaker G: Do you mean eliciting particular constructions?\nSpeaker G: What kinds of things we want to get people talking about?\nSpeaker C: Well, semantically speaking.\nSpeaker C: Both of us constructions are primarily semantic, right?\nSpeaker G: Right.\nSpeaker G: So from my point of view, I'm trying to care about this syntax.\nSpeaker C: Well, that too.\nSpeaker C: But if we make sure that we get them talking about temporal order, that would be great.\nSpeaker C: And if they use prepositional phrases or subordinate clauses or whatever.\nSpeaker C: Whatever form they use is fine.\nSpeaker C: So we're probably going to try to look at it as what semantic constructions do we want them to cause motion?\nSpeaker C: I don't know.\nSpeaker C: Something like that.\nSpeaker C: But this is actually a conversation you and I have to have about your thesis fantasies and how all this fits into that.\nSpeaker E: Well, I will send you the German tourist data because I have not been able to dig out all this stuff out of the 30 DVDs.\nSpeaker G: Is that roughly the equivalent of what I've seen in English?\nSpeaker G: No, not at all.\nSpeaker E: Dialogues.\nSpeaker E: Smartcom human, Wizard of Oz.\nSpeaker G: Okay, same.\nSpeaker G: Got it.\nSpeaker G: What have I got now?\nSpeaker G: What are the files that you sent me are the user side of some interaction with Faye?\nSpeaker G: Is that what it is?\nSpeaker G: With nothing.\nSpeaker G: Just talking into a box and not hearing anything back.\nSpeaker E: Some data I collected in a couple of weeks for training recognizers that you can't have a way back when.\nSpeaker E: Nothing to write home report.\nSpeaker E: And this ontology note is probably something that I will try to expand once we have the full ontology API.\nSpeaker E: What can we expect to get from the ontology?\nSpeaker E: And hopefully you can also try to find out sooner or later in the course of the summer what we can expect to get from the discourse that might, not the discourse, the utterance, as it were, in terms of...\nSpeaker C: Right, but we're not expecting keys to actually build a parser.\nSpeaker C: No, no, no, no.\nSpeaker C: Okay.\nSpeaker C: We are expecting Jono to build a parser.\nSpeaker G: By the end of the summer, too.\nSpeaker G: No.\nSpeaker G: No.\nSpeaker C: But sort of...\nSpeaker C: He's hoping to do this for his master's thesis by a year from now.\nSpeaker C: Still pretty formidable actually.\nSpeaker C: Absolutely.\nSpeaker C: Limited.\nSpeaker C: I mean, the idea is...\nSpeaker C: Well, the hope is that the parser itself is pretty robust, but it's not popular.\nSpeaker C: Right, right.\nSpeaker G: All business proof, you know, so that's what I talked about.\nSpeaker G: It's like infrastructure, yeah.\nSpeaker G: Right.\nSpeaker G: Sometimes I have to talk to some subset of the people in this group at least about what sort of construction I'm looking for.\nSpeaker G: I mean, you know, obviously, like just, again, looking at this one thing, you know, I saw things from sort of as general as argument structure constructions.\nSpeaker G: Oh, you know, I have to do verb phrase.\nSpeaker G: I have to do unbounded dependencies, you know, which are a variety of constructions in instantiate that.\nSpeaker G: On the other hand, I have to have, you know, this particular fixed expressions or semi-fixed expressions like get plus path expression.\nSpeaker G: Or, you know, how do I get there?\nSpeaker G: How do I get in?\nSpeaker G: How do I get away?\nSpeaker G: Right.\nSpeaker G: And all that kind of stuff.\nSpeaker G: So there's a variety of sort of different sorts of constructions.\nSpeaker G: And it is sort of like anything goes like.\nSpeaker C: Okay, so this is, I think, we're going to mainly work on with George.\nSpeaker C: Okay.\nSpeaker C: And here, let me put some of what I think is, is, so the idea is, first of all, I misspoke when I said we thought you should do the constructions.\nSpeaker C: Because apparently for a linguist, that means to do completely and perfectly.\nSpeaker C: So what I meant was do a first cut app.\nSpeaker C: Okay, because we do want to get them perfectly, but I think we're going to have to do a first cut of a lot of them to see how they interact.\nSpeaker G: Right, exactly.\nSpeaker G: Now, we talked about this before, right?\nSpeaker G: And I mean, it would be completely out of the question to really do more than say, like, oh, I don't know, 10 over the summer.\nSpeaker G: But, you know, obviously we need to get sort of a general view of what things look like.\nSpeaker C: Right. So the idea is going to be to do sort of like Nancy did in some of these papers where you do enough of them so you can go from top to bottom.\nSpeaker C: So you can do, you know, have a complete story of some piece of dialogue.\nSpeaker C: And that's going to be much more useful than having all of the causal constructions and nothing else.\nSpeaker C: Or something like that. So the trick is going to be to take this and pick a sort of lattice of constructions, some lexical and some phrasal.\nSpeaker C: And, you know, whatever you need in order to be able to then by hand, you know, explain some fraction of the utterances.\nSpeaker C: And so exactly which ones that partly depend on your research interests and a bunch of other things.\nSpeaker G: Sure. Okay. But I mean, in terms of the sort of level of analysis, you know, these don't necessarily have to be more complex than like the out of construction and the BCP paper.\nSpeaker G: Correct. It's just like, you know, half a page on your drawing.\nSpeaker C: Oh, yeah, yeah. Half a page is what we'd like.\nSpeaker C: Yeah. And if there's something that really requires a lot more than that, then it does and we have to do it.\nSpeaker G: Yeah. For the first cut, that should be fine.\nSpeaker E: Yeah. We could sit down and think of sort of the the ideal speaker utterances.\nSpeaker E: And I mean, two or three that follow each other. So where we can also sort of once we have everything up and running, show the tremendous insane inferencing capabilities of our system.\nSpeaker E: So, you know, as the smart people have just there's standard demo dialogue, which is what the system survives and nothing but that we could also sort of have the analogy of our sample sentences, the ideal sentences where we have complete construction coverage.\nSpeaker E: And sort of they mention I say sort of the how do I get to X?\nSpeaker E: Yeah. It's definitely going to be a major one.\nSpeaker G: Yeah. That's about six times in this little one here.\nSpeaker E: Where is X? Might be another one, which is not too complicated.\nSpeaker E: And tell me something about X.\nSpeaker E: Yeah.\nSpeaker E: And hey, that's already covering 80% of the system's functionality.\nSpeaker C: Right. That's not covering 80% of the intellectual interest.\nSpeaker E: Yeah.\nSpeaker E: No, we can throw in an out of film construction if you want.\nSpeaker C: No, no, no. Well, the thing is there's a lot that needs to be done.\nSpeaker C: To get this right. Okay.\nSpeaker C: Are we done?\nSpeaker E: I have one bit of news.\nSpeaker E: Yes.\nSpeaker E: The action planner guy has wrote has written a lengthy proposal on how he wants to do the action planning.\nSpeaker E: And I responded to him also rather lengthy how he should do the action planning.\nSpeaker C: And action planning meaning discourse modeling.\nSpeaker E: Yes. And I tacked on a little paragraph about the fact that the whole world calls that much of a dialogue manager.\nSpeaker E: Right.\nSpeaker E: Wouldn't it make sense to do this here too?\nSpeaker E: And also, Rainer Mumalaka is going to be visiting us shortly, most likely in the beginning of June.\nSpeaker E: How've we gone?\nSpeaker E: Yeah. He's just in a conference somewhere and he uses to make it through town.\nSpeaker E: And then making me incapable of going to an AACL for which I had funding.\nSpeaker E: But no no Pittsburgh this year.\nSpeaker E: What is the center Barbara who's going to show a lot of people?\nSpeaker E: That's something I would probably should go.\nSpeaker C: That's one you should probably go to.\nSpeaker G: How much does it cost?\nSpeaker C: I haven't planned to go.\nSpeaker C: Probably we can pay for it.\nSpeaker C: Student rate shouldn't be very high.\nSpeaker C: So if we all decided it's a good idea for you to go then you're right.\nSpeaker C: I mean, I don't have a feeling one way or the other at the moment.\nSpeaker C: Okay.\nSpeaker C: Probably is.\nSpeaker C: Okay. Great.\nNone: Thank you.\n"}, {"meeting_id": "ES2011c", "summary": "This meeting was mainly about conceptual design. Industrial Designer, Marketing and User Interface gave presentations mainly on component design, current trend and interface design respectively, based on their research on the website incorporating the ideas from last meeting. A lot of discussions was centred around the fruit and vegetable theme, which was the current marketing trend. In the meanwhile, they had some discussions on physical attributes and the use of energy. They agreed that the screen was not necessary, and TV could be used as the screen instead. In addition, Kryptonite became their preferred energy.", "dialogue": "None: Hi everyone.\nSpeaker B: Hope you have a nice lunch.\nSpeaker B: Alright, we're moving on to conceptual design.\nSpeaker B: I'll just review what we did in our last meeting.\nSpeaker B: Under marketing we targeted our audience and yeah, that was generally how helpful that was.\nSpeaker B: Then we considered some design options with how it's the book. We discussed an iPod-like button system which we haven't concluded but we're good.\nSpeaker B: If you all have presentations to do, we can see where you've come from our last meeting. Does everyone have presentations? Yes.\nSpeaker B: Then everybody likes to go first. Sure.\nSpeaker C: So I've been looking at the components design. Mostly by consulting remote control diagrams from the internet and also by incorporating design ideas from the last project meeting.\nSpeaker C: So we need some custom design parts and other parts which we'll just use standard.\nSpeaker C: I assume we'll be custom designing our case. Probably a hard plastic or some other material case to protect the remote and the locator.\nSpeaker C: We'll need to custom design a circuit board because a circuit board has to take the button input and send it to the output.\nSpeaker C: We have to design that each time. But once we come up with a design we'll send it to the circuit people and they'll just print it out.\nSpeaker C: Standard parts include the buttons and the wheels, the iPod style wheel.\nSpeaker C: The infrared LED is actually going to be included in the circuit board that comes with it.\nSpeaker C: We need a radio sender and receiver. Those are standard. We also need a beeper or buzzer or other sort of noise thing for locating the remote.\nSpeaker C: So we have some material options. We can use rubber, plastic, wood or titanium.\nSpeaker C: I'd recommend against titanium because it can only be used in the flat cases and it's really heavy.\nSpeaker C: The rubber case requires rubber buttons so if we definitely want plastic buttons we shouldn't have a rubber case.\nSpeaker B: Why not wood? Why not wood?\nSpeaker C: Well we can use wood. I don't know why we want to.\nSpeaker C: Also we should know that if we want an iPod style wheel button it's going to require a slightly more expensive chip.\nSpeaker C: We can't use the minimal chip. We need the next higher grade which is called regular. I don't think it's much more expensive but it is more expensive.\nSpeaker C: So that's what I've got on design.\nSpeaker B: That's good.\nSpeaker A: Can I do next? Can I do next? Can I have to say something about the material?\nNone: Yes.\nSpeaker A: It's quite shocking.\nSpeaker A: I have been searching the current trends both on the web and via fashion watches.\nSpeaker A: The first thing to aim for is a fancy look and feel.\nSpeaker A: Next comes technology and innovation with that.\nSpeaker A: Last thing is the easy to use factor.\nSpeaker A: Fancy look and feel goes far beyond the functionality of the thing but I suppose that is included in the ease of use.\nSpeaker A: Our fashion watches in Milan and Paris have decided, well notice that the fruit and vegetable theme is the current trend and therefore we need to go for that if we want whatever our motto is.\nSpeaker A: We want to put the fashion electronics. We need to go for vegetables and also go for a spongy feel.\nSpeaker A: The question of our technology is industrial designer.\nSpeaker A: As to the material should be limited to, I don't know how a spongy can be, I don't know how a spongy can be achieved but apparently that's the way to go.\nSpeaker A: I have been thinking about this fruit and vegetable thing and I prefer fruits to vegetables but that's just a personal opinion.\nSpeaker A: I think people like to have fruit instead of vegetables in the sitting room, those are just suggestions and also we need to decide whether this should be printed so that still has to do with the material discussion.\nSpeaker A: Should we print the fruit stuff or should the actual remote look like a fruit?\nSpeaker A: And finally again with the spongy.\nSpeaker A: It has to be technologically innovative so maybe again our industrial designer should look into that or come up with a solution that's better than mine.\nSpeaker A: To summarize these are the points that need to be touched in order to get a good decision and hopefully our user interface has not to say about the matter.\nSpeaker A: Thank you for your attention.\nSpeaker B: I think it's so much more.\nSpeaker D: I said that fruit and vegetable thing.\nSpeaker D: I actually wasn't aware of the new trends in electronics.\nSpeaker D: You're not going to wear your remote control.\nSpeaker D: Here we go.\nSpeaker D: Conceptual user interface.\nSpeaker D: I'm going to be using this iPod-ish remote control based on fruit vegetable design.\nSpeaker D: This is a touch based graphical interface system.\nSpeaker D: I assume we're still on the screen.\nSpeaker D: If we have this in order to have someone going around and using the wheel.\nSpeaker C: It seems like you would need a screen with music because you're looking for a specific song like that band or whatever.\nSpeaker D: If you think about it.\nSpeaker D: We need to have a screen.\nSpeaker D: I think you're going to have to have some kind of a...\nSpeaker D: I guess that's why I was thinking you need to have some kind of a screen because people can go back and forth and choose it.\nSpeaker D: If you just scroll and scroll and get to like 5 or 12.\nSpeaker B: What if you have satellite and you have 200 channels?\nSpeaker B: You have to like, we can quickly pass.\nSpeaker D: I think the wheel goes through like 100 channels at least on theirs.\nSpeaker D: What kind of range we need to have on the wheel.\nSpeaker D: You're doing this motion to control the channels.\nSpeaker D: Once you stop that you can tap for different functions like volume or you can tap to get to different channels.\nSpeaker D: If you just wanted to go from 5 to 6 you can tap.\nSpeaker D: There's also the concern about how do you get to the menu if you want to change the brightness of your television or if you want to switch around these different modes.\nSpeaker D: Turn on the timer or something like that.\nSpeaker B: With that many options I think that the screen would be better.\nSpeaker B: I would think so too.\nSpeaker D: It does seem a little silly to have this screen if you hardly are ever using it.\nSpeaker D: But that again it does make kind of if the screen is just like an option that is just there and you're not really using it.\nSpeaker C: It's more expensive to the design people.\nSpeaker C: You have to get an advanced chip if you want to have a screen in which is more expensive than the regular chip which is more expensive than the other.\nSpeaker D: It has to have some way to get to a mode on the television where you're doing...\nSpeaker D: Maybe it will be that central button that you hit that and then it brings up the menu on the TV and you can just scroll around.\nSpeaker B: To do the timer to do that.\nSpeaker B: It would have all these different options.\nSpeaker D: We don't want the screen I guess.\nSpeaker D: It does seem like it would be incredibly expensive.\nSpeaker D: So really all you need is this little wheel then you can control everything.\nSpeaker B: If you're thinking of the design of it now, the physical attributes and you just have this and it's just a long silver thing or whatever.\nSpeaker B: Are you going to have any buttons on that besides power in this thing?\nSpeaker D: It doesn't seem that you would need anything besides in the power button could even be like hold down the menu button for longer than one second and it turns on the TV.\nSpeaker B: Can we imagine that this would be smaller than the remote controls that you showed us before?\nSpeaker D: Yeah, definitely. I think we're looking at something that could be...\nSpeaker D: Even maybe... I'm seeing almost a circular sort of handheld thing.\nSpeaker D: It needs to be easy to manipulate and use your...\nSpeaker D: I mean, how do I not really... When I use an iPod, I end up just kind of using my index finger to control it.\nSpeaker B: I've seen some people just going like that one.\nSpeaker C: Or your thumb or something.\nSpeaker C: When we had the wheelie remote control, we just on the top I think.\nSpeaker B: With their buttons on there as well.\nSpeaker C: Yeah, well it had the wheelie thing and then it had those 18 different buttons that I don't know what they do.\nSpeaker D: We just used the top part.\nSpeaker D: I think it could be pretty small.\nSpeaker B: You want it to be large enough that you had a cover that went over buttons that you don't use very much.\nSpeaker B: You could slide it up if you needed to change the contrast or something like that.\nSpeaker B: So the options are there but they're not interfering with the design and the practicality of it.\nSpeaker B: You can't...\nSpeaker B: You just get...\nSpeaker C: You know what I'm talking about though?\nSpeaker C: Yeah, maybe something on the side where you need to flip over.\nSpeaker D: Yeah, but I mean, do you need that if you can get to...\nSpeaker D: If so long as you're able to bring up the menu on the TV screen.\nSpeaker B: Yeah, I mean I guess that's the thing.\nSpeaker B: If we can do this, that probably would be...\nSpeaker D: I just don't think you would even need it.\nSpeaker B: Yeah.\nSpeaker B: So I guess we have to look into the programming.\nSpeaker B: The programming, how they actually program these things.\nSpeaker C: Oh, how they make the menu show up on the TV?\nSpeaker D: I mean you can do it.\nSpeaker C: It doesn't seem that hard.\nSpeaker C: It doesn't seem that hard.\nSpeaker C: It's going to be inside the TV and outside the remote.\nSpeaker D: I mean, I've never bought a remote.\nSpeaker D: It sounds like this remote's going to be purchased separately from the television.\nSpeaker D: Well, they usually are.\nSpeaker D: Well, I've never bought just a remote.\nSpeaker D: So I don't really know.\nSpeaker C: But I guess that's right.\nSpeaker C: It almost comes with a TV.\nSpeaker D: So, but I mean I've never had a hard time with my remotes, like bringing up the menu screen if you need to change the date or whatever.\nSpeaker D: I don't think that should be too hard.\nSpeaker C: Yeah, most of the ones we've had have had the menu button.\nSpeaker C: It's not like you need to have a button access to change the contrast or something.\nSpeaker C: Yeah, so it just didn't come up every day or something.\nSpeaker D: Yeah.\nSpeaker D: Also, do we need...\nSpeaker D: I guess we have to think about it.\nSpeaker D: But I mean, you just basically need the output signal to bring it up.\nSpeaker A: It also, if you have it on the screen, you can actually write everything out because the problem with buttons is they have these sort of abbreviations.\nSpeaker A: Yeah, that you're supposed to have stuff.\nSpeaker A: You don't know if you've never got it.\nSpeaker D: Yeah.\nSpeaker D: So, but you mean if we have the screen like the iPod screen?\nSpeaker D: Well, on the telly.\nSpeaker D: On the telly, okay, yeah.\nSpeaker D: So, yeah, I mean, I think the touch-based graphical interface is a really cool idea because, you know, it is so obnoxious to have to push this.\nSpeaker D: Like, okay, now I hit this, you know, you have your little guy that, you know, hit this button twice.\nSpeaker B: And it is technologically innovative in a way.\nSpeaker D: Yeah, I guess it's with the...\nSpeaker D: And it is trendy.\nSpeaker D: The iPods are really hot right now.\nSpeaker A: Did you get that picture on the day provided about pictures?\nSpeaker D: Yeah, my web research.\nSpeaker A: That interesting.\nSpeaker A: What are we going to do about this virtual thing?\nSpeaker A: Oh, I know.\nSpeaker C: Oh, I was going to say, you said people want spongy.\nSpeaker C: Yeah.\nSpeaker C: One of the material options is a sort of rubber that's not unlike those stress ball things.\nSpeaker C: Oh, okay, that would be fun.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah, that would be kind of...\nSpeaker D: Usually, the touchpad things are kind of a hard plastic mouse type thing.\nSpeaker D: I mean, definitely the area of that.\nSpeaker D: What if we had a spongy sort of like stress bally kind of...\nSpeaker D: Yeah, I think it could work.\nSpeaker D: What if we integrated the...\nSpeaker D: Yeah, what if the whole thing about the fruit of vegetables, we somehow made it tactily?\nSpeaker D: You know, we could tap into that, so it feels like vegetable.\nSpeaker A: It could work.\nSpeaker A: It could work.\nSpeaker A: It could just put a copper.\nSpeaker A: Yeah.\nSpeaker A: It's a small thing.\nSpeaker A: Instead of creating an object for it that looks like a banana, which is fun.\nSpeaker A: It could be fun, though.\nSpeaker A: You could just have copper, and then your mobile...\nSpeaker A: It's like a mobile phone thing.\nSpeaker A: It was a time when they had all these different copies.\nSpeaker C: You could do the computers, or they have the grapefruit apple machine, and they have the blueberry...\nSpeaker C: Yeah, yeah.\nSpeaker C: Like all the colors were named after fruit.\nSpeaker C: Yeah.\nSpeaker D: You could name it after.\nSpeaker D: It's a pressure machine.\nSpeaker A: It could be the color conflict, so if you have red city, or you can have strawberry, and then you have the green one with that.\nSpeaker A: Yeah.\nSpeaker A: So what if...\nSpeaker B: So what if...\nSpeaker B: This is...\nSpeaker B: I'm just forming this idea in my head of how this thing is looking.\nSpeaker B: If you have that stress ball material, as what you're actually holding in your hand, so what you're feeling is comfortable, and then there's more of a hard plastic thing.\nSpeaker B: Yeah.\nSpeaker B: Where that thing is.\nSpeaker B: Yeah.\nSpeaker B: And on that hard plastic thing, you can change either the color or the fruit or vegetable that's on there.\nSpeaker B: Yeah.\nSpeaker B: Because I'm thinking silver because those are our company colors.\nSpeaker B: Yeah.\nSpeaker B: Unfortunately, you don't have silver.\nSpeaker D: It's a bunch of people's.\nSpeaker D: I don't know.\nSpeaker D: Um...\nSpeaker B: I mean, how are you all envisioning by what we've just...\nSpeaker B: The two back we've just got about the...\nSpeaker A: Maybe a ball.\nSpeaker B: A ball.\nSpeaker A: No, a squashy ball.\nSpeaker A: A relaxing squashy ball.\nSpeaker D: That's in the shape of a fruit.\nSpeaker A: Well, I see you're thinking... it's weird, you're thinking you're opposite.\nSpeaker A: You're thinking you changed the hard thing.\nSpeaker A: Yeah.\nSpeaker A: And I'm thinking, how do you change the hard... would you put a sticker sort of?\nSpeaker A: Because I was thinking, if you have a cover for the squashy bit, like...\nSpeaker B: This is just...\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: So that's the squashy bit.\nSpeaker B: Squashy ball.\nSpeaker A: Yeah, I was thinking of getting a cover for...\nSpeaker B: See, I was thinking this... sorry.\nSpeaker B: I was thinking this bit here would be the cover.\nSpeaker B: And that's your actual thing.\nSpeaker B: And this you can have cherries and...\nSpeaker B: Yeah.\nSpeaker B: Things are in there.\nSpeaker B: But...\nSpeaker C: I was thinking of a single ball shape.\nSpeaker C: I was thinking of who is like this.\nSpeaker B: Because the way you were describing the iPod and the roll thing.\nSpeaker B: It's like it has to be...\nSpeaker B: It's almost like your thumb is further up.\nSpeaker B: So if you could squish it lower.\nSpeaker A: Yeah.\nSpeaker A: So it wouldn't be very big.\nSpeaker A: And like how big this big and then you just did it up.\nSpeaker A: Yeah.\nSpeaker D: So you're saying the squishy part is detachable.\nSpeaker D: And you can... so maybe one... you can have the broccoli squishy thing.\nSpeaker D: And then you can have the banana squishy thing.\nSpeaker D: And you can have your choice.\nSpeaker A: Well, just... I can't just... like...\nSpeaker A: I can't think of this as a cover.\nSpeaker A: But... well, the question is, which one is the easiest to change?\nSpeaker A: And we can just contact our relevant department for that.\nSpeaker A: Yeah.\nSpeaker A: And see what the cost is for covering that or for now we can do two prototypes maybe.\nSpeaker A: Yeah.\nSpeaker A: Try and ask users what the best is.\nSpeaker A: Yeah.\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker A: That's nice.\nSpeaker A: So I think it's nice to have a drawing because it's neat.\nSpeaker B: Well, that's not very neat.\nSpeaker B: Yeah.\nSpeaker D: I mean, I think... and I think the hand... I think the hand help part is definitely.\nSpeaker D: So you could make that into the food vegetable part.\nSpeaker A: If it's a bit like there's juggling balls, you can change shape according to your...\nSpeaker A: to the way you hold it.\nSpeaker A: Yeah.\nSpeaker A: If it's got sun then it may be or something.\nSpeaker A: It just molds to your hand.\nSpeaker D: Yeah.\nSpeaker D: So where are the fruit and vegetables now?\nSpeaker C: I guess it would be either in the color of that plastic...\nSpeaker C: A silver front or in the color of the squishy thing underneath.\nSpeaker A: Yeah.\nSpeaker A: And the rest is the company... the company color silver.\nSpeaker B: It was silver and yellow.\nSpeaker B: It looks like...\nSpeaker D: We could promote the banana one.\nSpeaker D: Yeah.\nSpeaker D: Like...\nSpeaker B: I mean, that's another question where we're gonna... or we should have the logo somewhere on it.\nSpeaker B: Hmm.\nSpeaker A: Should also fit the box in the switch around.\nSpeaker C: I think the batteries would have to go right under the plastic case.\nSpeaker B: Yeah.\nSpeaker C: And that would...\nSpeaker C: Especially if you're switching out the squishy part, then you need to have the other part just be sort of a single unit that you can snap off of it.\nSpeaker C: Mm-hmm.\nSpeaker D: Yeah.\nSpeaker D: I think...\nSpeaker D: It'd be interesting to have the squishy bit... the part that you can change into the different trendy vegetables and fruits.\nSpeaker C: But, uh...\nSpeaker C: Well, I know you spend so much time squishing it to your own personal hand.\nSpeaker C: You didn't get a new one, you didn't have to do it all the time again.\nSpeaker A: Yeah.\nSpeaker A: A bit of...\nSpeaker A: Oat-mastic.\nSpeaker A: Does your...\nSpeaker A: I don't know.\nSpeaker B: I don't know what the rest of my notes mean, because they were made for me.\nSpeaker B: But if someone...\nSpeaker B: Caponins concept.\nSpeaker B: Question mark.\nSpeaker B: Energy.\nSpeaker B: Question mark.\nSpeaker B: Is that you?\nSpeaker B: Yes.\nSpeaker B: Alright, alright, alright.\nSpeaker B: Um, so what... but what do we know about...\nSpeaker B: Energy?\nSpeaker B: I mean, we're gonna use batteries, right?\nSpeaker C: Uh, we actually have an option of batteries, solar power, and...\nSpeaker C: A dynamo, which is something I don't know what it is.\nSpeaker A: A dynamo is...\nSpeaker A: It's a bicycle.\nSpeaker A: It's a bicycle mechanism.\nSpeaker A: It's like, if something moves.\nSpeaker A: Oh, okay.\nSpeaker A: When it moves.\nSpeaker C: Yeah, the other one was...\nSpeaker C: The other one was a kinetic thing where you basically have to wind it yourself.\nSpeaker C: So, I sort of picked battery.\nSpeaker C: We could have talked about doing a windup or a dynamo or a solar power.\nSpeaker C: Oh, but I think solar power is not available with a rubber case anyways.\nSpeaker B: Battery is something.\nSpeaker D: It seems a little weird but dynamo would be interesting.\nSpeaker A: The fact that dynamo is the moment you move it, it creates energy on its own.\nSpeaker A: What about a kryptonite?\nSpeaker A: It's like, we'll save you, throw it.\nSpeaker A: It's gonna store loads of energy and you don't need to buy a battery because I find them in the right.\nSpeaker A: But we need to find the cost.\nSpeaker A: Yeah.\nSpeaker A: I don't know what the cost is.\nSpeaker C: It has a data to actually.\nSpeaker C: All it said was it gave sort of a relative some chips are more expensive than other sort of things.\nSpeaker C: It didn't give me any actual cost.\nSpeaker C: But most of the stuff is pretty cheap though, but in bulk, so I think it's not much of a problem.\nSpeaker C: Like the chip is probably the most expensive part.\nSpeaker B: What does chip on print mean?\nSpeaker C: For things like remote controls, a stamp out of chip.\nSpeaker C: Calculated to I think.\nSpeaker C: So you can mass produce them pretty cheap.\nSpeaker C: But it's not like a computer, you can't reprogram your remote controls like stamp onto the chip.\nSpeaker B: So chip on print, it just means like that.\nSpeaker B: That means for this.\nSpeaker B: And case, I guess that's why we've been talking about it.\nSpeaker B: I guess it's just a case.\nSpeaker B: I'm thinking of like some tactic case.\nSpeaker B: Let's see.\nSpeaker B: Is there anything else we need to talk about?\nSpeaker B: When we move on, you two are going to be playing with PlayDoh.\nSpeaker B: And working on what can fill with design and user interface design.\nSpeaker B: You're going to be doing product evaluation.\nSpeaker B: So you get more instructions from your personal coach.\nSpeaker B: Cool.\nSpeaker A: See you soon.\nSpeaker B: Does it matter that I am early?\nSpeaker B: I think it'll pop up.\n"}, {"meeting_id": "Bed013", "summary": "An idea for future work was suggested during the visit of the german project manager: the possibility to use the same system for language generation. Setting up certain inputs in the Bayes-net would imply certain intentions, which would trigger dialogues. There is potential to make a conference paper out of presenting the current work and the project aspirations within a parsing paradigm. The focus should be the Bayes-net, to which all other modules interface. Situation, User, Discourse and Ontology feed into the net to infer user intentions. It was suggested that they start analysing what the Discourse and Ontology would give as inputs to the Bayes-net.", "dialogue": "Speaker C: That's usual. Yes\nSpeaker D: I was forgot it was a meeting Oh good 20 minutes ago. Thank you. What did I forget?\nSpeaker B: Scridge how the brain sort of does that\nSpeaker D: And turn all the alarms\nSpeaker B: Okay, so the news from me is a my forthcoming travel plans Yes two weeks from today Yeah, more or less I'll be off to Sicily in Germany for a couple three days\nSpeaker D: And what are you doing there forget?\nSpeaker B: Okay, I'm flying to Sicily basically to drop off Simon There's this great verse and then I'm flying to Germany to go to a Mojo to Athens which is the meeting of all the module responsible people in smart com And represent ici I It myself I guess there and That's the actual reason and then I'm also going up to email for a day and then I'm going to meet the very big boss of convoyster in sub-routine and the system integration people in kaiserslautern And then I'm flying back via Sicily They dropped my son coming here on the 4th of July\nSpeaker C: And a great time to be coming back to the us of a God bless America you see the fireworks from your plane coming in\nSpeaker B: I'm sure all the people at the airport will be happy to work on that day\nSpeaker C: Yeah We're not even better service than usual\nSpeaker D: We're aren't you flying on the tons of them I'll tell you That's not a big deal What you get to the United States it'll be a problem then\nSpeaker B: And That's that bit of news and the other bit of news is we had you know I was visited by my German project manager Who a did like what we did What we're doing here and B is planning to come here either three weeks in July or three weeks in august To actually work on us oh And we sat around and we talked and he came up We came up with a pretty strange idea And that's what I'm going to lay on you now And maybe it might be ultimately the most interesting thing for Eva because he has We know to complain about the fact that the stuff we do here is not weird enough So this is so weird that should even make you happy Okay Imagine if you will That we have a system that does all that understanding that we wanted to do Uh-huh based on utterances It should be possible To make that system produce questions So if you have the knowledge of how to interpret where is x Undergiven conditions situational user discourse on logical Conditions You should also be able to make that same system ask Where is x In a certain way based on certain intentions So instead of just being able to observe phenomenon Um and guess the intention we Might be able just Sort of Give it an intention and make it produce an utterance\nSpeaker D: But like in AI they generally do the Take in and then they also do the generation phase like Nancy's thing or You know in the hand Thing in 182 like not only was able to recognize which is also to generate based upon situations I mean that sort of thing absolutely\nSpeaker B: Once you've done that What we can do is have the system ask It's self An answer Ask something else and enter a dialogue with itself The basic the same idea is having two chess computers take\nSpeaker C: Except this smacks a little bit more of a skits of random computer\nSpeaker B: Yeah, you if you want you can have two parallel I'm asking each other what would that give us would a be something completely weird and strange and be If you look at all the factors We will never observe People Let's say in wheelchairs under you know under all conditions, you know when they say it x And there is a right at the goal and the parking is good. We can never collect enough data It's not possible, right But maybe one could do some learning If you get the system to speak to itself you may find breakdowns and errors And you may be able to learn And make it more robust maybe learn new things and And so there is no no end of potential things one could get out of it if that works And he would like to actually work on that for this\nSpeaker D: So they probably should be coming back a year from now\nSpeaker B: Yeah, I See that the generation bit making the system generate generate something this shouldn't be too hard\nSpeaker D: Well, once the system understands things. Yeah, no problem I just don't think I think we're probably a year away from getting the system to understand things\nSpeaker B: Yeah, well if we can get it to understand one thing like our worries run through We can also maybe a make it say or ask Where is X Or not? I don't know\nSpeaker C: I'm sort of having impression that getting it to say the right thing and the right circumstances is much more difficult than Getting it to understand something given the circumstances and so on You know, I mean just as it's sort of harder to learn to Speak correctly in a foreign language rather than learning to understand it right?\nSpeaker C: I mean just the fact that we'll get the point is that getting it to understand one construction doesn't mean that it will always know exactly When it's correct to use that\nSpeaker B: It's it's Well, I've done generation and language for ductile research for four four and a half years and so it's you're right It's not the same as the understanding. It's in some ways easier in some ways harder. Yeah, but um I think it'd be a Fun to look at it or into that question. It's a pretty strange idea\nSpeaker D: That's the basic idea. I guess it'd be to give a lot of the system to have intentions basically Is that's basically what needs to be added to the system for it?\nSpeaker B: Yeah, I think even think even what it would be the the prior intention So let's say we have to see that I mean no, let's we have to we have some some top down processing given certain setting Okay, now we change nothing and just say ask something Right\nSpeaker D: What would it ask? It wouldn't know what to ask It Unless it was in a situation we'd have to set up a situation where you didn't know where something was and it wanted to go there Yeah, which means it we'd need to set up an intention inside of the system right Which is basically I don't know where something is and I need to go there\nSpeaker B: Do we really need to do that?\nSpeaker D: because No, I guess not\nSpeaker B: It's I know it's it's strange, but look at it Look at our baseness if we don't have Let's assume we don't have any input from the language Right so there's also nothing we could query the ontology, but we have a certain user setting If you just ask what is the likelihood of that person wanting to enter something it'll give you an answer Right sure they are And so whatever that is is the generic default intention That it would find out which is wanting to know where something is maybe I don't know what it's going to be but there's going to be something that well you're not going to you're going to get a variety of\nSpeaker C: Intentions out of that then I mean you're just talking about like given this user What's the what is it what is that user most likely to want to do?\nSpeaker B: Well, you can have it sort of some user and context stuff and ask what's the Posterior probabilities of all of our decision notes You could even say let's take all the priors that's observed nothing query all the posterior probabilities It's going to tell us something right\nSpeaker D: Well it will Assign values to all the notes. Yes\nSpeaker B: And with posterior probabilities for all the values of the decision notes Which if we have an algorithm that filters out whatever the the best or the most consistent Answer out of that will give us the Intention x nihilo And that is exactly what would happen if we ask it to produce an utterance it would be based on that extension x nihilo Which we don't know what it is but it's there So we wouldn't even have to to kickstart it by giving it a certain attention or Observing anything on the decision note And Whatever that maybe that would lead to what is\nSpeaker D: Undershastle or what is I guess what I'm afraid of is if we don't you know set up a situation we'll just get a bunch of garbage out like You know everything's exactly 30%\nSpeaker B: Yeah, so what we actually then need to do is write a little script that changes all the settings You know go goes through all the permutations which is we did it didn't we calculate that one?\nSpeaker D: That was that was absurdly low and the last meeting because I went and looked at it because I was thinking That could not be right and it would it was on the order of 20 output nodes and something like 20 30 input nodes So to test every output node I would at least See said we two to 30 for every output node Which is very very large oh\nSpeaker B: That's nothing for those neural guys. I mean they trained for millions and millions of epochs. Well, I'm talking about\nSpeaker D: I was gonna think of my water Talking about billions and billions and billions and A number two to 30 is like it Boscow said we calculated out and Boscow believes that it's larger than the number of particles in the universe\nSpeaker C: And I don't know if that's right or not It's big It's just that's it's a billion right two to the 30\nSpeaker D: Well, it's 30 is a billion, but if we have to do it two the 20 times Okay, then that's a very very large number. Oh, okay. Yeah, that's big because we have to query the node for every all right Uh, or query the net to the 20 times or not to excuse me 20 times. Okay. So it comes to 20 billion or something Yes, that's pretty big. That's big actually. Oh, we calculated a different number before. How do we do that?\nSpeaker C: Hmm. I remember there being some other one flittering around but anyway Yeah, yeah, it's anyway the point is given all of these different factors It's it's still going to be impossible to run through all of the possible situations or whatever, but I think this will get us a bit closer at least right? I mean\nSpeaker D: If it takes us a second to do it for each one and let's say it's 20 billion then that's 20 billion seconds, which is Ava do the math Long I've heard some hours and hours and hours but we can do randomized testing\nSpeaker C: Yeah\nSpeaker D: Which problem willistically will be good enough\nSpeaker B: Yeah, so it'd be it's an idea that one could for for example run run past um What's that guys for me No, you're usually here\nSpeaker C: You're in the group Jerry Jerry Felder. Yeah, that's Yeah, yeah, that would be the bald guy. Oh my advisor\nSpeaker B: And um, so this is just an idea that's floating around and we'll see what happens and What other news do I have Yeah, we fix some more things from the smart com system, but that's not really a general interest Oh questions. Yeah, I'll ask Eva about the e-base and she's working on that. How is the generation XML\nSpeaker D: Thing I'm going to work on that today and tomorrow. Okay. No need to do it today or tomorrow. You can do it next week or I'm gonna finish it today. I hopefully Okay I want to do one of the things where I stay here because if I go home I can't finish it. I've tried About five times so far where I work for a while and then I'm like I'm hungry So I go home and then I think I'm not going back Yeah, that or I think to myself I can work at home and then I try to work at home, but I fail miserably Like I ended up at Blake's last night I'm not conducive. No, I almost got into a brawl But I did not finish the smart com but I've been looking into it. I said it's not like it's like a blank slate I found everything that I need and I It's the furthermore I told Jerry that I was gonna finish it before I get back so That's approaching. He's coming back when next. Oh, I think we think we'll see him definitely on Tuesday for the nice or no Wait the meek's are on Thursday\nSpeaker C: Maybe maybe okay, but we'll see him next week. All right That's good\nSpeaker B: Yeah The paper\nSpeaker D: Hmm I was thinking about that I think I Will try to work on the smart com stuff and I will if I can finish it today I'll help you with that tomorrow if you're working I don't have a problem with us working on it though\nSpeaker B: So okay, so you would say it's funky cool\nSpeaker D: I mean, it wouldn't hurt to write up a paper because I mean yeah, I was talking with Nancy and I said you don't know Whether you have a paper to write up until you write it up. So Well, since Jerry's coming back we can run it by him too, so yeah\nSpeaker B: What's your input?\nSpeaker C: Well, um, I don't have much experience with conference papers for competing the computer science realm So when I looked at what you had which was apparently a complete submission or just sort of said I didn't really know what to do with it like this is the sort of the basic outline of the system or whatever Or here's an idea, right? That's what that paper was here's here's one possible thing you could do Short eight pages and I just don't know What you have in mind for expanding like I what I didn't do is go to the website of the conference and look at what they're looking for or whatever\nSpeaker B: Well, it seems to me that\nSpeaker D: um Wait, is this a computer science conference or is it a\nSpeaker B: um Well, it's more it's both right it's it's sort of cognitive neuro psycholinguistic But all for the sake of doing computer science So it's sort of cognitive psycho Neuro plausibly Motivated architecture so natural language processing So it seems pretty interdisciplinary and I mean the keynote speaker is Tomasello Right so The question is what could we actually Do and and keep a straight face while doing it and I really can't keep a straight face doing anything My idea is well you can say we have done a little bit and that's it and Sort of the right position paper we want to also do that which is not too Good might be more interesting to do something like Let's assume We're right we have a steric closet a delusion of adequacy and take our where is accentance and say we will just talk about this and how we cognitively norally So Psycholinguistically construction grammar really Motivated in vision Uh, understanding that So we can actually show how we parse it That should be able to we should be able to come up with sort of a A parse It's on just just put it on\nSpeaker D: Yeah, that Ben harass you. Yeah, good\nSpeaker A: Yes\nSpeaker B: Oh, you're will suffer in hell you know\nSpeaker C: There's a Diagram somewhere which tells you how to put that\nSpeaker D: That you have to put it on exactly like that so put that those things over your ears like that See the pathopuss the things are I shouldn't like that there you go\nSpeaker A: Okay, it hurts it hurts real bad\nSpeaker C: But that's what you get for coming late to the\nSpeaker A: Okay, this is your mic on yeah Talking about\nSpeaker B: Um, we're talking about this and it paper that we made just sort of Oh, which gentleman yeah and I just sort of Brought forth the idea that we take a sentence where is the part of tower and we we Pretend to parse it we pretend to understand it and we write about it\nSpeaker C: And on how all of these things Okay, then we pretend to submit into a major international conference\nSpeaker B: Oh It's the whatever architectures You know where there's this conference the seventh already international conference on Neuroly cognitively motivated architectures of natural language processing And keynote speakers are Tomasalo Oh, Mike why me make when he make when he I think so interesting like child language people. Yep Okay, so maybe you want to write something too\nSpeaker A: Why are they thinking of it? It says it normally like Like dialogue systems or you know\nSpeaker B: Even Neuro and\nSpeaker A: Uh learning and like\nSpeaker B: Comprehension production you can look at the website Okay, and the deadline is a 15th of June 10 year time why we've got over a week It would be nice to go right two papers actually. Yeah, one one from your perspective and one from or a pivot\nSpeaker A: I mean that's the kind of thing that maybe like um the general Can sort of like N T L is to like whatever the previous simulation based version maybe you're talking on the same kind of thing a general paper about The approach here would probably be good. Yeah, good to do some point anyway. Yeah\nSpeaker B: Well, I also think that if we sort of Write about what we have done in the past six months We we could Sort of craft a nice little paper that if it gets rejected which could happen Doesn't hurt because it's something we having it as a good good thing. It's a nice exercise. It's I usually enjoy writing papers. It's not I don't regard it as a painful thing And we should all do more for our publication lists and never hurts and Keith and or John O will go probably Will I in case of So the 22nd of September in South Brook and Germany\nSpeaker A: Okay, so is the what you're just talking about\nSpeaker C: What would one possibly put in center?\nSpeaker B: What to write about What is on what's our take home message? What what do we actually because I mean it I don't like papers where you just talk about what you plan to do I mean It's obvious that we can do any kind of evaluation and have no you know, we can't write an ACL type paper We say okay, we've done this and now we're whatever percentage better than everybody else, you know It's part too early for that But we can tell them what we think I mean it's never heard to try And Maybe even that's maybe the time to introduce the the new formalism that you guys have cooked up\nSpeaker D: In the process, don't they need to finish the formalism? It's not even\nSpeaker A: Okay, so it's a little thing\nSpeaker D: We said it was 4,000 lines is that it?\nSpeaker B: I don't know. Did you look at it?\nSpeaker C: It depends on the problem. Oh my gosh. Oh, I thought you were I thought we were talking about something which was\nSpeaker B: No, that's I mean, there's actually a problem. It's difficult. It's more difficult to write on four pages than eight\nSpeaker A: Yeah, and it's also difficult to even if you had a lot of substance. It's hard to demonstrate that in four pages\nSpeaker C: Yeah, that would be hard\nSpeaker B: Well, maybe it's just 4,000 lines. I don't they don't want any they don't have a tech style They just want us key pure ask you Lines whatever why for whatever reason I don't know not including figures and such Very unspecific\nSpeaker D: Well, just let's say that's closer to six pages actually 4,000 lines of asking okay 4,000 lines. I mean\nSpeaker C: Isn't it isn't it about 50 55 60 lines to a page?\nSpeaker B: I don't quote me on this this is numbers I have from looking how many characters are online? Okay, let's let's watch what we should should we discuss this over T and all of us look at the web I can't I'm wizarding today Okay, look at the web page look at the web page and let's talk about it maybe tomorrow afternoon For us to find it are like neural journal world and you will think you have a link. Okay. I got an email\nSpeaker D: By the way Keith is comfortable with us calling a cool Keith\nSpeaker C: He decided I'm chilling in the 510 Yeah Thank you and I'm also from the 212\nSpeaker B: Sorry, yeah, I'm like to see sleep next In two weeks or not and a week of business in Germany I mentioned that for you and otherwise you haven't missed much except for a really weird idea But you'll hear about that\nSpeaker A: I do that you and I are you about that you already told me?\nSpeaker B: No, no, no Yeah, that is something for the rest of the game. The thing with the goats and the helicopters Change the watch bed. It's time to walk the sheep Um Did you catch that illusion? No Presumably one of the water gate codes Anyways Um Don't make any plans for spring break next year That I'm shooting Tell me you're gonna do an interview in the your internal workshop in Sicily\nSpeaker A: That's what that's what he said\nSpeaker B: I've already got the funding\nSpeaker A: You'll get your flyers there Yeah, that's what it means. Okay, cool. You'll put us up too. I know I know about that part. I know about the almond trees Oh Too easy coconut High-neighborcy mango\nSpeaker B: Too easy too easy I'm mango school everywhere so do kiwi okay. I was trying to find something that he didn't grow in his farm But cook another pineapple that's that's tricky. Yeah\nNone: Sorry So but we have to decide what like sort of the general I need is um I mean we're gonna have an example case Um, right the point is to like this where is case? Yeah, maybe you have it would be kind of the paper hat would have in my vision a nice flow\nSpeaker B: If we could say here is the Here's parsing if you want to do it right here is Understanding if you want to do it right and you know without going to the technical\nSpeaker A: And we're not doing like those things right. Yeah, right would that be clear in the paper or not?\nNone: That would be clear we would I Made around a little paper that I have yeah, we can sort of see this is No, I don't think you got it. See this if you if you're not around you do and for taking the discussion\nSpeaker A: Okay, come on So parsing done right so we can leave it say this is what it's sort of\nSpeaker B: state-of-the-art today Right and say this is bad. Yeah, and then we can say well what we do is this Okay\nSpeaker A: Parsing done right interpretation done right example\nSpeaker B: Mm-hmm\nSpeaker A: Yeah, and I'm actually getting to the cognitive neural part\nSpeaker D: That's the only that's the question mark. Don't you need to reduce it if it's a Or reduce it if it's a finite of neuro\nSpeaker A: I mean the conference may be cognitive neural doesn't mean that every paper has to be both\nSpeaker B: Yeah, and you can you can just point to the to the literature you can say the construction base\nSpeaker A: You know so this paper wouldn't particularly deal with that side although it could reference the anti-allage sort of like The fact that the meant is you're Where designed to be compatible with Yeah, I guess four pages you could I mean you could definitely It's definitely possible to do it It's just be small like introducing the formalism might be not really possible in detail\nSpeaker C: But you can look in a yeah looking at that paper that that you had I mean, you know like you didn't really Explained in detail what was going on in the XML cases whatever just said really you know Here's the general idea some stuff gets put in there You know hopefully you can you can say something like a constituent tells you what the construction is made out of you know Without going into this intensity Yeah Give him the one paragraph whirlwind tour of what this is for\nSpeaker B: This would be sort of documenting what we think and documenting what we have in terms of the base net and stuff and since this never a bad idea to document things Yes, that's definitely a good idea There would be my We should sketch out the details maybe tomorrow afternoon if everyone is around You know you probably Wouldn't be part of it. Maybe you want think about it Um You may may ruin your career forever\nSpeaker D: Hey you might give blacklisted and\nSpeaker B: The Other thing Yeah, we actually have we made any progress on what we decided last week I'm sure you read the transcript of last week's reading right so Up to date We decided that we're gonna take a where is something question and pretend we have parsed it and see what we could possibly Hope to observe on the first one\nSpeaker D: I came in and I started asking you about how we were sort of going to sort out the decision nodes\nSpeaker A: Yes\nSpeaker D: I remember you talking to me just that what you said Where it was like we needed to or in my opinion we need to design a base another sub-base net You know, it was whether it was whether we would have a base net of the output and on the input or whether the construction was gonna be in the base net Oh, yeah, and outside of it and\nSpeaker A: Was that the question was that what well that was it was related to what we were talking about\nSpeaker B: Should I introduce to the pseudo square? Yeah, sure We have to put this in the paper This is this is my only constraint So the pseudo square I Saw the diagram in the other situation user Discourse right autonomy oh my god, that's amazing No way way\nSpeaker C: So\nSpeaker A: So once you start making full calls\nSpeaker C: Oh god, what you know like su studio that horrible\nSpeaker D: I've blocked every aspect of football as out of my mind. Sorry. I haven't\nSpeaker C: He names in here\nSpeaker B: Oh, but also he's talking about suicide and that's that's not a notion I want to have it both\nSpeaker A: Yes, really I didn't really listen to it. I was too young. It sounds too rocking for that anyway\nSpeaker C: So what what so I wanted our constraints time and time\nSpeaker B: Okay, so we have tons of little things here and we can't believe that that's never been thought of before\nSpeaker D: What are the dots I don't remember what those little bugs\nSpeaker B: Okay, you know these are our whatever believe that decision notes and they all contribute to these Oh Things down here That's you that's you we ask When the moment it's a base net and it has sort of 50 not yet specified interfaces Okay, I have taken care that we actually can build little interfaces To other modules that will tell us whether the user likes these things and the all these things and he whether\nSpeaker A: He's in a wheelchair or not supposed to be the international sign for interface. I think so yeah\nSpeaker D: I'd never seen it before either\nSpeaker C: Because things fit onto that And frankly, I've seen fashion\nSpeaker B: No, this is a hard-eat corporate agent Design, I don't know\nSpeaker C: There's maybe a different so wait what are these letters again the situation use or discourse and what about the utterance\nSpeaker A: That's the discussion of discourse. Yeah, this course is all things linguistic yeah\nSpeaker B: So this this includes the the current utterance plus all the previous utterances And for example Irina Gouriri which is going to be here And of July she's a new linguist working for email and what she would like to do for example is great for us. She would like to Take the end to on toilet so We have discussed in terms of the Eva Think of back at the Eva vector and John O coming up with the idea that if the person discussed the Discuss the admission fee in previously that might be a good indication that How do I get to the castle actually he wants to enter?\nSpeaker B: You know how do I get to X Discussing the admission fee in the previous utterance is a good indication So we don't want a hard code a set of leg seams or things that persons No to filter or search the discourse history So what would be kind of cool is That if we encounter a concept such a castle tower bank hotel we run it through the ontology and the ontology tells us it has Admission opening times it has admission fees it has this it has that and then we we we make a thosaurus Lexicon look up and then search dynamic the through the discourse history for Currencies of these things in a different window of utterances And that might you know give us additional input to believe a versus B or key versus a\nSpeaker A: Okay, so you're looking for a few keys that you know are cues to Sorry, a few specific cues to some attention. You can dynamic look up keys. Yeah, oh, so wait\nSpeaker C: So um grip since since the sort of technical stuff is going over my head The point is that you that When someone's talking about a castle you know that it's the sort of thing that people are likely to want to go into or Is it the fact that if there's an admission fee then one of the things we know about admission fees is that you pay them in order to go in and then the idea of entering is active in the discourse or something and then\nSpeaker B: Well, blah blah the idea is even more general the idea is to say We encounter a certain entity in a in a in an utterance. So let's look up everything we the ontology gives us About that entity what stuff it does what roads it has what Parts whatever it has functions and then we look In the discourse whether any of that or any surface structure corresponding to these roads function has ever occurred And then the discourse history can tell us yeah or no, okay, and then it's up for us to decide what to do with it Okay, so So we may think that if you say Where's the theater Um whether or not he has talked about tickets before then we he's probably want to go there to See you something okay, where's the opera and perperes and lots of people go to the opera to take pictures of it and to look at it lots of people go to attend the performance and The discourse can maybe tell us what's more likely If we know what to look for in previous statements And so we can hard code for opera look for tickets do look for this look for that We look for Mozart look for this But the smarter way is to go via the ontology and dynamically then look up\nSpeaker C: Okay, but you're still doing look up so that when the person so the point is that when the person says Where is it then you sort of say Let's go back and look at Other things and then decide rather than the other possibility, which is that all through discourse as they talk about different things you know like Prior to the where is it question they say, you know how much does it cost to get in you know to see a movie around here um Where's it close to theater the the point is that by mentioning admission fees that just sort of stays active now You know that becomes part of like those sort of current ongoing active conceptual structure and then Over in your base net or whatever when when the person says where is it you've already got you know since they were talking about admission And that evokes the idea of entering um then when they go and ask where is it then your enter note is already active Because that's what the person is thinking about. I mean, that's the sort of cognitive linguistic way and yeah\nSpeaker B: Ultimately, that's also what we want to get at I think that's that's the correct way. So of course we have to keep Memory of what was the last intention and how does it fit to this and what does it tell us in terms of the what we're examining I have further more I mean we can idealize that you know people don't change topics But they do but right even the for that there's a student of ours who's doing a dialogue act um recognition module so Maybe we're even in a position where we can take your approach which is of course much better Just to say how much harder is this program?\nSpeaker B: Mm-hmm and much harder to program. Yeah, all these pieces fit together and And but okay, nevertheless, so these are issues, but what we actually decided last week is to this your benefit is to pretend we have observed and parsed an utterance such as where is the powder tower or where is the zoo and specify what what we think the the output uh observe out input notes for our baseness for the sub sub D for the discourse bit should be so that and I will I will then come up with the ontology side uh bits and pieces so that we can say okay we we always just look at this utterance that's the only utterance we can do it's hard coded like three needs sort of hand parsed handcrafted but this is what we hope to be able to observe in general from utterances and from ontologies and then we can sort of fiddle with these things to see what it actually produces in terms of output so we need to find out what the various x construction will give us in terms of semantics and sem spec type things okay just where's x or any variance of that no um look at it this way yeah what did we decide we decided sort of the the prototypical where is x where you know we don't really know does he want to go there or just want to know where it is so the difference of where is the railway station versus where where where is green net\nSpeaker D: as i was just dancing sir we're not videotypical you're this so\nSpeaker C: so um we're supposed to the i mean we're talking about sort of anything that has the semantics of request for location right actually or i mean anyway the node in the uh the ultimate uh in the base net thing when you're done the the node that we're talking about um is one that says request for location true or something like that right um and and exactly how that gets activated you know like whether we want the sentence how do i get there to activate that node or not you know that's that's sort of the\nSpeaker B: issue that sort of the linguistics side has to deal with right well actually more the other way around we wanted something that represents uncertainty what in terms of going there or just wanting to know where it is for example some to direct information and so this is prototypically founded the where is something question surface structure what which can be you know should be maps to something that activates both i mean the idea is to i don't know right okay let's have it fit nicely with the paper i don't see and how we would be\nSpeaker D: able to distinguish between the two intentions just from the utterance the i mean before before\nSpeaker B: we don't before we cranked it through the base net and we we wouldn't that's exactly what we want\nSpeaker D: we wouldn't no we wouldn't okay but then so basically it's just a very construction we have a node in the net right and we turn on yeah that node what what does this exactly what is the uh well given that we know that the construction has these two things we can set up probabilities we can basically define all the tables for every further it should be so we have\nSpeaker B: um let's assume we we call something like a lock x node and a path x node and what we actually get if we just look at the discourse where is x should activate or should should be both whereas maybe where is x located we find from the data is always just ask for the person once and no where it is and how do i get to is always asked when the person just wants to know how to get there right so we want to sort of come up with what gets input and how in case of a various question so what what what what the outcome of of your parser look like and what other discourse information from the discourse history could we hope to get squeeze out of that at once so define the the input into the base that based on what the utterance where is x gives us so definitely have an entity node here which is activated via the autology so where is x produces something that stands for x whether it's castle bank restroom toilet whatever and then the autology will tell us\nSpeaker A: that it has a location or the logical cells where actually it is located no not at all where it\nSpeaker B: is located we have a user proximity node here somewhere which tells us how far the user how far away the user is respect to that entity so you're talking about for instance the construction\nSpeaker A: obviously involves this entity or refers to this entity and from the construction also you know that it is a location is or a thing thing that can be located right autology says this thing has a location slot sure in that so thing that is being that is the content of the question that's being queried by one interpretation of where is x and another one is path from current user current location to that location so so is the question I'm not sure what the is the question for this particular construction how we specify that that's the information it provides or or castle work both sides right yeah you don't need to you do that it's just sort of what\nSpeaker B: what would be observed in that case observed when you heard the speaker say where is x or when\nSpeaker A: that's been parsed so these little circles you have by the d is that that's exactly what we're looking\nSpeaker D: for I just I don't like having characterizing the constructions with location and path or characterizing them like that because you don't it seems like in the general case you wouldn't know you wouldn't had to characterize them I mean or for when there could be an interpretation that we don't have a node for in the I mean it seems like make no more sense to have a node for the construction and then let the chips fall where they may versus uh saying this construction either can mean location or path and in this case since it can mean either those things relate both of those\nSpeaker C: up same thoughts questions I'm thinking about it it would be the same so I think\nSpeaker B: in here we have our goal there right and we have our info on so in my my case this would sort of make this happy and this would make the goal there happy what you're saying is we have a worries question worries node that makes both happy right that's what you're proposing but just my mind just as fine so if we have a construction node worries x it's gonna both get the posterior probability that it's informed up informed is true up and that goal there is true up as well which would be exactly analogous to what I'm proposing is this makes makes something here true and this makes something also something here true and this makes this true up and this makes this\nSpeaker C: true up as well I kind of like it better without an extra level of indirection to you know with this points to this point to that and so on because yeah because we get we get tons of\nSpeaker B: constructions I think because you know people have many ways of asking yeah sure I changed my mind\nSpeaker A: actually so I agree with that I have a different kind of question might be related which is okay so implicitly everything in you were always referring to the speaker intent right like what they want you know the information that they want for it's always information that they want probably some kind right all right this doesn't massage you or no okay so let's see so I don't know if the I mean if just there's more here that's not shown that you it's already like part of the system whatever but whereas X like the fact that it is you know speech act whatever it is a question it's a question that queries on some particular thing X and X is that location there's like a lot of structure in representing that yeah so that seems different from just having the no location X yeah it goes into it right pretty fast being this that's so that's that what you're\nSpeaker B: exactly we have some we have specified to okay the next one would be here just for mood the next one would be what we can squeeze out of the I don't know maybe we want to observe the the length of the words used and or the prosody and make conclusions about the user's intelligence\nSpeaker A: I don't know so in some ways in the other sort of parallel set of more linguistic meetings we've been talking about possible semantics of some construction right where it was the simulation that's according to you know that that corresponds to it and as well that it's discourse whatever contact discourse information such as the boot and you know other stuff so are we looking for a sort of abbreviation of that that's tailored to this problem because that that has you know basically you know it's in progress does in development still but definitely has various\nSpeaker B: features slots attributes bindings between things that's exactly why I'm proposing it's too early to have to think of them of all of these discourse things that one could possibly observe so let's just assume for the subset of human beings I'm not allowed to ask anything but where is X okay this is the only utterance in the world what could we observe from that okay that exactly where's X not\nSpeaker A: the the choices of where is X or how do I get to just where is X just where is X okay and but\nSpeaker B: you know do it do it in such a way that we know that people can also say is the town hall in front of the bank so that we need something like a WH focus should be should be there that you know this\nSpeaker A: whatever we do or do not take other kinds of constructions into account well if you if you can\nSpeaker B: okay that's not do where possible right if if if it's not at all triggered by our thing then it's irrelevant and it doesn't hurt to leave it out for the moment but okay it seems like for instance\nSpeaker A: where's X the fact that it might mean tell me how to get to X like so would you want to say that those two are both like those are the two interpretations right the the ones that are location or path so you could say that this construction is a question asking about this location and then you can additionally infer if they're asking about the locations because they want to go to that place in which case the you're jumping a skeptic step and saying oh I know where it is but I also know how to get they wanted seem they seem to want to get their song and I tell them so there's like structure\nSpeaker C: right this it's not it's not that this is sort of like semantically ambiguous between these two it's really about this but why would you care about this well it's because you also want to know this\nSpeaker A: or something so it's like you infer the speaker attend and then for a plan a larger plan from that for which you have the additional information you're just being extra helpful\nSpeaker B: think with this is just a mental exercise if you think about focusing on this question how would you design that is it do you feel confident about saying this is part of the language already to to detect those plans and why would anyone care about the location if not you know and so forth or do you actually I mean this is perfectly legitimate and I would not have any problems with erasing this and say that's all we can activate based on the utterance out of context right and then the the miracle that we get out the attention go there happens based on what we know about that entity about the user about this really believes goes to the address blah blah absolutely fine but this is sort of thing I propose that we think about so that we actually end up with um um notes for the discourse and ontology so that we can put them into our base net never change them so we all there is as well as x and if I can play around with the observed things and we can run our better travel base and have it produce some output and for the first time in in in the world we look at our output and um let's see whether it's any good you know I mean\nSpeaker C: here's hoping right across your fingers yeah I mean for me this is just a bit better of curiosity\nSpeaker B: I wouldn't I would like to look at what this at-hark process of designing a belief net actually produce if if we ask it where is something and maybe it also enables you to think about certain things more specifically um come up with interesting questions to which you can find interesting answers and additionally you might fit in really nicely with the paper because if we run an example for the paper I suggest there it is yeah so this might be a nice opening paragraph for the paper is saying you know people look at kinds of ambiguities and um in the literature verse bank and whatever kinds of garden path phenomenon and we can say well that's all nonsense hey these things are never really ambiguous in discourse be don't ever occur really in discourse but normal statements that seem completely unabiguous such as where is the blah blah actually art terribly complex and completely ambiguous and so whatever everybody has has been doing so far and you know has been completely nonsensical and can all go into the waste paper but it's always a good way to begin yeah\nSpeaker D: and the only I am great all other useless nice overture but you know just not really\nSpeaker B: okay I mean exactly but that might be you know saying hey you know some stuff is is actually complex if you look at it in in in the vacuum and and ceases to be complex in reality and some stuff that's that's absolutely straightforward in the vacuum is actually terribly complex in reality would be nice sort of also a nice um bottom up linguistics um type message versus the old top down scroll I'm running out of time okay when do you need to start wizarding at 410 okay this is the other bit of news the subjects today no face or she can't be here and do the wizarding so I'm gonna do wizarding until it's gonna do the instructing also we're getting a person who just got fired from her job person from Oakland who is interested in maybe continuing the wizard bit once they leaves in August and um she's gonna look at it today which is good news in the sense that if we want to continue after the third after July we can we could and um and that's also maybe interesting four keys and whoever if you want to get some more stuff into the data collection I mean there's we can completely change the setup anytime we want okay look at the results we've gotten so far for the first whatever 50 some subjects 50 had 50 so far no we're approaching 20 now but until phase yeah leaving we surely will hit some higher numbers and so that's cool do more fun stuff yeah I'll\nSpeaker C: have to look more into that data is that around like because that's pretty much getting posted or\nSpeaker B: something right away when you get it or it has to be transcribed we have found someone here who's hand transcribed the first 12 okay first thousand subjects just so we can build a language model for the regularizer okay but um so those should be available soon okay the first 12\nSpeaker C: and I can I mean you know that I that I looked at the first the first one and got enough data to keep me going for you know the first July so yeah but you can listen to all of them from your\nSpeaker B: Solaris box if you want it's always fun\n"}, {"meeting_id": "TS3003c", "summary": "The whole meeting was focused on the conceptual design of the new remote control product. After Project Manager briefly reaffirmed the agreements reached in previous meetings, Marketing, User Interface, and Industrial Designer each gave a presentation about trend-watching, interface design, and components design respectively. Then, Project Manager started a group discussion about important points just covered, including energy source, voice recognition, LCD screen, as well as case design, on which more emphasis was paid in the last half of the meeting. Finally, the group roughly drew out a specific case design.", "dialogue": "Speaker D: Just received a little beep.\nSpeaker D: I don't know what it was.\nNone: Okay.\nSpeaker D: Now good afternoon.\nSpeaker D: This is our third meeting already.\nSpeaker D: I hope you enjoyed your lunch.\nSpeaker D: I did anyway.\nSpeaker D: Let's see.\nSpeaker D: Presentation three.\nSpeaker D: Okay. This is the second phase we're going to discuss today.\nSpeaker D: It's the conceptual design meeting and a few points of interest in this meeting.\nSpeaker D: Are the conceptual specification of components, conceptual specification of design, and also trend watching.\nSpeaker D: These are hopefully the points you addressed in your presentations.\nSpeaker D: You're going to show me in a few minutes.\nSpeaker D: But first, I'll show you the agenda.\nSpeaker D: First the opening.\nSpeaker D: Then we have three presentations. After that we have to come to a decision on remote control concepts.\nSpeaker D: Are we going to make it? And then we're closing. We have about 40 minutes.\nSpeaker D: So I suggest let's get started.\nSpeaker D: Did someone encounter any problems during the preparation?\nSpeaker D: No. Everything fine. It's nice.\nSpeaker D: Then a little thing about the last meeting.\nSpeaker D: These are the points we agreed on. The requirements and the target market.\nSpeaker D: Our requirements are data text, docking station, audio signal, small screen with some extra button information.\nSpeaker D: And we're going to use default materials.\nSpeaker D: Does somebody have any comments on these requirements?\nSpeaker D: No. These are just the things we thought of.\nSpeaker D: So maybe if you figured something else or thought of something else, just let me know.\nSpeaker D: Maybe you can work it out.\nSpeaker D: And we're going to target the 60 to 80 year old customers.\nSpeaker D: So now everybody knows what we're doing.\nSpeaker D: I suggest let's get started with the presentations.\nSpeaker D: So should we keep the same line up last time?\nSpeaker D: Sure. Okay. I'll start off now. Good luck.\nSpeaker B: I'm going to inform you about the trend watching I've done over the past few days.\nSpeaker B: We've done some market research. We distributed some more on cats, questionnaires.\nSpeaker B: And besides that, I deployed some trend watchers to Milan and Paris to, well, get a hold of the news trends.\nSpeaker B: And I've consulted some additional trend watchers after the original trend watchers return about what the best design would be.\nSpeaker B: Okay. These are some overall findings.\nSpeaker B: The most important thing is the fancy design. The research indicated that that was by far the most important factor.\nSpeaker B: Inovativeness was about half as important as the fancy design. By inovativeness, this means functions which are not featured in other remote controls.\nSpeaker B: And half as important as the inovativeness was easy to use.\nSpeaker B: For our group, we're focusing on the people of 60 to 80 years old.\nSpeaker B: These factors are slightly more equal.\nSpeaker B: Okay. These are some more group specific findings. The older people prefer dark colors.\nSpeaker B: They like recognizable shapes and familiar material and our surveys have indicated that especially wood is pretty much the material for older people.\nSpeaker B: This image will give you a little bit of an impression about the look and feel that the remote should have.\nSpeaker B: This leads us to some personal preferences. The remote control and the docking station should blend in the room.\nSpeaker B: This would mean no eye-catching designs. Just keep it simple.\nSpeaker B: The docking station and the small screen would be your main points of interest because this would be the inovativeness in the remote control.\nSpeaker B: This would be very important that we at least include these features.\nSpeaker B: The trend watchers I consulted advised that remote control and the docking station should be telephone-shaped.\nSpeaker B: You could imagine that remote control will be standing up straight in the docking station.\nSpeaker B: This is pretty much a new shape to older people. They would prefer a design where the remote control just lies flat in the docking station.\nSpeaker B: It would be more telephone-shaped. Besides that, we would advise to bring two additions.\nSpeaker B: One with a wood-like color and maybe feel. One with a gray-black color.\nSpeaker B: The wood-like for the more exclusive people, people with more money. The gray-black color for well.\nSpeaker B: People with less means. That would be all.\nSpeaker D: Okay. Thank you.\nSpeaker D: Any questions?\nSpeaker D: I'll go to the next one.\nNone: Okay.\nSpeaker A: Yeah.\nSpeaker A: This is some research about designing of interface.\nSpeaker A: Last meeting we had about using a few buttons.\nSpeaker A: That's what I want to do in our design.\nSpeaker A: I found a way to control the remote control.\nSpeaker A: I found something about speech recognition. Maybe we can use that.\nSpeaker A: Yeah. And using a little display.\nSpeaker A: I found some of the findings.\nSpeaker A: Just to focus on the primary functions.\nSpeaker A: So only buttons for sound.\nSpeaker A: For one of shifting up again in the channel or that shifting down.\nSpeaker A: We need some new interactive functions which attract people for using it.\nSpeaker A: Like speech recognition.\nSpeaker A: And a special button for selecting a sub-title.\nSpeaker A: And overall user-friendly.\nSpeaker A: It's possible to make a quite cheap system for speech recognition.\nSpeaker A: You can think about when you lost your remote control.\nSpeaker A: You can call it and give it a sync.\nSpeaker A: And for shifting up a channel or for putting out a sound or something.\nSpeaker A: You can just give a sound off.\nSpeaker A: And the elevation of the sound off.\nSpeaker A: I was thinking about a special button for subtitles.\nSpeaker A: Just one button to keep it simple.\nSpeaker A: One push on the button.\nSpeaker A: You get small subtitles.\nSpeaker A: Double push.\nSpeaker A: Double click.\nSpeaker A: You get big subtitles for people.\nSpeaker A: Which can read smaller subtitles.\nSpeaker A: We have to keep in general buttons.\nSpeaker A: The buttons we have to use, the sound higher and lower.\nSpeaker A: The general button, more one button for shifting down a channel.\nSpeaker A: Also we want to use a little display for displaying the functions of the buttons.\nSpeaker A: We can build in a function which shows the channel.\nSpeaker A: A little picture of it.\nSpeaker A: We can put on off button in this corner.\nSpeaker A: All remote controls are using on off button on the place.\nSpeaker A: People will recognize the button.\nSpeaker A: This display of it is just a small display.\nSpeaker A: We can put it on top.\nSpeaker A: It is the most place where people most often look at.\nSpeaker A: A special button for shifting up and shifting down a channel.\nSpeaker A: On the top of the button, you can easily shift up or shift down.\nSpeaker A: Quite any place.\nSpeaker A: All the functions for subtitle.\nSpeaker A: We have to discuss about the form of it.\nSpeaker D: That's it.\nSpeaker D: Thank you.\nSpeaker C: The components design for the energy source, we can use a basic battery or as an optional thing, a kinetic energy like in a watch which you just shake and it produces energy.\nSpeaker C: If we choose for that option, the docking station would become obsolete.\nSpeaker C: I don't think it's really an option.\nSpeaker C: For the casing, the manufacturing department can deliver flat casing, single or double curved casing.\nSpeaker C: It's really up to the design that we're going to use.\nSpeaker C: It doesn't imply any technical restrictions.\nSpeaker C: As a case supplement, I thought of that later.\nSpeaker C: A rubber belt like anti-slip.\nSpeaker C: For the buttons, we can use plastic or rubber.\nSpeaker C: The chipset, it's as simple here, but it should be advanced because we're using an LCD screen.\nSpeaker C: As the trend watch presentation showed, people like wood, but it raises the price and it doesn't really fit the image unless we would start to product lines.\nSpeaker C: The form should follow function overall.\nSpeaker C: The kinetic energy source is rather fancy, but it depends on what we want.\nSpeaker C: I think we should discuss that.\nSpeaker C: For the case, the supplement and the buttons really depends on the designer and the chipset really should be advanced because otherwise it would really be a simple remote control.\nSpeaker C: That's it.\nSpeaker D: Okay.\nSpeaker D: Thank you.\nSpeaker D: That brings us to the discussion about our concepts.\nSpeaker D: These are the points we have to discuss.\nSpeaker D: First, I think we can talk about the energy source since it has a pretty big influence on production price and image.\nSpeaker D: First of all, we have to see it is possible to introduce kinetic energy.\nSpeaker C: There are four options.\nSpeaker C: We could use the basic normal battery, a hand dynamo, but I don't think that's really an option.\nSpeaker C: We don't want to swing before you can watch television.\nSpeaker C: Solar shells, but not every room is very light, so it's not a very good option, or the kinetic\nNone: energy. How exactly does the kinetic energy work?\nSpeaker B: You basically shake your remote and then it powers up.\nSpeaker B: Personally, I don't think that older people like to shake their remote control before they use it.\nSpeaker B: That's true.\nSpeaker B: Besides that, you mentioned it would make the docking station obsolete.\nSpeaker B: I think our docking station could be one of the marketing issues with which we could get great popularity for our product.\nSpeaker A: What's the function for allowing the batteries?\nSpeaker B: Yeah, you could load up batteries.\nSpeaker B: You could insert, find a lost remote control function in there.\nSpeaker A: I won't use much energy, I believe, just a small display, so I believe it will run on one battery for six months or more.\nSpeaker A: I believe one battery is just enough.\nSpeaker D: Well, I think elderly people just like to have everything in place, and I don't think they like remote, just laying everywhere in their rooms.\nSpeaker D: Maybe a docking station will help them give the remote a place.\nSpeaker D: Also, what you said, you can introduce a voice recognition by finding back-ear remote, but I think it's more efficient and cheaper to put it in the docking station.\nSpeaker D: You can put a little bit of a button on your docking station, which you can push, and then it starts beeping.\nSpeaker D: We can still use the voice recognition, but maybe then for only the channels.\nSpeaker B: I'm wondering, what will the voice recognition mean for the production price?\nSpeaker D: Yeah, that's a good point.\nSpeaker C: I don't have any informational pricing, so I'll have to ask the manufacturing department.\nSpeaker B: In our earlier market research, if you would allow me to push in the front board.\nSpeaker B: Go ahead.\nSpeaker B: So it was open here.\nSpeaker B: We also asked if people would pay more for speed recognition in a remote control.\nSpeaker B: You can see here our target group would not do that.\nSpeaker B: If they would increase the price for which we're selling our remote control, I would greatly advise not to do it.\nSpeaker B: I think it would be better to insert in our other product that is meant for the younger\nSpeaker C: people. That would also go for the LCD screen then, I guess.\nSpeaker C: It's a bit higher percentage.\nSpeaker B: Well, this is, yeah, but this is, here the question was, would you prefer it?\nSpeaker B: So that doesn't really mean they wouldn't pay extra for it.\nSpeaker B: And on top of that, the LCD screen would help in making the remote control easier to use.\nSpeaker B: Yeah, and I think the voice recognition function would not make the remote control much easier\nSpeaker D: to use. No, I think that's a good point.\nSpeaker A: But as Richard about a large LCD display or just a small one, we want to use.\nSpeaker B: Well, this was for like an LCD screen like you would have on the most advanced mobile phones.\nSpeaker B: Okay.\nSpeaker B: So pretty large.\nSpeaker B: Yeah.\nNone: Yeah.\nSpeaker D: I personally think the LCD screen we want to use with the extra information, I think nobody has anything against it because it's just some extra information and it's easy to ignore as well.\nSpeaker D: So if you don't want to use it, you just don't use it.\nSpeaker D: And yeah, I think maybe we have to discard the voice recognition because it will increase cost significantly.\nSpeaker D: And I don't think it will be a lot easier to use as well.\nSpeaker D: So that brings us back to the energy.\nSpeaker D: If we don't have voice recognition, it won't use a lot of energy to use.\nSpeaker D: So in that case, we could use kinetic energy.\nSpeaker D: But I think just a simple battery, which you can reload on a docking station, is just as good and much cheaper as well.\nSpeaker D: And that's the best choice.\nSpeaker D: Okay.\nSpeaker D: Let me just choose for the battery.\nSpeaker D: That brings us to the chip.\nSpeaker C: Well, there isn't any choice there because we're using the display.\nSpeaker C: So it's got to be advanced.\nSpeaker D: Okay.\nSpeaker D: That's fun.\nSpeaker D: Chip.\nSpeaker D: And then we get to the point of the case, which brings us a little bit back to marketing as well.\nSpeaker D: If you want to choose for wood or the black and gray or both.\nSpeaker D: As we saw, there is not, yeah, wood is a lot more expensive to produce.\nSpeaker D: But I think it will attract elderly people who want to have something exclusive, which they can show up to their grandkids.\nSpeaker D: Got a new remote control.\nSpeaker B: I know.\nSpeaker B: I think the most important factor there is the wooden collar.\nSpeaker B: So it wouldn't actually have to be wood.\nSpeaker B: Yeah, if it's just, that's right.\nSpeaker B: Wood-colored.\nSpeaker D: Wood-colored was a lot more expensive.\nSpeaker D: Or?\nSpeaker D: I don't know.\nSpeaker C: You don't know.\nSpeaker C: I'll have to.\nSpeaker C: I think so.\nSpeaker C: Because, yeah.\nSpeaker D: Probably.\nSpeaker D: It's a lot more difficult to handle and to get in the right shape.\nSpeaker A: It's possible to make changeable.\nSpeaker A: Okay.\nSpeaker A: You've got mobile phones.\nSpeaker A: Like Nokia mobile phones.\nSpeaker A: When you can change the case of it.\nSpeaker A: So it's the case.\nSpeaker A: Yeah.\nSpeaker A: Maybe it's possible.\nSpeaker A: Possibility.\nSpeaker A: So you have just to make one standard remote control.\nSpeaker A: And you can sell a few.\nSpeaker D: You can sell the case.\nSpeaker D: Yeah.\nSpeaker D: I think that's a very good option.\nSpeaker D: Because then you can advertise as well with the give your grandfather a new case for his remote control or whatever.\nSpeaker D: Because it's something extra.\nSpeaker D: It's something other remotes don't have.\nSpeaker D: Which we can get a greater advantage point.\nSpeaker D: So and then you can make a smooth collar, black and gray.\nSpeaker D: Other colors as well.\nSpeaker B: Yeah.\nSpeaker B: So you can put carefully into the design though.\nSpeaker B: Because we would have to make one control which would fit in with a wooden cover and a plastic cover or original one.\nSpeaker B: Or standard one.\nNone: Yeah.\nSpeaker D: So you suggest we should design two different telephones.\nSpeaker D: Or we can supply.\nSpeaker D: But remote control.\nSpeaker D: Which you can apply different case covers.\nSpeaker D: Well no.\nSpeaker B: For example.\nSpeaker B: We should just we should then just design one remote.\nSpeaker B: But it would have to be fancy with either the wood cover or the plastic one.\nSpeaker B: Okay.\nSpeaker B: So.\nSpeaker B: But that shouldn't be too much of a problem.\nSpeaker D: So everybody's okay with the changing covers.\nSpeaker D: That's a good option.\nSpeaker D: Changing case covers.\nSpeaker B: I heard our industrial designer talk about flat single and double curved.\nSpeaker B: Yes.\nSpeaker B: Could you explain that a little more?\nSpeaker C: Well the general like most older remotes are flat just straight.\nSpeaker C: And are the manufacturing department can also do they have a single curved or double curved cases.\nSpeaker C: And would single curves and double curved mean?\nSpeaker C: It would just only affect the form for far as I know.\nSpeaker C: So it's really just up to the design department what we're going to use.\nSpeaker C: It doesn't really matter for the price or the functionality.\nSpeaker B: So we can pretty much just pick one you like.\nSpeaker B: Okay.\nSpeaker B: That's good.\nSpeaker D: But the form has to be it has to be possible to stand up or just only to let.\nSpeaker B: Just lay down.\nSpeaker D: And the cover of the docking station is also on top of the television then or not.\nSpeaker B: Well or besides it.\nSpeaker D: And you can just yeah click it in.\nSpeaker D: That's okay.\nSpeaker D: So the interface what type of interface do we want to use.\nSpeaker D: Maybe you can make a little drawing of it on the board.\nSpeaker D: Somebody have ideas for a form or?\nSpeaker A: We can just use the regular form of it but it's not quite fancy.\nSpeaker B: You said you wanted to put the changing channels button on the right side so your thumb would be easily.\nSpeaker B: Well I think that was a very good point because I pointed out earlier that a lot of remote caused air as I said it would be great for that.\nSpeaker B: I thought maybe we could just make one of those buttons on both left and the right side.\nSpeaker B: For left handed users.\nSpeaker B: Also.\nNone: Yeah.\nSpeaker A: Is it possible to program it so you've got on the left side or on the right side buttons for shifting open shifting up.\nSpeaker A: Now on the other side buttons for shifting for the sound.\nSpeaker B: Well that could.\nSpeaker B: Yeah we could do that but I'm not sure if that would be very good for the ease of use.\nSpeaker D: Yeah ease of use will be a lot more difficult than that.\nSpeaker B: If we would make changing channels and changing volume button on both sides so it certainly yield great options for the design of the remote.\nSpeaker B: Because it could be made all symmetrical and stuff.\nSpeaker D: Yeah but you have extra buttons so people can get confused.\nSpeaker B: That is true.\nSpeaker D: Yeah.\nSpeaker D: Especially if they have the same writing so not.\nSpeaker A: Especially or you have to make left.\nSpeaker A: Contemaker.\nSpeaker D: Contemaker remote which you can flip over and use on the same functions as the normal one.\nSpeaker A: You mean.\nSpeaker A: Yeah.\nSpeaker A: Let's see.\nSpeaker D: Okay.\nSpeaker D: Blank one.\nSpeaker D: Then you get.\nSpeaker D: There is a little LCD screen.\nSpeaker D: Now I have to think.\nSpeaker D: It's a plus and a min.\nSpeaker D: No it's not very handy I think.\nSpeaker D: Because the plus and a min will be opposite and all kinds of.\nSpeaker D: No that's not going to work.\nSpeaker D: I guess.\nSpeaker D: Maybe you should.\nSpeaker D: Yeah.\nSpeaker D: But is it the problem that left handed persons use a different hand?\nSpeaker D: I think the functions are that basic that nobody should have any problems with choosing a channel.\nSpeaker A: That's true.\nSpeaker A: It's just you think you're a thumb.\nSpeaker D: Yeah.\nSpeaker D: I think we could just leave it in normal shape.\nNone: Yeah.\nSpeaker D: But maybe we have to make it a bit more fancy in one another way.\nSpeaker C: I think we should start by choosing a case because that's the basis you're building on.\nSpeaker D: Yeah.\nNone: Yeah.\nSpeaker C: So I could draw them out.\nSpeaker C: Let's get this flat case.\nSpeaker C: Oh.\nSpeaker C: From the side so.\nSpeaker C: Just right on normal.\nSpeaker C: Yeah.\nSpeaker C: The single girth so I'm not really sure what they're going to look like but I think it's something like this so this type should be better for you or better should prevent.\nSpeaker C: Better the straight injury a bit and the double girth looks something like this I guess.\nSpeaker C: So those are the three options we have.\nSpeaker D: Okay.\nSpeaker D: I should just the single girth because maybe the girth is pretty good to put the screen in so that elderly people can use the remote control and at the same time look easily at the screen because it's a bit.\nSpeaker D: That's a bit of an angle.\nSpeaker A: You say it's.\nSpeaker A: You got like a belief.\nSpeaker A: Yeah.\nSpeaker A: So you want to put this play over here or not?\nSpeaker D: Yeah.\nSpeaker D: I think so.\nSpeaker D: Yeah.\nSpeaker D: But now it's can make the upside down or do you have a.\nSpeaker A: Let's talk.\nSpeaker A: It's down.\nSpeaker A: Okay.\nSpeaker A: Maybe it's possible to make this site like.\nSpeaker A: See.\nSpeaker A: Right.\nSpeaker A: Like so in the form of your hand.\nSpeaker A: So.\nNone: Yeah.\nSpeaker A: So the remote control I have to.\nSpeaker A: Lay in your hand so it's possible.\nSpeaker A: Yeah.\nSpeaker A: So get your mouse.\nSpeaker A: Yeah.\nSpeaker A: And the.\nSpeaker A: And the buttons for.\nSpeaker A: Changing the channel over here.\nSpeaker D: Yeah.\nSpeaker D: That's a good one.\nSpeaker D: But I think it's better to put the screen on top.\nSpeaker D: So just flip it 180 degrees around.\nSpeaker A: And you get.\nSpeaker A: This place.\nSpeaker D: That's about here.\nSpeaker D: If you can have this one you turn it like this and then flip it upside down.\nSpeaker A: I don't know.\nSpeaker D: Because maybe your hand is in the way if you have to display here.\nSpeaker C: It's more logical to have it on top as well because I think it's like on your mobile phone.\nSpeaker C: It's always a ball.\nNone: So.\nNone: Yeah.\nSpeaker C: Maybe you should just start on a blank page.\nSpeaker D: Yeah.\nSpeaker D: So then we get.\nSpeaker D: This is the curve.\nSpeaker A: Life minutes.\nSpeaker D: And then we can.\nSpeaker D: And then the rest of the buttons over here.\nSpeaker A: But I'm the enough button still on the top.\nSpeaker D: Yeah.\nSpeaker B: And I'd prefer the corners to be round.\nSpeaker D: Okay.\nSpeaker D: Supplements.\nSpeaker D: Okay.\nSpeaker D: That's my mouse.\nSpeaker D: Then we've got the general idea of the concepts and the materials we're going to use.\nSpeaker D: So now for the next meeting we'll have to look at the look at field design.\nSpeaker D: It's important that the corporate design image is going to be in the remote.\nSpeaker D: So check out the corporate website maybe.\nSpeaker D: The user interface design is the same story and product evaluation.\nSpeaker D: So the industrial designer and user interface designer are going to work together on this one.\nSpeaker D: But you're going to get your instructions I think, send it by the coach.\nSpeaker D: So just I will put these minutes in the folder.\nSpeaker D: Okay.\nSpeaker D: And then we're going to try to finish our project and make a good design for all the grandfathers and grandmothers.\nSpeaker D: I think we should...\nSpeaker D: Let's see.\nSpeaker D: I'm not sure if you're going to start right away to work together or I think you're going to fill in the questionnaires first.\nSpeaker D: And then you'll get a message.\nSpeaker D: So that's basically it.\nSpeaker D: Maybe this one.\nSpeaker D: And you can save this one in the folders group.\nSpeaker D: It's here.\nSpeaker D: Smartboard.\nSpeaker D: There it is.\nSpeaker D: So if you want to have a look at it, it's over there in the project folder.\nSpeaker D: And then I guess we will start in 30 minutes again.\nSpeaker D: Thank you.\nSpeaker D: Good.\nSpeaker D: Okay.\nSpeaker D: Right on time.\nSpeaker B: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker A: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\n"}, {"meeting_id": "Bmr011", "summary": "The group discussed recording equipment and setup issues, recent developments in the transcription effort, other potential types of tagging to be assigned to transcribers, and the post-processing of waveforms. The discussion was largely focused on efforts to facilitate transcriptions, including the improvement of strategies for transcribing overlapping speech, and achieving greater uniformity in the type of equipment used during recordings and the manner in which recording devices are worn by speakers.", "dialogue": "Speaker G: Are we on?\nSpeaker C: We're on.\nSpeaker C: We're down.\nSpeaker G: Okay, so we can have a cent around the agenda.\nSpeaker G: Any agenda items that he has to talk about?\nSpeaker H: What's going on?\nSpeaker H: Has everyone met at dawn?\nSpeaker H: Yeah, doing.\nSpeaker G: Okay, agenda item one, introduce Don.\nSpeaker D: Well, I had a good question, but I know there was a discussion of it at a previous meeting that I missed, but just about the wish list item of getting good quality close packing mics on every speaker.\nSpeaker G: Okay, so let's do agenda building right now.\nSpeaker G: Okay, so let's talk about that a bit.\nSpeaker G: Close talking mics, better quality.\nSpeaker E: We can talk about that. You were going to start to say something.\nSpeaker E: Well, you already know about the meeting that's coming up, and I don't know if this is appropriate.\nSpeaker G: No, no, it's okay.\nSpeaker G: So we can talk.\nSpeaker G: So NIST folks are coming by next week, and so we'll talk about that.\nSpeaker G: Who's coming?\nSpeaker G: John Fiskis, and I think George Dunn and we'll be around as well.\nSpeaker G: Okay, so we can talk about that.\nSpeaker G: I guess just hear about how things are going with the transcriptions, right?\nSpeaker G: So, maybe, maybe this is going to discuss.\nSpeaker G: Anything else?\nSpeaker G: It's striking, but...\nSpeaker D: We've started running recognition on one conversation, but it isn't working yet.\nSpeaker D: But if anyone has... the main thing would be if anyone has knowledge about ways to post-process the waveforms that would give us better recognition, that would be helpful to know about.\nSpeaker K: Talk conversation.\nSpeaker I: Yeah, so what about, is there anything new with the speech? Non-speech?\nSpeaker B: No, I'm working on it, but it's not finished.\nSpeaker G: All right, that seems like a good collection of things, and we'll end up with other things.\nSpeaker E: I had thought under my topic that I would mention the four items that I put out for being on the agenda on that meeting, which includes the presugmentation of the developments and multi-trans.\nSpeaker G: Oh, under the NIST meeting.\nSpeaker G: All right, when I start off with this... I guess the order we wrote up seems fine.\nSpeaker G: So, better quality, close talking mics.\nSpeaker G: So, the one issue was that the lapel mic isn't as good as you would like, and so it'd be better if we had close talking mics for everybody, right?\nSpeaker G: Is that basically the point?\nSpeaker D: Yeah, and actually in addition to that, that the close talking mics are worn in such a way as to best capture the signal.\nSpeaker D: The reason here is just that for the people doing work, not on microphones, but on sort of like dialogue and so forth, or even on prosody, which Donna's going to be working on soon, it adds this extra variable for each speaker to deal with when the microphones aren't similar.\nSpeaker D: So, and I also talked to Mary this morning, and she also had a strong preference for doing that, and in fact she said that that's useful for them to know and starting to collect their data too.\nSpeaker D: Right, so one...\nSpeaker G: Well, one thing I was going to say was that we could get more of the head-mounted microphones, even beyond the number of radio channels we have, because I think whether it's radio or wire is probably second order in the main time.\nSpeaker G: So, the main thing is having the microphone close to the mic, not too close.\nSpeaker H: So, actually the way it was A's is correct.\nSpeaker H: So, it's not correct?\nSpeaker H: Yes, it's good.\nSpeaker H: So, it's in the corner of your mouth, so that the sound doesn't get on it, and then just sort of a thumb on the half away from your mic.\nSpeaker D: Right. But we have more than one type of, I mean, Princeton here.\nSpeaker H: This one isn't very adjustable, so this is good as I get it, instead of fixed.\nSpeaker D: But if we could actually standardize, you know, the microphones as much as possible, that would be really helpful.\nSpeaker G: Well, I mean, it doesn't hurt to have a few extra microphones around, so why don't we just go out and get an order of if this microphone seems okay to people?\nSpeaker H: I just get a half dozen of them. The only problem with that is right now, some of the gen ones aren't working.\nSpeaker H: The box is under the table.\nSpeaker H: And so, I've only been able to find three jacks that are working.\nSpeaker I: Can we get these?\nSpeaker D: No, but my point is...\nSpeaker D: I could just record these signals separately and timeline in what the start of the meeting.\nSpeaker H: Right.\nSpeaker G: Right now, we've got two microphones in the room that are not quote-unquote standard.\nSpeaker G: So why don't we replace those?\nSpeaker G: Okay, just two.\nSpeaker G: Well, however many we can plug in, if we can plug in three, let's plug in three.\nSpeaker G: Also, we've talked before about getting another radio.\nSpeaker G: And so then that would be three.\nSpeaker G: So we should go out to our full complement of whatever we can do, but have them all be the same mic.\nSpeaker G: I think the original reason that it was done the other way was because it was sort of an experimental thing.\nSpeaker G: And I don't think anybody knew whether people would rather have more variety or more uniformity.\nSpeaker H: Sounds like you're more into the lens.\nSpeaker D: Well, for short-term research, there's just so much effort that would have to be done upfront.\nSpeaker I: You're probably going to be great.\nSpeaker I: You're saying for dialogue purposes, so that means that the transcribers are having trouble with those mics, is that what you mean?\nSpeaker D: They would know more about the transcriber.\nSpeaker E: And that's true.\nSpeaker E: I mean, we could discuss this.\nSpeaker E: And a couple times.\nSpeaker E: So, yeah, the transcribers notice that in fact there's somewhere...\nSpeaker E: Well, I mean, it's the double thing.\nSpeaker E: It's the equipment and also how it's worn.\nSpeaker E: And they always...\nSpeaker E: They just write about how wonderful Adam's at his class.\nSpeaker D: So does the recognize there.\nNone: No, really.\nSpeaker D: Yeah, that's a problem.\nSpeaker D: I'm a fah!\nSpeaker D: Yeah, it's not just that.\nSpeaker D: It's your talking to someone else's mic.\nSpeaker E: It's not so loud with no breathing.\nNone: No, you know, it's like...\nNone: It's...\nNone: It's just...\nSpeaker E: The transcribers point at you and also the one he's just pointing...\nSpeaker H: The point of doing the close-talk mic is to get a good quality signal.\nSpeaker H: We're not doing any search-drunken close-talk mics.\nSpeaker H: So we might as well give it a good view.\nSpeaker G: Now, this is locked in the barn door after the horse was stolen.\nSpeaker G: We do have 30 hours of speech, which is on this way.\nSpeaker G: But, yeah, for future ones, we can get it a bit more uniform.\nSpeaker G: So I think we just do a field trip at some point.\nSpeaker G: Yeah, probably.\nSpeaker G: Yeah, to the store we talked about.\nSpeaker E: And there was some talk about maybe the headphones that are uncomfortable.\nSpeaker H: Yeah, so this is...\nSpeaker H: We'll do a field trip and see if all of the same mic that's more comfortable than these things, which I don't know.\nSpeaker C: Great.\nSpeaker C: Thank you very much.\nSpeaker D: It makes our table a lot easier.\nSpeaker H: Okay, we're researchers.\nSpeaker H: And we'll have big hits.\nSpeaker H: Yeah.\nSpeaker G: Okay, second item was the NIST visit.\nSpeaker G: It's going on there.\nSpeaker G: Okay, so...\nSpeaker E: John, this is coming on the second of February.\nSpeaker E: There's a lot of people here, not everyone.\nSpeaker E: And he expressed an interest in saying the room and in saying a demonstration on the modified multi-trans, which I'll mention in a second.\nSpeaker E: And also, it was interested in the pre-segmentation.\nSpeaker E: And then he's also interested in transcription conventions.\nSpeaker E: And so it seems to me in terms of like...\nSpeaker E: Okay, so the room is things like the audio, audio, audio, acoustic properties, the room and how the recordings are done and that kind of thing.\nSpeaker E: Okay, in terms of the multi-trans, well, that's being modified by Dave Gilbert to a hand-roll multi-channel recording.\nSpeaker H: I see this thing mentioned in my interview to this meeting.\nSpeaker H: I've got to do it.\nSpeaker C: Yeah, okay.\nSpeaker C: Sorry.\nSpeaker C: Yeah, well that's a game.\nSpeaker C: Yeah, it looks really great.\nSpeaker E: He has a prototype.\nSpeaker E: I didn't see it yesterday, but I didn't see it today.\nSpeaker E: And that's that will enable us to do nice, tight, time-marking of the beginning and ending of overlapping segments at present.\nSpeaker E: It's not possible with limitations of the original design and software.\nSpeaker E: And so in terms of like pre-segmentation, that continues to be a terrific asset to the transcribers.\nSpeaker E: Do you...\nSpeaker E: I know that you're also supplementing it for the do you want to mention something about that, too?\nSpeaker B: Yeah, what I'm doing right now is I'm trying to include some information about which channel there's some speech in, but that's not working at the moment.\nSpeaker B: I'm just trying to do this by comparing energies, normalizing energies and comparing energies of the different channels.\nSpeaker B: And so to give the transcribers some information in which channel there's the speech.\nSpeaker B: In addition to the thing we did now, which is just speech non-speech detection on the mixed file.\nSpeaker B: So I'm relying on the segmentation of the mixed file, but I'm trying to subdivide the speech portions into different portions if there is some activity in different channels.\nSpeaker B: Excellent.\nSpeaker E: So it'd be like providing also a speaker idea.\nSpeaker G: Something I guess I didn't put in the list, but on that same day later on, and maybe it's...no, actually, this week, they've go over it and I will be visiting with John Canny, who is a CS professor who's interested in array microphones.\nSpeaker G: Oh, he's doing array.\nSpeaker G: Thanks.\nSpeaker G: And so we want to see what commonality there is here.\nSpeaker G: Maybe they'd want to stick an array of microphones when we're doing things.\nSpeaker G: Or maybe it's not a specific array microphone they want, but they might want to just...you could imagine them taking the four signals from these table mics and trying to do something with them.\nSpeaker G: I also had a discussion.\nSpeaker G: So we'll be over there talking with him after the class I'm fighting, which you know what goes on with that.\nSpeaker G: I also had a completely unrelated thing I had a discussion today with Berger Kolmeier, who's a German scientist.\nSpeaker G: He's got a fair-sized group doing a range of things, sort of a lot of Tory related, largely for hearing aids and so on.\nSpeaker G: But does stuff with auditory models and he's very interested in directionality and location and head models and paper film things.\nSpeaker G: He and possibly a student, there's a student of his who gave a talk here last year, may come here in the fall for sort of a five month sabbatical.\nSpeaker G: So it might be around to get him to give some talks and so on.\nSpeaker I: So anyway, he might be interested.\nSpeaker I: There reminds me of a thought of an interesting project that somebody could try to do with the data from here.\nSpeaker I: Either using the mics on the table or using signal energies from the head worn mics.\nSpeaker I: And that is to try to construct a map of where people were sitting.\nSpeaker H: Oh, did he?\nSpeaker H: Oh, that's interesting.\nSpeaker H: So that's the cross-correlation stuff.\nSpeaker I: And so you could plot out who was sitting next to who?\nSpeaker G: I mean, he didn't do a very extreme thing, but it was a sort of given that the block of wood with the two mics on either side.\nSpeaker G: If I'm speaking, or if you're speaking or someone over there speaking, you're looking across correlation functions.\nSpeaker G: If someone was on the axis between the two is talking, then you get a big peak there.\nSpeaker G: And if someone's talking on one side or the other, it was the other way.\nSpeaker G: It even looks different if the two people are on the other side are talking, then if one in the middle, it actually looks different.\nSpeaker I: So I was just thinking, you know, as I sit here next to Tilo, when he's talking, my mic probably picks it up better than your guy's's mic.\nSpeaker I: So if you just looked at the energy on my mic, you could get an idea about who's closest to who.\nSpeaker I: Or who talks the last.\nSpeaker G: Yeah, well, you have to, the appropriate normalizations are tricky.\nSpeaker D: You just search for Adam's voice on each individual mic.\nSpeaker G: Yeah, we switched positions recently.\nSpeaker G: So those are just a little couple of news items.\nSpeaker E: And then one thing, so John Fiskus expressed an interest in microphone arrays is there.\nSpeaker E: And I also want to say his e can't stay all day, he needs to leave for from here to make a 245 white.\nSpeaker E: No, it's just morning from all night.\nSpeaker E: Right.\nSpeaker E: Makes a sketch on a little bit type.\nSpeaker E: But do you think that John Canny should be involved in this somehow or not?\nSpeaker G: Probably not, but I'll know better after I see him this Friday.\nSpeaker G: What kind of level he wants to get involved.\nSpeaker G: He might be excited to and it might be very appropriate for him to or he might have noticed once or ever.\nSpeaker G: I guess really don't know.\nSpeaker H: Is he involved in, I'm blanking on the name of the project.\nSpeaker H: NIST has done a big meeting room, instrumented meeting room with video and microphone arrays and very elaborate solvers.\nSpeaker H: Well, that's what they're starting up.\nSpeaker H: Okay.\nSpeaker G: Yeah, no, I mean, that's what all this is about.\nSpeaker G: They haven't done it yet.\nSpeaker G: Okay, they wanted to do.\nSpeaker H: The papers that look like they had already done some work.\nSpeaker G: Well, I think they've instrumented the room, but I don't think they haven't started recordings yet.\nSpeaker G: They don't have the transcription standard.\nSpeaker G: Are they going to do a video as well?\nSpeaker G: Yeah, I think.\nSpeaker H: I think so.\nSpeaker H: What I had read was a fairly large amount of software infrastructure recording, and also live room where you're interacting with the computer and with the video and lots of stuff.\nSpeaker G: Well, I'm not sure.\nSpeaker G: All I know is that they've been talking to me about a project that they're going to start out recording people meeting and meetings.\nSpeaker G: And it is related to ours.\nSpeaker G: They were interested in ours.\nSpeaker G: They wanted to get some uniformity with us about the transcriptions and so on.\nSpeaker G: And one notable difference, actually, I can't remember whether they were going to routinely click video or not.\nSpeaker G: But one difference from the audio side was that they are interested in using raymines.\nSpeaker G: So, you know, you just tell you the party line on that.\nSpeaker G: The reason I didn't go for that here was because the focus, both the buy interest and the madam's interest, was in impromptu situations.\nSpeaker G: And we're not recording a bunch of impromptu situations, but that's because it's different to get data for research than to actually imply it.\nSpeaker G: And so, for scientific reasons, we thought it was good to instrument this room as we wanted it.\nSpeaker G: But the thing we ultimately wanted to aim at was a situation where you were talking with one or more other people in an impromptu way, where you didn't actually know what the situation was going to be.\nSpeaker G: And therefore, it would not be highly unlikely that room would be outfitted with some very carefully designed array of microphones.\nSpeaker G: So, it was only for that reason.\nSpeaker G: You know, yet another piece of research seemed like you had enough to know portable array of mic.\nSpeaker G: No, so there's a whole range of things. There's a whole array of things that people do on this.\nSpeaker G: So, the big arrays, places like Rutgers and Brown and other places, they have big arrays with 100 mikes or something.\nSpeaker G: And so, there's a wall of mics.\nSpeaker G: You get really, really good beampointing that sort of thing.\nSpeaker G: And in fact, at one point, we had a proposal in with Rutgers where we were going to do some of the sort of per-channel signal processing, and they were going to do multi-channel stuff.\nSpeaker I: I've seen demonstrations of the microphone array. It's amazing.\nSpeaker I: Yeah, how they can cut out noise.\nSpeaker G: And then they have the little ones.\nSpeaker G: They don't have a block of wood.\nSpeaker G: Yeah, our block of wood is unique.\nSpeaker G: But no, there are these commercial things now. You can buy that have four mics or something.\nSpeaker G: So, yeah, there's a range of things that people do.\nSpeaker G: So, if we connect that with somebody who was interested in doing that sort of thing, that's a good thing to do.\nSpeaker G: I mean, whenever I've described the people who are interested on the acoustic side, that's invariably the question they ask.\nSpeaker G: Just like someone who's interested in the general dialogue thing, we always ask, are you recording video?\nSpeaker G: Right.\nSpeaker G: And you could see people always say, well, are you doing a ray of microphone?\nSpeaker G: So, it's a good thing to do.\nSpeaker G: But it doesn't solve the problem of how do you solve things when there's one mic or at best two mics in this imagined VBA that we have.\nSpeaker G: So, maybe we'll do some more of that.\nSpeaker E: I mean, I know that having a ray of imagined would be more expensive than a ray of mic and mic and mic and mic.\nSpeaker E: But couldn't you kind of approximate the natural situation by just shutting off channels when you're later on?\nSpeaker E: I mean, it seems like if the mic and mic falls down and affect each other, then couldn't you just record them within a ray and then just not use all the data?\nSpeaker H: It's just a lot of infrastructure for our particular purpose, recalculate the center.\nSpeaker G: Yeah, if 99% of what you're doing is shutting off most of the mics and going through that.\nSpeaker G: But if you get someone who has that as a primary interest then that drives the right answer.\nSpeaker G: That's right.\nSpeaker H: Someone came in and said, you really want to do it?\nSpeaker H: I mean, we don't care.\nSpeaker I: So, to save that data, you have to have one channel recording per mic in the array?\nSpeaker G: Well, at some level, at some level, but then, you know, there's...\nSpeaker I: What you save, I mean, if you're going to do research with that.\nSpeaker G: I don't know what they're going to do and I don't know how big their array is.\nSpeaker G: Obviously, if you were going to save all of those channels for later research, you'd use up a lot of space.\nSpeaker H: Well, their software infrastructure had a very elaborate design for plugin helpters and mixers and all sorts of processing so that they can do stuff in real time and not save that each channel.\nSpeaker H: Yeah, that was...\nSpeaker G: But I mean, for optimum flexibility later, you want to save each channel, but I think in practical situations, you would have some engine of some sort doing some processing to reduce this to the equivalent of a single microphone that was very directional.\nSpeaker I: Oh, okay.\nSpeaker I: Sort of saving the result of the beam plugin.\nSpeaker D: It seems to me that there's, you know, there are good political reasons for doing this, just getting the data because there's a number of sites like...\nSpeaker D: Right now, SRI is probably going to invest a lot of internal funding into recording meetings also, which is good, but they'll be recording with video and they'll be...\nSpeaker D: You know, it'd be nice if we can have at least make use of the data that we're recording as we go.\nSpeaker D: This is the first site that has really collected these really impromptu meetings and just have this other information available.\nSpeaker D: So if we can get the investment in just the infrastructure and then, I don't know, save it out or have whoever's interested, save that data out, transfer it there, it would be good to have the recording.\nSpeaker H: You mean to actually get a microphone or a radio?\nSpeaker H: Well, even if we're not...\nSpeaker D: I'm not sure about video that sort of a video has a little different nature since, right now we're all being recorded, but we're not being taped.\nSpeaker D: But definitely in the case of microphone array since if there's a community interested in this.\nSpeaker H: Well, I think we need to reset our hears interested in it.\nSpeaker F: It's a bunch of long.\nSpeaker G: See, the problem is it took at least six months for Dan to get together the hardware and the software and debug stuff and the microphones and the boxes and it was a really big deal.\nSpeaker G: And so I think we could get a microphone array in here pretty easily and have it mixed to one channel of some sort.\nSpeaker G: But I think for...\nSpeaker G: I mean, how we're going to decide...\nSpeaker G: For maximum flexibility later, you really don't want to end up with just one channel that's pointed in the direction of the person with the maximum energy or something like that.\nSpeaker G: I mean, you want actually to have multiple channels being recorded so that you can...\nSpeaker G: And to do that, we're going to end up greatly increase the this space that we use up.\nSpeaker G: We also only have boards that will take up to 16 channels and this meeting we've got eight people and six mics in there were already using 14.\nSpeaker H: And we actually only have 15.\nSpeaker D: 15, 16. Well, if there's a way to say time to sort of solve each of these those...\nSpeaker D: So suppose you can get an array in because there's some person that Berkeley who's interested and has some equipment.\nSpeaker D: And suppose we can...\nSpeaker D: As we save it, we can transfer it off to some other place that holds this data.\nSpeaker D: Who's interested, even if it's the itself isn't.\nSpeaker D: And it seems like as long as we can time align the beginning, do we need to mix it with the rest? I don't know.\nSpeaker G: Yeah, so I think you need a separate set up and the assumption that you could time align the two.\nSpeaker D: I mean, it's just... it's worth considering as sort of once you make the upfront investment and can sort of save it out each time and not have to worry about the dis space factor than it might do with having the data.\nSpeaker G: It's not so much worried about this space. I mentioned that as a practical matter, but the real issue is that there's no way to do a recording extended to what we have now with low skew.\nSpeaker G: So you would have a completely separate set up, which would mean that the sampling times and so forth would be all over the place compared to this.\nSpeaker G: So it would depend on the level of processing you're doing later.\nSpeaker G: The kind of person who's doing array processing you actually care about funny little times.\nSpeaker G: So you actually would want to have a completely different set up than we have one that would go up to 32 channels or something.\nSpeaker G: So basically...\nSpeaker G: So I'm kind of skeptical.\nSpeaker G: But I don't think we can share the resource in that way.\nSpeaker G: What we could do is if there's someone else that's interested, they could have a separate set up which they wouldn't be trying to sync with ours, which might be useful for...\nSpeaker G: And then we can offer up the rule.\nSpeaker G: Yeah, we can offer the meetings and the physical space and... yeah, the transcripts.\nSpeaker D: Right, I mean, it'd be nice if we have more information on the same data.\nSpeaker D: Yeah.\nSpeaker D: But if it's impossible or if it's a lot of effort then you have to just balance it too.\nSpeaker G: Yeah, the thing will be... and talking to you, these other people to see what we can do.\nSpeaker I: Is there an interest in getting video recordings for these meetings?\nSpeaker I: Right, so we had some...\nSpeaker H: It's actually the same problem that you have an infrastructure problem.\nSpeaker H: You have a problem with people not wanting to be video-tabied with your problem.\nSpeaker H: And no one who's currently involved in the project is really hot to do it.\nSpeaker I: So there's not enough interest to overcome it.\nSpeaker D: Not entirely, but I know there is interest from other places that are interested in looking at meeting data and having the video.\nSpeaker D: So it's just...\nSpeaker E: Yeah, well, well, I have to mention the human subjects problems that increase with video.\nSpeaker E: Right.\nSpeaker G: Yeah, so people getting shy about it, there's this human subjects problem, there's the fact that then...\nSpeaker G: I heard comments about this.\nSpeaker G: Well, why don't you just put on a video camera?\nSpeaker G: You know, it's sort of like saying, well, we're primarily interested in some dialogue things.\nSpeaker G: But why don't we just throw a microphone out there?\nSpeaker G: I mean, the thing is, once you actually have serious interest in any of these things, then you actually have to put a lot of effort in.\nSpeaker G: And you really want to do it right.\nSpeaker G: So I think NIST or LBC or something like that, I think is much better shape to do all of that.\nSpeaker G: There will be other meeting recordings.\nSpeaker G: We won't be the only place to be meeting recordings.\nSpeaker G: We're doing what we're doing, and hopefully it will be useful.\nSpeaker E: It's pretty much done in the scientific form.\nSpeaker H: Probably not.\nSpeaker H: Did you sign up?\nSpeaker H: What did you do actually?\nSpeaker H: Did you read in the industry?\nSpeaker J: Yeah, I was...\nSpeaker I: You were here to meet you before.\nSpeaker J: I was here before once.\nSpeaker J: You were here to sign up for it.\nSpeaker J: Did you sign up for it?\nSpeaker J: I think so.\nSpeaker J: I'll get it done before we meet.\nSpeaker E: Thank you.\nSpeaker E: Yeah.\nSpeaker E: You don't have to leave me behind.\nSpeaker E: I just...\nSpeaker F: I can't really leave you behind.\nSpeaker F: You're being recorded, isn't it?\nSpeaker C: We don't perform electroshock during this meeting.\nSpeaker C: That's fine.\nSpeaker C: Usually.\nSpeaker F: Okay.\nNone: Transcriptions.\nSpeaker G: Okay.\nSpeaker E: I thought about the many three aspects of this.\nSpeaker E: So first of all, I got eight transcribers.\nSpeaker E: Seven of them were linguists.\nSpeaker E: One of them was a graduate student in psychology.\nSpeaker E: I gave each of them their own data set.\nSpeaker E: Two of them have already finished the data sets.\nSpeaker E: And the meetings around, you know, it would say an hour.\nSpeaker E: Sometimes it's not just an hour and a half.\nSpeaker E: What I mean is one meeting.\nSpeaker E: Each person got their own meeting.\nSpeaker E: I don't want to have any conflicts of, you know, when to stop transcribing this one.\nSpeaker E: So I wanted to keep it clear whose data was.\nSpeaker E: And so...\nSpeaker E: And meetings, you know, I think that they go as long as two hours in some cases.\nSpeaker E: And that means, you know, if we've got two already finished and they're working on right now, all of them have additional data sets.\nSpeaker E: That means potentially as many as ten might be finished by the end of the month.\nSpeaker E: Also, the pre-signitation really helps a huge number.\nSpeaker E: And also, Dan Ellis's innovation of the multi-channel to hear really helped a lot, in terms of clearing up your earrings that evolve over labs.\nSpeaker E: But just out of curiosity, I asked one of them how long it was taking here.\nSpeaker E: One of these two is already finished her data set.\nSpeaker E: She said it takes about 60 minutes transcription for every five minutes of real time.\nSpeaker E: So it's about 12 to one, which is what we were making.\nSpeaker E: It's well in the finish.\nSpeaker E: Okay.\nSpeaker E: At least still, when they're finished, that means that they're finished with their pastor.\nSpeaker E: They still need to be edited and all, but it's word-level, the speed for change, the things that we were mentioned.\nSpeaker E: Okay. Now I wanted to mention the teleconference I had with John Fiskus.\nSpeaker E: I was sold for an hour and a half, and had not bought a lot of things in common.\nSpeaker E: He indicated to me that he's been spending a lot of time with, quite sure the connection, but spending a lot of time with the Atlas system.\nSpeaker E: And I guess that, I mean, I need to read up on that, and there's a website that has lots of papers.\nSpeaker E: But it looks to me like that's the name that has developed a system that birthed the leave room and developed for the annotated graphs approach.\nSpeaker E: So what he wants me to do, and what we will do, is to provide them with an RE transcribe meeting for him to be able to experiment with in this Atlas system.\nSpeaker E: And they do have some sort of self-draight, my impression, related to Alice, and that he wants to experiment with taking our data and putting in that form and see how that works out. I explained to him in tail the conventions that we're using here in this word-level transcript. And I explained the reasons that we were not coding more elaborately, and they focus on reliability.\nSpeaker E: He expressed a lot of interest in reliability. He's really up on these things.\nSpeaker E: He's very independently, yes, well, what about reliability?\nSpeaker E: He's interested in the consistency of the encoding and that sort of thing.\nSpeaker D: So can you explain what the Atlas and that move?\nSpeaker E: Well, at this point, I think, well, Adam's read more in more detail than I have on this.\nSpeaker E: I need to equate myself more with it, but there's a way of viewing, whenever you have coding categories, and you're dealing with taxonomy, then you can have branches that have alternative choices that you could use for each of them.\nSpeaker E: And it just sends up looking like a graph or a presentation.\nSpeaker H: Is Atlas the annotated transcription graph stuff?\nSpeaker H: I don't remember the acronym. The one I think you're referring to, they have this concept of an annotated transcription graph representation.\nSpeaker H: And that basically, when I based the format that I did, I based it on their work almost directly in combination with the TEI stuff. And so it's very, very similar.\nSpeaker H: And so it's a data representation and set of tools for manipulating transcription graphs of various types.\nSpeaker I: Is this the project that's sort of between NIST and a couple of other places?\nSpeaker I: I think you know the CIFA stuff.\nSpeaker E: Yeah, yeah.\nSpeaker E: There's a website that has lots of papers. I'll look through them and they've mainly had to do with this tree structure annotated tree.\nSpeaker E: And I have a thing.\nSpeaker E: So, in terms of the conventions that I've adopted, there's no conflict at all.\nSpeaker E: And he was very interested in, oh, how did you handle this one?\nSpeaker E: I said, well, you know, this way. And we had a really nice conversation.\nSpeaker E: Okay, now I also wanted to say a different direction is Brian Kingsbury.\nSpeaker E: So I correspond briefly with him.\nSpeaker E: He still has an account here. I told him he could SSH on use multitrans and have a look at the already done transcription.\nSpeaker E: And he did. And what he said was that what they'll be providing is will not be as fine-grained in terms of the time information.\nSpeaker E: And that's, you know, I need to get back to him and explore that a little bit more and see what they'll be giving us.\nSpeaker I: So, the specific piece of the folks that they're subcontracting out the transcription to are they like court reporters?\nSpeaker E: Apparently, well, I get the sense they're kind of like that. It's like a pool of somewhat secretarial.\nSpeaker E: I don't think that they're court reporters. I don't think they have a special keyboard and that type of training.\nSpeaker E: I get the sense they're more secretarial. And that what they're doing is giving them medical transcriptionist types people.\nSpeaker I: So, it's for their speech recognition products. So, they're hiring them. They're coming in. It's not a service they send the tapes out to.\nSpeaker H: Well, they do send it out, but my understanding is that that's all that's coming. It's transcription.\nSpeaker H: So, most of it is via voice people reading. I see. They're trying to real support them.\nSpeaker E: I see. After that, it's been monologues for us. And what they're doing is, Brian himself downloaded.\nSpeaker E: So, Adam, something to see, Brian himself downloaded. We wanted to have it so that they were familiar with terms.\nSpeaker E: What they wanted to do. He downloaded from CD on to audio tapes. Apparently, he did it one channel per audio tape.\nSpeaker E: So, each of these people is transcribing from one channel. And then what he's going to do is check it.\nSpeaker E: Before they go beyond the first one, check it.\nSpeaker G: So, each person gets one of these channels. So, if they hear something off in the distance, they don't...\nSpeaker G: They just go...\nSpeaker H: That's okay because you'll do all of them in the end of the mind.\nSpeaker I: But there could be problems, right? Like that.\nSpeaker E: I think it would be difficult to do it that way. I really...\nSpeaker I: Well, if you got that channel right there...\nSpeaker H: No, no, close talk.\nSpeaker H: Not the desktop. Are you?\nSpeaker E: Yes. I sure am. I really foolish to do otherwise.\nSpeaker E: I think it would be hard to come up with.\nSpeaker D: I think it's hard just playing the...\nSpeaker D: Just having played the individual files.\nSpeaker D: I mean, I know what your voice sounds like. I'm familiar with it.\nSpeaker D: It's pretty hard to follow, especially...\nSpeaker D: There are a lot of words that are so reduced phonetically that make sense when you know what the person was saying before.\nSpeaker D: Yeah, that's it.\nSpeaker D: Especially to define these where you are in...\nSpeaker H: We've had this discussion many times. The answers we don't actually know the answer because we haven't tried both ways.\nSpeaker E: Well, except I can say that my transcribers use the mix signal mostly.\nSpeaker E: Unless there's a huge disparity in terms of the volume of the mix, in which case, you know, they...\nSpeaker E: They wouldn't be able to catch anything except from the channel.\nSpeaker H: That might change in one of really fine time markings.\nSpeaker H: Well, okay.\nSpeaker D: But they're not giving really fine markings.\nSpeaker D: So, are they giving any time markings?\nSpeaker D: I'm not even asking.\nSpeaker D: And I stress my email to you and that needs to be with Compa.\nSpeaker E: But I didn't want to say that it's hard to follow one channel of a conversation even if you know the people.\nSpeaker E: And if you're dealing furthermore with a highly abstract network concepts you've never heard of.\nSpeaker E: One of these people was transcribing the network's group talk of the city.\nSpeaker E: I don't really know what a lot of these abbreviations are.\nSpeaker E: But I just put them in parentheses.\nSpeaker E: I just don't know if you're interested in that.\nSpeaker H: Just that of curiosity.\nSpeaker H: I mean, a lot of heavy accents.\nSpeaker I: Given all of the effort that is going on here in transcribing, why do we have IBM doing it?\nSpeaker I: Why not just do it all ourselves?\nSpeaker G: It's historical.\nSpeaker G: I mean, at some point ago we thought that, boy, we'd really have to ramp up to do that.\nSpeaker G: Like we just did.\nSpeaker G: And here's a collaborating institution that's volunteering to do it.\nSpeaker G: So that was a contribution they can make in terms of time, money.\nSpeaker G: I'm just wondering now.\nSpeaker I: I'm wondering now.\nSpeaker D: Yeah, my heart is asking the same question as sort of talking about more e-mail layers.\nSpeaker D: Yeah.\nSpeaker F: Yeah.\nSpeaker G: So, let's see.\nSpeaker G: I mean, I think they've proceeded long a bit.\nSpeaker E: Let's see what comes out of it and have some more discussions with them.\nSpeaker E: It's very, a real benefit having Ryan involved because his knowledge of what the\nSpeaker H: have a data need to be used and so what is useful to have. Yeah.\nSpeaker D: So Liz with the SRI recognize or can it make use of some time marks?\nSpeaker D: I think this is what Don has been, he's already been really helpful in chopping up these.\nSpeaker D: So first of all, before the SRI front end, we really need to chop things up into pieces that are not too huge.\nSpeaker D: But second of all, in general, because some of these channels, I'd say like, I don't know, at least half of them, probably on average, have a lot of cross talk.\nSpeaker D: It's good to get sort of short segments if you're going to do recognition, especially forced alignment.\nSpeaker D: So, Don has been taking a first stab actually using James first, the first meeting that James transcribed, which we did have some problems with and Tilo, I think told me why this was, but that people were switching microphones around in the very beginning.\nSpeaker D: So, yes, right.\nSpeaker B: And they were not switching them, but they were adjusting them.\nSpeaker B: So, after a minute or so, it's way better.\nSpeaker D: So, we have to sort of normalize the front end and so forth and have these small segments.\nSpeaker D: So, we've taken that and chopped it into pieces, based always on your cuts that you made on the mix signal.\nSpeaker D: And so, every speaker has the same cuts, and if they have speech in it, we run it through.\nSpeaker D: And if they don't have speech in it, we don't run it through and we face that knowledge on the transcription.\nSpeaker D: The problem is if we have no time marks, then for forced alignment, we actually don't know where, you know, in the signal, the transcriber heard that word.\nSpeaker D: And so, if it's a whole conversation and we get a long, you know, paragraph of talk, I don't know how they do this, we actually don't know which piece goes where.\nSpeaker I: Well, you need to, like, a forced alignment before you do the chopping, right?\nSpeaker D: No, we use the fact that, so when Jane transcribes in the way she has transcribers doing this, whether it's with a pre-segmentation or not, they have a chunk and then they transcribes the words in the chunk.\nSpeaker D: And maybe they choose the chunk or now they use a pre-segmentation and then correct it, necessary, but they're supposed to chunk and then a transcription, then a chunk and a transcription.\nSpeaker D: That's great, because the recognizer can...\nSpeaker H: It's all to be good sized for the right recognizer roles.\nSpeaker D: Right, and it helps that it's made based on sort of heuristics and human ear.\nSpeaker D: But there's going to be a real problem, even if we chop up based on speech silence, these, the transcripts from IBM, we don't actually know where the words were, which segments they belong to.\nSpeaker I: That's sort of what we're worried about. A forced alignment. That's what she's saying, is that they can't.\nSpeaker D: If you do a forced alignment on something really...\nSpeaker D: Well, even if you do it on something really long, you need to know...\nSpeaker D: You can always chop it up, but you need to have a reference of which words went with which chop.\nSpeaker D: So...\nSpeaker G: I think that they are...\nSpeaker G: Yeah, I'm sure they will, and so we have to have a dialogue with them.\nSpeaker D: It sounds like we just have concerns.\nSpeaker D: Maybe actually there is some, even if they're not fine-grained, maybe the transcribers, I don't know, maybe it's saved out in pieces or something, that would help.\nSpeaker D: But it's just an unknown right now.\nSpeaker E: I need to know a right to it. I just think it was that I got over 10.\nSpeaker D: But it is true that the segments... I haven't tried the segments that T-Lo gave you, but the segments that in your first meeting are great.\nSpeaker D: That's a good length.\nSpeaker D: Good size, good size.\nSpeaker E: I was thinking, would you find it to win line?\nSpeaker E: Give us a pre-signitation.\nSpeaker E: Maybe you have one already, at the meeting that the first transcrib meeting, the one that I transcried?\nSpeaker B: Sure, I have some, but that's the one where I'm training on.\nSpeaker E: Oh, I see.\nSpeaker B: A little bit of dog soup.\nSpeaker D: And actually, as you get transcribes for new meetings, we can try...\nSpeaker D: The more data we have to try the alignments on the better.\nSpeaker D: So it'd be good for... just to know as transcriptions are coming through the pipeline from the transcribers, we're playing around with parameters on the recognizer.\nSpeaker D: That would be helpful.\nSpeaker D: Especially as you get more voices.\nSpeaker D: The first meeting had, I think, just four people.\nSpeaker E: It wasn't nice, but it was suddenly on Tuesday.\nSpeaker E: And I was planning to do just a preliminary look over a two that are finished and then give it to you.\nSpeaker E: Okay.\nSpeaker C: That's great.\nSpeaker G: I guess the other thing, I can't remember if we discussed this in the meeting, but I know you and I talked about this a little bit.\nSpeaker G: There was an issue of...\nSpeaker G: Suppose we get in the... I guess it's in a vehicle position, although it is just saying where the weak link is in the chain.\nSpeaker G: Where we have all the data transcribed and we have these transcribers and we're still a bit slow on feeding.\nSpeaker G: At that point we've caught up and the weak link is recording meetings.\nSpeaker G: Okay.\nSpeaker G: Two questions come is, you know, how do we...\nSpeaker G: It's not really a problem one because we haven't reached that point, but how do we step out the recorded meetings?\nSpeaker G: And the other one is, is there some good use that we can make of the transcribers to do other things?\nSpeaker G: So I can't remember how much we talked about this in this meeting, but...\nSpeaker E: And there is one use that also we discussed, which was when Dave finishes the...\nSpeaker E: And maybe it's only been the modification to a multitrans which will allow fine grain encoding of overlaps, then it would be... These people would be very good to shift over to finer grain encoding of overlaps just when providing.\nSpeaker E: So right now you have two overlapping segments in the same time, and one would be improvement in the database, and in the answering interface, be possible to just do a click and drag thing and get the specific place of each of those at the time of the time.\nSpeaker E: So say, at the beginning of each segment.\nSpeaker G: Right, so I think we talked about three things. One was, it's had some discussion that has to have some very high level...\nSpeaker G:...labelaps, types of overlaps and so forth, that someone could do.\nSpeaker G: Second was, some lower level of just doing these more precise timings.\nSpeaker G: And the third one is just completely wild airbrained idea that I have, which is that if we have time and people are able to do it, take some subset of the data and do some very fine grain analysis of the speech, for instance, marking in some overlapping, potentially overlapping fashion, the value of articulatory features.\nSpeaker G: You know, just sort of say, okay, it's voiced from here to here, there's nasal from here to here, and so forth.\nSpeaker G: And as opposed to doing phonetic analysis and assuming articulatory feature values for those things, obviously it's extremely time consuming. That would be really valuable.\nSpeaker G: But we could do it on some small subset.\nSpeaker E: Also, could you do anyone's consciousness that would be easier than a balance with it?\nSpeaker E: I don't think that being able to code that there's a fricative extended from here to here would be a lot easier than classifying precisely which a vowel that was.\nSpeaker E: I think a balanced bowser, I think harder.\nSpeaker G: Well, yeah, but I think also it's just the issue that when you look at the switchboard, for instance, very close up, where places where whether it's a consonant or a vowel, you still have trouble calling it a particular phone at that point.\nSpeaker G: Because there's this movement from here to here and it's...\nSpeaker I: You're saying sort of remove the high level of constraints and go bottom up.\nSpeaker G: Yeah, describe it now. I'm suggesting articulatory features.\nSpeaker G: Maybe there's even a better way to do it, but that's sort of a traditional way of describing these things.\nSpeaker G: And I mean, actually, this might be a neat technique to talk to.\nSpeaker I: Acoustic features versus psychological categories.\nSpeaker G: I mean, it's still some sort of categories, but something that allows for overlapping change of these things.\nSpeaker G: And then this would give some more groundwork for people who are building statistical models that allow for overlapping changes, different timing changes, as opposed to just click.\nSpeaker G: You're now in a state which, of course, allows to this speech.\nSpeaker D: So this is like gestural.\nSpeaker G: And actually, if we get into that, it might be good to haul John O'Hall into this, as his views on it.\nSpeaker D: But is the goal there to have this on meeting data so that you can do far-field stories of those gestures?\nSpeaker D: Or is it because you think there's a different kind of actual production in meetings that people use?\nSpeaker G: No, I think it's for that purpose.\nSpeaker G: I'm just viewing meetings as being a neat way to get people talking naturally.\nSpeaker G: And then it's natural in all senses, in the sense that you have microphones that are at a distance that one might have.\nSpeaker G: And you have the close mics, and you have people talking naturally.\nSpeaker G: And the overlap is just a digger of the fact that people are talking naturally.\nSpeaker G: So I think that given that it's that kind of corpus, if it's going to be a very useful corpus, if you say, OK, we've limited the use by some of our sensor choices.\nSpeaker G: We don't have a video, we don't, so forth.\nSpeaker G: But there is a lot of use that we could make of it by expanding the annotation choices.\nSpeaker G: And most of the things we've talked about have been fairly high level, and kind of a bomb up person.\nSpeaker G: I thought there'd be some of the others.\nSpeaker E: It's a really nice offer, those things that might range.\nSpeaker G: Hopefully someone would make use of it. I mean, people made a lot of use of Timit and did its markings.\nSpeaker G: And then the switchboard transcription thing, I think it's been very useful for a lot of people.\nSpeaker D: I guess I wanted to make a pitch for trying to collect more meetings.\nSpeaker D: Yeah.\nSpeaker D: Actually, I talked to Chuck Filmer, and I think they've what vehemently said no before, but this time he wasn't vehement, and he said, well, Liz, come to the meeting tomorrow and try to convince people.\nSpeaker D: So I'm going to try, go to their meeting tomorrow and see if we can try.\nSpeaker D: Because they have something like three or four different meetings.\nSpeaker D: And we have very interesting meetings from the point of view of a very different type of talk than we have here, and definitely than the front-end meeting, probably.\nSpeaker I: In terms of the topic.\nSpeaker D: Well, yes, and in terms of the fact that they're describing abstract things, and it just dialogue-wise, right?\nSpeaker D: So I'll try.\nSpeaker D: And then the other thing is, I don't know if this is at all useful, but I asked Lila if I can maybe go around and talk to the different departments in this building to see if there's any groups that, for free lunch, if we can still offer that.\nSpeaker D: You're not going to see nonexionic, academic, government, I don't know.\nSpeaker H: The problem is so much that it's done as confidential. It would be very hard for them.\nSpeaker H: Also, I think it takes its way to the point of that.\nSpeaker E: I mean, it seems like we had this idea before of having like linguistic students brought down for a few lectures.\nSpeaker D: Right, and we could also, we might try advertising again, because I think it would be good if we can get a few different sort of non-internal types of meetings, and just also more data.\nSpeaker I: Does John O'Hill have weekly phonetic reactions?\nSpeaker D: So I actually wrote to him and he answered, great, that sounds really interesting, but I never heard back, because we didn't actually advertise openly.\nSpeaker D: I told I'd asked him privately.\nSpeaker D: And it is a little bit of a trek for campus folks.\nSpeaker H: It would be nice if we got someone other than me who knew how to set it up to do the recording, so I didn't have to.\nSpeaker D: Exactly, and I was thinking.\nSpeaker C: He's supposed to be trained.\nSpeaker D: Plus, we could also get a student, and I'm willing to try to learn.\nSpeaker D: I mean, I would do my best.\nSpeaker D: The other thing is that there's a number of things at the transcription side that transcribers can do, like dialogue, tagging, this fluency tagging, things that are in the speech that are actually something we're working on for language modeling.\nSpeaker D: And Mari is also interested in Andreas as well.\nSpeaker D: If you want to process utterance, the first thing they say is, well, and that well is coded as some kind of interrupt tag, and things like that.\nSpeaker E: Of course, some of that can be done like so clearly.\nSpeaker E: And I also heard a lot of utility done to tagging.\nSpeaker D: Great, so a lot of this kind of, I think there is a second pass, and I don't really know what would exist in it, but there's definitely a second pass worth doing to maybe encode some kinds of, you know, is it a question or not?\nSpeaker D: That maybe these transcribers could do.\nSpeaker E: Maybe really good.\nSpeaker E: Well, while we're interested in just briefly to this question of more meeting data, two questions.\nSpeaker E: One of them is Jerry Falman's group.\nSpeaker E: They know that they recorded one meeting.\nSpeaker G: I think they're open to it.\nSpeaker G: I think all these things, I think there's, we should go beyond XE, but I mean there's a lot of stuff having an XE that we're not getting out of.\nSpeaker D: Okay, I thought that all these people had sort of said no twice already.\nSpeaker G: No, no, no, so there was the thing in Film Wars Group, but even there he hadn't what he'd said no to was for the main meeting, but they have several smaller meetings a week.\nSpeaker G: And the notion was raised before that that could happen, and just, you know,\nSpeaker I: and the other thing too is when they originally said no, they didn't know about this post-editing capability.\nSpeaker G: Yeah, so I mean there's possibilities there. I think Jerry's group, yes.\nSpeaker G: There's the network's group.\nSpeaker G: Do they still meet regularly?\nSpeaker H: Well, I don't know if they meet regularly or not.\nSpeaker H: They're no longer reporting.\nSpeaker G: But I mean, have they said they don't want to anymore?\nSpeaker H: What was his name?\nSpeaker H: Yeah, when with him gone, it sort of tripled off.\nSpeaker F: Okay, so they're down to three or four people, but the thing is three or four people is okay.\nSpeaker E: We might be only hearing this.\nSpeaker H: Well, he was sort of like contact, so I just need to find out who's running in the house.\nSpeaker E: Okay.\nSpeaker E: I see that Leva has a much meeting period.\nSpeaker D: Yeah, I mean, one thing that would be nice, and it sounds bizarre, but I'd really like to look at, to get some meetings where there's a little bit of heated discussion like arguments, or emotions, and things like that.\nSpeaker D: And so I think if there's any like Berkeley political groups, I mean that would be perfect.\nSpeaker D: So, yes, we might.\nSpeaker H: Who's really good at running in the street?\nSpeaker F: Well, and I saw that in the political party.\nSpeaker F: Yeah, with potential use from the defense department.\nSpeaker D: Well, maybe student groups or filmmakers or something, a little bit of color.\nSpeaker E: If we give them a chance to excite later, we might end up with like five minutes out of the club.\nSpeaker D: I don't mean that they're angry, but just something with some more variation in prasadic contours and so forth would be neat.\nSpeaker D: So, if anyone has ideas, I'm willing to do the legwork to go try to talk to people, but I don't really know which group there were.\nSpeaker E: What is this pursuing idea?\nSpeaker G: Yeah, there's a problem there in terms of the commercial value of the press.\nSpeaker G: It turned out to be a bit of a problem.\nSpeaker E: And I had one other aspect of this, which is John Biscuits expressed a major interest in having meetings which were all English speakers.\nSpeaker E: Now, he wasn't trying to shape us in terms of what we gathered, but that's what he wanted me to show him.\nSpeaker E: So, I'm giving him our initial meeting because he asked for all English.\nSpeaker E: And I think we don't have a lot of all English meetings right now.\nSpeaker E: Did he mean that non-British?\nSpeaker E: No, if he meant non-British, I think he said British was okay.\nSpeaker I: British is okay.\nSpeaker G: British is okay.\nSpeaker G: Well, I don't think if he didn't say that.\nSpeaker E: I bet he meant native speaking for a better.\nSpeaker F: I bet he did.\nSpeaker F: Oh, really?\nSpeaker F: That's why I wouldn't care.\nSpeaker I: Knowing the population.\nSpeaker G: I remember studying the BBN where they trained on, this was in Wall Street Journal days, or something they trained on American English, and then they tested on different native speakers from different areas.\nSpeaker G: And the worst match was people whose native tongue was Mandarin Chinese.\nSpeaker G: The second worst was British English.\nSpeaker G: That's fine.\nSpeaker F: So, it's...\nSpeaker F: German was much better.\nSpeaker F: It was Swiss.\nSpeaker G: So, I think if he's thinking in terms of recognition kind of technology, I think he would probably want to...\nSpeaker G: Are we on the issue?\nSpeaker G: Yeah, unless we're going to train with the whole country.\nSpeaker E: I think that the elements may be more that way, and they sort of feel like they have...\nSpeaker G: Maybe, so.\nSpeaker H: Yeah.\nSpeaker H: And maybe there were a few with us where Dan wasn't there in for us, as he worked on it.\nSpeaker G: It's pretty tough, this group.\nSpeaker G: So, what are the people who are involved in some artistic endeavor?\nSpeaker G: I mean, filmmaking.\nSpeaker G: Is that great?\nSpeaker D: I think that they would be...\nSpeaker D: Something where there is actually discussion, where there's no right or wrong answer, but...\nSpeaker D: But it's a matter of opinion, kind of thing.\nSpeaker D: Anyway.\nSpeaker D: Yeah, I do.\nSpeaker I: You're also going to get to have a political discussion.\nSpeaker I: Any department that calls itself science?\nSpeaker I: Yeah, I'm going to get a computer science.\nNone: I'm going to say...\nSpeaker D: I'm actually serious because...\nSpeaker D: Yeah, yeah.\nSpeaker D: We have the setup here, and that has...\nSpeaker D: Chance to give us some very interesting fun data.\nSpeaker D: So, if anyone has ideas, you know any groups that are...\nSpeaker D: Well, I guess some of the students...\nSpeaker D: Student groups like clubs, things like that.\nSpeaker G: A little ad-up saying, come here and argue.\nSpeaker D: If you're really angry, someone use our conference room.\nSpeaker H: The business school might be good.\nSpeaker H: I actually spoke with some students up there, and they express willingness back when they thought they would be doing research on speech.\nSpeaker H: But when they lost interest in speech, they also stopped answering my email about other stuff.\nSpeaker G: Or people who are wrong.\nSpeaker G: What about tax cuts or something?\nSpeaker G: I heard that.\nSpeaker G: How tech they have a special room.\nSpeaker K: Someone said that it's a special room to get all your frustrations out.\nSpeaker K: You can go through a little throw up things and break things.\nSpeaker K: So, we don't know if that is not actually what we...\nSpeaker D: Yeah, to that extent.\nSpeaker D: Well, a far-fledged likes can pick up where the recruits are.\nSpeaker F: But we don't want them to throw the far-fledged mixes.\nSpeaker H: Please throw everything in that direction.\nSpeaker H: And itself.\nSpeaker H: I think I think it looked good.\nSpeaker H: There was a dorm room at Tech that someone had coded the walls and the ceiling and the floor with mattresses.\nSpeaker H: The entire room.\nSpeaker G: I had this my fourth thing here, processing of waveforms.\nSpeaker G: What did we mean by that?\nSpeaker H: Liz wanted to talk about menacing, improving accuracy.\nSpeaker D: Well, I think that was just sort of an IRA-ass deal.\nSpeaker D: Oh, where did that?\nSpeaker D: But it would be helpful if I can stay in the loop somehow with people who are doing any kind of post-processing, whether it's to separate speakers or to improve the signal, to noise ratio, or both, that we can sort of try out as we're running recognition.\nSpeaker D: So, is that...\nSpeaker D: Who else's work, I guess, Dan Ellis?\nSpeaker D: Yeah.\nSpeaker G: And Dave.\nSpeaker G: And Dave.\nSpeaker G: Again, he's interested in, in fact, we're looking, starting to look at some macro-canceration kind of things.\nSpeaker G: Okay.\nSpeaker H: Which is how much that is.\nSpeaker H: An issue with a close talking mic.\nSpeaker G: Well, isn't that what you want?\nSpeaker D: I don't know.\nSpeaker G: No, so what you want, when you're saying improving waveform, you want the close talking microphone to be better.\nSpeaker G: And the question is, to what extent is it getting hurt by any room acoustic source, it's just given that it's close, it's not a problem?\nSpeaker D: It doesn't seem like big room acoustic problem to my ear, but I'm not an expert.\nSpeaker D: It seems like a problem with crosstalk.\nSpeaker H: I bet there's plenty of room acoustic.\nSpeaker D: That may be true, but I don't know how good it can get either by those methods.\nSpeaker D: That's true.\nSpeaker H: I think it's just, yeah, what you said.\nSpeaker D: All I meant is just that as sort of, as this pipeline of research is going on, we're also experimenting with different ASR techniques, and so it'd be good to know about it.\nSpeaker I: So the problem is like, on the microphone of somebody who's not talking, they're picking up signals from other people, and that's...\nSpeaker D: Right, although if they're not talking, using the in-house transcriptions, we're sort of okay because no one transcribed any words there, and we throw it out.\nSpeaker D: But if they're talking at all, and they're not talking the whole time, so you get some speech, and then some more speech, so that whole thing is one chunk.\nSpeaker D: And the person in the middle who said only a little bit is picking up the speech around it, that's where it's a big problem.\nSpeaker E: You know, this does seem like it would relate to someone one of those ASR, who's working on this wall, and coding them.\nSpeaker E: And he also...\nSpeaker D: The energy, right?\nSpeaker E: Exactly.\nSpeaker E: I was trying to remember you had this interface where you showed us one time when you left off that you had different visual displays.\nSpeaker A: Yeah.\nSpeaker A: I only displayed different colors for the different situation, but for me, for my problems, it's enough, because it's possible in a simple view to compare with the equipment, the kind of assessment, what happened with the different parameters, only with different bands of colors for the few situations I consider for acoustic even, is enough to...\nSpeaker A: I see that you are considering now a very sophisticated set of graphics, as symbols to transcribe, no?\nSpeaker A: A lot.\nSpeaker A: Because before you are talking about the...\nSpeaker A: the possibility to include in the transcribed word program, a set of symbol, or a fixed symbol, to market the different situation during the transition.\nSpeaker E: So symbols for differences between lab and cell, and...\nSpeaker E: Yes.\nSpeaker A: The symbol you talk before, no?\nSpeaker A: To...\nSpeaker E: To mal...\nSpeaker E: The symbols so much.\nSpeaker E: The main change that I see in the interface is just that we'll be able to...\nSpeaker E: to more money in time and things.\nSpeaker E: But I...\nSpeaker E: There was another aspect of you, I was thinking about this topic, which is that it's not much to me, but it's not to me, so partly inels, is that you're doing involves taking segments, which are of a particular type, and putting them together.\nSpeaker E: And so if you have like a...\nSpeaker E: You know, a speech from one speaker, then you cut out the part that's not that speaker, and you combine segments from that same speaker to...\nSpeaker E: and run them through the recognize as that.\nSpeaker D: Well, we try to find as close of start and end time as we can to the speech from an individual speaker, because then we're more guaranteed that the recognizer will...\nSpeaker D: for the forced alignment, which is just to give us the time boundaries, because from those time boundaries then the plan is to compute prosotic features.\nSpeaker D: And the sort of more space you have that isn't the thing you're trying to align, the more errors we have.\nSpeaker D: So, you know, that it would help to have either a pre-processing of a signal that creates very good signal noise ratio, which I don't know how possible this is for the lapel, or to have closer time, you know, synced times, basically, around the speech that gets transcribed or both.\nSpeaker D: And it's just sort of an open world right now of exploring that.\nSpeaker D: So I just wanted to see, you know, on the transcribing end, from here, things look good.\nSpeaker D: The IBM one is more... is an open question right now, and then the issue of like global processing of some signal, and then, you know, before we chop it up, is yet another way we can improve things.\nSpeaker I: What about increasing the flexibility of the alignment?\nSpeaker I: Do you remember that thing that Michael Fink had did? That experiment? He did a wild bath?\nSpeaker D: Right. You can...\nSpeaker D: The problem is just that the acoustic, when the signal to noise ratio is too low, you'll get an alignment with the wrong duration path.\nSpeaker D: So that's the problem, is the... Yeah, it's not the fact that you have like...\nSpeaker D: I mean, what he did is allow you to have words that were in another segment, move over to the... at the edges of segmentation.\nSpeaker D: Or even words inserted that weren't there.\nSpeaker D: Right. Things here, the boundaries, where if you got your alignment wrong, because what they had done there is a line and then chop.\nSpeaker D: And this problem is a little bit more global.\nSpeaker D: It's that there are problems even inside the alignments, because of the fact that there's enough acoustic signal there to recognize or to eat as part of a word, and it tends to do that.\nSpeaker D: So...\nSpeaker D: But we'll probably have to do something like that in addition.\nSpeaker D: Anyway, so yeah, bottom line is just I wanted to make sure I can be aware of whoever's working on these signal processing techniques for detecting energies because that'll really help us.\nSpeaker G: Okay, T is started out there, I suggest we went through our digits.\nSpeaker G: So...\nSpeaker G: Transcript 3031-3050.\nSpeaker G: 0368.\nSpeaker G: 04960.\nSpeaker G: 17050293462.\nSpeaker G: 7204640.\nSpeaker G: 8415281.\nSpeaker G: 9.\nSpeaker G: 0937762.\nSpeaker G: 0.\nSpeaker G: 109376028.\nSpeaker G: 425495.\nSpeaker G: 677890.\nSpeaker H: Transcript 30713090.\nSpeaker H: 004217337.\nSpeaker H: 4325.\nSpeaker H: 5639826.\nSpeaker H: 670740.\nSpeaker H: 810.\nSpeaker H: 051406.\nSpeaker H: 623964.\nSpeaker H: 507971.\nSpeaker H: 7176171.\nSpeaker H: 8503872.\nSpeaker H: 9704299.\nSpeaker H: 099093.\nSpeaker I: Transcript 3111-3130.\nSpeaker I: 2878024.\nSpeaker I: 304080.\nSpeaker I: 6316884.\nSpeaker I: 76685989.\nSpeaker I: 010665.\nSpeaker I: 33496408.\nSpeaker I: 590.\nSpeaker I: 689798.\nSpeaker I: 804.\nSpeaker I: 990.\nSpeaker I: 03.\nSpeaker I: 160281.\nSpeaker B: Transcript 3151-3170.\nSpeaker B: 4499.\nSpeaker B: 550606780.\nSpeaker B: 0770394.\nSpeaker B: 0470616824.\nSpeaker B: 2939.\nSpeaker B: 4507608494.\nSpeaker B: 5107260.\nSpeaker B: 84084567101.\nSpeaker B: 0810101.\nSpeaker B: 20184264019.\nSpeaker J: Transcript 3333113333.\nSpeaker J: 0671307474.\nSpeaker J: 0723864507284.\nSpeaker J: 96045050081.\nSpeaker J: 1 2 4 7 397.\nSpeaker J: 5 4 8 2 6 6 4 5 7.\nSpeaker J: 7 9 8 8 9 071231.\nSpeaker D: Transcript 28912910.\nSpeaker D: 4185277240.\nSpeaker D: 65347.\nSpeaker D: 7 9 1 6 8 6 3.\nSpeaker D: 9015.\nSpeaker D: 012524.\nSpeaker D: 3475840.\nSpeaker D: 457.\nSpeaker D: 56094.\nSpeaker D: 7056932.\nSpeaker D: 9509.\nSpeaker D: 06031910523.\nSpeaker A: Transcript 2931.\nSpeaker A: That's 2950.\nSpeaker A: 503.\nSpeaker A: 6 81690008.\nSpeaker A: 91987.\nSpeaker A: Old 605.\nSpeaker A: 048.\nSpeaker A: 028.\nSpeaker A: 4123.\nSpeaker A: 053185216.\nSpeaker A: 5607400.\nSpeaker A: 8 9 309 9 9 6 4 4.\nSpeaker A: 0 0 3752110889.\nSpeaker A: 3 14 5 0 7 4.\nSpeaker A: 5 6 6 3.\nSpeaker E: Transcript 291-3010.\nSpeaker E: 8 5 9 0 0 1 3 15 0 5 4 4 0 9 4 2.\nSpeaker E: 5 6 7 4 6 8 3 6 7 7 8 4 8 9 0 1.\nSpeaker E: 0 1 6 1 3 5 0 1 5 7 2 8 9.\nSpeaker E: 8 8 9 3 4 5 0 8 7 4 7 1 1 8 4 0 305 1.\nSpeaker A: Did your feel go back?\n"}, {"meeting_id": "Bmr005", "summary": "The group discussed topics including a potential collaboration with another ICSI member regarding the analysis of inference structures, efforts by speaker mn005 to detect speaker overlap, the current status of recordings and transcriptions, and future efforts to collect meeting data. In addition to weekly meetings by the BMR group, efforts are in progress to record meetings by other ICSI research groups, as well as routine discussions by non-ICSI members.", "dialogue": "Speaker G: So, okay, it doesn't look like a crash, that's great.\nSpeaker I: So I think maybe what's causing it to crash is I keep starting it and then stopping it to see if it's working.\nSpeaker I: And so I think starting it and then stopping it and starting it again causes it to crash.\nSpeaker I: I won't do that anymore.\nSpeaker C: And it looks like you've found a way of mapping the location to that without having people have to give their names each time.\nSpeaker C: It's like you have the...\nSpeaker C: No.\nSpeaker C: So you know that...\nSpeaker C: I mean, are you going to write down that I said here?\nSpeaker I: I'm going to collect the digit forms and write it down.\nSpeaker I: So they should be right with what's on the digit forms.\nSpeaker I: Okay, so I'll go ahead and start with the digits.\nSpeaker I: Reading transcript.\nSpeaker I: 1, 2, 5, 1, dash, 1, 2, 7, 0.\nSpeaker I: 7, 5, 3, 9, 1, 2, 0, 1, 2, 7, 3, 4, 5, 2, 6, 9, 6, 3, 9, 3, 0, 6, 4, 7, 8, 2, 0, 0, 6, 3, 0, 9, 0, 2, 9, 3, 4, 9, 5, 0, 4, 5, 4, 6, 7, 8, 0, 4, 2, 2, 2, 1.\nSpeaker I: And I should say that you just read each line and then pause, really.\nSpeaker I: Okay.\nSpeaker G: Start by giving the transcript.\nSpeaker A: OK.\nSpeaker A: Number 1, 9, 1, dash, 1, 2, 1, 0, 8, 0, 1, 1, 0, 9, 7, 4, 0, 2, 6, 1, 6, 2, 8, 4, 8, 3, 2, 7, 3, 4, 5, 0, 5, 0, 8, 5, 6, 7, 3, 8, 8, 2, 9, 3, 9, 7, 8, 4, 0, 8, 6, 3, 0, 9, 0 9 4 9 7 4 4 1 2 0 8 4 1 5 4 8 4 0 6 8 3 0 3 8 7 7 8\nSpeaker G: Transcript 1 2 1 1 dash 1 2 3 0 9 0 1 5 2 3 1 2 5 9 1 3 7 0 0 4 5 0 1 4 5 3 5 6 7 9 4 1 9 0 0 0 2 7 0 5 1 4 2 3 0 2 4 3 8 6 2 4 7 4 9 8 4 1 8 7 5 0 9 6 0\nSpeaker D: Transcript 9 5 1 9 7 0 9 0 0 0 6 2 0 1 0 6 1 4 9 5 3 5 0 4 2 0 5 6 8 7 9 8 5 8 9 0 1 0 2 9 9 0 0 5 9 1 6 0 0 2 6 6 2 0 0 6 3 9 3 1 0 4 5 6 0 7 7 0 9 1 4 3 9 2\nSpeaker H: Transcript 1 2 3 1 dash 1 2 5 0 0 0 1 0 0 0 1 3 2 4 4 5 3 6 4 5 5 6 8 0 6 7 8 9 0 5 0 2 9 1 0 3 7 0 1 6 5 9 9 2 9 8 1 1 2 7 3 6 4 5 6 0 5 7 7 8 0 8 1 3 6 2 9 2 1 0 0 0 0\nSpeaker C: Transcript number 1 1 3 1 dash 1 1 5 0 6 5 2 9 7 8 8 9 0 0 1 1 8 2 4 9 2 3 5 7 8 0 4 5 0 2 5 6 8 6 4 6 7 9 9 9 686-867-914-062-068-19829-234-6106059\nSpeaker J: Transcript 1151-117-0, 73868-936-02002-10329-6068-4300-56508-790-638-902225-01515-2535-345-6\nSpeaker E: Transcript 971-919-436-0790-5403-120033-305-51915-657-2790-89343-0090-0, 0000927-213-01002-103, 00003-6505-6505-660810-617-9405-6505-6802-103-103305-6405-8405-6405-6805-6405-6405-6505-8406-710-8405-6405-6605-6405-6405-6405-6805-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405-6405- IS-M-198-444-5\u05de\u05d9\u05ddcyr Mysticor Over\u7be6\nSpeaker G: Comm\u043d\u0438\u0431\u0443\u0434\u044c\nSpeaker I: We should add a feel to the form here for native language.\nSpeaker G: You know, it doesn't seem like a bad idea. I think I've probably forgot that.\nSpeaker I: But I think that would be a good thing.\nSpeaker I: After I just printed out a Zillion of them.\nSpeaker G: Yeah, well.\nSpeaker G: So I do have a agenda suggestion.\nSpeaker G: I think the things that we talk about in this meeting tend to be a mixture of procedural mundane things and research points.\nSpeaker G: And I was thinking, I think it was a meeting a couple of weeks ago that we spent much of the time talking about the mundane stuff because that's easier to get out of the way.\nSpeaker G: And then we sort of drifted into the research and maybe five minutes into that.\nSpeaker G: Andreas had to leave.\nSpeaker G: So I suggest we turn around.\nSpeaker G: And sort of we have, anyway, has some mundane points that we could send in an email later, hold them for a bit, and let's talk about the research, kind of, things.\nSpeaker G: So the one thing I know that we have on that is we had talked a couple of weeks before about the stuff you were doing with attempting to locate events.\nSpeaker G: We had little go around trying to figure out what you meant by events.\nSpeaker G: But I think what we had meant by events, I guess, was points of overlap between speakers.\nSpeaker G: But I gathered from our discussion a little earlier today that you also mean interruptions with something else like some other noise.\nSpeaker G: Yes?\nSpeaker G: You mean that is an event also?\nSpeaker G: Senator, you have done some work on that.\nSpeaker G: And then the other thing might be nice to have a preliminary discussion of some of the other research areas that we're thinking about doing.\nSpeaker G: I think, especially since you haven't been, and this means a little bit, maybe, in some discussion of some of the plausible things to look at how we're starting to get data.\nSpeaker G: And one of the things I know that also came up is some discussions that Jane had with Lucandra about some work about.\nSpeaker G: I don't want to try to say it, because I'll say it wrong.\nSpeaker G: But anyway, some potential collaboration there about the working with these data.\nSpeaker G: Oh, sure.\nSpeaker G: So don't around.\nSpeaker G: I don't know if this is sort of like everybody has something contribute sort of thing.\nSpeaker G: I think there's just a couple of people primarily.\nSpeaker G: But why not?\nSpeaker G: Actually, I think that last one I just said we could do fairly quickly.\nSpeaker G: So when you start with that.\nSpeaker G: OK.\nSpeaker C: Child.\nSpeaker C: Yeah, just explain a little more.\nSpeaker C: So he was interested in the question, relating to the research that he presented recently, of inference structures, and the need to build in this sort of mechanism for understanding language.\nSpeaker C: And he gave the example in his talk about how, I'm remembering it just off the top of my head right now, but it's something about how Joe Slip, John had washed the floor or something like that.\nSpeaker C: And I don't have it quite right, but that kind of thing, where you have to draw the inference that, OK, this time sequence, but also the causal aspects of the floor and how it might have been the cause of the fall, and that it was the other person who fell, and the one who cleaned it, and these sorts of things.\nSpeaker C: So I looked through the transcript that we had so far, and found identified a couple different types of things of that type.\nSpeaker C: And one of them was something like, during the course of the transcript, we had gone through the part where everyone said which channel they were on and which device they were on.\nSpeaker C: And the question was raised, well, should we restart the recording at this point?\nSpeaker C: And Daniela said, well, we're just so far ahead of the game right now, we really don't need to.\nSpeaker C: Now, how would you interpret that without a lot of inference?\nSpeaker C: So the inferences that are involved are things like, so how do you interpret ahead of the game?\nSpeaker C: So it's metaphorically.\nSpeaker C: What you draw, the conclusions that you need to draw are that space is involved in recording, that we have enough space, and he continued.\nSpeaker C: It's like we were so ahead of the game, because now we have built-in down sampling.\nSpeaker C: So you have to get the idea that ahead of the game is speaking with respect to space limitations that in fact down sampling is gaining us enough space, and that therefore we can keep the recording we've done so far.\nSpeaker C: But there are a lot of different things like that.\nSpeaker I: So do you think his interest is in using this as a data source or training material or what?\nSpeaker G: Well, I should have maybe interject, because this started off with a discussion that I had with him.\nSpeaker G: So we were trying to think of ways that his interest could interact with our resume.\nSpeaker G: And I thought that if we were going to project into the future when we had a lot of data and such things might be useful for that, in order before we invested too much effort into that, you should have looked into some of the data that we already have, and see, is there anything to this at all?\nSpeaker G: Is there any point to which you think that you could gain some advantage and some potential use for it?\nSpeaker G: Because it could be that you looked through it and you'd say, well, this is just the wrong task for him to pursue his.\nSpeaker G: And I got the impression from your mail that in fact there was enough things like this, just a little sample that you looked at that it's possible at least.\nSpeaker C: It's possible.\nSpeaker C: You know, we met and he was going to go and look through them more systematically, and then meet again.\nSpeaker C: So it's not a matter.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I think.\nSpeaker G: Anyway, that's quite different than anything we've talked about that might come out from this.\nSpeaker J: So he gave me his text.\nSpeaker J: I mean, that's his major.\nSpeaker J: That's his major.\nSpeaker C: Just using text.\nSpeaker C: I mentioned several that had to do with implications drawn from international contours.\nSpeaker C: And that wasn't as directly relevant to what he's doing.\nSpeaker C: He's interested in these knowledge structures, inferences that you draw from.\nSpeaker G: I mean, he certainly could use text.\nSpeaker G: We were, in fact, looking to see if there is there something in common between our interested meetings and his interest in this stuff.\nSpeaker G: So.\nSpeaker I: And I imagine that transcripts of speech, text that his speech probably has more of those than sort of prepared writing.\nSpeaker I: I don't know whether or not that it means.\nSpeaker G: Probably depends on what the prepared writing was.\nSpeaker C: Yeah, I don't think I would make that leap.\nSpeaker C: Because in narratives, you know, I mean, if you spell out everything in a narrative, it could be really tedious.\nSpeaker I: Yeah, I'm just thinking, you know, when you're face to face, you have a lot of back channel.\nSpeaker I: And oh, that aspect.\nSpeaker I: Yeah.\nSpeaker I: And so I think it's just easier to do that sort of broad inference-drumping if it's face to face.\nSpeaker I: I mean, so if I just read that Dan was saying we're ahead of the game in that context, I might not realize that he was talking about discuses and opposed anything else.\nSpeaker C: I had several that had to do with back channels.\nSpeaker C: And this wasn't one of them.\nSpeaker C: This one really does make you leap from, so he said, we're ahead of the game, we have built in down sampling.\nSpeaker C: And the inference, if you had it written down, would be, but there are others that have back channels.\nSpeaker C: It was less interested in those.\nSpeaker D: Sorry to interrupt.\nSpeaker D: A minute, seven minutes ago, I briefly was not listening.\nSpeaker D: And so who is he in those context?\nSpeaker D: Yeah, there's a lot of people.\nSpeaker D: OK, so I was just realizing, you guys have been talking about E for at least three, three, four minutes without ever mentioning the person's name again.\nSpeaker D: So this is going to be a big, big problem if you want to later do indexing or speech understanding of any sort.\nSpeaker D: I wrote this down.\nSpeaker D: You guys know this?\nSpeaker J: As Morgan will say, well, you had some ideas.\nSpeaker J: And he never said, he looked.\nSpeaker I: Well, I think he's doing that intentional.\nSpeaker I: Right, yes, great.\nSpeaker J: So this is really great because the thing is, because he's looking at the first, even for addresses in the conversation, I bet you could pick that up in the acoustics just because your gaze is also correlated with the directionality of your voice.\nSpeaker I: Oh, that would be interesting.\nSpeaker J: Yeah, so that, I mean, to even know, when, yeah, if you have the PZM, you should be able to pick up what a person is looking at from their voice.\nSpeaker I: Well, especially with Morgan, with the way we have the microphones arranged, I'm sort of right on axis.\nSpeaker I: And it would be very hard to tell.\nSpeaker J: Right.\nSpeaker C: But you have things like this.\nSpeaker C: You'd have Fainter.\nSpeaker C: Wouldn't you get Fainter reception out here?\nSpeaker I: Sure, but I think if I'm talking like this, right now I'm looking at Jane and talking now.\nSpeaker I: I'm looking at Chuck and talking.\nSpeaker I: I don't think the microphones would pick up that difference.\nSpeaker I: But you don't have this problem.\nSpeaker G: Morgan is someone who does.\nSpeaker I: So if I'm talking to you, or I'm talking to you.\nSpeaker G: I've probably been affected by too many conversations where we were talking about lawyers and talking about concerns about, oh, I'm going to say something bad and so on.\nSpeaker G: And so I'm tending to stay away from people's names, even though.\nSpeaker J: Even though you could pick up later on just from the acoustic.\nSpeaker J: Who you really did mention who he was.\nSpeaker D: But I missed it.\nSpeaker D: Can I say it?\nSpeaker G: Yeah, no, no, no, it isn't sensitive at all.\nSpeaker G: I was overreacting just because we've been talking about it.\nSpeaker C: And in fact, it is sensitive.\nSpeaker C: I came up with something from the Kimmer's Abjects People that I wanted to mention.\nSpeaker C: I mean, it fits into the area of Monday.\nSpeaker C: And but they did say, I asked her very specifically about this cause of how it says, no individuals will be identified in any publication using the data.\nSpeaker C: OK, well, individuals being identified.\nSpeaker C: Let's say you have a snippet that says, Joe thinks such and such about this field, but I think he's wrong-headed.\nSpeaker C: Now, I mean, we're going to be careful not to have the wrong-headed part in there.\nSpeaker C: But let's say we say Joe used to think so and so about this area.\nSpeaker C: And in his publication, he says that, but I think he's changed his mind or whatever.\nSpeaker C: Then the issue of being able to trace Joe because we know he's well known in this field and all this and tie it to the speaker, whose name was just mentioned a moment ago, can be sensitive.\nSpeaker C: So I think it's really kind of adaptive and wise to not mention names anymore of them.\nSpeaker C: We have to because if there's a slander's aspect to it, then how much do we want to be able to have to remove?\nSpeaker G: Yeah, well, there's that.\nSpeaker G: But I mean, I think also to some extent it's just educating the human subjects people in a way.\nSpeaker G: Because if there's court transcripts, there's transcripts of radio shows.\nSpeaker G: I mean, people say people's names all the time.\nSpeaker G: So I think it can't be bad to say people's names.\nSpeaker G: It's just that, I mean, you're right that there's more potential.\nSpeaker G: If we never say anybody's name, then there's no chance of slandering anybody.\nSpeaker G: But then it won't.\nSpeaker G: It's not me.\nSpeaker J: If we, yeah, I mean, we should do whatever is natural in a meeting if we weren't being recorded.\nSpeaker G: Right.\nSpeaker G: So my behavior is probably not natural and so.\nSpeaker C: I'm going to ask, my feeling on it wasn't really important who said it.\nSpeaker D: Well, since you have to go over the transcripts later anyway, you could make it one of the jobs of the people who\nSpeaker I: do that to mark about this during an anonymization. If we wanted to go through and extract from the audio and the written every time someone says a name.\nSpeaker I: And I thought that our conclusion was that we didn't want to do that.\nSpeaker G: Yeah, we really can't.\nSpeaker G: But actually, I'm sorry, I really would like to push.\nSpeaker C: How did you say?\nSpeaker C: No, I just was suggesting that it's not a bad policy for potentially.\nSpeaker C: So we need to talk about this later.\nSpeaker G: Yeah, I didn't tend to this policy.\nSpeaker G: It was just unconscious.\nSpeaker G: Well, semi-conscious behavior.\nSpeaker G: I sort of knew who he is.\nSpeaker J: I remember who he is.\nSpeaker J: No, I didn't say you still know who he is.\nSpeaker G: With that presence.\nSpeaker G: We were talking about Dana one point.\nSpeaker G: We were talking about Lucindra, another point.\nSpeaker G: And I remember which part.\nSpeaker G: I don't know.\nSpeaker D: Well, the difference was Lucindra.\nSpeaker D: OK, that makes sense.\nSpeaker J: Yeah.\nSpeaker J: Down trembling.\nSpeaker J: Yeah.\nSpeaker F: Good.\nSpeaker F: Yeah.\nSpeaker F: Yeah, you can do all these inferences.\nSpeaker F: Yeah.\nSpeaker G: Yeah.\nSpeaker G: I would like to move it into what Jose has been doing.\nSpeaker G: He's actually doing something.\nSpeaker E: He's supposed to the rest of us.\nSpeaker E: OK.\nSpeaker E: I remember that my first objective in the projects is to study different parameters.\nSpeaker E: To find a good solution.\nSpeaker E: To detect the overlapping zone in speech record.\nSpeaker E: But in that way, I am speaking to study and to analyze the different sessions.\nSpeaker E: To find, to locate, to mark, to different overlapping zones.\nSpeaker E: So I was, I had transcluing the first session.\nSpeaker E: And I had found 1,000 acoustic events.\nSpeaker E: Besides the overlapping zone, I mean the spreads, expression, tag, club.\nSpeaker E: I don't know what is the difference.\nSpeaker E: You used to name the.\nSpeaker I: Landspeach zone.\nSpeaker I: Which.\nSpeaker I: I don't think we've been doing it at that level of detail.\nSpeaker I: So.\nSpeaker E: Yeah.\nSpeaker E: I don't need to level the different acoustic.\nSpeaker E: I prefer because I would like to study.\nSpeaker E: I will find a good parameter to detect the overlapping.\nSpeaker E: I would like to test this parameter with another acoustic events.\nSpeaker E: To find what is the false.\nSpeaker E: The false hypothesis which I produced that when we use this parameter, a big difference.\nSpeaker H: You know, so I think some of these that are the non-speech overlapping events may be difficult even for humans to tell what there's to there.\nSpeaker H: If it's a tapping sound, you wouldn't necessarily or something like that.\nSpeaker H: It might be hard to know that it was two separate events.\nSpeaker I: You weren't talking about just overlaps, or you were just talking about acoustic events.\nSpeaker E: Someone started, someone stopped.\nSpeaker E: I talked about acoustic events in general.\nSpeaker E: Yeah.\nSpeaker E: But my objective will be to study overlapping zones.\nSpeaker G: How many overlaps were there?\nSpeaker E: In 12-15 minutes, I found 1,000 acoustic events.\nSpeaker E: No, no.\nSpeaker E: How many of them were the overlaps of speech, though?\nSpeaker E: How many are most 300 in one session?\nSpeaker E: In fight in 45 minutes?\nSpeaker E: 300 overlapping.\nSpeaker E: 300 overlapping.\nSpeaker E: Overlapped speech with different duration.\nSpeaker C: Sure.\nSpeaker C: So if you had an overlap involving three people, how many times was that counted?\nSpeaker E: Yeah, three people to people.\nSpeaker E: I would like to consider one people with different noise in the background.\nSpeaker G: No, no.\nSpeaker G: I think what she's asking is, if at some particular stretch, you had three people talking instead of two, did you call that one event?\nSpeaker E: I consider one event for all the songs.\nSpeaker E: Well, I consider an acoustic event.\nSpeaker E: The overlapping songs, the period where three speakers are talking together.\nSpeaker I: So let's say me and Jane are talking at the same time, and then Liz starts talking also overall of us.\nSpeaker I: How many events would that be?\nSpeaker E: Should they be?\nSpeaker I: So two people are talking.\nSpeaker I: And then a third person starts talking.\nSpeaker I: Is there an event right here?\nSpeaker E: No.\nSpeaker E: For me, is there overlapping songs?\nSpeaker E: So if two or more people are talking?\nSpeaker E: You have more voice, if they're using them in moments.\nSpeaker G: Yeah, so I think we just wanted to understand how you're defining it.\nSpeaker G: So then in the region, since there is some continuous region, in between regions where there is only one person speaking, and one continuous region like that, you're calling an event.\nSpeaker G: Is it, are you calling the beginning or the end of it the event, or are you calling the entire length of it the event?\nSpeaker E: I consider the entire, all the time where the voices are overlapping.\nSpeaker E: But I don't distinguish between the numbers of speakers.\nSpeaker E: I'm not considering the fact of, for example, what you say, first two jokers are speaking, and third person joined to that.\nSpeaker E: For me, it's all that song with several numbers of speaker is the same acoustic event.\nSpeaker E: But without any mark between the song or the overlapping song, with two speaker speaking together, the song with the three speakers.\nSpeaker C: That would just be one.\nSpeaker E: One.\nSpeaker E: With the beginning mark and the ending mark.\nSpeaker E: Because for me, it's a song with some kind of distortion in the special.\nSpeaker E: Well, by the moment, by the moment.\nSpeaker I: But you could imagine that three people talking has a different spectral characteristic.\nSpeaker E: Yeah, good.\nSpeaker E: I have to start somewhere.\nSpeaker E: Yeah, we have to start somewhere.\nSpeaker E: We're just right away.\nSpeaker E: I don't know what we're talking about.\nSpeaker E: Yeah.\nSpeaker G: So again, that's three hundred in 45 minutes that are speaker, just speakers.\nSpeaker G: OK, yeah.\nSpeaker G: So that's about eight per minute.\nSpeaker C: For 1,000 events in 12 minutes, that's that can actually be tapped.\nSpeaker C: Well, 1,000 taps in eight minutes is a lot of fun.\nSpeaker E: I consider acoustic events the silent two.\nSpeaker E: Silence, starting or silent two things.\nSpeaker E: To detect because I consider acoustic events, all the things are not speech.\nSpeaker E: You're in the general point of view.\nSpeaker E: OK, so how many of the speech?\nSpeaker D: That speech or too much speech.\nSpeaker D: That's it.\nSpeaker G: So how many of the 1,000 were silent?\nSpeaker G: Silence, actually.\nSpeaker E: This is silent.\nSpeaker E: I haven't.\nSpeaker E: I would like to do a statistic study with you with a report from the study from the session, one session.\nSpeaker E: And I found another thing.\nSpeaker E: When I was a look at the difference speech file, for example, we use the Mrs. file to transcribe the events and the words.\nSpeaker E: I saw that the speech signal collected by this kind of mic or this kind of mic are different from the Mrs.\nSpeaker E: signer we collect by a microphone.\nSpeaker E: Right.\nSpeaker E: It's right.\nSpeaker E: But the problem is the following.\nSpeaker E: I knew that the signal will be different.\nSpeaker E: But the problem is we detected difference events in the speech file collected by a mic compared it with the Mrs. file.\nSpeaker E: And so if you try to crack only using the Mrs. file, it's possible if you use the transcription to evaluate a different system, it's possible.\nSpeaker E: And you use the speech file collected by the Fed mic to do experiments with the system if possible to evaluate or to consider acoustic events that which you marked in the Mrs. file at the donor pure in the speech signer collected by.\nSpeaker I: The reason that I generated the mix file was for IBM to do word level transcription.\nSpeaker I: It's a double the VAT transcription.\nSpeaker I: So I agree that if someone wants to do speech event transcription that the mix signals here, I mean, if I'm tapping on the table, it's not going to show up on any of the mics.\nSpeaker I: But it's going to show up rather loudly on the PZM.\nSpeaker E: So I say that this only because I, in my opinion, it's necessary to put the transcription on the speech file collected by the objective signer.\nSpeaker E: I mean the signer collected by the real mic in the future, the prototype, to correct the initial segmentation with the real speech you have to analyze, you have to process.\nSpeaker E: Because I found it a difference.\nSpeaker G: Yeah, well, just in that one 10 second or whatever was example that Adam had that we passed on to others a few months ago, there was that business where, I guess, was Adam and Jane were talking at the same time.\nSpeaker G: And in the close talking mics, you couldn't hear the overlap.\nSpeaker G: And then the distant mic, you're good.\nSpeaker G: So yeah, it's clear that if you want to study, if you want to find all the places where they're overlap, it's probably better to use a distant mic.\nSpeaker G: On the other hand, there's other phenomena that are going out at the same time before which it might be useful to look at the close talking mic.\nSpeaker J: So it's...\nSpeaker J: But why can't you use the combination of the close talking mic?\nSpeaker I: If you use the combination of the close talking mics, you would hear Jane interrupting me, but you wouldn't hear the paper rustling.\nSpeaker I: And so if your interest is...\nSpeaker G: So have it's masking, Matt.\nSpeaker J: If you're interested in speakers, you're working on those things.\nSpeaker J: And not the other kind of non speech.\nSpeaker I: Right.\nSpeaker I: Right.\nSpeaker I: So all of the other issue is that the mixed close talking mics, I'm doing weird normalizations and things like that.\nSpeaker J: But it's known.\nSpeaker J: Yep.\nSpeaker J: I mean, the normalization you do is over the whole conversation.\nSpeaker J: Right.\nSpeaker J: Over the whole meeting.\nNone: Yep.\nSpeaker J: So if you wanted to study people overlacking people, that's not a problem.\nSpeaker F: Right.\nSpeaker E: I saw the...\nSpeaker E: But I have a...\nSpeaker E: In your response, I saw this pitchfai...\nSpeaker E: I was really collected by the fed mic.\nSpeaker E: And the signal to noise, a relation is low.\nSpeaker E: It's very low.\nSpeaker E: It's very...\nSpeaker E: We compare it with the headphone.\nSpeaker E: Yep.\nSpeaker E: I found that...\nSpeaker E: Probably, in not sure by the moment.\nSpeaker E: But this is probably that a lot of...\nSpeaker E: For example, in the overlapping song, and in several parts of the files where you can find a smooth speech from one talking in the meeting, is probably that those files, you cannot find.\nSpeaker E: You cannot project because it confuses with noise.\nSpeaker E: And there are a lot of...\nSpeaker E: I see.\nSpeaker E: I have to study with more detail.\nSpeaker E: But my idea is to process only this...\nSpeaker E: This is the most of the speech.\nSpeaker E: The fact that it is more realistic, it's more realistic, but it'll be a lot harder.\nSpeaker G: Well, it'll be hard, but on the other hand, as you point out, if your concern is to get the...\nSpeaker G: Overliving people, people's speech, you will get that.\nSpeaker G: It's more better.\nSpeaker G: Are you making any use?\nSpeaker G: You were working with the data that had already been transcribed.\nSpeaker G: Does it?\nSpeaker G: Yes.\nSpeaker G: Now, did you make any use of that?\nSpeaker G: Because we have these 10 hours of other stuff that is not yet transcribed.\nSpeaker E: The tradition by Jane, I want to use to put...\nSpeaker E: It's a reference for me.\nSpeaker E: But the tradition, for example, I don't...\nSpeaker E: I'm not interested in the words, the tradition, the words, for the speech file.\nSpeaker E: But Jane, for example, puts a mark at the beginning of each talk in the meeting.\nSpeaker E: She includes information about the song where there is a overlapping song.\nSpeaker E: And there is any mark, time, temporal mark, to...\nSpeaker E: To...\nSpeaker E: To...\nSpeaker E: To Laver.\nSpeaker E: Right, so she is...\nSpeaker E: I think we need this information to...\nSpeaker G: Right.\nSpeaker G: So the 12 hours...\nSpeaker G: Of course, this included maybe some time where you were learning about what you wanted to do.\nSpeaker G: But...\nSpeaker G: But the talk is something that 12 hours marked the 45 minutes.\nSpeaker G: 12 minutes.\nSpeaker G: 12 minutes.\nSpeaker G: 12 minutes.\nSpeaker G: 12 minutes.\nSpeaker E: I thought you did 45 minutes.\nSpeaker E: No, 45 minutes is the session.\nSpeaker E: All the session.\nSpeaker E: Oh, you haven't done the whole session.\nSpeaker E: This is just 12 minutes.\nSpeaker E: Oh.\nSpeaker E: 12 hours of work.\nSpeaker E: Do... to segment a level, 12 minutes processes.\nSpeaker E: So let me back up again.\nSpeaker G: So when you said there were 300 speaker overlaps, that's in 12 minutes.\nSpeaker E: No, no, no.\nSpeaker E: I consider all the session because I count the overlapping's marked by...\nSpeaker E: Oh, okay.\nSpeaker E: In the...\nSpeaker E: In the 45 minutes.\nSpeaker G: So it's 345 minutes, you have... you have time marked 12 minutes.\nSpeaker G: The... the...\nSpeaker G: The...\nSpeaker G: Overlaps in 12 minutes.\nSpeaker G: Got it.\nSpeaker G: So...\nSpeaker D: Can I ask...\nSpeaker D: Can I ask whether you found...\nSpeaker D: You know how accurate...\nSpeaker D: James...\nSpeaker D: Labels were as far as...\nSpeaker D: You know, did you miss some overlaps or did you...\nSpeaker E: By the moment, I...\nSpeaker E: I don't computer my...\nSpeaker E: Temporal Mark with James.\nSpeaker E: I want to do it because...\nSpeaker E: Perhaps I have errors in the marks.\nSpeaker E: If I... I compare with James, it's probably... I can correct...\nSpeaker E: To get more...\nSpeaker E:...create the...\nSpeaker E:...tractition.\nSpeaker I: Also, James was doing work level.\nSpeaker I: Yeah.\nSpeaker I: So we weren't very...\nSpeaker J: Right, right.\nSpeaker D: I'm not expecting...\nSpeaker J: I'm not in work level, but actually...\nSpeaker J: No.\nSpeaker J: I didn't need to...\nSpeaker J:...show the exact point of interruption.\nSpeaker J: You just were showing at the level of the phrase...\nSpeaker J: Or the level of the speech spurt or...\nSpeaker C: Well, yeah, I would say time been.\nSpeaker C: So my goal was to get words with reference to a time been...\nSpeaker C: Beginning and end point.\nSpeaker C: And sometimes, you know, I was like, You could have an overlap where someone said something in the middle.\nSpeaker C: But it was just wasn't important for our purposes to have it that...\nSpeaker C: Disrupt that unit in order to have, you know, the words in the order in which they were spoken.\nSpeaker C: It would have been hard with the interface that we have.\nSpeaker C: Now, my app...\nSpeaker C: I'm working on course, and I realized...\nSpeaker C: It's an overlapping interface, but...\nSpeaker E: You said we'd work, but I think we need to...\nSpeaker E: No, of course.\nSpeaker D: I expected to find more overlap than...\nSpeaker D: No.\nSpeaker D:...because you're looking at it at a much more detailed level.\nSpeaker E: I want to computer it.\nSpeaker E: I hope 60 to 1.\nSpeaker G: Well, but I have a suggestion about that.\nSpeaker G: Obviously, this is very, very time consuming...\nSpeaker G:...and you're finding lots of things which I'm sure...\nSpeaker G:...are going to be very interesting.\nSpeaker G: But in the interest of making progress...\nSpeaker G:...my name is...\nSpeaker G:...how would it affect your time if you only marked the speaker overlaps?\nSpeaker G: Only.\nSpeaker G: Yes. Do not mark any other events, but only marked speaker.\nSpeaker G: Do you think that would speed it up quite a bit?\nSpeaker G: Do you think that would speed it up?\nSpeaker G: Speed up your marking?\nSpeaker E: I... I...\nSpeaker E: I wanted to...\nSpeaker E: I don't understand the...\nSpeaker G: It took you a long time to mark 12 minutes.\nSpeaker G: Oh, yeah.\nSpeaker G: Now, my suggestion was for the other 33...\nSpeaker E:...only to mark only 12 minutes.\nSpeaker G: Yeah, and my question is, if you did that...\nSpeaker G:...if you followed my suggestion...\nSpeaker G:...would it take much less time?\nSpeaker E: Oh, yeah. Sure.\nSpeaker E: Yeah, okay.\nSpeaker E: That I think it's a good idea.\nSpeaker E: Sure.\nSpeaker E: That I think it's a good idea because...\nSpeaker E: I need a little time to put the label on to...\nSpeaker G: Yeah, I mean, we know that there's noise.\nSpeaker G: There's continual noise from fans and so forth.\nSpeaker G: And there is more impulsive noise from taps and so forth.\nSpeaker G: And something in between with paper rustling.\nSpeaker G: We know that all of that's there and it's a worthwhile thing to study.\nSpeaker G: But obviously, it takes a lot of time to mark all of these things.\nSpeaker G: Whereas, I would think that we can study more or less as a distinct phenomenon...\nSpeaker G:...the overlapping of people talking.\nSpeaker G: Okay, so then you can get...\nSpeaker G:...because you need...\nSpeaker G:...if it's 300...\nSpeaker G:...it sounds like you probably only have 50 or 60 or 70...\nSpeaker G:...events right now that are really...\nSpeaker G:...and you need to have a lot more than that to have any kind of...\nSpeaker G:...even visual sense of what's going on...\nSpeaker G:...much less than any kind of reasonable statistics.\nSpeaker J: Now, why do you need to mark...\nSpeaker J:...speaker overlap by hand if you can infer it from the relative energy...\nSpeaker I:...that's why I was going to bring up.\nSpeaker J: You shouldn't need to do this completely.\nSpeaker G: Okay, yeah, so let's back up because you went here for an earlier conversation.\nSpeaker G: So the idea was that what he was going to be doing...\nSpeaker G:...was experimenting with different measures...\nSpeaker G:...such as the increase in energy...\nSpeaker G:...such as the energy and the LPC residuals...\nSpeaker G:...I mean, there's a bunch of things...\nSpeaker G:...I mean, increased energy is sort of an obvious one.\nSpeaker G: Yeah.\nSpeaker G: And it's not obvious.\nSpeaker G: I mean, you could do the dumbest thing and get it 90% of the time.\nSpeaker G: But when you start going past that and trying to do better...\nSpeaker G:...it's not obvious what combination of features...\nSpeaker G:...is going to give you the right detector.\nSpeaker G: So the idea is to have some ground truth first.\nSpeaker G: And so the idea of the manual marking was to say,...okay, it's really here.\nSpeaker H: But I think Liz is saying why not get it out of the time...\nSpeaker J: Get it from the close talking mics.\nSpeaker J: Yeah, or get it first pass.\nSpeaker J: We talked about that.\nSpeaker J: And then go through sort of a lot better.\nSpeaker G: We talked about that.\nSpeaker G: So it's a bootstrapping thing.\nSpeaker G: The idea was we thought it would be useful for him to look at the data anyway.\nSpeaker G: And then whatever he could mark would be helpful.\nSpeaker G: And it's a question of what you bootstrapped from.\nSpeaker G: You know, do you bootstrapped from a simple measurement...\nSpeaker G:...which is right most of the time and then you do better...\nSpeaker G:...or do you bootstrapped from some human being looking at it...\nSpeaker G:...and then do your simple measurements from the close talking mic.\nSpeaker G: Even with a close talking mic, you're not going to get it right all the time.\nSpeaker J: Well, that's what I wonder because...\nSpeaker I:...or how bad it is.\nSpeaker I: I'm working on a program to do that.\nSpeaker J: Because that would be interesting, especially because the bottleneck is the transcription.\nSpeaker J: Right?\nSpeaker J: We've got a lot more data than we have the solutions.\nSpeaker J: We have the auto-experts.\nSpeaker J: We have the close talking mic.\nSpeaker J: So I mean, it seems like one kind of project is not perfect.\nSpeaker J: But that you can get the training data for pretty quickly is...\nSpeaker J:...if you infer from the close talking mics where the on-off points are, of speech...\nSpeaker J:...we discuss that.\nSpeaker J: And how can we detect that from a far film?\nSpeaker I: I've written a program to do that.\nSpeaker I: And sorry, I'm in it.\nSpeaker I: It's okay.\nSpeaker I: But it's doing something very, very simple.\nSpeaker I: It just takes a threshold based on...\nSpeaker D: Or you can set the threshold low and then we'd out the false alarms by hand.\nSpeaker I: And then it does a median filter and then it looks for runs.\nSpeaker I: And it seems to work. I'm sort of fiddling with the parameters to get it to actually generate something.\nSpeaker I: And I haven't...\nSpeaker I: I don't...\nSpeaker I: What I'm working on, what I was working on was getting it to a form where we can import it into the user interface that we have into transcriber.\nSpeaker I: And so I told...\nSpeaker I: I said it would take about a day.\nSpeaker I: I've worked on it for about half a day.\nSpeaker I: So give me another half day and I will have something we can play with.\nSpeaker G: See, this is where we really need the meeting record to query stuff to be working.\nSpeaker G: Because we had these meetings and we had this discussion about this and I'm sort of remembering a little bit about what we decided.\nSpeaker G: But I could remember all of it.\nSpeaker G: So I think it was partly that...\nSpeaker G: You know, give somebody a chance to actually look at the data and see what these are like.\nSpeaker G: Partly that we have some ground truth to compare against when he gets his thing going.\nSpeaker J: It was definitely good to have somebody look at us thinking as a way to speed up.\nSpeaker G: But that was exactly the notion that we discussed.\nSpeaker C: Another thing we discussed was that...\nSpeaker C: I think that's the idea.\nNone: I think that's the idea.\nSpeaker C: The idea was that there was this already a script, I believe, that Dan had written, that handles bleed through.\nSpeaker C: I mean, because you have this close...\nSpeaker C: You have contamination from other people who speak loudly.\nSpeaker I: Yeah, and I haven't tried using that.\nSpeaker I: It would probably help the program that I'm doing to first feed it through that.\nSpeaker I: It's a cross-correlation filter.\nSpeaker F: Yeah.\nSpeaker I: So I haven't tried that.\nSpeaker I: So it might be something...\nSpeaker I: It might be a good way of cleaning it up a little.\nSpeaker C: Having that be a pre-processor and then run it through yours.\nSpeaker C: But that's a refinement.\nSpeaker G: I think we want to see try the simple thing first because you have this complex thing up afterwards that does something good.\nSpeaker G: You sort of want to see what the simple thing does first.\nSpeaker G: But having somebody have some experience again with marking it from a human standpoint...\nSpeaker G: I don't expect Jose to do it for 50 hours of speech.\nSpeaker G: But I mean, if he could speed up what he was doing by just getting the speaker overlapped so that we had, say, for 45 minutes, then at least we'd have 300 examples of it when Adam was doing his automatic thing.\nSpeaker G: He could then compare it to that and see what it was doing.\nSpeaker H: I did something almost identical to this at one of my previous jobs.\nSpeaker H: And it works pretty well.\nSpeaker H: I mean, almost exactly what you described, an energy detector with a biggie and filter, you look for runs.\nSpeaker H: And you think like the right thing to do.\nSpeaker H: Yeah, I mean, you can get...\nSpeaker H: I mean, you get them pretty far with the literature, sir.\nSpeaker H: And so I think doing that to generate these possibilities and then going through and saying yes or no on them would be a quick way to...\nSpeaker I: That's good validation.\nSpeaker I: Yeah.\nSpeaker B: Is this proprietary?\nSpeaker I: If you have a patent.\nSpeaker F: No.\nSpeaker H: This one I was working for the government.\nSpeaker H: Nicey.\nSpeaker C: Oh, then everybody owns it.\nSpeaker C: Is it something that we could just co-op to?\nSpeaker G: Well, he's pretty close.\nSpeaker G: Yeah, anyway.\nSpeaker C: I just thought if it was tried and true, then he's gone through additional models.\nSpeaker I: Although, if you have some parameters like what's a good window size for the median filter?\nSpeaker H: I have to remember.\nSpeaker H: I'll think about it.\nSpeaker D: And it might be different for government people.\nSpeaker I: I didn't know for government work, because they say different bandwidth.\nSpeaker I: I was doing pretty short, you know, 10th of a second.\nNone: Sort of numbers.\nSpeaker G: Okay.\nSpeaker G: I don't know.\nSpeaker G: If we want to...\nSpeaker G: So, maybe we can move on to other things in the meantime.\nSpeaker C: One question about statistics.\nSpeaker C: 12 minutes.\nSpeaker C: If we took 300 and divided it by 4, which is about the length of 12 minutes, I'd expect, like, to be 75 overlaps.\nSpeaker C: Did you find more than 75 overlaps in that period?\nSpeaker C: More than?\nSpeaker C: More than how many overlaps in your 12 minutes?\nSpeaker E: Not even.\nSpeaker E: All the...\nSpeaker E: I skype only 12 minutes from the...\nSpeaker E: I don't count the...\nSpeaker E: I consider...\nSpeaker E: I consider to be...\nSpeaker I: I bet there more, because the beginning of the meeting had a lot more overlaps than...\nSpeaker I:...the middle ran.\nSpeaker I: Because we're dealing with the...\nSpeaker I: In the early meetings, we're recording while we're saying, who's talking on what microphones and things like that.\nSpeaker I: And that seems to be a lot of overlap.\nSpeaker C: I think it's an empirical question.\nSpeaker C: I think we could find that out.\nSpeaker C: I'm not sure that the beginning had more.\nSpeaker G: So, I was going to ask, I guess, about any other things that...\nSpeaker G: That you do want to talk about, especially since...\nSpeaker G:...I'm very misleading in five minutes.\nSpeaker G: But...\nSpeaker J: I just asked about the data, like, very straightforward questions, where we are on the amount of data and the amount of transcribed data, just because I...\nSpeaker J: I wanted to get a feel for that to sort of be able to know what...\nSpeaker J:...can be done for us.\nSpeaker J: Right, so there's this...\nSpeaker J: How many meetings are we recording?\nSpeaker G: There's this 45-minute piece that Jane transcribed.\nSpeaker G: That piece was then sent IBM.\nSpeaker G: So they could transcribe as we have some comparison points.\nSpeaker G: Then there's a larger piece that's been recorded and...\nSpeaker G:...put on CD-ROM and sent IBM.\nSpeaker G: Right?\nSpeaker G: And then...\nSpeaker G: We don't know.\nSpeaker G: What's that?\nSpeaker G: That was about 10 hours and it was about...\nSpeaker G: Like, 10 meetings?\nSpeaker J: Yeah, something like that.\nSpeaker I: And then 10 meetings that have been sent to IBM.\nSpeaker I: Well, I haven't sent them yet because I was having this problem with the missing files.\nSpeaker I: Oh, that's right.\nSpeaker G: That has...\nSpeaker G:...those have not been sent.\nSpeaker H: How many total have we recorded now?\nSpeaker H: All together?\nSpeaker G: It's saying about...\nSpeaker G: About 12...\nSpeaker G: 12 or 12 or 13.\nSpeaker J: So we're recording only this meeting, like, continuously, we're only...\nSpeaker J: No.\nSpeaker G: No, no.\nSpeaker G: So that's the morning one.\nSpeaker G: That's the biggest one, chunk so far.\nSpeaker G: But there's at least one meeting recorded of the...\nSpeaker G:...natural language guys.\nSpeaker G: Do they mean every week?\nSpeaker G: They do.\nSpeaker G: And we talked to them about recording some more and we're going to...\nSpeaker G: We started having a morning meeting today, starting a week or two ago...\nSpeaker G:...on the front-end issues.\nSpeaker G: And we're recording those.\nSpeaker G: Okay.\nSpeaker G: There is a network services and applications group here who's agreed to have their meetings recorded.\nSpeaker G: And we're going to start recording them.\nSpeaker G: They're meeting on Tuesdays.\nSpeaker G: We're going to start recording them next week.\nSpeaker G: So actually we're going to start having a pretty significant chunk.\nSpeaker G: And so, you know, Adam's sort of struggling with trying to get things to be less buggy...\nSpeaker G:...and come up quicker when they do crash and stuff like that.\nSpeaker G: Now that the things are starting to happen.\nSpeaker G: So right now, I'd say the data is predominantly meeting meetings.\nSpeaker G: But there are scattered other meetings in it and that amount is going to grow...\nSpeaker G:...so that the meeting meetings will probably ultimately...\nSpeaker G:...if we collect 50 or 60 hours, the meeting meetings will probably be 20 to 30 percent of it.\nSpeaker J: So there's probably...\nSpeaker J:...there's three to four a week.\nSpeaker I: That's what we're aiming for.\nSpeaker J: That we're aiming for.\nSpeaker J: And they're each about an hour.\nSpeaker I: Yeah, although we'll find out tomorrow whether we can really do this or not.\nSpeaker G: Yeah, and the other things I'm not paused.\nSpeaker G: I'm sort of thinking as we've been through this a few times...\nSpeaker G:...that I really don't know.\nSpeaker G: Maybe you want to do it once for the novelty.\nSpeaker G: But I don't know if in general we want to have meetings...\nSpeaker G:...that we record from outside this group to do the digits.\nSpeaker G: Because it's just an added bunch of weird stuff.\nSpeaker G: And we're highly motivated.\nSpeaker G: In fact, the morning group is really motivated because they're working on connected digits.\nSpeaker I: So it's something I wanted to ask.\nSpeaker I: I have a bunch of scripts to help with the transcription of the digits.\nSpeaker I: We don't have to hand-transcribe the digits because we're reading them and I have those.\nSpeaker I: So I have some scripts that let you very quickly extract the sections of each utterance.\nSpeaker I: But I haven't been doing that.\nSpeaker I: If I did that, is someone going to be working on it?\nSpeaker G: Definitely something of interest.\nSpeaker G: Absolutely.\nSpeaker G: Whoever we have working on the acoustics for the meeting recording...\nSpeaker G: I'm interested in that.\nSpeaker I: I just don't have time to do it now.\nSpeaker D: I'm sure someone thought of this, but this reading of the numbers...\nSpeaker D:...would be extremely helpful to do adaptation.\nSpeaker D: Yep.\nSpeaker I: I would really like someone to do adaptation.\nSpeaker I: So if we got someone interested in that, I think it would be great for meeting recorder.\nSpeaker G: I mean, one of the things I wanted to do...\nSpeaker G: I talked to Don about one of the possible things you could do...\nSpeaker G:...or us, we have someone else do it...\nSpeaker G:...is to do block echo cancellation to try to give it some of the effects of the particle effects.\nSpeaker G: I mean, we have...\nSpeaker G: The party line has been that echo cancellation is not the right way to handle the situation...\nSpeaker G:...because people move around.\nSpeaker G: And if it's not a simple echo, like, a cross-talk kind of echo...\nSpeaker G:...but it's actually room acoustics.\nSpeaker G: You can't really do inversion.\nSpeaker G: And even echo cancellation is going to be something...\nSpeaker G:...someone may be moving enough that you are not able to adapt quickly.\nSpeaker G: And so the tack that we've taken is more...\nSpeaker G:...let's come up with feature approaches and multi-stream approaches and so forth...\nSpeaker G:...that will be robust to it for the recognizer...\nSpeaker G:...and not try to create a clean signal.\nSpeaker G: That's the party line.\nSpeaker G: But it occurred to me a few months ago that party lines are always sort of dangerous...\nSpeaker G:...and it's good to sort of test them, actually.\nSpeaker G: And so we haven't had anybody try to do a good serious job on echo cancellation...\nSpeaker G:...and we should know how well that can do.\nSpeaker G: So that's something I like somebody to do at some point.\nSpeaker G: Just take these digits, take the far field signal...\nSpeaker G:...and apply really good echo cancellation.\nSpeaker G: There was a nice talk recently by Lucen.\nSpeaker G: The block echo cancellation particularly appealed to me.\nSpeaker G: I'm not trying to change it sample by sample, but you have some reasonable sized blocks.\nSpeaker H: What is the artifact you've tried to recognize?\nSpeaker G: So you have a direct...\nSpeaker G:...what's the difference in...\nNone:...if you were trying to recognize the familiar filter...\nSpeaker G:...that was signing off.\nSpeaker G: That would subtract off the parts of the signal...\nSpeaker G:...that were the aspects of the signal that were different between the close talk and the distant.\nSpeaker G: So I guess the most echo cancellation...\nSpeaker G:...you've given that...\nSpeaker G:...you're trying to...\nSpeaker G:...there's a distance between the close and the distant mic...\nSpeaker G:...so there's a time delay there.\nSpeaker G: And after the time delay there's these various reflections.\nSpeaker G: And if you figure out what's the...\nSpeaker G:...there's a least-grays algorithm that adjusts itself...\nSpeaker G:...or adjust the weight so that you try to subtract...\nSpeaker G:...essentially subtract off different reflections.\nSpeaker G: So let's take the sample case where you just had...\nSpeaker G:...you had some delay in a satellite connection or something...\nSpeaker G:...and then there's an echo.\nSpeaker G: And you want to adjust this filter so that it will maximally reduce the effect of this echo.\nSpeaker H: So that would mean like if you were listening to the data that was recorded on one of those...\nSpeaker H:...just the raw data.\nSpeaker H: You might hear kind of an echo...\nSpeaker H:...and then this noise cancellation.\nSpeaker G: Well, I'm saying that's a simplified version of what's really happening.\nSpeaker G: What's really happening is...\nSpeaker G: Well, when I'm talking to you right now...\nSpeaker G:...you're getting the direct sound from my speech...\nSpeaker G:...but you're also getting the indirect sound that's bounced around the room a number of times.\nSpeaker G: Okay.\nSpeaker G: So now if you try to completely remove the effect of that...\nSpeaker G:...it's sort of impractical for a number of technical reasons.\nSpeaker G: But not to try to completely remove it that is invert the room response...\nSpeaker G:...but just to try to eliminate some of the effect of some of the echoes.\nSpeaker G: A number of people have done this...\nSpeaker G:...so that if you're talking to a speaker phone...\nSpeaker G:...it makes it more like it would be if you were talking right up to it.\nSpeaker G: So this is sort of the straight forward approach.\nSpeaker G: You say, I want to use this item...\nSpeaker G:...but I want to just track it off various kinds of echo.\nSpeaker G: So you construct a filter...\nSpeaker G:...and you have this filtered version of the speech...\nSpeaker G:...get subtracted off from the original speech...\nSpeaker G:...and you try to minimize the energy in some sense.\nSpeaker G: And so...\nSpeaker G: It's kind of a cleanup thing.\nSpeaker G: It's a cleanup thing, right?\nSpeaker G: So echo canceling is commonly done in Tollophany...\nSpeaker G:...and it's sort of the obvious thing to do in this situation...\nSpeaker G:...if you know you're going to be talking some distance from the mic.\nSpeaker H: In fact, when I would have meetings with the folks in Cambridge...\nSpeaker H:...when I had the BDN over the phone...\nSpeaker H:...they had some kind of a special speaker phone...\nSpeaker H:...and when they would first connect me...\nSpeaker H:...it would hear all this noise.\nSpeaker H: And then it would come on and it was very clear.\nSpeaker G: So it's taking samples, it's doing adaptations, adjusting weights...\nSpeaker G:...and then it's getting some.\nSpeaker G: So...\nSpeaker G:...anyway that's kind of a reason for something like that somebody tries...\nSpeaker G:...somebody like...\nSpeaker G:...and the digits would be a reasonable thing to do that with...\nSpeaker G:...I think there'd be enough data to do it with...\nSpeaker G:...and for that sort of task you wouldn't care...\nSpeaker G:...but it was largely a vocabulary speech or anything.\nSpeaker C: Is Brian King's worries work related to that?\nSpeaker C: There's a different type of evaluation.\nSpeaker G: Brian King's worries work is an example of what we did...\nSpeaker G:...from the opposite dogma, right?\nSpeaker G: And what I was calling a party line, which is that...\nSpeaker G:...doing that sort of thing is not really what we want.\nSpeaker G: We want something more flexible...\nSpeaker G:...where people might change their position...\nSpeaker G:...and there might be...\nSpeaker G:...there's also...\nSpeaker G:...oh yeah, noise.\nSpeaker G: So that echo cancellation does not really allow for noise.\nSpeaker G: If you have a clean situation but you just have some delays...\nSpeaker G:...then we'll figure out the right set of weights for your taps...\nSpeaker G:...for your filter in order to reduce the effect of those echoes.\nSpeaker G: But if there's noise, then the very signal that it's looking at...\nSpeaker G:...is corrupted so that it's decision about what the right...\nSpeaker G:...right delays are is in correct.\nSpeaker G: And so in a noisy situation, also in a situation...\nSpeaker G:...that's very, very, very long reverberation times...\nSpeaker G:...really long delays, it's sort of typically impractical.\nSpeaker G: So for those kind of reasons...\nSpeaker G:...and also a complete inversion.\nSpeaker G: If you actually, I mentioned that it's kind of hard to really do...\nSpeaker G:...the inversion of the room acoustics.\nSpeaker G: That's difficult because...\nSpeaker G:...often times the...\nSpeaker G:...the system transfer function is such that when it's inverted...\nSpeaker G:...you get something that's unstable.\nSpeaker G: And so if you do your estimate of what the system is...\nSpeaker G:...then you try to invert it, you get a filter that actually...\nSpeaker G:...rangs and goes to infinity.\nSpeaker G: So there's that sort of technical reason...\nSpeaker G:...and the fact that things move, there's error currents...\nSpeaker G:...I mean there's all sorts of reasons why that's not really practical.\nSpeaker G: So for all those kinds of reasons...\nSpeaker G:...include we didn't want to do inversion...\nSpeaker G:...and we even pretty skeptical of echo cancellation...\nSpeaker G:...which isn't really inversion.\nSpeaker G: And we decided to do this approach of taking...\nSpeaker G:...just picking features...\nSpeaker G:...which will give you something more stable...\nSpeaker G:...in the presence of or absence of room reverberation...\nSpeaker G:...and that's what Brian was trying to do.\nSpeaker G: So let me just say a couple things...\nSpeaker G:...that I was going to bring up.\nSpeaker G: Let's see, I guess you actually already said...\nSpeaker G:...this thing about the consent forms...\nSpeaker G:...which was that we now don't have to.\nSpeaker G: So this was the human subject's focus, you said this?\nSpeaker C: Apparently, we're going to do a revised form, of course.\nSpeaker C: But once a person is signing it once...\nSpeaker C:...then that's valid for a certain number of meetings.\nSpeaker C: She wanted me to actually estimate how many meetings...\nSpeaker C:...and put that on the consent form.\nSpeaker C: I told her that would be a little bit difficult to say.\nSpeaker C: So I think from a practical standpoint...\nSpeaker C:...maybe we could have them do it once every 10 meetings...\nSpeaker C:...or something that won't be that many people who do it that often...\nSpeaker C:...but just, you know, something else...\nSpeaker C:...they don't forget that they've done it, I guess.\nSpeaker G: Okay.\nSpeaker G: Back on data thing...\nSpeaker G:...so there's this sort of one hour, ten hour, a hundred hours sort of thing...\nSpeaker G:...that we have...\nSpeaker G:...we have an hour...\nSpeaker G:...that is transcribed.\nSpeaker G: We have 12 hours that's recorded, but not transcribed.\nSpeaker G: And at the right we're going by the end of the semester...\nSpeaker G:...we'll have, I don't know, 40 or 50 or something...\nSpeaker G:...if this really...\nSpeaker G:...but we have that one, so see what happens.\nSpeaker J: It's three to four per week.\nSpeaker G: So, eight weeks...\nSpeaker G:...not a lot of hours.\nSpeaker G: Eight weeks, ten to three, so it's 24.\nSpeaker G: So that's, yeah, so like 30 hours?\nSpeaker J: I mean, is there, I know this sounds tough, but we've got the room set up.\nSpeaker J: I was starting to think of some projects where you would use...\nSpeaker J:...well, similar to what we talked about with the energy detection...\nSpeaker J:...on the close talking mic.\nSpeaker J: There are a number of interesting questions that you can ask...\nSpeaker J:...about how interactions happen in the meeting.\nSpeaker J: They don't require any transcription.\nSpeaker J: So what are the patterns of the energy patterns over the meeting?\nSpeaker J: And I'm really interested in this.\nSpeaker J: But we don't have a lot of data.\nSpeaker J: So I was thinking, you know, we've got the room set up.\nSpeaker J: And you can always think of also for political reasons if the XC collected, you know, 200 hours...\nSpeaker J:...that looks different than 40 hours, even if we don't transcribe it ourselves.\nSpeaker G: But I don't think we're going to stop at the end of the semester.\nSpeaker G: So I think that if we are able to keep that up for a few months...\nSpeaker G:...we are going to have more like 100 hours.\nSpeaker J: Are there any other meetings here that we can record, especially meetings...\nSpeaker J:...that have some kind of conflict in them or some kind of...\nSpeaker J:...that are less...\nSpeaker J:...that have some more emotional aspects to them or...\nSpeaker J: We had some good ones earlier.\nSpeaker J: There is laughter.\nSpeaker J: I'm talking more about strong differences of the meeting...\nSpeaker J:...maybe with manager types or...\nSpeaker I: I think it's hard to record this.\nSpeaker C: To be allowed to record.\nSpeaker C: It's also a lot of people cancel out afterwards.\nSpeaker C: But I wonder if they can keep AID anyway.\nSpeaker I: Yeah, I was going to mention that.\nSpeaker I: That's a good idea. That would be a good match.\nSpeaker G: Yeah, so I'd mention to Adam and...\nSpeaker G:...that was not the thing I was going to talk about.\nSpeaker G: I'd mention to them before that...\nSpeaker G:...it occurred to me that we might be able to get some additional data...\nSpeaker G:...by talking to acquaintances and local broadcast media.\nSpeaker G: Because we had talked before about the problem...\nSpeaker G:...but using found data that it's just set up...\nSpeaker G:...however they haven't set up.\nSpeaker G: They don't have any say about it.\nSpeaker G: It's typically one microphone.\nSpeaker G: And so it doesn't really give us the character six we want.\nSpeaker G: And so I do think we're going to continue recording here...\nSpeaker G:...and record what we can.\nSpeaker G: But it did occur to me that we could go to friends and broadcast media...\nSpeaker G:...and say, hey, you have this panel show...\nSpeaker G:...or this discussion show.\nSpeaker G: And can you record multi-channel?\nSpeaker G: And they may be willing to record it with...\nSpeaker G: Tell them I could.\nSpeaker G: Well, they probably already used the PAL...\nSpeaker G:...but they might be able to have it...\nSpeaker G:...it wouldn't be that weird for them to have another mic that was somewhat distant.\nSpeaker G: It wouldn't be exactly this setup, but it'd be that sort of thing.\nSpeaker G: And what we were going to get from UW, assuming they start recording,\nSpeaker J:...is also not going to be this exact setup.\nSpeaker G: No, I think that would be great. So I was thinking of looking into that.\nSpeaker G: And the answer we had that discussion, in fact, is that it's even possible...\nSpeaker G:...since, of course, many radio shows are not live...\nSpeaker G:...that we could invite them to record some of their shows here.\nSpeaker J: Wow.\nSpeaker J: The thing is, they're not as averse to wearing one of these.\nSpeaker J: I mean, they're on the radio, right?\nSpeaker J: Right.\nSpeaker J: I think that would be fantastic, because those kind of panels and those have interest in.\nSpeaker J: Yeah.\nSpeaker J: That's a side of a style that we're not collecting here.\nSpeaker G: And the other side to it, which is where we're coming from, I'll talk to you more about later, is that there's...\nSpeaker G:...the radio stations and television stations already have stuff worked out presumably related to illegal issues and permissions and all that.\nSpeaker G: I mean, they already do what they do.\nSpeaker G: So it's another source.\nSpeaker G: So I think it's something we should look into. We'll collect what we collect here.\nSpeaker G: Hopefully, they will collect more UW also.\nSpeaker G: And maybe we have this other source, but yeah, I think it's not unreasonable to aim at...\nSpeaker G:...getting significantly in excess of 100 hours.\nSpeaker G: I mean, that was sort of our goal.\nSpeaker G: The thing was, I was hoping that we could...\nSpeaker G:...in the under this controlled situation, we could at least collect, you know, 30 to 50 hours.\nSpeaker G: And if the rate we're going, we'll get pretty close to that, I think, this semester.\nSpeaker G: And if we continue to collect some next semester, I think we should...\nSpeaker J: Right, yeah, I was mostly trying to think, okay, you start a project within, say, a month.\nSpeaker J: How much data do you have to work with?\nSpeaker J: You want to sort of freeze your data for a while.\nSpeaker J: So right now, and we don't have the transcripts back yet from IBM.\nSpeaker J: Well, we don't have it for this, you know, 45 minutes that way.\nSpeaker J: So not complaining, I was just trying to think, what kinds of projects can you do now,...versus six months from now, and they're pretty different.\nSpeaker G: Yeah, so I've seen it right now. It's sort of this exploratory stuff where you look at the data, you do some primitive measures, and get a feeling for what the scatter plots look like.\nSpeaker G: Right, right.\nSpeaker G: And meanwhile, we collect, and then it's more like, yeah, three months from now or six months from now, you can do a lot more.\nSpeaker J: Because I'm not actually sure just logistically that I can spend...\nSpeaker J: I don't want to charge the time that they have on the project too early before there's enough data to make good use of the time.\nSpeaker J: And especially with the student, for instance, this guy who seems to...\nSpeaker J: Anyway, I shouldn't say too much, but if someone came that was great, and wanted to do some real work, and they have to end by the end of the school year in the spring, how much data will I have to work with that person?\nSpeaker J: Right.\nSpeaker G: Yeah, so I would think exploratory things now.\nSpeaker G: Three months from now, I mean, the transcriptions are...\nSpeaker G: I mean, the transcriptions, I think, are a bit unknown, because we haven't gotten those back yet, as far as the timing.\nSpeaker G: But I think, as far as the collection, it doesn't seem to me unreasonable to say that, in January, roughly, just roughly three months from now, we should have at least something like 25, 30 hours.\nSpeaker G: So that's...\nSpeaker J: And you just don't know about the transcription part of that.\nSpeaker J: And we need to...\nSpeaker C: I think that there's the possibility that the transcript will need to be adjusted after words and...\nSpeaker C: especially since these people won't be used to dealing with multi-channel transcription.\nSpeaker C: So I think that we'll need to adjust some...\nSpeaker C: And also, if we want to add things like...\nSpeaker C: well, more refined coding of overlaps than definitely, I think we should count on having extra pass through.\nSpeaker C: I wanted to ask another aspect of the data collection.\nSpeaker C: There'd be no reason why a person couldn't get together several, you know, friends and come and argue about a topic if they wanted to, right?\nSpeaker G: If they really have something they want to talk about as opposed to something...\nSpeaker G: I mean, what we're trying to stay away from was artificial constructions, but I think if it's a real...\nSpeaker G: Why not?\nSpeaker G: I'm thinking...\nSpeaker G: Stakes in political debates?\nSpeaker J: Well, yeah, or just if you're...\nSpeaker J: If you have...\nSpeaker J: There are meetings here that happen that we can record, even if we don't...\nSpeaker J: have them do the digits, or maybe have them do a shorter digit thing.\nSpeaker J: Like, you know...\nSpeaker I: We don't have to do the digits at all if we don't want to.\nSpeaker J: One string of digits is something that probably will indeed.\nSpeaker J: Then having the data is very valuable, because I think it's politically better for us to say we have as many hours of audio data, especially with the ITR, if we put in a proposal, it'll just look like EXU's collected a lot more audio data, whether it's transcribed or not, is another issue, but there are research questions you can answer without the transcriptions, or at least that you can start to answer.\nSpeaker B: It seems like you could hold some meetings, you know, and maybe you could hold some additional meetings.\nSpeaker H: Would it help at all?\nSpeaker H: I mean, we're already talking about sort of two levels of detail, and meetings.\nSpeaker H: One is without doing the digits...\nSpeaker H: I guess the full-blown one is where you did the digits and everything, and then talk about doing it without digits.\nSpeaker H: What if we had another level just to collect data, which is without the headsets, and we just did the table amount.\nSpeaker H: We need the close-talking line.\nSpeaker J: I mean, absolutely.\nSpeaker I: Yeah, I'm really scared.\nSpeaker I: It seems like it's a big part of this corpus, is that what it was like?\nSpeaker J: Or at least, like me personally, I couldn't use that data.\nSpeaker C: Okay, and Murray also, we had this camera-bunship issue here, that's important.\nSpeaker J: Yeah.\nSpeaker J: So it's a great idea, and if it were true, then I would just do that, but it's not that bad.\nSpeaker J: Like, the room is not the bottleneck, and we have enough time in the room, and it's getting the people to come in and put on the...\nSpeaker G: Okay, by the way, I don't think the transcriptions are actually, in the long run, such a big bottleneck.\nSpeaker G: I think the issue is just that we're blazing that path.\nSpeaker G: And do you have any idea when you'd be able to send the...\nSpeaker I: Well, I've been burning two CDs a day, which is about all I can do with the time I have.\nSpeaker I: Yeah.\nSpeaker I: So it'll be early next week.\nSpeaker G: Yeah, okay.\nSpeaker G: So early next week is sent it to them, and then we check with them to see if they've got it, and we start asking about timing for it.\nSpeaker G: So I think once they get sorted out about how they're going to do it, which I think they're pretty well long on, because they were able to read the files and so on, right?\nSpeaker G: Yeah, but...\nSpeaker G: Well...\nSpeaker G: Yeah, when those were, they aren't.\nSpeaker G: Have they ever responded?\nSpeaker G: No.\nSpeaker G: Yeah, but, you know, so they have, you know, they're volunteering at time, and they have a lot of other things to do, right?\nSpeaker G: Yeah, they have a lot of other things to do, right?\nSpeaker G: But anyway, though, I think once they get that sorted out, they're making cassettes there, and then they're handing it to someone who is doing it, and I think it's not going to be...\nSpeaker G: I don't think it's going to be that much more of a deal for them to do 30 hours than to do one hour.\nSpeaker G: Yeah, it's not going to be 30- So it's the amount of...\nSpeaker G: It's just getting it going.\nSpeaker I: It's pipeline, pipeline issue.\nSpeaker I: So what's the pipeline?\nSpeaker I: What about these lunch meetings?\nSpeaker J: I mean, I don't know.\nSpeaker J: If there's any way without too much more overhead, even if we don't ship it right away to IBM, even if we just collected here for a while to record, you know, two or three more meetings a week, just to have the data, even if they're not doing the digits, but they do wear the...\nSpeaker G: But the lunch meetings are pretty much one person getting up.\nSpeaker J: No, I meant...\nSpeaker J: Sorry, the meetings where people eat their lunch downstairs, maybe they don't want to be recorded, but...\nSpeaker I: Oh, and we're just chatting?\nSpeaker I: Yeah, we have a lot of the things.\nSpeaker J: Actually, I actually think that's useful data.\nSpeaker I: Yeah, the problem with that is...\nSpeaker I: I think I would feel a little constrained to...\nSpeaker I: Okay, you know, some of them are soccer ball meetings.\nSpeaker I: I guess none of you were there for a soccer ball meeting.\nSpeaker J: Throw it out there with anyone knows of one more or two more meetings per week that happen at XC that we can record.\nSpeaker J: I think it would be worth it.\nSpeaker G: Yeah, well we should also check with Maria again, because they were really intending, maybe it just didn't happen, but they were really intending to be duplicating this in some level.\nSpeaker G: So then that would double what we had.\nSpeaker G: And there's a lot of different meetings that UW mean really a lot more than we have here, because we're not right on campus.\nSpeaker H: Is the notion of recording any of Chuck's meetings dead in the water?\nSpeaker H: They seem to have some problems,\nSpeaker G: whether we can talk about that later. But again, Jerry's open.\nSpeaker G: So I mean, we have two speech meetings, one network meeting.\nSpeaker G: Jerry was open to it, but I...\nSpeaker G: One of the things that I think is a little bit of a limitation there, is I think when people are not involved in our work, we probably can't do it every week.\nSpeaker G: I think that people are going to feel...\nSpeaker G: are going to feel a little bit constrained.\nSpeaker G: I might get a little better if we don't have them do the digits all the time.\nSpeaker G: Yeah.\nSpeaker G: And so then they can just really sort of chart you, put the mics on, they just charge in.\nSpeaker J: What if we give people, you know, we cater a lunch in exchange for them having their meeting here?\nSpeaker C: I do think eating while you're doing a meeting is going to be increasing my rights.\nSpeaker C: I had another question, which is, you know, in principle, I know that you don't want artificial topics, but it does seem to me that we might be able to get subjects from campus to come down and do something that wouldn't be too artificial.\nSpeaker C: I mean, we...\nSpeaker C: Political discussions or something, rather.\nSpeaker C: And, you know, people who are...\nSpeaker C: Because, you know, there's also this constraint.\nSpeaker C: It's like, you know, the...\nSpeaker C: Goldy bars, Goldy blocks.\nSpeaker C: It's like, you don't want meetings that are too large, but you don't want meetings that are too small.\nSpeaker C: And it just seems like maybe we could exploit human subjects in the positive sense of...\nSpeaker H: Well, even, I mean, coming down from campus is sort of a big thing.\nSpeaker H: But what about...\nSpeaker H: Could pay subjects.\nSpeaker H: Or what about people in the building?\nSpeaker B: Yeah, I don't get it.\nSpeaker B: They need credit.\nSpeaker H: Yeah, I was saying there's only other people.\nSpeaker H: And, you know, there's a lot of people in California downstairs.\nSpeaker I: And I just really doubt that any state of California meetings would be recordable and then really so both of the general public.\nSpeaker I: So, I mean, I talked with some people at the Hoss Business School who are interested in speech recognition.\nSpeaker I: And they sort of hummed and said, well, maybe we could have some meetings down here, but then I got a email from them and said, no, we decide we're not really interested, and we don't want to come down to hold meetings.\nSpeaker I: So, I think it's going to be a problem to get people regularly.\nSpeaker G: But, yeah, I can maybe he can...\nSpeaker G: Yeah, we can get some scattered things from this and that.\nSpeaker G: And I do think that maybe we can get somewhere with the radio.\nSpeaker G: I better context radio.\nSpeaker G: You can get a lot of lively discussions from those radio.\nSpeaker J: Well, and they're already...\nSpeaker J: Yeah.\nSpeaker J: These things are already recorded.\nSpeaker J: We don't have to ask them to.\nSpeaker J: Even, and I'm not sure how they recorded, but they must need some individual.\nSpeaker G: No, I'm not talking about ones that are already recorded.\nSpeaker G: I'm talking about new ones, because we would be asking them to do something different.\nSpeaker J: Well, we can find out...\nSpeaker J: I know, Princess Mark, living in this interested in L.D.C.\nSpeaker J: getting data...\nSpeaker G: Right, that's the found data idea.\nSpeaker G: But what I'm saying is, if I talk to people that I know who do these and who produce these things, we could ask them if they could record an extra channel, let's say, of a distant mic.\nSpeaker G: And I think routinely they would not do this.\nSpeaker G: So, since I'm interested in the distant mic stuff, I want to make sure that there is at least that somewhere.\nSpeaker G: But if we ask them to do that, they might be intrigued enough by the idea that they...\nSpeaker G: I might be able to talk them into it.\nSpeaker I: We're getting towards the end of our disc space, so we should think about...\nSpeaker I: Okay, well, why don't we...\nSpeaker G: When we...\nSpeaker I: Okay, leave them on for a moment until I turn this off, because that's what I mean.\nSpeaker I: Crash last time.\nSpeaker G: Turning off the microphone and crash.\nSpeaker G: Yeah.\nSpeaker B: Let's question them.\nSpeaker G: Okay.\n"}, {"meeting_id": "IS1007c", "summary": "The team discussed the conceptual design and some specific details in this meeting. Industrial Designer talked about the components concept of industrial design, such as the different choice of energy and the different material of the remote control. The team would find out which one to use in the future. User Interface designer gave the presentation about system design, mainly about programme design for different users and to help them use the device. Then the team decided to discuss the marketing part at first, then discuss what to design. Marketing talked about specific details found in the market survey. The team agreed to use fancier design, less buttons and focus on voice recognition. They would find out whether using LCD on remote controls was too expensive or not. Team members agreed to use a voice commander and fixed charger, which would prevent the remote control getting lost. Then the team talked about the set of vocabularies for the speech recogniser.", "dialogue": "Speaker C: So, we come to the third meeting, I hope the lunch was good.\nSpeaker C: So, in the last meeting we have discussed the functional design and now we will talk about the conceptual design.\nSpeaker D: So, I think I will do my presentation on the components concept. So, can you please open the slide? I'm participant 2, couple of these design.\nSpeaker D: Okay, so the first thing I have done is to make a review together with the manufacturer department, which components were available to build remote control.\nSpeaker D: So, for energy sources we have to choose between the solar energy, hand-in-a-mo and kinetic, well, kinetic technique to store the energy.\nSpeaker D: We also can put regular battery in the remote control. This is what we have decided in the last meeting. But if we use battery.\nSpeaker D: Yeah, well, I meant by battery, I would not have a wire between the remote control and an energy source. But we didn't decide yet which kind of battery we will put inside the remote.\nSpeaker D: So, it's a point to discuss. Then the case material we have also several choices like hood, rubber, titanium or latex.\nSpeaker D: But it's not a really issue from the technical point of view. Concerning the interface, we can put just simple buttons or scrolls. Our buttons are much more complicated.\nSpeaker D: But it also requires that the chip to process the button is more complicated. So this is the last points, the choice of chips.\nSpeaker D: So, what I have found is that I think basic battery or kinetic energy collection is the better way to provide energy because I think solar energy won't work in the accretal environment.\nSpeaker D: So, I think we can start with these two main things. For the case, I think that a titanium is a good choice because it's trendy and it's a well, it's modern.\nSpeaker D: So, the user will be very happy to have a nice remote. For the interface, I think that we can achieve all the desired functionalities by just using rubber buttons, simple buttons.\nSpeaker D: And does this allow to use a regular chip? That's alright, well, cheaper.\nSpeaker D: So, we can move to the next slide. Sorry. What is a single curved? What does it mean?\nSpeaker D: The shape of the remote, you will have the curve will fit into your hand when you grab the...\nSpeaker B: When you hold on it, it's comfortable with the hole.\nSpeaker D: It's more comfortable that if it is completely flat.\nSpeaker B: And the battery is kind of rechargeable or it doesn't matter.\nSpeaker D: That's the point. The kinetic one is you can recharge by the...\nSpeaker B: That's what it means by kinetic?\nSpeaker D: Well, but just by moving your arm, the remote will accumulate energy.\nSpeaker D: But I don't know if it is feasible because I don't know if the user will move enough to provide the remote or the necessary energy.\nSpeaker B: We might check with our R&D department to see if they have this product.\nSpeaker B: Yeah. Ready for market.\nSpeaker D: And so can you go to the next slide, please?\nSpeaker D: So, that's summarizing what I have said.\nSpeaker D: So you're right. We can see our R&D.\nSpeaker B: The department.\nSpeaker D: If the kinetic method is sufficient to provide enough energy. That's it.\nSpeaker B: So I'll keep in touch with the R&D department.\nSpeaker D: Oh yeah, I take care of it.\nSpeaker B: So the titanium case is the normal case.\nSpeaker B: I'll show you some pictures that I have and you tell me whether they are titanium case or not.\nSpeaker B: Or something very...\nSpeaker B: Plastic, titanium or whatever.\nSpeaker B: There's another point I want to make is that the... Well, you have seen my presentation later on.\nSpeaker B: I point out some wide buttons and not the only ways you can use.\nSpeaker C: Maybe we can work it.\nSpeaker B: So the user interface is...\nSpeaker B: It gives you the aspect of a computer system, a program which can be seen or heard or otherwise perceived by the human user in the commands and mechanism the user uses to control its operation and input data.\nSpeaker B: So it just gives you the ways to input data and we have an emphasized mode on the graphical user interface here.\nSpeaker B: The idea is to represent buttons as figures, diagrams, symbols and so you can easily...\nSpeaker B: When you look at the symbols you understand what it is doing.\nSpeaker B: Watch the function of this.\nSpeaker B: It makes the exact first way.\nSpeaker B: Is it to use...\nSpeaker B: So next one?\nSpeaker B: This is function 5.\nSpeaker B: So I can use the button, the mouse maybe.\nSpeaker C: The graph you can use on the entire first surface I still use.\nSpeaker B: So next night.\nSpeaker B: So here are some examples.\nSpeaker B: So they cluster the buttons together.\nSpeaker B: They group them into color and they have different forms as well.\nSpeaker B: But these interface are kind of confusing.\nSpeaker B: Basically they are too many buttons.\nSpeaker B: Next one?\nSpeaker B: So some people propose voice recognition and...\nSpeaker B: So by the way I received an email from one of the departments saying that the voice recognition has been used in the coffee machine.\nSpeaker B: Produced by a company.\nSpeaker B: When you tell the user, good morning coffee machine and the machine will reply to you.\nSpeaker B: So I just got an email saying that...\nSpeaker B: And it seems like this voice recognition technology is ready to be used.\nSpeaker B: So we might consider that.\nSpeaker B: Yeah, firstly.\nSpeaker B: The next one?\nSpeaker B: So some people use a spinning wheel with the LC display.\nSpeaker B: So instead of using the buttons, you have an LCD screen.\nSpeaker B: And then there you can use that as buttons.\nSpeaker B: You can use that as wheel.\nSpeaker B: So there could be an option as well.\nSpeaker B: Touch screen.\nNone: Next one?\nSpeaker B: And some people propose a scroll button integrated with push button.\nSpeaker B: So you may have scroll button instead of just the push button like the one we have here.\nSpeaker B: Next one?\nSpeaker B: So there are a few aspects that I collected here.\nSpeaker B: So basically these deals with special users, children, handicapped people, old people, and...\nSpeaker B: Basically they are programmable, especially for children.\nSpeaker B: And then they also secure covers to protect secured and hidden programming and battery covers.\nSpeaker B: They will protect your settings.\nSpeaker B: But we don't have to integrate all these complicated features.\nSpeaker B: I'm just saying that currently in the market, they are remote controllers customizable for different people.\nSpeaker B: So that's the point.\nSpeaker B: Next one?\nSpeaker B: And you see this is the one where you have the protection cover.\nSpeaker B: It may be useful for children, they only see the buttons outside.\nSpeaker B: And for adults where you have more control, you can see the one inside.\nSpeaker B: So the adults might want to have a key to lock that.\nSpeaker B: So children will not touch the button inside.\nSpeaker D: Yeah.\nSpeaker D: That's a good idea.\nSpeaker B: Next one?\nSpeaker B: So this guy, this is another company that provides big buttons.\nSpeaker B: I see that it's useful for all people and then you don't get lost.\nSpeaker B: But for our product, we don't need a big one because you have voice recognition.\nSpeaker B: Eventually it is used and you can call your remote controller if you don't know where it is.\nSpeaker B: TV remote controller, where are you?\nSpeaker B: And then your BIPs and to say that I'm here.\nSpeaker B: For example, is it possible?\nSpeaker C: We should think of speech sentes in this case.\nSpeaker D: But as no man safe, there is already a commercial product available to do this.\nSpeaker D: We can check to integrate it into our new remote control.\nSpeaker B: Yeah.\nSpeaker B: And this is another one where you can...\nSpeaker B: The parts that are V standing for the volume.\nSpeaker B: So there's up arrow and down arrow.\nSpeaker B: But you see that in the V appears to be the down arrow on the top.\nSpeaker B: On the top up arrow, there's a V.\nSpeaker B: As if it's turning down, so it's confusing in the phrase.\nSpeaker B: So you want to avoid this kind of thing in the design.\nSpeaker B: And here is a short summary that I compiled.\nSpeaker B: After the findings I found.\nSpeaker B: Big buttons are convenient, voice recognition helps.\nSpeaker B: Push buttons, scroll buttons, spinning wheels can be used as navigation tools.\nSpeaker B: And user-customizable is important.\nSpeaker B: And finally, simplicity is the key.\nSpeaker B: So we have many concepts there, but we have to choose later on which ones are important to be used.\nSpeaker D: Well, I think it's fine to have reviewed all the possibilities.\nSpeaker D: But if we consider that the user interface is displayed on the TV screen, I don't think we need much buttons in the remote.\nSpeaker D: Since we just have to navigate and to have an OK or an enter key or things like that.\nSpeaker D: Because adding wheels or scrolls makes the thing more complicated and more expensive also.\nSpeaker C: Or maybe we can include the user in the remote model and we should have just a button like help and you can get your button help.\nSpeaker C: And maybe you can see the user.\nSpeaker D: That's a good idea to have a help button.\nSpeaker B: A help button.\nSpeaker B: So you have to play on the screen.\nSpeaker B: So the user is removed.\nSpeaker B: So the element is the complicated documentation.\nSpeaker B: So people are afraid of looking at the help.\nSpeaker A: Once the user has a help button, they say this is a complicated stuff.\nSpeaker A: In the case of where they need it.\nSpeaker C: It's a psychological.\nSpeaker C: Why are they doing the marketing?\nSpeaker A: OK.\nSpeaker A: And let us see what the market demands.\nSpeaker A: We could just go to my presentation.\nSpeaker C: Well, I think it's just for user.\nSpeaker C: Testimizes.\nSpeaker C: Yeah, that's why.\nSpeaker C: For kids or old people.\nSpeaker A: I mean, it's just short as the remote with uncap which could be used for kids and if you remove the.\nSpeaker C: So it's the same.\nSpeaker C: Same remote with some can be used by both kids and old people.\nSpeaker B: Well, what I propose is that, you know, a remote controller, it could be a cube.\nSpeaker B: It's a small device that looks like a cube.\nSpeaker B: And maybe you can just change the buttons.\nSpeaker B: If you turn one side, you get one one button.\nSpeaker B: You turn the other side, you get the other buttons.\nSpeaker B: So for maybe new generation people who get used to the computer, they want lots of controls.\nSpeaker C: Maybe for kids, they like a small.\nSpeaker C: So let's see what the.\nSpeaker D: Let's see the market demand.\nSpeaker B: One, we can decide what the market.\nSpeaker B: Yes.\nSpeaker A: So we just made an marketing survey of what people need from our remotes and how it could be special from the other remotes.\nSpeaker A: And we got the based on the responses from the questionnaires.\nSpeaker A: We also have some prizes for the most creative suggestions and we found the following suggestions which we could, which would be helpful for our design.\nSpeaker A: So 75% of the users, they find their remote controls very ugly.\nSpeaker A: They don't find it pleasant to use in the size or usage or anything.\nSpeaker A: And 80% of the people, they are always, I mean, they are willing to spend more money if their remote control would look fancy.\nSpeaker A: And the current remote controls do not match well the operating behavior of the user.\nSpeaker A: And 75% of the users said there's a lot.\nSpeaker A: And 50% say they use only 10% of the buttons.\nSpeaker A: So the rest of the 90% of the buttons, they are not used most of the times.\nSpeaker A: Yes.\nSpeaker A: So these were the findings which we found.\nSpeaker A: And also these are the frustrations with the present remote controls.\nSpeaker A: Most of 50% of the time the remote controls are lost somewhere in the room and people are always searching for them rather than watching the TV.\nSpeaker A: And while the time they found the remote control, the program is finished.\nSpeaker A: So they have frustrated a lot.\nSpeaker A: And if the remote control is too complicated, it takes much time to learn the functionality of it.\nSpeaker A: So you can just see the percentage, 50% people responded that they always lose the remote, and 34% they say that it's quite difficult to learn if it's too complex.\nSpeaker A: So keeping in view of all these findings and the frustrations, I think this should be the solution for them.\nSpeaker A: We should have an LCD on the remote control.\nSpeaker D: Oh.\nSpeaker D: Well, I don't really see the advantage of having LCD on the remote control if we have a big screen and display on the screen.\nSpeaker D: Yeah, of course it's fancy, trendy and so on, but it's expensive to produce.\nSpeaker D: And it's not really easy.\nSpeaker A: As our service is that people are willing to pay more if we are remote or fancy.\nSpeaker A: So if you have an LCD on the remote, rather than looking onto the TV, you just look into your remote and navigate it.\nSpeaker A: We just say menu.\nSpeaker A: As we saw that iPod remote control, we just play around.\nSpeaker D: Yeah, but when you play with the iPad, you don't have a big screen in front of you.\nSpeaker C: You can't use the screen instead of the big screen.\nSpeaker B: The existing screen, you eliminate the LCD.\nSpeaker B: After all, the LCD is just to display.\nSpeaker B: And if you have the colorful screen, you can make the display colorful, fancy, a fancy SD-1 on the LCD, maybe even better.\nSpeaker B: So...\nSpeaker A: I mean, these were the points which we got from the market demands.\nSpeaker D: So...\nSpeaker D: I think we can focus on the fancy look on the...\nSpeaker B: Yeah, more on the fancy design.\nSpeaker D: On the speech recognition, if the technology is available, but I think LCD will make us spend a lot of money for not so big results.\nSpeaker B: Remember we have a budget for the cost of producing the remote controller.\nSpeaker B: Yeah, we have it.\nSpeaker B: So the thing is you can find out how much an LCD will cost.\nSpeaker B: And then we will decide again.\nSpeaker A: I mean, that should be found out by the industry designers.\nSpeaker B: Oh, we can find out the price in the previous time.\nSpeaker D: So the price of LCD display.\nSpeaker A: Yeah.\nSpeaker A: And it's always good to have an voice recognition for the remote controller.\nSpeaker B: And also the cost for the speech recognition.\nSpeaker B: Yeah.\nSpeaker C: That's our R&D department.\nSpeaker C: It's just for our sport vocabulary.\nSpeaker A: Yeah, it's only for a limited vocabulary.\nSpeaker A: Yeah.\nSpeaker A: Say, IT commands or so.\nSpeaker A: Yeah, okay.\nSpeaker B: And also the scroller button, how much will it cost?\nSpeaker D: Well, compared to the SIMPRA, SIMPRA, SIMPRA.\nSpeaker A: The scroll button, anyway, from the survey, we never see that people would like to have some scrolling button.\nSpeaker A: Because they are just frightened to use the scrolling or help button.\nSpeaker D: Yeah, I think that as we have seen in the presentation, we have about 50% of the person.\nSpeaker D: Don't use the button.\nSpeaker D: Don't use the button.\nSpeaker D: So I think to have a five simple button is sufficient for our functionality.\nSpeaker D: Yeah.\nSpeaker C: It doesn't mean that the other buttons are not necessary or important.\nSpeaker C: Yeah.\nSpeaker C: But they are just...\nSpeaker A: They are not those much.\nSpeaker D: The thing is that we can add a functionality on the TV screen, like a list of functions.\nSpeaker D: And then you choose with the button to navigate.\nSpeaker C: So they add more power?\nSpeaker C: Or maybe we can make this 10% of the pattern more bigger than the others.\nSpeaker C: Yeah.\nSpeaker D: But if we could have a display, a user interface that is very complete on the TV screen, I think that just five buttons are sufficient.\nSpeaker D: You want to up left, right, down, and enter.\nSpeaker D: And you just select the functionality you want to access, things like that.\nSpeaker D: Yeah.\nSpeaker A: Or it could be like this as the people say if they have an LCD on the remote, not on the television.\nSpeaker A: Because when you have the LCD onto the television screen, you miss the picture in the background.\nSpeaker A: We are most focused on the commands.\nSpeaker A: So if you have an LCD in the remote, you just have a menu and increasing and lower the signs here to change the programs.\nSpeaker A: And this menu, when you press the menu, in the LCD, it displays as you go on pressing the menu.\nSpeaker A: It first displays volume, then the program, then the brightness, contrast and all this stuff.\nSpeaker A: And accordingly you can just increase or decrease.\nSpeaker D: Yeah, but if you look at the LCD, you don't look at the TV screen.\nSpeaker D: So it's not really worth to get to have the image if you don't look at it.\nSpeaker C: And I think it changes the cost of the remote control.\nSpeaker C: Yeah, that has to be checked out.\nSpeaker B: I think there is no contradiction here.\nSpeaker B: Because if there are a few buttons, you don't have to look at your controller anymore because you know where the buttons are.\nSpeaker B: So if you want to control the screen sharpness, you just say sharpness.\nSpeaker B: And then you just press left, increase or decrease button.\nSpeaker B: And the same for the volume.\nSpeaker B: Yeah.\nSpeaker B: And the channel, if you have the speech recognition day, just show your channel.\nSpeaker B: And then you don't even have to look at the controller.\nSpeaker B: So finally, that eliminates the need for LCD.\nSpeaker B: With the help of speech recognizer, you can...\nSpeaker A: I mean, it would be better if you could just check out the cost with LCD and also with the speech recognition.\nSpeaker A: Yeah.\nSpeaker A: And then we could find which would be the most suitable for this case.\nSpeaker A: Yeah.\nSpeaker B: And the third problem was to find the remote control.\nSpeaker A: Always. So 50% of the people say they lose the remote.\nSpeaker D: Well, so we can think about a vocal cover and like find and when the remote control is signed.\nSpeaker D: Yeah.\nSpeaker D: And you can sign to a peep.\nSpeaker D: Yeah.\nSpeaker A: Just to make him beep.\nSpeaker D: Yeah, that's right.\nSpeaker A: That's exactly what I mean by voice commander.\nSpeaker A: Or it could be also something like this.\nSpeaker A: It's always boring to change the batteries of the remote control.\nSpeaker A: So we have someone charger there and whenever we don't use the remote control, we put it in the charger.\nSpeaker A: Yeah.\nSpeaker A: And when we are using that remote and if you miss player somewhere, in the charger we have a small button.\nSpeaker A: And just by pressing the button in the charger, the remote control beeps.\nSpeaker A: Yeah.\nSpeaker A: Wherever it is.\nSpeaker D: That's a good idea.\nSpeaker D: That's simple.\nSpeaker D: Yeah.\nSpeaker A: I mean, it also doesn't require a voice commander.\nSpeaker A: Because there are problems with the voice commander.\nSpeaker C: Sometimes we have to move the charger.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker A: I mean, charge it with the fixed speakers.\nSpeaker A: Because it's always with the electricity plug.\nSpeaker A: Yeah.\nSpeaker D: If there is not enough battery.\nSpeaker D: Also, and the remote is lost.\nSpeaker D: Yeah.\nSpeaker B: That's right.\nSpeaker B: What we can do is we can program a function whereby when you press the switch off TV button, the off button, the remote, there will be an instructional screen.\nSpeaker B: Please charge me.\nSpeaker B: Yeah.\nSpeaker B: You never get it lost because every time you off the computer, the TV, you have asked the command, the TV command remote control will tell you to put it back to where the charger is.\nSpeaker B: That's a good reminder.\nSpeaker A: Yeah.\nSpeaker B: That's right.\nSpeaker B: So you never get lost.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker C: Maybe for some people, but for my lazy people.\nSpeaker B: Yeah, because everything is programming inside.\nSpeaker B: Yeah.\nSpeaker B: So it's all about strategy.\nSpeaker A: And of course, the final point is the fancy look.\nSpeaker A: As we have seen earlier, the remotes which were displayed by Norman, they were on fancy.\nSpeaker A: Lovely.\nSpeaker A: Very big or something with lot of buttons.\nSpeaker A: I think we should have something.\nSpeaker D: Well, the last one with the...\nSpeaker D: Yeah, the two parts was in the original.\nSpeaker A: I mean, you say, if it's like that, even a kid who wants to have a control, he could just plug it and use it.\nSpeaker A: You can't avoid him.\nSpeaker A: But you can have a button for child lock.\nSpeaker E: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So just by pressing the button with some code, you put a lock onto the remote so that he can't use even if I...\nSpeaker D: No, we can think about having on the user interface when you switch on the TV.\nSpeaker D: You can write a code or choose a category if it is kids.\nSpeaker C: Yeah, that's right.\nSpeaker C: Or maybe you have to choose some specific programs for kids.\nSpeaker C: And then...\nSpeaker C: Oh, did the...\nSpeaker C: Just push a kid's button so it's automatically...\nSpeaker C: I'm sorry.\nSpeaker C: So if he carries you much, we...\nSpeaker A: I think these are the four points where the market demands.\nSpeaker A: And so it's for the user interface, the design and industrial design.\nSpeaker D: So for my part, I will check the prices, the prices difference of what to use, where to use.\nSpeaker A: I think it's a little clearer for us in the next meeting that this poverty school being ordered.\nSpeaker B: I think we need to define also the set of vocabularies for the speech recognizer.\nSpeaker B: Yeah.\nSpeaker B: Because if you want this, say we can sort by channels or sort by TV programs, you have to decide the category of vocabularies for them.\nSpeaker B: If numbers, they are easy.\nSpeaker B: But if you want to name the channel by name.\nSpeaker D: Well, I think we can have just numbers for channels and you can say to your remote control like sports.\nSpeaker D: And then on the TV, you have a list with sports program playing now.\nSpeaker A: No, you have a problem there.\nSpeaker A: You see, if you have a voice, comments, and you are watching a score on a basketball score or something.\nSpeaker A: And if the score comes 25, 35, you just say 25.\nSpeaker A: And suddenly, the channel goes to 25.\nSpeaker A: So I think there should be a prefix to some numbers.\nSpeaker D: Well, every possible world has a probability to come about the TV.\nSpeaker A: You just check out the probability that saying TV 25 and just ordinary 25.\nSpeaker A: Ordinary 25, almost there's a probability of being said about 60, 70%.\nSpeaker A: And TV 25, I don't know, it should be around one or two percent.\nSpeaker A: So it's better to have some prefix code before the number.\nSpeaker B: But I think that the user would like to associate the channel, we'll call the channel rather than the number.\nSpeaker A: Yeah, something, some code.\nSpeaker B: You say number channel number five of your TV corresponds to something else in the channel.\nSpeaker B: So some people may want to say I want to see this channel.\nSpeaker A: That would be too big and it would be difficult for the vocabulary also.\nSpeaker B: Check with the R&D department, the capability I recognize.\nSpeaker C: If you have the name of the channel, it would be difficult to say just the name of the channel.\nSpeaker C: Well, it's convenient for the user.\nSpeaker C: Yeah, but you have to have all the name of the channel in your vocabulary.\nSpeaker A: Also, it might be you just forgot the channel name, you'd only know the number.\nSpeaker C: Or maybe the user can create his own vocabulary just for announcing the name of channels and include the vocabulary.\nSpeaker B: I think there's another way you can do is that when the user press a button to choose the channel, for example, then what you can do is that you can make the TV screen to split them into small little squares of images where you have a snapshot of every channel.\nSpeaker B: So let's say it's a 4x4 matrix of the images.\nSpeaker B: So now what you do is looking at all the 16 channels available at one time, you just use the control button there.\nSpeaker B: You just choose the option you want and then you just hit the button and then you go to that channel.\nSpeaker C: Or let's say the user create his own vocabulary.\nSpeaker B: So you don't use the speech recognizer in that way?\nSpeaker C: No, just you have in the beginning you have to train, you have to create a vocabulary.\nSpeaker C: Oh, okay.\nSpeaker C: Yeah.\nSpeaker C: I also think each channel with the name.\nSpeaker D: I also think about another problem if there is more than one person watching TV.\nSpeaker A: Yes, that's why I just wanted to recognize this.\nSpeaker D: One of the speakers recognizer should be able to distinguish between the two because I remember when I was a young child with my sister.\nSpeaker D: I want to watch this, I want to watch this.\nSpeaker D: We always want to watch different emission at the same time.\nSpeaker D: It could be a problem if someone passes by when you are watching TV and say, oh, TV, 30.\nSpeaker D: And just try.\nSpeaker A: And in the same lines we would have one more problem.\nSpeaker A: If you are using the television in the different environments, say in the factories or in the shops, there is a lot of noise and these voice commands if they fail to work, that would bring a bad name by a repetition for our company.\nSpeaker A: So I think we should specify some pre-requirements if we want to use the voice commands.\nSpeaker A: Say that it should be used in a silent, homely environment or something.\nSpeaker C: I will switch on or switch off the speech.\nSpeaker C: Yeah, that should be an option.\nSpeaker D: Yeah, I think that is a good option because it is simple and simple to implement.\nSpeaker A: These are the practical problems.\nSpeaker A: So we need to take care of them in the design.\nSpeaker C: We will discard them in the next meeting. It is time to close this meeting.\nSpeaker C: We still have five minutes.\nSpeaker B: Oh, it came around three minutes back.\nSpeaker A: So we are releasing.\nSpeaker A: I think we are just plastic.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker E: Okay.\nSpeaker D: So I see.\nSpeaker A: Okay.\nSpeaker A: Save another.\nSpeaker A: 30 minutes.\nNone: Okay.\nSpeaker C: It is time to make this order or after the meeting.\nNone: What is it?\nSpeaker C: When we are going to meet.\nNone: Don't take that word.\nNone: No, after the next one.\nNone: Yes, it is on your person.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker E: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "ES2007c", "summary": "This meeting is mainly about three presentations from three team members to talk about what the team is supposed to be doing. First, it was the Industrial Designer's presentation on energy source and case material ; then, User Interface introduced potential buttons, the layout design, and the color of the remote control ; and at last, Marketing made a presentation on three important marketing points that the team should pay attention to, which are fancy appearance, innovative technology and user-friendliness. Various discussions went on among team members concerning their presentations.", "dialogue": "Speaker F: I feel like Madonna with one of these on.\nSpeaker E: I feel like Madonna with one of these on.\nSpeaker F: It's cool, we're dancing, they're saying it's like.\nSpeaker F: I've always wanted one of these, I really have.\nSpeaker F: Where did you buy them from?\nNone: Hello everybody.\nSpeaker D: Back again for another wonderful meeting.\nSpeaker D: Everyone ready?\nSpeaker D: Okay we can hold on for a minute.\nSpeaker F: Oh my gosh.\nSpeaker E: I figured with the spam thing if you can't.\nSpeaker B: I enjoyed it.\nSpeaker D: Are you ready?\nSpeaker D: Okay, right.\nSpeaker D: First off we'll recap from our last meeting.\nSpeaker E: We got together to basically decide on what it is that we were actually supposed to be doing and who we all are and stuff like that.\nSpeaker E: We'll get a bit more of an idea together of what's going on.\nSpeaker E: What we're going to talk about in this meeting is now that we know what it is that we're doing, now we know that it's a TV remote and stuff.\nSpeaker E: You guys have just been off doing some R&D for that.\nSpeaker E: Research and development, I haven't heard that before.\nSpeaker E: There are all sorts of new terms and technology today.\nSpeaker D: Whether they be on computer or on the white board or whatever you want, if you have any preference or order.\nSpeaker E: I'd like to hear on the Catherine actually first.\nSpeaker G: I think she's still finishing her.\nSpeaker G: I think she's still in the same position.\nSpeaker G: For the energy sources we can have a basic battery, hand in a mode which was used in the 50s for torches.\nSpeaker G: I think any of us remember the 50s.\nSpeaker G: It's like a crank thing.\nSpeaker G: You can have a kinetic provision of energy which is used on some watches these days.\nSpeaker B: There are sometimes combinations that calculate the two combinations of battery, but also using some solar panel.\nSpeaker G: Does anybody know?\nSpeaker G: I think the kinetic provision of energy could be nice, but I don't know if it's worth the cost.\nSpeaker G: For the case of the remote itself, there can be a general case which is just a flat one.\nSpeaker G: It kind of suits the palm of your hands.\nSpeaker G: The case material itself can be either plastic or latex rubber or titanium.\nSpeaker G: For each of them you have cases where, for example, titanium you can't use it if we're choosing a double curved case.\nSpeaker G: We can't choose titanium and if we're choosing solar cells then we can't choose latex for the case material.\nSpeaker G: We just have to take that into account.\nSpeaker G: But if we're choosing just the flat case then we can go for anything.\nSpeaker G: I think we discussed earlier, Rombi, or as I probably think so.\nSpeaker D: That might be an idea of using a rubber, but then it's a squeezable ring.\nSpeaker G: It also doesn't break as easily.\nSpeaker E: When a TV program is watching the match and teams just lost you can fuzz it across the room and it will bounce off the wall back at you.\nSpeaker D: I like that idea.\nSpeaker G: So rubber is the thing.\nSpeaker F: I think rubber is...\nSpeaker E: Rubber, we're all going with it, we're all like that idea.\nSpeaker E: You think you can mark it now?\nSpeaker F: But after my fashion thing, I think you'll realise that rubber is more...\nSpeaker D: Oh, we like rubber.\nSpeaker G: Okay.\nSpeaker G: And then there are the push buttons.\nSpeaker G: So you can have basic push buttons or a squall wheel like you have on a mouse.\nSpeaker G: Or you can have LCD to give you a display.\nSpeaker G: Squall buttons as well.\nSpeaker G: So if you use a rubber double curved case, you must use rubber push buttons.\nSpeaker G: So if we're going for rubber, then we have to decide for the case.\nSpeaker G: And if we choose the double curved, then we have to go for rubber push buttons.\nSpeaker E: Well, I think we've decided that it's going to be a rubber case.\nSpeaker G: Yeah, but if it's a double curved one or not.\nSpeaker G: If it's not a double curved, then we've got the choice for the push buttons.\nSpeaker G: If it's a double curved, we've got to go for rubber push buttons.\nSpeaker G: If that makes sense.\nSpeaker E: Well, push buttons instead of the wheel.\nSpeaker B: Yeah.\nSpeaker B: If it's rubber, isn't it malleable anyways?\nSpeaker B: I mean, it doesn't matter if it's double...\nSpeaker B: I mean, isn't a rubber case, I mean, it's completely flexed.\nSpeaker B: I mean, it flexes to whatever they want.\nSpeaker B: It's the difference between a normal rubber case and a rubber double.\nSpeaker B: A rubber double double double.\nSpeaker D: You see, you've got...\nSpeaker D: Okay, the energy.\nSpeaker D: That's one thing.\nSpeaker D: I'll have a big back, please.\nSpeaker G: Then you have...\nSpeaker G: The case is whether it's flat or curved.\nSpeaker G: And that's...\nSpeaker G: We don't care if it's rubber or not, but then we've decided that we're going for rubber for the case material.\nSpeaker G: So if we've chosen rubber, and now we have the choice for the case whether it's flat, single curved or double curved.\nSpeaker G: And I'm just saying, if we choose it to be double curved, then we need to go...\nSpeaker G: I don't know why, but we need to go for rubber push buttons.\nSpeaker G: Okay.\nSpeaker G: So either...\nSpeaker G: I don't know, we just need to decide on the...\nSpeaker E: Let's have rubber push buttons, hi.\nSpeaker B: Okay.\nSpeaker B: Go rubber.\nSpeaker G: Let's go crazy.\nSpeaker G: And then do I have a last slide?\nSpeaker G: Yes, I do.\nSpeaker G: So the push buttons themselves, they can be just simple, they can be...\nSpeaker G: So that's just the electronics behind the push buttons.\nSpeaker G: Yeah.\nSpeaker G: And the price that goes with it, so the simple push buttons are going to be the cheapest.\nSpeaker G: If we get a scroll wheel that's a higher press range, if we get an advanced chip, which is used for the LCD, the display thing, then that's even more expensive.\nSpeaker A: Okay.\nSpeaker A: Simple.\nSpeaker A: Chip on print.\nSpeaker A: It's a long print in through the infrared center.\nSpeaker E: Okay.\nSpeaker E: One I'm not understanding here is...\nSpeaker E: Okay.\nSpeaker E: Advanced chip on print, which I presume is like one PCB.\nSpeaker E: And that's got all the electronics on one board, including the infrared center.\nSpeaker E: Yeah.\nSpeaker E: Right.\nSpeaker E: What alternatives do we have to that?\nSpeaker G: Well, if it's not chip on print, then I guess you get different chip components and you build them separately and it doesn't include the infrared.\nSpeaker G: It's less expensive.\nSpeaker G: So technically speaking, it's not as advanced, but it does the job too.\nSpeaker E: So why would we not go for that?\nSpeaker E: If it's something that's inside the unit.\nSpeaker E: It doesn't, yeah, it doesn't affect whether the customer is going to buy it or not.\nSpeaker E: Yep.\nSpeaker E: So let's not go for it.\nSpeaker E: And it's as long as it works.\nSpeaker G: Yeah, yeah.\nSpeaker G: I agree.\nSpeaker E: So let's not bother with the chip on print.\nSpeaker G: So it's either the scroll wheel or the push buttons.\nSpeaker A: Yeah.\nSpeaker A: So yeah, push buttons.\nSpeaker F: What about the just the now sample sensor?\nSpeaker D: What about what?\nSpeaker F: The sample sensor.\nSpeaker E: Well, what do we need to speak before in the remote control unit?\nSpeaker F: Oh, I don't know.\nSpeaker F: Be cool.\nSpeaker G: It'd be cool, but they're saying they've just developed it.\nSpeaker G: I'm just guessing both.\nSpeaker G: It's going to be the most expensive option probably.\nSpeaker B: The speech recognition option is it doesn't seem really very promising for us.\nSpeaker B: Yeah.\nSpeaker E: It's not something that we want to go into with the product.\nSpeaker B: Yeah, the example that they're already using it for is with the coffee machine where basically you can program a sample that when you say something with a response and you program the response as well, it's just clips that you record yourself so you can program your coffee maker that when you say good morning to it, it says hello Rick or whatever.\nSpeaker B: Yeah.\nSpeaker B: I mean, it's not.\nSpeaker B: It would be one thing if it was speech recognition where you say something and it turns the TV on like turn the TV on them.\nSpeaker B: It comes on, but it's not that it's just gives you an answer and she gives you a verbal response.\nSpeaker B: So, yeah, I mean, like what's the point of hello remote?\nSpeaker F: I mean, I thought it was when they say really lonely.\nSpeaker F: Yeah, really lonely.\nSpeaker F: I thought when they said voice recognition, they meant like channel five.\nSpeaker F: Channel five.\nSpeaker F: No, that would be promising.\nSpeaker B: It's just a remote that talks to you.\nSpeaker B: Oh, forget about it.\nSpeaker B: Okay.\nSpeaker B: I mean to certain cues.\nSpeaker G: Yeah.\nSpeaker G: Okay, so I'll go back maybe to the previous slide and we can decide for each problem what we should choose.\nSpeaker G: So for the energy source, do we go for the battery or the.\nSpeaker B: Yeah, I'm fine with the basic battery.\nSpeaker E: Basic battery.\nSpeaker E: It's cheap.\nSpeaker E: It's cheerful.\nSpeaker E: It's worked.\nNone: It does work.\nSpeaker G: Cheap option.\nSpeaker G: Are you happy with that?\nSpeaker G: Yeah.\nSpeaker G: Okay.\nSpeaker G: So we'll go for the battery.\nSpeaker G: Then the case.\nSpeaker G: Do you want it flat or curved or single or double?\nSpeaker E: We were going with the late with the RSI rubber.\nSpeaker B: Yeah.\nSpeaker B: So we wanted rubber double curve.\nSpeaker B: Do the yeah.\nSpeaker F: So it will look like something like this.\nSpeaker G: Double wafer please.\nSpeaker G: Okay.\nSpeaker G: So then if we use double curved case, then we have to use.\nSpeaker E: We go for the simple buttons.\nSpeaker E: So rubber push buttons.\nSpeaker E: Yeah.\nSpeaker E: That's fine.\nSpeaker E: So it's cheaper still rounds.\nSpeaker E: It sounds kind of funky.\nSpeaker E: And we can also mark it as because we were saying earlier in your research.\nSpeaker E: Do you have the people getting the complaining of RSI?\nSpeaker E: And this is anti RSI.\nSpeaker G: Yeah.\nSpeaker E: Yeah.\nSpeaker E: So that's another marketing point that we can use.\nSpeaker F: Oh, the rubber push button.\nSpeaker F: So.\nSpeaker F: Do you have to move?\nSpeaker E: Anything is going to have buttons.\nSpeaker E: Even if it's a jog wheel, it's still repetitive.\nSpeaker G: I thought they would give an option of flat buttons or.\nSpeaker G: You can still get.\nSpeaker E: You still get repetitive strain injury, whether you're pressing a button or pressing a flat bit of screen.\nSpeaker E: It's the fact that you're pressing the same doing the same movement.\nSpeaker E: It's not actually what you're doing.\nSpeaker E: But the fact that this rubber is actually used in these anti RSI.\nSpeaker E: It's specific.\nSpeaker B: We're giving them a way to burn off steam.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker B: So they can sit there and not that watching TV should be that stressful.\nSpeaker E: And you can fuzz it across the room.\nSpeaker B: Throw it at your children.\nSpeaker B: Yeah.\nSpeaker G: Alright.\nSpeaker G: That's me.\nSpeaker E: Don.\nSpeaker E: Okay, Gabrielle.\nSpeaker E: Let's hear from you about the interface.\nSpeaker B: Yeah.\nSpeaker B: Some of what I have to say ties into what Katherine was just talking about.\nNone: Sorry.\nNone: Wow.\nNone: Okay.\nSpeaker B: So I'm continuing with the user interface.\nSpeaker B: The topic.\nSpeaker B: And so basically I consulted with our manufacturing division.\nSpeaker B: It sounded like Katherine was also speaking with them.\nSpeaker B: I also took Reese's marketing findings from the last meeting into consideration.\nSpeaker B: Because I think that's crucial as far as what keys we're going to include, what buttons we're going to include and how they're laid out.\nSpeaker B: And so the manufacturing division sent some samples of interface components that we might be interested in using that have been used in other products like the coffee machine.\nSpeaker B: So I already mentioned the speech recognition interface.\nSpeaker B: I guess we basically vetoed that idea.\nSpeaker B: It's pointless.\nSpeaker B: It's just a sample sensor, sample output.\nSpeaker B: It would just be probably the most expensive part of our remote without any actual interesting functionality as far as operating the TV.\nSpeaker B: So yeah, they also give the, they suggested the idea of using a spinning wheel like you use on the side of an mp3 player like iPod.\nSpeaker B: So we've already addressed that.\nSpeaker B: I think that would actually be worse for something like RSI.\nSpeaker B: And I have a feeling that we're interested in something more general, but they suggested going a little bit into a niche like either gearing a remote towards kids or you could have a lot of colors.\nSpeaker B: It might be funny or playful in some way or something for the elderly where the remote's very large and the buttons are very large and there's only a few buttons.\nSpeaker B: But we can discuss this, but it sounded like from our last meeting, we really wanted something that was general, but done well.\nSpeaker B: So the key layout and design are really crucial. You don't want, you want people to be able to quickly access the buttons that they use a lot without always pressing a long one.\nSpeaker B: And I didn't mention that we need a power button in our last.\nSpeaker B: I can give you an example here of a good layout and bad layout.\nSpeaker B: So this would be an example of bad layout where you have volume down, but they have a V on both of them.\nSpeaker B: It's sort of confusing for the user.\nSpeaker B: Do we have a good one?\nSpeaker B: This is the example of the giant remote that's impossible to lose.\nSpeaker B: Brilliant.\nSpeaker B: Hello.\nSpeaker B: Something for kids.\nSpeaker B: Yeah.\nSpeaker B: I think my personal preference is we've all talked about and seem to be on the same page.\nSpeaker B: So I was against the speech recognition and against going towards anything in the niche sense.\nSpeaker B: I think it should be more general.\nSpeaker B: I didn't think the spinning wheel or the LC display were crucial for us.\nSpeaker B: And that's it for me.\nSpeaker E: Okay.\nSpeaker E: Well, what are our definite decisions on this then as a team?\nSpeaker E: The interface type we're going for.\nSpeaker B: So we're not going to have any sort of display.\nSpeaker B: It's just going to be buttons.\nSpeaker B: I think we should have a limited number of buttons, ideally, I mean, power, channel up, channel down, volume up, volume up, numerical, keypad.\nSpeaker B: And some sort of, we'll either have a lock button like we mentioned or a cover or something like that.\nSpeaker B: Because it's to the point where we need to decide about it.\nSpeaker B: Okay.\nSpeaker E: And we're not.\nSpeaker E: Yeah.\nSpeaker E: Well, now that we've decided.\nSpeaker E: Are we going to, are we, are we decided on whether we're going to supplement it with anything, you know, colors or, you know, gimmicky bits to it.\nSpeaker E: We're not.\nSpeaker B: We're just going to go for something.\nSpeaker B: It seems like we wouldn't want to make it too busy and to sort of.\nSpeaker B: Dottie, but.\nSpeaker B: Yeah, I would say maybe a couple of colors like a, like a black with yellow in somewhere like maybe the R.\nSpeaker B: Yeah.\nSpeaker B: Yellow.\nSpeaker B: Something like that.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Okay, Ducky.\nSpeaker E: Yeah, I don't have any other questions on this.\nSpeaker B: I guess the fact also that we are having a rubber case would prevent us from having the cover function that we thought of before.\nSpeaker B: I mean, yeah, sure.\nSpeaker A: If we want to have a lock of some sort of.\nSpeaker B: Yeah, I think that should be.\nSpeaker B: I mean, I can speak with the button department, but I think that it should be easy to have a button that just prevents the other buttons from operating.\nSpeaker B: So that should be simple.\nSpeaker A: Cool.\nNone: Right.\nSpeaker F: Well, I'm just basically letting you know what's happening in the.\nSpeaker F: Yeah.\nSpeaker F: And what the fashions are for next year.\nSpeaker F: So yes.\nSpeaker F: So from looking at this is trends and fashions and also recent investigation that we've done in the road control market.\nSpeaker F: We have found that from the road control market.\nSpeaker F: These are like most important aspects like we really need to, which we've already probably discussed.\nSpeaker F: The most important aspect is look and feel so the rock control has to look and feel fancier than the ones that already that we already have.\nSpeaker E: So it has to be why should people buy this when everybody got a remote that came with the TV.\nSpeaker F: Second should be technologically innovative.\nSpeaker E: So what's that mean?\nSpeaker F: Technologically, it should be like.\nSpeaker F: Work.\nSpeaker F: Basically, I guess it should work.\nSpeaker B: It should be it should be maybe cutting edge in some sense.\nSpeaker B: I mean, have something that's a little more technologically advanced.\nSpeaker E: Okay.\nSpeaker E: The trouble is, is we've already decided that we're going with the stuff that works already.\nSpeaker E: That's cheap.\nSpeaker B: Actually, I mean, these first two points we've already sort of gone away from because our rubber one is not fancy.\nSpeaker B: I mean, it's different, but I wouldn't say like a rubber remote is fancy.\nSpeaker B: If that's what people want, then we we're going in the wrong direction and it's not technologically innovative.\nSpeaker B: You know, maybe we could.\nSpeaker F: That's why I think in Bluetooth, because if you like put up Bluetooth and there's a Bluetooth remote control, everybody's going like, oh, because Bluetooth is in thing nowadays, like really.\nSpeaker F: People and when it comes to marketing, like that's what people go for.\nSpeaker F: They don't really care whether, you know, at the end of the day, where they works properly or not.\nSpeaker F: Of course they do.\nSpeaker F: Well, they do, but it's not.\nSpeaker E: 100% that's your first thing you go, oh, I'm not going to buy that because I don't know if it works or not.\nSpeaker F: Yeah, but it looks good.\nSpeaker F: If it looks good and it's it can just be there for decoration.\nSpeaker E: Okay.\nSpeaker E: Well, what do you two think about this?\nSpeaker B: Is the advantage of Bluetooth that you can just like synchronize it with other electronics?\nSpeaker E: Yeah, I don't understand what.\nSpeaker B: It's basically what it allows you to do.\nSpeaker E: Yeah, and then this is just going to do all this is being used for is your television.\nSpeaker F: Yeah, but I mean, people...\nSpeaker E: That would mean you'd need a television that has Bluetooth in it, which no television does, does it?\nSpeaker F: If you're looking at something that's going to be bought by people, you have to make it new, you have to make it.\nSpeaker E: That would mean we'd have to make it television as well.\nSpeaker G: Bluetooth would, for example, in a volume thing to connect.\nSpeaker G: For example, if you get a call on your mobile phone, but your mobile phone is downstairs or something, you would get on your television.\nSpeaker G: You being called by this person right now.\nSpeaker G: Things like that.\nSpeaker E: No, that would be your telephone syncing in with your television.\nSpeaker B: Yeah, that wouldn't be the remote.\nSpeaker B: Yeah.\nSpeaker G: But if you get Bluetooth on the remote, you'd be able to...\nSpeaker E: The television would have to be a Bluetooth.\nSpeaker E: It's just trying to kind of advance it.\nSpeaker E: Well, it doesn't have to be a Bluetooth.\nSpeaker E: And there is no such thing.\nSpeaker F: Like, it doesn't have to be, you know, Bluetooth. That was just an idea.\nSpeaker F: But like, it needs to be something that, you know, is new.\nSpeaker F: Whether it's a battery, it could be something really...\nSpeaker F: Really minor, you know? Like, but I think we are really keeping to what is already out there.\nSpeaker F: And people have already seen it. People have already got it.\nSpeaker F: If we want something new, we need to move away from what we already have.\nSpeaker F: And...\nSpeaker F: Maybe the Canadian...\nSpeaker F: Just go creative.\nSpeaker G: For the vision of energy, then.\nSpeaker G: It's been done for watches, but I haven't seen that for...\nSpeaker G: Okay.\nSpeaker D: Yeah.\nSpeaker D: That's very good.\nSpeaker F: I think it can market it. Never have to change the battery again.\nSpeaker F: And...\nSpeaker E: And this is all tying in very nicely.\nSpeaker E: The fact that it's made out of this rubber, we can throw it about...\nSpeaker E: We should encourage people to throw their remote controls about, because it charges itself up.\nSpeaker G: And in little characters, you say, yeah, but not too much.\nSpeaker G: But yeah, by the squeezing it...\nSpeaker B: Then we can make the squeezing of the rubber be the generating...\nSpeaker B: Yeah.\nSpeaker E: Yeah.\nSpeaker E: That's a great idea. Well done.\nSpeaker F: Third.\nSpeaker F: The most important aspect is...\nSpeaker F: Is it's easy to use.\nSpeaker F: And I think we've all...\nSpeaker F: All about that.\nSpeaker F: What's that?\nSpeaker F: Okay.\nSpeaker F: In the fashion, how it's supposed to look, next is fashion very much in...\nSpeaker F: Frozen vegetables.\nSpeaker F: Other than are like being themed for cloths, shoes and furniture.\nSpeaker F: So next year, people will be buying.\nSpeaker F: I found this really funny.\nSpeaker F: You know, strawberry shaped chairs.\nSpeaker B: So we've got our keys that are like a broccoli key.\nSpeaker B: Yeah.\nSpeaker B: And then I'll kind of...\nSpeaker B: I want to watch the pineapple channel.\nSpeaker F: And as it's rubber, the feel, which is in this year, it's spongy.\nSpeaker F: So it's not quite spongy.\nSpeaker F: Well, it's spongy.\nSpeaker E: That's where we got that.\nSpeaker E: Yeah.\nSpeaker E: We're ahead of the game there.\nSpeaker F: Yeah.\nSpeaker F: And so, personal...\nSpeaker F: What I was just saying, like...\nSpeaker F: Move away from the current.\nSpeaker F: Remote controls like the look and the feel of the current ones.\nSpeaker F: And change the look and feel while still keeping to the company's image.\nSpeaker F: Basically.\nSpeaker F: So yeah.\nSpeaker F: Okay.\nSpeaker F: This is what I have to say.\nSpeaker F: So we're moving in the right direction.\nSpeaker F: All right.\nSpeaker E: Yeah.\nSpeaker E: No, this is good.\nSpeaker E: So through all that, we're right.\nSpeaker E: So we're going to go back to going with the kinetic thing.\nSpeaker B: Yeah.\nSpeaker E: That's great.\nSpeaker E: Using the spongy rubber that we were talking about, that's anti-RSI.\nSpeaker E: As you squeeze it, you're not only therapeuticising yourself, you are charging the batteries.\nSpeaker E: Mm.\nSpeaker E: And I'm not sure about the buttons being in the shape of fruit though.\nSpeaker E: Yeah.\nSpeaker E: I don't know how we...\nSpeaker D: No vegetables.\nSpeaker D: Oh, but yeah.\nSpeaker F: Well...\nSpeaker F: Maybe make it like fruity colours or something.\nSpeaker F: Or...\nSpeaker B: The power button could be like a big apple.\nSpeaker G: Yeah, but apple would sue you.\nSpeaker B: Yeah, this is true.\nSpeaker B: They don't own all images of apples.\nSpeaker B: So sue the Beatles, though.\nSpeaker B: Okay.\nSpeaker B: We'll make it a pomegranate.\nSpeaker B: A big pomegranate.\nSpeaker E: Yeah.\nSpeaker E: Well...\nSpeaker E: Okay, it seems like the only thing that we haven't really finally agreed on is its image.\nSpeaker E: Like, yeah, we're saying no, we don't want it to be fruit and vegetables, but we don't know what it should be.\nSpeaker E: Or like, we're going, yeah, it should look slick, but what do we mean by slick sort of thing?\nSpeaker F: I think if it's rubber, it needs to be...\nSpeaker E: I mean, you said earlier on that it should be funkeys.\nSpeaker F: I think it should be...\nSpeaker F: I mean, what do you associate with rubber, you know, like really different colours?\nSpeaker F: Keep it clean.\nSpeaker F: Okay, sorry, I used the wrong word.\nSpeaker F: What do you associate with the material?\nSpeaker F: The material.\nSpeaker F: That material.\nSpeaker F: Um...\nSpeaker F: Like, I'm just thinking bright colours.\nSpeaker F: Bright natural colours.\nSpeaker F: Nothing too...\nSpeaker E: Bright, but not too bright.\nSpeaker F: Yeah.\nSpeaker F: Like, no lime green or bright yellow or bright pink.\nSpeaker F: It'll make it different colours, so anybody can do like...\nSpeaker E: Like, volume buttons should be all the same colour and the channel button should be one colour and stuff like that, do you mean?\nSpeaker F: Yeah.\nSpeaker F: And on the back of it, have the logo.\nSpeaker A: Okay.\nSpeaker A: What do you do?\nSpeaker C: No.\nSpeaker A: Sure.\nSpeaker C: Why not?\nSpeaker A: Okay.\nSpeaker B: The one thing I'm wondering about, I hope that we're not going too much down a gimmicky road of...\nSpeaker B: Yeah, I mean, if somebody goes into this store, they're going to see like three or four normal remotes and then a big spongy pink tomato.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: What a nice present.\nSpeaker E: It's the remote controlled bar though.\nSpeaker F: Well, I can say in this country, you'll get, you know, lots of people wanting something really funky and cool.\nSpeaker F: Like, and kids will be walking in with their parents saying, Well, I mean, I want to buy that one.\nSpeaker F: And parents will see the pro in it as well, because it's like kids won't break it.\nSpeaker F: It's not breakable.\nSpeaker F: We've thrown around.\nSpeaker F: Especially with younger kids.\nSpeaker F: You know, they can pick it up.\nSpeaker F: And yeah, the only thing is that really small kids might chew on it.\nSpeaker E: Well, it's going to be too proof.\nSpeaker F: Yeah.\nSpeaker F: So, I'm going to write that down.\nSpeaker F: So it's quite, it's quite like user friendly and also for different families, like, like families as well.\nSpeaker F: Little kids and older, like teenagers will like it.\nSpeaker F: I think especially maybe younger girls, if it's in pink, they'll be like, Oh yeah, I think we're not controlled for their room.\nSpeaker F: So what do you say?\nSpeaker E: Maybe we should mark it in different colors.\nSpeaker E: Yeah.\nSpeaker E: For different.\nSpeaker E: So we could do like the pink range, the blue range, the green range.\nSpeaker F: So like you walk in, you're like, Oh, I like that remote control because it's so bright.\nSpeaker F: And then, and then the shop assistant comes up and says, Oh, what color would you like it?\nSpeaker F: And they go like, Oh, I can do the color.\nNone: Wow.\nSpeaker F: So it puts, I think even the customer into more control of what they're buying instead of you got all color.\nSpeaker F: It's either that or nothing.\nSpeaker F: So they also get to pick.\nSpeaker F: Well, personally, I like walking into a shop and choosing a color.\nSpeaker B: I mean, that, that seems to work well with products like iPod, where you know you have a variety of colors that people feel like they're customizing it when they buy it.\nSpeaker G: You see how many, you know, there are some colors that I wouldn't, I would never choose and I'd be curious how many people choose that.\nSpeaker E: Well, I suspect can be down to market researching, you know, that's easy enough to find out what colors are more popular.\nSpeaker G: Produce less of the colors.\nSpeaker G: Okay.\nSpeaker D: All right.\nSpeaker D: Well, um,\nSpeaker E: all right, that's what we're going to aim to get together by the next meeting then. Um, is from you, Catherine, you know, you're going to be working on the look and feel design.\nSpeaker E: Gabrielle, you're going to be working on, you know, come up with the user interface design, then basically you two are going to be working together.\nSpeaker E: Unless you won't be going off to separate offices.\nSpeaker E: Um, and I think you're going to get a chance to play with some clay.\nSpeaker E: All right.\nSpeaker E: I think you know, come on, kick, you know, be it for the next meeting, be able to come in and show us some, some, some physical ideas.\nSpeaker E: Cause at the moment, you know, it's, it's hard. We were kind of going, yeah, it's got to feel nice.\nSpeaker E: It's got to look cool and that it's, you know, now we can actually start, you know, some sort of physical, something or other.\nSpeaker E: That sounds great.\nSpeaker E: Um, and you're going to be working on the product evaluation.\nSpeaker A: Um,\nSpeaker E: and I will be, uh, talking to the bosses, basically, and, uh, fielding off some more spam and, uh, that's it really.\nSpeaker E: Keeping things, well, you know, hopefully, keeping things together. Um, yeah, that's, this is, this is, uh, good.\nSpeaker E: So we know for definite, you know, we've now got some definite things going on.\nSpeaker E: We definitely know how it's powered. We definitely know that, um, it's going to be a simple buttons. We're not going to be going for the new technological chip on print, the expensive things. So we, we're keeping the costs down.\nSpeaker E: It's this rubber casing that we can sell as, um, fun and funky.\nSpeaker E: Don't have any other remote controls that are made out of this stuff.\nSpeaker E: People, you know, people are saying, oh, standard stuff gives you repetitive strain and injury.\nSpeaker E: So this remote control is designed to do the complete opposite.\nSpeaker E: Supposed to fix it rather than cause it.\nSpeaker E: Um, this, you know, we're going to, we're going to, as a sort of extra something on it, you know, what, what extra things, other extra things this product have.\nSpeaker E: We're looking to this lock key facility. I mean, whether or not it happens.\nSpeaker E: It's almost possible. I don't know, but something to look into.\nSpeaker E: Okay. I think that's, um, well done, everybody.\nSpeaker E: Anyone have any, uh, any questions? Everyone know what they're doing?\nSpeaker E: Cause if you don't, you'll, I'm sure you'll soon get any mail about it.\nSpeaker B: I think we all know what we need to do now.\nSpeaker F: This gives you a little detail.\nSpeaker E: Okay. Right. Well, it's, um, we're still a couple of minutes until our meeting's due to finish.\nSpeaker E: But, um, I got a note saying that you two will be working together.\nSpeaker E: So before you disappear off.\nSpeaker A: I'm not moving yet.\nSpeaker A: Hold, hold fire.\nSpeaker B: Um, does the pay raise immediately come into effect on our next paycheck or is that?\nSpeaker E: Um, I think it's, uh, yeah, I think it's going to be part, part more of a profit sharing on the product.\nSpeaker B: So we're buying, I mean, we're getting futures and they come.\nSpeaker E: Yeah, I think, I think that's, I think that's the way it's going to happen at the moment.\nSpeaker B: So we really have an incentive to make this remote work.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: I'd like to share in the coffee machine profits because that's really doing well.\nSpeaker E: I want to share in the spice rocket.\nSpeaker E: Did you see that this company sent a sign with my display ship?\nSpeaker B: This company has its fingers in a lot of different parts.\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: What is it?\nSpeaker B: The money making department.\nSpeaker C: Yeah.\nSpeaker B: I did notice looking at, I mean, they told you that they wanted whatever our product to represent the, the rr brand and to be immediately.\nSpeaker B: Yeah.\nSpeaker B: I mean, if you look like at the coffee maker at the corporate webpage, it's not obvious.\nSpeaker B: Our, our, well, it is to us because we worked here for years and we use that coffee machine.\nSpeaker B: But it's doesn't have the color scheme of the company anywhere on it.\nSpeaker B: Obviously it's just silver and black.\nSpeaker G: That's true.\nSpeaker E: And the spice ship doesn't have the, you think it have a massive hour and all down the side of it.\nSpeaker E: It doesn't.\nSpeaker E: I still want one.\nSpeaker E: Okay.\nSpeaker E: I've just had a sign, but I should not say finish the meeting.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I guess that's us.\nSpeaker E: Yeah.\nSpeaker E: It's not telling, it's not saying do anything in particular just yet.\nSpeaker E: So maybe you should go back to your own offices.\nSpeaker B: Okay.\nSpeaker B: All right.\nSpeaker G: Are we taking this off?\nSpeaker A: Yeah.\nNone: Yeah.\nSpeaker E: Yeah.\nNone: So it's YouTube.\n"}, {"meeting_id": "ES2007b", "summary": "This meeting is mainly about three presentations from three team members. First, it was Industrial Designer's presentation on the working design of the remote control; then, User Interface introduced the layout design of the remote control; and at last, Marketing made a report on the user experience survey and potential user-friendly functionalities. At the same time, various discussions among team members went on concerning their presentations, on the industrial design, user interface, appearance and functionality of the remote control.", "dialogue": "Speaker B: I think first.\nNone: Okay.\nNone: Who's going to do PowerPoint presentation?\nNone: I think we are.\nNone: You will as well.\nNone: I thought we are.\nNone: Yeah, I have one too.\nNone: Okay.\nSpeaker A: I thought we are.\nSpeaker A: Yeah, I have one too.\nSpeaker C: Yeah.\nSpeaker A: Yep.\nSpeaker B: I just want to, because basically I can't really crap it remembering everyone's names so I just want to rather than go out and miss marketing and miss this and miss that.\nSpeaker B: I want to know your names again.\nSpeaker B: I just can leave this up here because I'll, you know.\nSpeaker C: So, Katherine, C-A-T-H-E-R-I-N.\nSpeaker B: Okay, and Gabriel.\nSpeaker B: E-L-G-L.\nSpeaker B: Okay, and your-\nSpeaker D: I am Ryser, R-E-I-D-B-L-S-A.\nSpeaker B: R-E-I-S-A-N. Double S-A.\nSpeaker D: Yeah, yeah, yeah.\nSpeaker D: Thanks.\nSpeaker B: Oh, gosh.\nSpeaker B: It's just a bit nicer calling people by their names, I think.\nSpeaker B: Okay, right, welcome to meeting B.\nSpeaker B: This is going to go a lot better than the last meeting, basically.\nSpeaker B: Because I know what I'm supposed to be doing now.\nSpeaker B: I am your project manager.\nSpeaker B: And, yeah, I'm just here to, you know, sort of liaise between the three of you and get things going, get people talking, and I'm going to be making notes and sending them off to the powers that be and stuff, basically.\nSpeaker B: Right, this, for the purposes of this meeting, what this meeting is all about is I'm going to have some presentations from all three of you what you've been working on for the last wee while when you haven't been getting hit with spam on your computers and, you know, filling out silly questionnaires and things.\nSpeaker B: But hopefully you've been actually been doing something productive, so we're going to, each of you are going to give us a little presentation.\nSpeaker B: Then we're going to work, you know, from each of your presentations, we'll talk about what we actually need as a final coming together of it all.\nSpeaker B: And then we'll, yeah, we'll do a sort of concluding, anything else comes up at the end of it.\nSpeaker B: How long is the meeting?\nSpeaker B: This meeting, it's not very long, it's probably down to about 35 minutes now.\nSpeaker B: So I want each of your presentations to not be too long, five minutes, something like that.\nSpeaker B: If you haven't done a PowerPoint thing, it doesn't matter, it's just says that you, that's just one particular medium.\nSpeaker B: If you haven't had time to prepare one, you can draw stuff on the note board, you can talk to us, you can, you know, however you want to do your presentation, basically, you can don't feel pressurized into using this thing, because I don't.\nSpeaker B: Okay, so, you're okay over there?\nSpeaker B: I'm fine, yeah. Are you joining in with this meeting?\nSpeaker B: I'm going to do your first presentation.\nSpeaker B: You're just doing some internet shopping there.\nSpeaker B: No, no, no, no, I'm done.\nSpeaker B: Okay, jolly good.\nSpeaker B: All right, let's have, well, we all know that it's a remote control that we're going to be dealing with.\nSpeaker B: I think the first thing we should look at is probably the, what it is that it's actually supposed to be, so that's going to be UCath, right?\nSpeaker B: Okay.\nSpeaker B: We're going to hear from you first.\nSpeaker C: Okay, just going to do this.\nSpeaker B: You've done that to worry about screwing it in just.\nSpeaker B: There you go.\nSpeaker C: Cool.\nSpeaker C: Okay.\nSpeaker C: I've got a very quick.\nSpeaker C: Okay.\nSpeaker C: So the working design, I've got a very quick presentation on this.\nSpeaker C: So, I've, oh no, you can't see a thing.\nSpeaker C: Oh, well, I'm going to join on the board then.\nSpeaker C: It's in blue and I couldn't change it.\nSpeaker C: It's fine on my screen, but nevermind.\nSpeaker C: So, the idea is that we've got the energy source, which in our case, oh, well, okay.\nSpeaker C: Nevermind.\nSpeaker C: So, I think maybe two batteries.\nSpeaker C: I don't know what they call a lot six or something like that.\nSpeaker C: And then, then on the remote control itself will have the sender for the signal, which could be an infrared signal.\nSpeaker C: Which will be sent by an electronic chip.\nSpeaker C: And the chip will be controlled by the user interface.\nSpeaker C: So, we'll hear about that later from Gabriel.\nSpeaker C: And the sender will send to the tele itself an infrared signal to tell it to switch on or switch channels.\nSpeaker C: And that's it really for the working design.\nSpeaker B: Great.\nSpeaker C: Sorry, the presentation wasn't very okay.\nSpeaker B: I prefer to put up further here than touch personally.\nSpeaker C: Should I erase this?\nSpeaker C: Do you want to just give us a moment?\nSpeaker B: I just want to copy this down.\nSpeaker B: I don't know if you guys have got any questions for Catherine on any of this.\nSpeaker C: Suggestions.\nSpeaker D: Is a battery like the only way of?\nSpeaker C: Well, it's just you don't want it plugged in really.\nSpeaker A: Yeah, alternate energy source like wind power.\nSpeaker C: No, it's not.\nSpeaker C: You blow on it.\nSpeaker D: No, because like, charge is always changing.\nSpeaker D: Batteries can get like annoying the battery down and maybe.\nSpeaker D: I don't know.\nSpeaker C: I think changing your batteries once every six months is not really a pain.\nSpeaker B: Yeah, it's worked for the last 50 years.\nSpeaker A: One question I have and I don't know how much control we have over this is as far as the infrared signal.\nSpeaker A: Do we have control over how far away you can be from the receiving unit that the TV and still have it be operational?\nSpeaker A: Maybe we want one with a strong signal straight.\nSpeaker B: How far away is your television?\nSpeaker B: It's never going to be more than it's never going to be.\nSpeaker B: Unless you've got a TV the size of a football pitch.\nSpeaker B: It's not doesn't have to go that far, does it?\nSpeaker B: Doesn't have to go through a wall because you're not going to be looking through a wall.\nSpeaker D: Yeah, but if like you're on the phone in the other room and you tell the television office and think you really want to go into the telephone down the garden.\nSpeaker C: But we can make the signal strong enough to go through walls if she finds it.\nSpeaker C: I do not think about that.\nSpeaker D: How about blue tea?\nSpeaker D: Instead of using infrared.\nSpeaker D: Why not?\nSpeaker C: I just think that it's going to cost more.\nSpeaker B: Yeah, it sounds like we don't want to overcomplicate things.\nSpeaker C: It's a fancy idea. It's quite nice.\nSpeaker C: But then I don't know.\nSpeaker C: Either if you want to watch the teller you're in the room, you're going to...\nSpeaker B: Yeah, exactly.\nSpeaker B: Basically we're designing and marketing television remote control unit.\nSpeaker B: We're not designing something that you can plug in a headset to and connect to your laptop computer and stuff.\nSpeaker C: We can keep the idea.\nSpeaker C: We can see the later stage maybe.\nSpeaker B: Okay.\nNone: Well done.\nNone: Go for real.\nNone: Let's hear from you on...\nSpeaker B: Do you need the board or six?\nSpeaker B: I'm just going to use the power case.\nSpeaker A: Technical struggles.\nSpeaker A: Okay.\nSpeaker A: So while this is warming up, there it is.\nSpeaker A: So I'm doing the user interface design.\nSpeaker A: And basically, as far as methods, I was looking at...\nSpeaker A: Looking at already existing remotes, trying to find some inspiration from designs that are already out there.\nSpeaker A: Thinking of what we can retain, what we can do away with, what we can perfect a little bit as far as design.\nSpeaker A: We don't want to do something that's too radical of a change, I guess.\nSpeaker A: People want a remote that's familiar, that has their favorite functionalities and does the basics.\nSpeaker A: So we can improve what's out there and maintain the basic functionality that we want.\nSpeaker A: So things that seem like absolute must-haves would be volume control to up-down keys for that.\nSpeaker A: Channel keys up-down, but then also numerical keypad so that they can just key directly to the channel that they want rather than doing up-down.\nSpeaker A: And a mute button.\nSpeaker A: One thing that I didn't include here that I forgot that we talked about last time was doing some sort of block function.\nSpeaker A: Okay. I don't know. That's one possibility.\nSpeaker A: And so in the research that I was doing, there's basically two types of remotes, one that are engineering-centered and one that are more user-centered.\nSpeaker A: Actually, I don't know if I can access the web page from here.\nSpeaker A: I can show you.\nSpeaker A: So this is an engineer-centered one. See, it's rather busy, but it also lets you play your movie, stop your movie, fast forward, all this.\nSpeaker A: It's a breeze frame, and this is a user-centered one.\nSpeaker A: It's easier to just glance at this and see what's possible to do. You're not going to be staring at it for five minutes.\nSpeaker A: All right. Judging from what we all talked about during our last meeting, I kind of gathered that that's what we were going after, or the direction that we were going in at least.\nSpeaker A: So the engineering-centered ones provide a lot of functionality, but it can be a little bit overwhelming.\nSpeaker A: So the user-centered ones just focus on ease of use.\nSpeaker A: And it sort of overlaps with the marketing person, Lisa, because we need to find out what people want before we make the same assignments.\nSpeaker A: I think we're all good.\nSpeaker A: So, yeah, that's me.\nSpeaker B: Great. Okay.\nSpeaker B: Now, I just have a question for you.\nSpeaker B: This research that you've been doing, looking at other existing units and stuff.\nSpeaker B: Have you found that anyone else just looked into the locking?\nSpeaker A: That seemed like a novel idea as far as I know.\nSpeaker A: I mean, obviously another that exists, like you said, in mobile phones.\nSpeaker A: Yeah.\nSpeaker A: I've never seen that in all my years in business.\nSpeaker A: Do you guys have a yay or a yay on that feeling about whether that's really necessary?\nSpeaker B: I would say if it's simple to do, which I think it probably should be, even if it's a physical, you know, like a physical switch or a physical cover for the remote, even something like that.\nSpeaker B: Then, yeah, it's like, you know, like you said earlier on a flip thing, something like that, but you know, being physical.\nSpeaker B: Look into.\nSpeaker B: I've had word down from head office that something that we should be centered at.\nSpeaker B: Well, something we should take into account is we've got to keep the corporate image within this remote control unit.\nSpeaker B: It's got to look like it's in the R&R.\nSpeaker B: You know, the company is it's from what I can see from our other products, a yellow with blue writing on them.\nSpeaker A: And our motto is we put the fashion in electronics.\nSpeaker A: There you go.\nSpeaker A: I think we have to carry that mental.\nSpeaker B: So it's kind of got to look, it's got to look new and, you know, something fashionable if remote control.\nSpeaker B: Well, if telephones can be fashionable, then maybe remote control units can be.\nSpeaker A: Yeah, these I think we can.\nSpeaker A: So we talked about the layout in my presentation, or why I didn't mention it really is just the sort of like the ergonomic.\nSpeaker A: Yeah, I think we can make big improvements over these two that you see here.\nSpeaker A: Yeah.\nSpeaker A: Everything is going to ergonomic.\nSpeaker A: You know, there's, you know, nice for your computers that are very ergonomic and keyboards and that could be one of our niches sort of in the market, I guess.\nSpeaker A: Okay, fantastic.\nSpeaker B: Well done Gabrielle.\nSpeaker B: Lisa.\nSpeaker B: Let's plug you in, baby.\nSpeaker B: Here.\nSpeaker B: The blue thing.\nSpeaker D: All right, Alice is getting a little tangled.\nSpeaker A: I guess that's the answer.\nSpeaker A: Is that in?\nSpeaker A: Yeah, then you just have to do function, effort, and it should come up.\nSpeaker D: Well, function, effort.\nSpeaker D: Yeah, it is.\nSpeaker B: It's really well.\nSpeaker D: All right.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Well, from our marketing report, we observed remote control users in the usability lab.\nSpeaker D: And also gave, so this is research, and we also gave participants questionnaires to fill out.\nSpeaker D: Total number of people tested were 100, so, you know, so that 100 people were tested.\nSpeaker D: And these were the findings.\nSpeaker D: So 75% of users find the remote control ugly.\nSpeaker D: Okay, so they don't like the look of the remote control.\nSpeaker D: 80% of them would spend more money if the remote control looked really cool and fancy.\nSpeaker D: I think we all agree with that.\nSpeaker D: Current remote controls do not match well with the operating behavior of the user.\nSpeaker D: So they don't like the way they operate.\nSpeaker D: It doesn't match how people behave.\nSpeaker D: And 30% of the users say that they only 10% of the buttons on the remote.\nSpeaker D: So probably if you have like one, two, three, four, five, the whole up to 10, they probably don't use those.\nSpeaker B: They only use up and down channels.\nSpeaker D: Because we've only got five channels.\nSpeaker D: That's another thing.\nSpeaker D: 75% of users say they zap.\nSpeaker B: Not quite sure what they mean, zap.\nSpeaker B: So you want to navigate the channels quickly?\nSpeaker D: Yeah.\nSpeaker D: It takes too much time to learn how to use a new remote, I think, especially for the older generation.\nSpeaker D: I know my grandmother doesn't like mobile phones.\nSpeaker D: It takes ages to work out.\nSpeaker D: Anyway, and they also get lost in the room.\nSpeaker D: So I know we can find them.\nSpeaker D: So maybe tracking devices is a good idea.\nSpeaker D: Personal.\nSpeaker D: Wow.\nSpeaker B: You are a child of technology, aren't you?\nSpeaker D: So yeah.\nSpeaker D: I was thinking something easy to use, especially for older people.\nSpeaker D: That's to look really cool, flash and groovy people to buy it.\nSpeaker D: And it's easy to find.\nSpeaker D: So I don't know whether maybe.\nSpeaker D: And also we asked them whether they wanted, whether they'd be interested in voice activating.\nSpeaker D: So voice activation.\nSpeaker D: So, and this was what we came up with.\nSpeaker D: And if you look 15 to 25, this is age.\nSpeaker D: Sorry.\nSpeaker D: It's out of the way.\nSpeaker D: Age groups.\nSpeaker D: So 15 to 25 said like 92, 91% of them.\nSpeaker D: So basically the majority, except for the 45.\nSpeaker D: 55 year olds for some reason didn't want a voice activated one.\nSpeaker D: And neither did the older generation, but the younger generation, who we are catering for, like who have lost the money nowadays, do want a voice like speech recognition.\nSpeaker B: But do the younger generation have the money?\nSpeaker B: No, I would absolutely not.\nSpeaker B: They don't.\nSpeaker B: It's older generation of the ones that have gone out.\nSpeaker D: Well, the 25 to 35 year olds, 35 and the 35 to 45, 45.\nSpeaker D: 45.7%.\nSpeaker B: People from the age of 30, there's a big drop off there.\nSpeaker B: For people up to the age of 35, they're kind of saying, yeah, they want it.\nSpeaker B: But no, they're not the sort of, most people that have the money, people from the age of 35 to 55.\nSpeaker B: Yeah, that would be my guess.\nSpeaker B: Because they're the ones that have been working for 20 years.\nSpeaker B: Wow.\nSpeaker B: And that's quite a minority there.\nSpeaker B: So yeah, it's not even like 50, 50.\nSpeaker B: That's 35%.\nSpeaker D: These guys are growing up.\nSpeaker A: What about just from the perspective of our manufacturing cost?\nSpeaker A: I mean, if it's 1250 per unit.\nSpeaker A: Yeah.\nSpeaker D: I mean, okay, there's voice activation might not be the best.\nSpeaker B: I would say scrap that straight off.\nSpeaker D: Also with buttons, I think called RSI.\nSpeaker D: So, repetitive strain injury.\nSpeaker D: Yeah, repetitive strain injury.\nSpeaker D: I think people who watch TV maybe too often keep changing the channel and hurt their wrist.\nSpeaker B: So, that's something we should have a look into.\nSpeaker B: When designing the ergonomics to see, have a look if there's any medical background we can find out about this.\nSpeaker C: But then it could be just touching.\nSpeaker B: We just want to need to cover our asses so that people aren't going to sue us in 10 years time.\nSpeaker B: So, your remote control gave me repetitive strain injury.\nSpeaker A: We should probably consult with our legal department.\nSpeaker B: Yeah.\nSpeaker B: They're having an lunch break at the moment.\nSpeaker B: Yeah.\nSpeaker B: We'll see if I can get elders for next meeting.\nSpeaker A: I think we can do some really in that department, the ergonomic department.\nSpeaker A: Yeah.\nSpeaker D: It makes some really good improvements.\nSpeaker D: Maybe the buttons not so high up so you don't have to press so much really just like flat buttons.\nSpeaker D: Yeah.\nNone: Okay.\nSpeaker B: So, that is me.\nSpeaker B: That's great.\nSpeaker B: Thank you very much for that, Risa.\nSpeaker B: Okay.\nSpeaker B: So, we've basically.\nSpeaker B: We've decided that it's going to, you know, we're going for basic television remote.\nSpeaker B: It's got to be safe to use.\nSpeaker B: It's got to look cool.\nSpeaker B: It's got to be cheap.\nSpeaker B: Now, going back to the industrial design event, you know, we're looking at whether to use the technology.\nSpeaker B: We're looking at whether to use maybe infrared or Bluetooth.\nSpeaker B: I think we should just go with the simplest option on everything.\nSpeaker B: And that would be infrared energy source and batteries.\nSpeaker B: When we can look into using the, you know, the little tiny, we need batteries or special long lasting batteries.\nSpeaker B: But in this note, I don't think there's any point in making a remote control unit that's going to last for 50 years because technology would have changed and, you know, we won't have televisions in 10 years time.\nSpeaker B: So, I think we're all pretty sussed on that.\nSpeaker B: Anyone have any questions?\nSpeaker B: Everybody happy in their work?\nSpeaker A: It seems like we're all pretty much on the same page.\nSpeaker B: Now we're pretty now.\nSpeaker B: This is good.\nSpeaker B: We've got a good structure going on.\nSpeaker B: We all know where we're going to.\nSpeaker B: Have you been, have any of you not been getting loads of crap spam on your computers?\nSpeaker A: It's probably just you because you're the project manager.\nSpeaker D: Just question.\nSpeaker A: Yeah.\nSpeaker B: Yeah, just stuff.\nSpeaker B: Okay.\nSpeaker B: Oh, did you guys find out if we can email stuff to each other?\nSpeaker B: Yeah, you can.\nSpeaker B: Right.\nSpeaker B: You all know my email address.\nSpeaker B: No, I don't.\nSpeaker C: In this project announcement, you've got the addresses, I think.\nSpeaker C: So project manager, it's a one at a.\nSpeaker B: Can you all email me your email addresses?\nSpeaker A: You have the message.\nSpeaker A: You have them.\nSpeaker B: But we'll send you an email.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker A: Sorry, we headed towards a big yellow and black remote as far as maybe that's our next meeting.\nSpeaker B: Is it yellow and black or is it yellow and blue?\nSpeaker A: I kind of thought it was blue writing on a yellow background, but I might be just kind of like white and blue on a black background with white with yellow borders.\nSpeaker A: Okay.\nSpeaker B: Maybe that's getting ahead of ourselves.\nSpeaker B: I think I'm up with a few with a couple of different ideas.\nSpeaker D: Can we have different colors and all of us are somebody can choose different colors?\nSpeaker B: Well, see, the thing is, is we've got to keep the company image.\nSpeaker B: It's got to say people have got to look at this remote control and instantly recognize that it's a real reaction.\nSpeaker D: But if it's R, R, it would be real reaction.\nSpeaker B: Like if that's a lot of companies that called R.\nSpeaker B: They just slow, but this is the slogan. This is the, the, the, the, the.\nSpeaker A: In a something that came on down from, from the higher.\nSpeaker A: Yeah.\nSpeaker A: I mean, we're sort of beholden to them.\nSpeaker D: So we have to have it one color.\nSpeaker D: Well, not necessarily.\nSpeaker A: Not one color, but we have to incorporate it.\nSpeaker B: Not necessarily even.\nSpeaker B: It's just has to, you have to look at the product and instantly know that it's one of our products.\nSpeaker B: As opposed to a Sony product or, you know, a Panasonic product.\nSpeaker B: It's got to.\nSpeaker B: So maybe so you can have a look at our other products and see if they all follow us.\nSpeaker B: Similar vein.\nSpeaker B: Perhaps.\nSpeaker C: But it could come in different colors and have the R, R colors just somewhere.\nSpeaker C: Like just around the lock button could be the R logo or colors.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker B: Well, this is all your department.\nSpeaker B: Okay.\nSpeaker B: Well done, everybody.\nSpeaker B: And I think we stop for lunch now.\nSpeaker D: Are we finished ahead of schedule?\nSpeaker B: We might possibly have done.\nSpeaker A: All right.\nSpeaker A: See you soon.\nSpeaker B: If we've finished five minutes before the meeting is supposed to finish then that means we get an extra five minutes for lunch.\nNone: Yeah.\nNone: There you go.\nNone: All right.\nSpeaker D: I just have to do a few little bits pieces.\nNone: I have to write down, but.\nNone: Take your head sets off.\nSpeaker B: Kick back.\nSpeaker B: Smoke them.\nNone: If you got a little bit of a kickback smoke them.\nNone: Yeah.\nSpeaker D: There you go.\nNone: All right.\nSpeaker B: I just have to do a few little bits pieces.\nSpeaker B: Smoke them.\nSpeaker B: If you got them.\nSpeaker B: All right.\n"}, {"meeting_id": "IS1007b", "summary": "In the discussion, team members did presentations and talked about the design of the remote control. Project Manager introduced the aim of the second meeting and new project requirement. Project Manager said that the team did not need to pay attention to teletext and lighting adaptive. Then Industrial Designer, User Interface and Marketing gave presentations respectively. Industrial Designer proposed that they could add software functionalities on programmable chips and could use the FPGA. User Interface focused on reducing the number of buttons and recommended navigation buttons. Marketing discussed a lot of ideas including using a speech recogniser, the size of the remote control and lighting adaptive options. In the end they discussed some topics like teletext and accessing the Internet.", "dialogue": "Speaker B: So we are coming again for the second meeting.\nSpeaker B: So for the aim of this meeting, now it's to make presentation about the world for each one and you will take the decision, the design and the functionality of the world.\nSpeaker B: And we have, I think I got a new project requirement.\nSpeaker B: So I think the teletext becomes outdated.\nSpeaker B: So the popularity of the, since the popularity of the internet and I think we don't need writing adaptive.\nSpeaker B: So the remote control should be only used for the telefiction.\nSpeaker B: And of course we shouldn't have our immersion in the design.\nSpeaker B: So let's start with the, the industry designer.\nSpeaker C: Yeah.\nSpeaker C: All right.\nSpeaker B: So you can use the whiteboard.\nSpeaker C: When I have a PowerPoint presentation stored in my, in my personal folder.\nSpeaker C: So I think you can reach it from here.\nSpeaker D: Just go to Explorer or Open.\nSpeaker D: The participant.\nSpeaker C: The participant to this one.\nSpeaker B: Yeah.\nSpeaker D: Open.\nSpeaker D: Okay.\nNone: No.\nNone: Open.\nSpeaker A: You can use up an email.\nSpeaker A: All right.\nSpeaker C: So I will talk about the working design.\nSpeaker D: Slideshow.\nSpeaker D: Will slideshow or this one.\nSpeaker C: And well, I will present my first idea on how to build the, our new remote control for television.\nSpeaker C: So can you go one page down please.\nSpeaker C: So I think the first thing to do is to define the hardware components needed to achieve what we want to do.\nSpeaker C: So I'm thinking, I think I'll do a survey about what is, what is available on the market and what is the cheapest possible thing we can use.\nSpeaker C: Then I will try with the technical team to, to build a prototype and to see how this little box would look like.\nSpeaker C: And how, and we can start troubleshooting first communication problems or things like that.\nSpeaker C: And when we are ready with this first prototype, I think we can have some software functionalities on a problemable chip like browsing by content or things like that.\nSpeaker C: Okay.\nSpeaker C: So can you go down there?\nSpeaker C: So what I think for now is we, we don't want to have a remote control which is wide.\nSpeaker C: So I think we can use that tree for the implementation.\nSpeaker C: Then two programmable chips for both software functionalities and communication.\nSpeaker C: And the communication with TV set is made through infrared communication.\nSpeaker C: So this is the schema of the story.\nSpeaker C: Of the future, I'm remote control.\nSpeaker C: So you can see the components, battery and the two chips goes to the infrared connection to the TV set.\nSpeaker A: What is the other chip for?\nSpeaker A: The one on top.\nSpeaker C: The one on top is for the, well, the functionality, the functionalities and the red, sorry, the green one is to, well, putting things together, transform the data into the format to protect the color, to communicate with the TV set.\nSpeaker A: Okay.\nSpeaker C: And that's it.\nSpeaker C: I think we should use a FPGA for the functionalities which is easy to...\nSpeaker C: What is FPGA?\nSpeaker C: It's field programmable something.\nSpeaker C: Get various.\nSpeaker D: So how is it different from the Bluetooth?\nSpeaker C: Well, FPGA is just cheap.\nSpeaker C: You can program it with whatever you want.\nSpeaker C: Okay.\nSpeaker C: And the Bluetooth chip is just responsible to make communication between the two devices.\nSpeaker A: So these are the, they have to work together or do they have to work together to separate choice?\nSpeaker C: The FPGA will produce the data to send.\nSpeaker D: Okay.\nSpeaker D: Or something like, is it hardware, the first one?\nSpeaker D: And the second one is for the software.\nSpeaker A: It's the software part.\nSpeaker A: Yeah, to run.\nSpeaker A: To make it run.\nSpeaker A: Okay.\nSpeaker A: So you can control it, you're not booking if you want, right?\nSpeaker A: Yeah.\nSpeaker C: All right.\nSpeaker C: And that's it for the working design.\nSpeaker C: So if you have any questions.\nSpeaker B: Okay.\nSpeaker D: And how about the battery power?\nSpeaker D: You mean that battery will be somewhere there and the remote control, the power to the battery comes through in private?\nSpeaker D: No, no, no.\nSpeaker C: I think we have to have a embedded battery in the...\nSpeaker C: Into the...\nSpeaker C: Into the...\nSpeaker C: Yeah, into the...\nSpeaker C: More compact.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: And I don't think it will need very much power to make it run.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: You can put it on a charger when you don't need to use it.\nSpeaker C: It's a good idea.\nSpeaker D: Yeah.\nSpeaker D: That's right.\nSpeaker D: Having a charger rather than putting a battery cell is always...\nSpeaker D: Yeah.\nSpeaker D: People don't like it to have...\nSpeaker D: To buy the batteries when they run out.\nSpeaker D: We just make a small charger and put it...\nSpeaker A: Yeah.\nSpeaker A: Because you're using...\nSpeaker A: Yeah.\nSpeaker A: Because I think if some people have PDA, they can use their PDA to control the TV if they want to.\nSpeaker C: Also, but...\nSpeaker C: But...\nSpeaker A: I don't know.\nSpeaker C: I think the goal is to sell our...\nSpeaker C: We don't want to make a PDA.\nSpeaker C: So...\nSpeaker C: Okay.\nSpeaker C: So...\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: So, charger for the identity.\nNone: Here's the key.\nSpeaker B: Um...\nSpeaker A: So, it's mine.\nSpeaker A: It's mine.\nSpeaker B: Oh.\nSpeaker B: Partition point one.\nSpeaker A: No?\nSpeaker A: Yeah.\nSpeaker B: Okay.\nSpeaker B: Oh, we have 14.\nSpeaker B: So, let's move to...\nSpeaker B: To use the entire first design.\nSpeaker A: Yeah.\nSpeaker A: So, you can open...\nSpeaker B: Dr. Sipa...\nSpeaker B: Three.\nSpeaker A: Yeah.\nSpeaker A: So...\nSpeaker A: So, I'm working on the technical functions design.\nSpeaker A: Okay.\nSpeaker A: Let's show the next line.\nSpeaker A: So, the purpose is to find the important questions to ask is what effects should the apparatus have.\nSpeaker A: So...\nSpeaker A: So, I found on the website...\nSpeaker A: On the internet.\nSpeaker A: Doing everything.\nSpeaker A: Yeah.\nSpeaker A: I spent a lot of time searching.\nSpeaker A: That's right.\nSpeaker A: And I found that the function of remote control is the same messages the television set.\nSpeaker A: For example, switch on, switch off, switch the next channel and so on and so on.\nSpeaker A: So, I found two very good prototypes for this interface from our competitors.\nSpeaker A: So, can you...\nSpeaker B: This are usual functionality.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, ours is a bit different.\nSpeaker A: So, these are two examples.\nSpeaker A: One is from...\nSpeaker A: OpaG, the other one is from...\nSpeaker A: Tasks.\nSpeaker A: Yeah.\nSpeaker A: Engineering Central.\nSpeaker A: Yeah.\nSpeaker A: This is the most competing prototypes I found.\nSpeaker A: But then, look.\nSpeaker A: But then I found, if you look at...\nSpeaker A: You see on the left one, there are too many buttons.\nSpeaker B: And they are small.\nSpeaker A: Yeah.\nSpeaker A: On the right, I tried to play with.\nSpeaker A: The problem is that if I have 100 channels, I have to press the other button to choose the 100 channels.\nSpeaker A: And I have to compose the number, so it's very lousy.\nSpeaker A: So, you move to the next...\nSpeaker A: The next one?\nSpeaker A: Yeah.\nSpeaker A: So, I talk about the problem.\nSpeaker A: And then I look at the user manual, the 100 page is thick, so we don't want that.\nSpeaker A: Yeah.\nSpeaker A: So, I propose that easy to use prototype.\nSpeaker A: You can navigate on TV screen, and we can prove it to the user's preference.\nSpeaker A: And we need to have the TV connected to the internet.\nSpeaker A: So, in order to access the TV programs by XML, and we need some...\nSpeaker A: To do some pre-processing, from the technical aspect, the processing should be done on the TV set, then on the remote controller.\nSpeaker A: Right.\nSpeaker A: And then, the speech recognition is a...\nSpeaker A: Harry says we may just put in... we may KIV.\nSpeaker B: What do you mean by the pre-processing will be done on the TV?\nSpeaker A: Yeah, all the processing is done.\nSpeaker A: The TV is a computer. It has some processing power there.\nSpeaker B: So, we should have specific TV or...\nSpeaker C: So, I use this...\nSpeaker C: A TV with the remote control.\nSpeaker A: Yeah, because...\nSpeaker A: I use this one remote controller.\nSpeaker A: Yeah, I think it is.\nSpeaker C: There is a technology like ShowView, who is already available on most of the TV set, on recorders or things like that, and we can try to get this information onto the remote control to do the processing on the remote controller.\nSpeaker A: Yeah, that's right.\nSpeaker A: So, we do the processing on the remote controller.\nSpeaker A: Yeah.\nSpeaker A: So, can you be using any TV, any conventional TV sets?\nSpeaker B: Yeah.\nSpeaker A: Okay.\nSpeaker A: Speech recognition?\nSpeaker B: Yeah, that's all.\nSpeaker A: The next one.\nSpeaker A: So, I cover a simple design.\nSpeaker A: Just keep the navigation buttons.\nSpeaker A: That's a good idea.\nSpeaker C: I think we don't need really much buttons to... if we have a screen to navigate on TV.\nSpeaker C: Yeah.\nSpeaker C: So, well, four or five buttons.\nSpeaker C: It's sufficient.\nSpeaker C: It's easy to build.\nSpeaker C: Yeah, that does not consume much power.\nSpeaker A: Okay.\nSpeaker A: That's all.\nSpeaker D: Oh, but you have a catch there.\nSpeaker D: Assume that if you want to go to... if you're watching channel two and if you want to go to channel 99, then...\nSpeaker C: Well, then you go to the main menu and you have a go to channel and then...\nSpeaker A: No, because you choose by channel and so...\nSpeaker A: Yeah.\nSpeaker A: You buy TV programs so you don't have 100 channels to choose from.\nSpeaker C: If you go by channel, you don't have to do that.\nSpeaker C: But I think he said if you want to make a big jump.\nSpeaker C: Oh, jump.\nSpeaker C: Yeah, that's right.\nSpeaker C: Well, you have to have a device when you could...\nSpeaker A: A mouse or...\nSpeaker C: Not a mouse, but something that says more clearly that right.\nSpeaker C: Right, left, up, down.\nSpeaker C: Okay.\nSpeaker C: To have the ability to write something to the navigator, maybe directly.\nSpeaker C: Okay.\nSpeaker C: We can think of buttons like in the telephone to send messages or things like that.\nNone: Mm-hmm.\nSpeaker B: But how do you...\nSpeaker B: We can see the external... this remote.\nSpeaker A: It's gonna be small.\nSpeaker A: Yeah, of course.\nSpeaker B: It's too small.\nSpeaker B: It's too small.\nSpeaker A: Under the sofa and we can't find it.\nSpeaker A: So, if you want to find it, you just...\nSpeaker A: You just... it just has to respond to you when you look for the device.\nSpeaker A: Yeah, this one.\nSpeaker A: I don't know how.\nSpeaker D: We'll just give it an aim and we call him.\nSpeaker A: And you respond to you.\nSpeaker A: And there...\nSpeaker B: The other side.\nSpeaker C: Okay, so, next presentation.\nSpeaker C: What is it, print for?\nSpeaker B: So, hi.\nSpeaker D: Okay, after having the inputs from industrial design and user interface, I think most of the points which I want to already covered here.\nSpeaker D: And to find the most interesting features, what the users would be interested in, what we have done is we have put our feedback forms in all the magazines.\nSpeaker D: And the users sent the feedbacks.\nSpeaker D: And based on that, these are the findings which we got.\nSpeaker D: And adding up a speech interface is always good for a TV remote.\nSpeaker D: But the technology, we already know that.\nSpeaker D: What is it?\nSpeaker C: I think it will be a bit complicated to make a speech recognized.\nSpeaker C: It runs on the small...\nSpeaker D: It has to be seen how feasible it is.\nSpeaker A: But I think if you want to recognize numbers, it's a well-studied problem.\nSpeaker A: If you just recognize numbers, it's limited.\nSpeaker D: We can put a limited vocabulary as in the mobile phones.\nSpeaker D: We just have the...\nSpeaker C: And this can allow to choose the program, for instance, without adding buttons.\nSpeaker C: So, it's a good idea.\nSpeaker D: It's not going to take much space also.\nSpeaker D: It's going to be very slim.\nSpeaker D: And next one was the size of the remote control.\nSpeaker D: It has to be, of course, a very slim and small one.\nSpeaker D: And, of course, most of the people are interested in if it's less expensive.\nSpeaker D: So, this is an important criteria here.\nSpeaker B: But do you think what they suggested is possible to...\nSpeaker B: I mean, we have to look for the trade-off.\nSpeaker D: The futures and the cost.\nSpeaker A: I think that if we go for quality, people may tolerate it for high quality.\nSpeaker A: And, of course, it comes with a reasonable cost.\nSpeaker A: Maybe not less, but they may be willing to pay.\nSpeaker A: A little bit more.\nSpeaker D: A little bit more for...\nSpeaker A: With extra futures.\nSpeaker A: Yeah, extra futures.\nSpeaker C: Yeah, speech is an important extra future.\nSpeaker C: Yeah, that's right.\nSpeaker B: But is it useful or not?\nSpeaker B: Well, there is a noisy thing.\nSpeaker D: I mean, for a lazy guys, they could just say nine.\nSpeaker D: And the nine channel comes.\nSpeaker D: Or just say a movie name.\nSpeaker D: I don't go for movie names, but only for the numbers on the channel, or volume up, volume down, brightness, contrast.\nSpeaker D: So, I think that should be a good idea to put these features.\nSpeaker D: And the fourth one was the teletext in various languages.\nSpeaker C: Well, we just have to find a mean how to write a microphone.\nSpeaker C: Well, there's a microphone into the remote control.\nSpeaker C: So...\nSpeaker C: I think it should be all right.\nSpeaker B: What is the teletext?\nSpeaker B: The next thing is that we can do that.\nSpeaker B: We can integrate small microphones in the remote.\nSpeaker B: So, it's not a really problem.\nSpeaker C: Okay.\nSpeaker B: What about lighting adaptive options?\nSpeaker B: Yeah, as this goes on the...\nSpeaker B:...alurementing.\nSpeaker B: But require more traffic.\nSpeaker D: I think not much people are really interested in this point.\nSpeaker D: If it's not really...\nSpeaker A: It is interesting, but we are only concerned with the remote controller.\nSpeaker A: So, you need special TV sets to do that.\nSpeaker A: It's done via this remote controller.\nSpeaker C: It's very complex.\nSpeaker C: Yeah, I think it's a bit complex too.\nSpeaker D: Yeah, it's for the industrial design and users, interface designers to decide if it's going to be working or not.\nSpeaker C: For our next product, our new TV set with...\nSpeaker C: Yeah, the automatic control...\nSpeaker A:...sound adjustment.\nSpeaker A: Yeah, but it's quite possible.\nSpeaker A: But maybe not in this project.\nSpeaker B: So, we have...\nSpeaker B: I think we have...\nSpeaker B:...still we have...\nSpeaker B:...capped off me.\nSpeaker B: So, anything to discuss?\nSpeaker D: I think as I discussed, that four points would be good for unmarketing.\nSpeaker D: The speech interface and the less reasonable...\nSpeaker D:...cause and the lighting adaptation and the teletext.\nSpeaker D: And regarding the teletext, these are the text information which you go on the televisions.\nSpeaker D: If you're watching a foreign movie, you get a subtitle sometimes.\nSpeaker D: And if you're a French guy watching an English movie, you would like to have it in French.\nSpeaker D: And if I'm a German, then I would like to have my options in German.\nSpeaker D: So, the remote...\nSpeaker D: It should have some...\nSpeaker D: It should give me some flexibility of choosing the languages.\nSpeaker D: That's a really good point.\nSpeaker C: Yeah, but what about the new projects requirements?\nSpeaker C: I think we should give up with teletext.\nSpeaker B: I guess it's the popular.\nSpeaker C: Well, so maybe we can already think about a way to access the internet.\nSpeaker C: We had a television.\nSpeaker A: Using the TV to access the internet.\nSpeaker A: Are you thinking about the internet?\nSpeaker C: Yeah, but we can think that in a few years there will be TV set that can access the internet.\nSpeaker C: Yeah, okay.\nSpeaker C: So, it's a good idea if we think about how to build up our remote control with this functionality.\nSpeaker C: Which would come in a few years.\nSpeaker C: So, if we already have it in our remote control...\nSpeaker B: So, you have to...\nSpeaker B:...remod control...\nSpeaker A:...the future.\nSpeaker A: Yeah.\nSpeaker A: The future demand, market demand.\nSpeaker A: Yeah, to design.\nSpeaker C: Well, not to implement it, but to find a way to add this functionality in a...\nSpeaker A: In the future?\nSpeaker C: Yeah.\nSpeaker C: In an easy manner, if it's possible.\nSpeaker A: Because all the TV sets will be digital, right? All the programs, everything will be in digital, then...\nSpeaker A:...and a lot.\nSpeaker A: The system will change and...\nSpeaker A: Yeah.\nSpeaker A:...we have to enter the bed for those changes.\nNone: Yeah.\nNone: Okay.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker B: So, let's go for the...\nSpeaker B:...the Nenshiprik.\nSpeaker B: And...\nSpeaker B: Okay.\nSpeaker B: Thank you.\nSpeaker A: Okay.\nSpeaker A: Thank you.\nNone: Okay.\nNone: All right.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: All right.\nNone: Now\n"}, {"meeting_id": "Bmr010", "summary": "The group talked about the ongoing transcription effort and issues related to the Transcriber tool, which despite its limitations for capturing tight time markings for overlapping speech, will continue to remain in use. Speaker mn014 explained his efforts to pre-segment the signal into speech and non-speech portions for facilitating transcriptions. Recording equipment and procedures were discussed, with a focus on audible breathing and the need for standards in microphone wear and use. ", "dialogue": "Speaker G: Okay, we seem to be recording.\nSpeaker G: All right.\nSpeaker G: Sorry about not pre-doing everything the lunch went a little later than I was expecting.\nSpeaker G: Chuck.\nSpeaker F: Chuck was selling too many jokes.\nSpeaker F: Yeah, pretty much.\nSpeaker F: Does anybody have an agenda?\nSpeaker H: Nope.\nSpeaker B: Well, I've sent a couple of items.\nSpeaker B: I thought somebody had a lot of your...\nSpeaker F: Yeah, we only watched the end of the video.\nSpeaker F: No, why don't we talk about practical things?\nSpeaker B: Well, I can give you an update on the transcription effort.\nSpeaker B: Great.\nSpeaker B: Maybe raise the issue of microphone procedures with reference to the finalyness of the recordings.\nSpeaker E: Okay, transcription, microphone issues.\nSpeaker B: And then, maybe ask these guys, we have great steps forward in terms of the non-speech speech pre-segmenting of the signal.\nSpeaker B: Okay.\nSpeaker G: Well, we have steps forward.\nSpeaker G: Yeah, we prefer this.\nSpeaker F: Okay.\nSpeaker I: We don't know if it's the results.\nSpeaker G: I have a little bit of I-RAM stuff, but I'm not sure if that's of general interest.\nSpeaker F: I-RAM?\nSpeaker F: I-RAM.\nSpeaker F: I-RAM by-gram.\nSpeaker F: Yeah, let's see where we are at 330.\nSpeaker C: Since I have to leave this usually at 330, can we do the interesting stuff first?\nSpeaker C: I'm sorry.\nSpeaker C: Well, I'm sorry.\nSpeaker F: I'm sorry.\nSpeaker F: Yeah, now you get to tell us what's the least specified.\nSpeaker C: Well, I guess the work that's been done on segmentation.\nSpeaker B: I think that was a good thing to start with.\nSpeaker E: Okay.\nSpeaker E: And the other thing, which I'll just say very briefly, that maybe relates to that a little bit, which is that one of the suggestions that came up in a brief meeting I had the other day when I was in Spain with Lola Parto and Javier Ferreros, who was here before, was when I started with what they had before, but adding the non-silence boundaries.\nSpeaker E: So in what Javier did before, when they were doing, he was looking for speaker change points.\nSpeaker E: As a simplification, he originally did this only using silence as a punitive speaker change point.\nSpeaker E: And he did not say, look at points where you were changing broads, for that class, for instance.\nSpeaker E: And for broadcast news, that was fine.\nSpeaker E: Here obviously it's not.\nSpeaker E: And so one of the things that they were pushing and discussing with me is why you're spending so much time on the feature issue, when perhaps if you sort of deal with what you were using before, and then just broad did it instead of just using silence as punitive change point.\nSpeaker E: So then you've got, you already have the superstructure with Gaussian and HM and simple HM and so forth.\nSpeaker E: So there was a little bit of a difference of opinion because I thought it was interesting to look at what features are useful.\nSpeaker E: But on the other hand, I saw that they had a good point that if we had something that worked for many cases before, maybe starting from there a little bit because ultimately we're going to end up with some kind of structure like that where you have some kind of simple HM, you're testing the hypothesis that there is a change.\nSpeaker E: So anyway, just reporting that.\nSpeaker E: So yeah, why don't we do this speech?\nSpeaker D: Speech?\nSpeaker D: Speech?\nSpeaker D: Okay.\nSpeaker D: So what we basically did so far was using the mixed file to detect speech or non-speech or something like that, and what I did so far is I just used our old Munich system which is a nation-based system with Gaussian mixtures for speech and non-speech.\nSpeaker D: And it was a system which used only one Gaussian for silence and the one Gaussian for speech and now I added multi-mixer possibility for speech and non-speech and I did some training on one dialogue which was transcribed by, yeah, we did a non-speech transcription atom, Dave and I, we did for that dialogue and I trained it on that and I did some pre-segmentations for Jane and I'm not sure how good they are or what other transcribes say.\nSpeaker D: They can use it or?\nSpeaker B: They think it's a terrific improvement and it really just makes it a world of difference.\nSpeaker B: And you also did something in addition which was for those in which there were quiet speakers in the mix.\nSpeaker D: Yeah, yeah, that was one one thing.\nSpeaker D: Why I added more mixtures for the speech?\nSpeaker D: So I saw that there were loudly speaking speakers and quietly speaking speakers and so I did two mixtures, one for the loud speakers and one for the quiet speech.\nSpeaker G: And did you hand label who was loud and who was quiet?\nSpeaker D: I did that for five minutes of one dialogue.\nSpeaker D: That was enough to train the system and so it adapts on while running.\nSpeaker C: So what kind of a front end processing did you do?\nSpeaker D: It's just our old, mionic loudness-based spectrum on male scale, 20, 20, 20, 20 bands and on loudness and for additional features which is energy, loudness, modified loudness and zero crossing grade.\nSpeaker D: So it's 24 features.\nSpeaker B: And you also provided me with several different versions which I compared.\nSpeaker B: And so you change parameters, do you want to say something about the perverse?\nSpeaker D: You can specify the minimum length of speech and silence portions which you want.\nSpeaker D: And so I did some modifications in those parameters basically changing the minimum length for silence to have more or less silence portions in the literature.\nSpeaker G: So this would work well for pauses and utterance boundaries and things like that before overlap I imagine that doesn't work at all, that you'll have plenty of sections that are.\nSpeaker B: That's true, but it saves so much time.\nSpeaker B: The transgoers just enormous, enormous savings.\nSpeaker E: That's great.\nSpeaker E: Just one quickly still on the features.\nSpeaker E: So you have these 24 features, a lot of my spectral features.\nSpeaker E: Is there a transformation, principal components, transformation or something?\nSpeaker D: No, we're just, we're sourced too.\nSpeaker D: Originally we did that, but we saw when we used it for our closed talking microphone which, yeah, for our recognition as a mionic, we saw that it's not as necessary.\nSpeaker D: It works as well without the LDA or something.\nSpeaker E: Okay, no, I was curious.\nSpeaker E: I think it's a big deal for this application.\nSpeaker E: Yeah, yeah.\nSpeaker B: Okay, then there's another thing that also Tilo has involved in, which is, okay, and also Dave Galbart.\nSpeaker B: So this is this problem.\nSpeaker B: So we had this meeting, also Adam, before you went away.\nSpeaker B: Regarding the representation of overlaps, because it's present, because of the limitations of the interface we're using, overlaps are not being encoded by the transgoers in this complete and detailed away as it might be.\nSpeaker B: And as might be desired, I think would be desired in the corpus, ultimately.\nSpeaker B: So we don't have start and end points at each point where there's an overlap.\nSpeaker B: We just have the overlaps encoded in the simple bin.\nSpeaker B: Well, okay, so the limits of the interface are such that we were, in this meeting, we were entertaining how we might either expand the interface or find other tools which already do what would be useful, because what would ultimately be ideal in my view, and I think, and I had the sense that it was consensus, is that a thoroughgoing musical score notation would be the best way to go, because you can have multiple channels.\nSpeaker B: There's a single timeline.\nSpeaker B: It's very clear, flexible, and all those nice things.\nSpeaker B: Okay, so I spoke, I had a meeting with Dave Galbart, and he had excellent ideas on how the interface could be modified to do this kind of representation.\nSpeaker B: But in the meantime, you were checking into the existence of already existing interfaces, which might already have these properties, so do you want to see something about that?\nSpeaker D: Yes.\nSpeaker D: I talked with the many guys from Ludwig-Manns-Millianz University who do a lot of transcribing and transliterations, and they basically said they have a tool they developed themselves and they can't give away if it's to our own and it's not supported.\nSpeaker D: But Zanipur, who was at CMU, who was formerly at Munich, and is now with CMU, she said she has something which she uses to do a trans-transferrations, a channel simultaneously, but it's running under Windows.\nSpeaker D: So, I'm not sure if we can use it.\nSpeaker D: She said she would give it to us, it wouldn't be a problem, and I've got some kind of manual down in my office.\nSpeaker G: Well, maybe we should get it, and if it's good enough, we'll arrange Windows machines to be available.\nSpeaker B: I also wanted to be sure, I've seen this is called PROT, the RAA, which I just mean speech\nSpeaker D: and speech. Yeah, but I'm not sure that's the right thing for us.\nSpeaker B: In terms of it being Windows versus Windows, but I'm just wondering if it's not.\nSpeaker G: No, no PROT.\nSpeaker B: Oh, I see.\nSpeaker B: Oh, I see, so PROT may not be.\nSpeaker D: That's not PROT, it's called trans-edit.\nSpeaker D: I see.\nSpeaker E: The other thing, to keep in mind, I mean, we've been very concerned to get all this rolling so that we would actually have data.\nSpeaker E: But I think our outside sponsor is actually going to kick in and ultimately, that path could be smoothed out.\nSpeaker E: I don't know if we have a long-term need to do lots and lots of transcribing.\nSpeaker E: I think we had a very quick need to get something out, and we'd like to be able to do some later because it's interesting.\nSpeaker E: It's interesting.\nSpeaker E: But as far with any luck, we'll be able to wind down the link to PROT.\nSpeaker G: What our decision was is that we'll go ahead with what we have with a not very fine time scale on the overlaps and do what we can later to clean that up if we need to.\nSpeaker B: I was just thinking that if it were possible to bring that in like this week, then when they're encoding the overlaps, it would be nice for them to be able to specify when the start points and then points of overlaps.\nSpeaker B: They're making really quick progress.\nSpeaker B: That's great.\nSpeaker B: So my goal was, my charge was to get 11 hours by the end of the month and it'll be, I'm clear that we'll be able to do that.\nSpeaker G: That's great.\nSpeaker G: And did you forward Morgan, Brian's thing?\nSpeaker B: I sent it to, who can I send that to?\nSpeaker B: I sent it to a list and I thought I sent it to the...\nSpeaker B: Meeting the court list?\nSpeaker B: Oh yeah, okay, so you probably forgot.\nSpeaker B: So Brian did tell me that in fact what you said that they are making progress and that he's going to check the output of the first transcription.\nSpeaker E: I mean, basically it's all a different world, and basically he's on it.\nSpeaker B: Oh, that's just a new development.\nSpeaker B: So this is, so it will happen.\nSpeaker B: Super.\nSpeaker E: Okay.\nSpeaker E: I mean, basically it's just saying one of our best people is on it.\nSpeaker E: It just doesn't happen to hear anymore someone else's face.\nSpeaker C: But about the need for transcription, I mean, don't we, didn't we previously decide that the IBM transcripts would have to be checked anyway?\nSpeaker C: Yes.\nSpeaker C: Positive, augmented, so I think having a good tool is worth something.\nSpeaker C: Yeah, okay.\nSpeaker C: That's a good point.\nSpeaker G: Yeah, Dave Galbert did volunteer and since he's not here, I'll repeat it to at least modify transcriber, which if we don't have something else that works, I think that's a pretty good way of doing.\nSpeaker G: And we discussed on some methods of doing it.\nSpeaker G: My approach originally, and I've already hacked on it a little bit, it was too slow because I was trying to display all the waveforms, but he pointed out that you don't really have to, and then it's a good point, that if you just display the mix waveform and then have a user interface for editing the different channels, that's perfectly sufficient.\nSpeaker B: Yeah, exactly, and just keep those things separate.\nSpeaker B: And Dan Ellis's hack already allows them to be able to display different waveforms to clarify overlabs the things, so that's what it is.\nSpeaker G: They can only display one, but they can listen to different ones.\nSpeaker B: Well, yes, but what I mean is that from transcribers perspective, those two functions are separate, and Dan Ellis's hack handles the choice, the ability to choose different waveforms from moment to moment.\nSpeaker G: But only to listen to, not to look at.\nSpeaker G: The waveform you're looking at doesn't change.\nSpeaker B: That's true, yeah, but that's okay, because they're focused on it anyway.\nSpeaker B: And then the hack to preserve the overlap better would be one which creates different output files for each channel, which then would also serve Liz's request of having single channel, separable, cleanly, easily separable, transcriber tied to single channel audio.\nSpeaker E: Have folks from this been in contact with you?\nSpeaker B: Not directly.\nSpeaker B: If I could have gotten it over list, I don't think so.\nSpeaker E: Okay, well, holidays may have interrupted things, because they seem to want to get absolutely clear on standards for transcription standards and so forth.\nSpeaker E: Oh, this was supposed to be for December.\nSpeaker E: Right, because they're presumably going to start recording next month.\nSpeaker G: Okay, okay.\nSpeaker G: So we should definitely get with them then and agree upon a format.\nSpeaker G: I don't remember your email on that, so Wes, I had not in the loop on that.\nSpeaker E: Yeah, I don't think I mailed anybody.\nSpeaker E: I just think I told them the contact gene.\nSpeaker E: Okay.\nSpeaker E: That's right.\nSpeaker E: That's the point person on it.\nSpeaker G: Yeah, I think that's right.\nSpeaker E: So yeah, maybe I'll hang in there a little bit about it.\nSpeaker E: Okay.\nSpeaker B: Keeping the conventions absolutely simple as possible.\nSpeaker E: Yeah, because with any luck, there will actually be collections at Columbia and collections at UW and Dan.\nSpeaker E: Dan is very interested in doing this.\nSpeaker E: Right.\nSpeaker E: Yeah.\nSpeaker G: Well, I think it's important both for the notation and machine representation to be the same.\nSpeaker G: Yeah.\nSpeaker B: So there was also this email from Dan regarding the speech nonsense presentation.\nSpeaker B: Yeah.\nSpeaker B: I don't know if we want to end Dan.\nSpeaker B: Dave Gowart is interested in pursuing the aspect of using amplitude as basis for the separation.\nSpeaker B: Oh, yeah.\nSpeaker E: He was a correlation.\nSpeaker E: I mean, we hit cross correlation.\nSpeaker E: I mentioned this a couple times before the commercial devices that do voice active miking.\nSpeaker E: Basically, look at the at the energy and each of the mics.\nSpeaker E: You basically compare the energy here to some function of all of the mics.\nSpeaker E: So by doing that, rather than setting any absolute threshold, you actually do pretty good selection of who's talking.\nSpeaker E: And those systems work very well.\nSpeaker E: So people use them in panel discussions and so forth with sound reinforcement.\nSpeaker E: And this, why the guy knew we built them like 20 years ago.\nSpeaker E: So the techniques worked pretty well.\nSpeaker B: It's fantastic.\nSpeaker B: Because here's one thing that we don't have right now.\nSpeaker B: And that is the automatic channel identifier.\nSpeaker B: That would help in terms of encoding of overlaps, the transcribers would have less than just dangling to do if that were available.\nSpeaker B: Yeah.\nSpeaker E: So I think basically you can look at some, you have to play around a little bit to figure out what the right statistic is, but you compare each microphone to some statistic based on the overall.\nSpeaker E: We also have the advantage of having distant mics too.\nSpeaker G: Now using the close talking I think would be much better with the...\nSpeaker E: I don't know.\nSpeaker E: If I was actually working on it, sit there and play around with it and get a feeling for it.\nSpeaker E: But you certainly want to use the close talking.\nSpeaker E: Right.\nSpeaker E: At least.\nSpeaker E: I don't know if the other would add some other helpful dimension or not.\nNone: Okay.\nSpeaker I: What is the difference that classes to code the overlap?\nSpeaker I: You would use?\nSpeaker B: To code some types of overlap?\nSpeaker B: Yeah.\nSpeaker B: So, and I mean that it wasn't transcribed.\nSpeaker B: We worked up a typology.\nSpeaker B: And...\nSpeaker I: It looked like you explained in the book.\nSpeaker I: Yes, exactly.\nSpeaker B: That hasn't changed.\nSpeaker B: So, that's basically a two-tiered structure where the first one is whether...\nSpeaker B: The person who's interrupt continues or not.\nSpeaker B: And then below that there's subcategories, have more to do with, you know, is it simply back channel or is it someone...\nSpeaker B: Completing someone else's thought or is it someone who introduced a new thought?\nSpeaker G: And I hope that if we do a forced alignment with the close talking mic, that will be enough to recover at least some of the time.\nSpeaker G: The time information of when the overlap occurred.\nSpeaker I: We hope.\nSpeaker I: Yeah, who knows.\nSpeaker C: Who's going to do that?\nSpeaker C: Who's going to do forced alignment?\nSpeaker G: Well, IBM was going to.\nSpeaker G: And I imagine they still plan to, but, you know, I haven't spoken with them about that recently.\nSpeaker E: Well, my suggestion now is that all of these things to contact my...\nSpeaker B: Okay, this is wonderful to have a direct contact like that.\nSpeaker B: Well, let me ask you this.\nSpeaker B: It occurs to me.\nSpeaker B: One of my transcribers told me today that she'll be finished with one meeting by...\nSpeaker B: Well, she sent tomorrow, but then she said, but, you know, let's just say maybe the day after just we just on the same side.\nSpeaker B: I consent Brian, the transcript.\nSpeaker B: I know these are...\nSpeaker B: I consent him that it would be possible or a good idea or not to try to do a forced alignment on what we're on the way we're encoding overlaps now.\nSpeaker E: I'll just talk to him about it.\nSpeaker E: I mean, you know, basically he's just a colleague, friend, and...\nSpeaker E: And, you know, the organization always did one.\nSpeaker E: I hope this was just a question of getting the right people connected to him at the time.\nSpeaker G: Is he on the mailing list?\nSpeaker G: Who should add him?\nNone: Yeah, I don't know for sure.\nSpeaker H: Did something happen?\nSpeaker H: Or can he put on this or was he already on it?\nSpeaker E: No, I think it...\nSpeaker E: Yes, something happened.\nSpeaker E: I don't know what it was.\nSpeaker E: Yes, for war.\nSpeaker B: That would be like, that would be great.\nSpeaker F: Right.\nSpeaker F: So, where would...\nSpeaker F: Maybe a brief...\nSpeaker F: Well, let's...\nSpeaker F: Why don't we talk about microponations?\nSpeaker F: Yeah.\nSpeaker G: That would be great.\nSpeaker G: So one thing is that I did look on Sony's for replacement for the mics...\nSpeaker G: for the head-worn ones, because they're so uncomfortable.\nSpeaker G: But I think I need someone who knows more about mics than I do, because I couldn't find a single other model that seemed like it was.\nSpeaker G: It seemed like it would fit the connector, which seems really unlikely to me.\nSpeaker G: Does anyone like no stores or...\nSpeaker G: know about mics who would know the right questions to ask?\nSpeaker E: Oh, I probably would.\nSpeaker E: I mean, my knowledge is just 20 years out of date, but some of it still disconnects.\nSpeaker E: So, where...\nSpeaker H: You couldn't find the right connector that go into these things?\nSpeaker G: Yep.\nSpeaker G: When I looked, they listed one microphone, and that's it.\nSpeaker G: As having that type of connector.\nSpeaker G: And guess is that Sony maybe uses a different number for their connector than everyone else does.\nSpeaker G: And so...\nSpeaker G: Oh, let's look at it.\nSpeaker G: It seems really unlikely to me that there's only one.\nSpeaker B: And there's no adapter for it?\nSpeaker B: It seems like it would be a...\nSpeaker G: Okay.\nSpeaker G: Does that who knows?\nSpeaker G: Who are we buying these for?\nSpeaker G: I have it downstairs.\nSpeaker E: I don't remember off the top of my head.\nSpeaker E: Yeah, we can try and look at that together.\nSpeaker G: And then...\nSpeaker G: Just in terms of how you wear them.\nSpeaker G: I had thought about this before.\nSpeaker G: When you use a product like Dragon Dictate, they have a very extensive description about how to wear the microphone and so on.\nSpeaker G: But I felt that in a real situation, we were very seldom going to get people to really do it.\nSpeaker G: And maybe it wasn't worth concentrating on.\nSpeaker E: Well, I think that that's a good back-off position that I was saying earlier.\nSpeaker E: We are going to get some recordings that are imperfect and pay that's life.\nSpeaker E: But I think that it doesn't hurt the naturalness of the situation to try to have people wear the microphones properly if possible.\nSpeaker E: Because the natural situation is really what we have with the microphones on the table.\nSpeaker E: Oh, that's true.\nSpeaker E: In the target applications, we're talking about people aren't going to be wearing headmine mics anyway.\nSpeaker E: So this is just for these headmine mics are just for use with research.\nSpeaker E: And it's going to make, you know, if Andreas plays around with language modeling.\nSpeaker E: He's not going to be want to be used up by people breathing into the microphones.\nSpeaker G: Well, I'll dig through the documentation to Dragon Dictate and see if they still have the little form.\nSpeaker E: But it does form.\nSpeaker C: It's interesting.\nSpeaker C: I talked to some IBM guys last January.\nSpeaker C: I think I was there. So people who are working on their V-Avoise vacation product.\nSpeaker C: And they said the breathing is really a terrible problem for them to not recognize breathing as speech.\nSpeaker C: So anything to reduce breathing is a good thing.\nSpeaker G: It seemed to me when I was using Dragon that it was really microphone placement helped an enormous amount.\nSpeaker G: So you want it enough to the side so that when you exhale through your nose, the wind doesn't hit the mic.\nSpeaker G: Everyone's adjusting their microphone of course.\nSpeaker G: And then just close enough so that you get good volume.\nSpeaker G: So you know, wearing it right about here seems to be about the right way to do it.\nSpeaker E: I remember when I used a prominent laboratory's speech recognizer.\nSpeaker E: This is a bonus as well, I guess, about 12 years ago or something.\nSpeaker E: And they were perturbed with me because I was breathing in instead of breathing out.\nSpeaker E: And they had models for the Markov models for the breathing out that they didn't have for breathing in.\nSpeaker B: How do I wonder, is whether it's possible to have to maybe use the display at the beginning to be able to judge how correctly.\nSpeaker B: Or how do I do that?\nSpeaker G: I mean, when it's on, you can see it.\nSpeaker G: You can definitely see it.\nSpeaker G: Absolutely.\nSpeaker G: And so, you know, I've sat here and watched sometimes the breathing and the bar going up and down and thinking, I can say something.\nSpeaker G: I don't want to make people self-conscious.\nSpeaker E: Stop breathing.\nSpeaker E: It's going to be imperfect.\nSpeaker E: And you can do some, you know, first-order thing about it, which is to have people move it away from being just directly in front of the middle, but not too far away.\nSpeaker E: And then, you know, I think there's not much because you can't, you know, interfere.\nSpeaker E: You can't fine tune the meeting that much.\nSpeaker B: Right.\nSpeaker B: That's true.\nSpeaker B: It just seems like if something simple like that can be tweaked and the quality goes, you know, dramatically up in my people.\nSpeaker G: And then also, the position of the mic also, if it's more directly, you'll get better volume.\nSpeaker G: So, like, there's this pretty far down, the lower mouth.\nSpeaker B: My feedback from the transcribers is he is always close to Chris McLaren, and just been fantastic.\nSpeaker G: I don't know why that is.\nSpeaker B: You're also, your volume is greater, but still, I mean, they say...\nSpeaker G: I've been eating a lot.\nSpeaker E: It makes it their job extremely easy.\nSpeaker B: I could say something about the why I don't you want to.\nSpeaker B: About what?\nSpeaker B: About the transcribers or anything?\nSpeaker C: Well, why don't we do that?\nSpeaker C: But just to one more remark concerning the SRI recognizer, it is useful to transcribe and then ultimately train models for things like breadth and also the laughter is very, very important.\nSpeaker C: So, in your transcribers mark them.\nSpeaker C: Mark very audible breaths and laughter especially.\nSpeaker B: They are putting... so in curly brackets, they put inhale or breath.\nSpeaker B: Oh, great.\nSpeaker B: And then in curly brackets, they say laughter.\nSpeaker B: Now, they're not being awfully precise.\nSpeaker B: So, the two types of laughter, they're not being distinguished.\nSpeaker B: One is, when sometimes someone will start laughing one there in the middle of a sentence.\nSpeaker B: And then the other one is when they finish the sentence and then they laugh.\nSpeaker B: So, I did some double checking to look through.\nSpeaker B: You need to have extra complications like time tags indicating the beginning and ending up laughing.\nSpeaker B: That's a lot of different than that.\nSpeaker B: What they're doing is, in both cases, just saying Craig is laughing at it.\nSpeaker C: As long as there is an indication that there was laughter somewhere between two words, I think that's a fish worth the most.\nSpeaker C: Actually, the recognition of laughter once you know, it's pretty good.\nSpeaker C: So, as long as you can stick a tag in there that indicates that there was laughter, that would probably be a very interesting, prismatic feature.\nSpeaker B: And I'm going to ask one thing about that.\nSpeaker B: So, if they laugh between two words, you'd get it in between the two words.\nSpeaker B: But if they laugh across three or four words, you get it after those four words, does that matter?\nSpeaker C: Well, the thing that's hard to deal with is when they speak while laughing.\nSpeaker C: And I don't think that we can do very well with that.\nSpeaker C: But that's not as frequent as just laughing between speaking.\nSpeaker G: So, do you treat breath and laughter as phonetically or as word models or what?\nSpeaker C: We tried both. Currently, we use special words.\nSpeaker C: There's actually a word for, it's not just breathing, but all kinds of mouth stuff.\nSpeaker C: And then laughter is a special word.\nSpeaker G: How would we do that with the hybrid system?\nSpeaker G: So, train a phone in the neural net?\nSpeaker C: Yeah. Oh, and each of these words has a dedicated phone.\nSpeaker C: So, the mouth noise word has just a single phone.\nSpeaker G: Right, so, in the hybrid system, we could train the net with a laughter phone and a breath sound phone.\nSpeaker E: So, it's the same thing, right? I mean, you could say, well, we now think that laughter should have three states.\nSpeaker E: So, some units, three states, different states, and then you would have three.\nSpeaker C: Do whatever you want.\nSpeaker C: The pronunciation, the pronunciations are somewhat non-standard. They actually are. It's just a single phone in the pronunciation, but it has a self-loop on it.\nSpeaker C: So, it can...\nSpeaker C: To go on forever.\nSpeaker G: And how do you handle it in the language model?\nSpeaker C: It's just a word.\nSpeaker C: It's just a word in the language of the other word.\nSpeaker C: Yeah.\nSpeaker C: We also tried absorbing these both laughter and actually also noise.\nSpeaker B: So, it's...\nSpeaker B: Yes. Okay.\nSpeaker C: Anyway, we also tried absorbing that into the pause model. I mean, the model that matches the stuff between words.\nSpeaker C: And it didn't work as well.\nSpeaker F: Yeah. Okay.\nSpeaker G: Can you hand me your digit form?\nSpeaker G: Sure.\nSpeaker G: It's on a mark that you did not read digits.\nSpeaker E: Say hi for me.\nNone: Yeah.\nSpeaker B: You didn't get me to thinking that. I'm not really sure which is more frequent, whether laughing.\nSpeaker B: I think maybe an individual thinks that people are more prone to laughing when they're speaking.\nSpeaker B: Yeah. I think...\nSpeaker G: I was noticing that with Dan and the one that we...\nSpeaker G: And Dan and the one that we're doing, we're not claiming to be getting the representation of mankind in these recordings.\nSpeaker E: We have very, very tiny sample of speech researchers.\nSpeaker G: Yeah.\nSpeaker E: It was really nice.\nSpeaker E: So...\nSpeaker E: Why we just... since we're on this main line, we just continue with what you're going to see about the transcriptions.\nSpeaker B: Okay. I'm really very...\nSpeaker B: I'm extremely fortunate with people who apply new...\nSpeaker B: Are transcribing for us. They are really perceptive.\nSpeaker B: And I'm not just saying that...\nSpeaker H: Because they're going to be transcribing it. That's the...\nSpeaker H: No, they're super.\nSpeaker H: Okay, turn the mic off and let's talk.\nSpeaker B: I know. I'm serious. They're just super.\nSpeaker B: So I brought them in and trained them in pairs because I think people can raise questions.\nSpeaker B: That's a good idea.\nSpeaker B: They think about different things and they think of different.\nSpeaker B: And I trained them on about a minute or two of the one that was already transcribed.\nSpeaker B: This also gives me a sense of...\nSpeaker B: I can use that later with ribbons to intercut your liability kind of issues.\nSpeaker B: And the main thing was to get them used to the conventions and the idea of the...\nSpeaker B: The size of the unit versus how long it takes to play it back.\nSpeaker B: So it's sort of calibration issues.\nSpeaker B: And then just set them loose.\nSpeaker B: And they're...\nSpeaker B: They all have already background and using computers.\nSpeaker B: They're trained in linguistics.\nSpeaker B: Oh, no, is that good or bad?\nSpeaker B: So one of them said, well, you know, he really said, not really.\nSpeaker B: And so what should I do with that?\nSpeaker B: And I said, oh, for our purposes, I do have a convention if it's a non-canonical...\nSpeaker B: That one, I think, you know, with Eric's work, I sort of figure we can just treat that as a variant.\nSpeaker B: But I told them if there's an obvious speech here, like I said in one thing and I gave my example.\nSpeaker B: I said, my phone instead of my phone, didn't bother...\nSpeaker B: I knew when I said it. I remember thinking, oh, that's not correctly pronounced.\nSpeaker B: But I thought it's not worth fixing because often when you're speaking, everybody knows what you mean.\nSpeaker B: But I have a convention that if it's obviously a non-canonical pronunciation, a speech error within the realm of resolution that you can tell in this American-English speaker, you know that I didn't mean to say microphone.\nSpeaker B: Then you put a little tick at the beginning of the word and that just signals that this is not standard and then curly-brack is a prong error.\nSpeaker B: And other than that, it's a word level. But when you know the fact that they noticed, you said, not end.\nSpeaker B: What shall I do with that? I mean, they're very receptive. And several of them are trained in IDEA.\nSpeaker B: They really could do phonetic transcription if we wanted to.\nSpeaker E: Right. Well, where were they when they wanted to do with some small subset of the whole thing?\nSpeaker E: I certainly would want to do everything.\nSpeaker B: I'm also thinking these people are terrific. Cool. I mean, so I told them that we don't know if this will continue past the end of the month.\nSpeaker B: And I also think they know that the data source is limited and I may not be able to keep them employed till the end of the month, even though I hope to.\nSpeaker E: The other thing we could do actually is use them for a more detailed analysis of the overall.\nSpeaker G: That would be so super. I mean, this is something that we were talking about. We could get a very detailed overlap if they were willing to transcribe each meeting four or five times.\nSpeaker G: Right one for each participant. So they could by hand.\nSpeaker E: Well, that's one way to do that. But I would say the other thing is just go through for the overlaps.\nNone: Yeah.\nSpeaker E: So instead of doing phonetic transcription for the whole thing, which we know from the Steve's experience with the source transcription is very, very, very, tank consuming. And it took them, I don't know how many months to get four hours.\nSpeaker E: So that hasn't been really our focus. We can consider it. But I mean, the other thing is this is something so much time thinking about overlaps is maybe get much more detailed analysis of the overlaps.\nSpeaker E: But anyway, I'm open to our consideration. I don't want to say that I feel I'm open to a consideration of what are some other kinds of detailed analysis that would be most useful.\nSpeaker E: And I think this year we actually do it. It says we have due to variations in funding. We seem to be doing very well on money for this this year next year with any have much less.\nSpeaker E: You mean 2001 calendar year or I mean calendar year 2001. Yeah, so it's it's we don't want to hire a bunch of people a lot of time staff because the funding that we've gotten is sort of a big chunk for this year.\nSpeaker E: But having temporary people doing some specific thing is actually perfect.\nSpeaker B: Wonderful. And then school start in the 60 on the 16th. Some of them will have to cut back their hours.\nSpeaker B: Yeah, they were people time now or some of them are. Wow.\nSpeaker B: Well, I wouldn't say 40 hour weeks. No, but what I mean is I shouldn't say that way because it does sound like 40 hour weeks. No, I would say they're probably they don't have they don't have other things that are taking away.\nSpeaker B: I don't see how someone to do 40 hours a week on transcription. No, you're right. It's it's too taxing but they're putting in a lot.\nSpeaker B: And I checked them over. I haven't checked them all but just spot checking. They're fantastic. I think it would be transcribing.\nSpeaker E: I mean, for Ron Tay volunteer to do some of that. The first first stuff he did was transcribing Chuck.\nSpeaker E: And he said, you know, I was thought Chuck spoke really well.\nSpeaker B: Well, you know, I also thought Liz has this, you know, and I do also this, this interest in the types of overlaps that are involved.\nSpeaker G: So would be great choices for doing coding of that type if we wanted or whatever. So I think it would also be interesting to have a couple of the meetings have more than one transcriber do because I'm curious about interanotator agreement.\nSpeaker B: Yeah, I think that's a good idea. And there's also in my mind, I think on Andreas was leading to the topic the idea that we haven't yet seen the type of transcript that we get from IBM.\nSpeaker B: And it may just be, you know, pristine, but on the other hand, given the lesser interface, because this is, you know, we've got a good interface. We've got great headphones.\nSpeaker E: It could be that they will, there's one that being a kind of first pass or something like that.\nSpeaker E: Maybe a lot of them because again, they probably are going to do these in one which will also be.\nSpeaker B: That's true. Although you have to, don't you have to start with the close enough approximation of the verbal part.\nSpeaker E: Well, that's, that's available, right? I mean, so the argument is that if your statistical system is good, it will in fact clean things up.\nSpeaker E: So it's got its own objective criterion. And so in principle, you could start up with something that was kind of raw.\nSpeaker E: I mean, give an example of something we used to do at one point back with, like this area, in the other times as we would take, take a word and have a canonical pronunciation.\nSpeaker E: And if there's five phones in a word, break up the word into five equal length pieces, which is completely rough.\nSpeaker E: Yeah.\nSpeaker E: I mean, the timing is off all over the place, just about any word.\nSpeaker E: It's okay.\nSpeaker E: But it's okay. You start off with that and the statistical system then lines things. Eventually, you get something that doesn't really look to bad.\nSpeaker E: Oh, excellent.\nSpeaker E: So, so I think using a good aligner actually can help a lot.\nSpeaker E: But, you know, they both help each other. If you have a better starting point than it helps the aligner, if you have a good aligner, it helps the human taking less time to graduate.\nSpeaker E: So, excellent.\nSpeaker B: I guess there's another aspect too. And I don't know. This is very possibly different topic.\nSpeaker B: But, just let me say, with reference to this idea of higher order organization within meetings. So, like, you know, the topics that are covered during a meeting with reference to the other uses of the data.\nSpeaker B: So, being able to find where so and so talked about such and such.\nSpeaker B: Then, I mean, I did sort of a rough pass on encoding, like, episode, like, level things on the transcribed meeting.\nSpeaker B: Already transcribed meeting.\nSpeaker B: I don't know if, or if that's something that we want to do with each meeting, sort of like a manifest when you get a box full of stuff. Or if that's, I mean, I don't know what level of detail would be most useful.\nSpeaker B: I don't know if that's something that I should do when I look over it. Or if we want someone else to do, or whatever, at this issue of the contents of the meeting in an outline form.\nSpeaker E: Meaning, really, my thing.\nSpeaker G: I think just whoever is interested can do that. So, someone wants to use that data.\nSpeaker E: I think it's a little short here. We've been trying to finish. Well, you know, the thing that sort of I wanted to do these digits and haven't heard from the beginning.\nSpeaker G: We could skip the digits. We don't have to read digits each time.\nSpeaker E: I think, you know, another, another, what you did, it's more than that is good.\nSpeaker E: So, I'd like to do that. I think you maybe, did you prepare some all thing you wanted to see? Or is prepared?\nSpeaker I: Yeah. How long?\nSpeaker I: I think it's fast because I have the results of the study of the training energy without lowering.\nSpeaker I: I'm just trying to do the medium, the average, dividing by the variance.\nSpeaker I: The last meeting, I don't know if you remain, we have problems with the parameters, with representations, the parameters.\nSpeaker I: Because the values of the peaks in the signal look like that's a follow to the energy in the signal.\nSpeaker I: And it was probably not with the scale. With what?\nSpeaker I: Scale.\nSpeaker I: And I change the score and we can see the variance.\nSpeaker E: But the bottom line is still not separating out.\nSpeaker E: That's an option.\nSpeaker E: There's no point in going through all that.\nSpeaker E: I think we have to start.\nSpeaker E: There's two suggestions really, which is what we said before, is that it looks like, at least you have probably the obvious way to normalize so that the energy is anything like a reliable indicator of the overlap.\nSpeaker E: I'm still low. I think that's a low funny.\nSpeaker E: It seems like it should be.\nSpeaker E: But you don't want to keep knocking at it if you're not getting any result of that.\nSpeaker E: But I mean the other things that we talked about is pitch related things and hominicity related things, which also should be some kind of reasonable indicator.\nSpeaker E: But a completely different tack on it is the one that was suggested by your colleagues in Spain, which is to say, don't worry so much about the features.\nSpeaker E: That is to say use, as you're doing with speech, non-speech use of very general features.\nSpeaker E: And then look at it more from the aspect of models.\nSpeaker E: Have a couple of mark-up models.\nSpeaker E: And try to determine when is the layer you're in an overlap when you're not in overlap.\nSpeaker E: And let the statistical system determine what's the right way to look at the data.\nSpeaker E: I think it would be interesting to find individual features put them together.\nSpeaker E: I think that you'd end up with a better system overall, but given the limitation in time, given the fact that heavier system already exists, doing this sort of thing.\nSpeaker E: But its main limitation is that again, it's only working at silences, which maybe that's a better place to go.\nSpeaker I: I think that the possibility can be that fellow working with a new class, not only non-speech and speech, that in the speech class, dividing speech from speaker and overlapping to do a fast experiment to prove that this general feature can solve the problem.\nSpeaker I: How far is...\nSpeaker I: I have prepared the pitch tracker now.\nSpeaker I: I hope the next week we have some results.\nSpeaker I: We will see the parameter of pitch tracking with the problem.\nSpeaker E: Have you ever looked at the heavier speech segmenter?\nSpeaker E: No.\nSpeaker E: Maybe you could show...\nSpeaker E: Yeah, sure.\nSpeaker E: The limitation there again was that he was only using it to look at silences as a punitive split point between speakers.\nSpeaker E: But if you included broad classes, then principle maybe you could cover the overlap cases.\nSpeaker D: But I'm not too sure if we can really represent overlap with the detector I use, up to now, the speech.\nSpeaker D: I think that's right.\nSpeaker G: I think that the heavier speech might be able to.\nSpeaker G: It doesn't have the same HMM modeling, which is a drawback.\nSpeaker E: It's just a Gaussian for each.\nSpeaker G: Yeah.\nSpeaker G: And then you choose optimal splitting.\nSpeaker E: What does it have? Does it have any temporal...\nSpeaker G: Maybe I'm misremembering, but I did not think it had a mark-off.\nSpeaker E: I guess I don't remember either.\nSpeaker E: It's been a while.\nSpeaker I: Yeah.\nSpeaker I: I could have looked at it.\nSpeaker I: You can see the same word with the mark-off.\nSpeaker I: Yeah, I didn't think so.\nSpeaker E: He just computes a Gaussian over a Gaussian.\nSpeaker G: And so I think it would work fine for detecting overlap.\nSpeaker G: What he does is, as a first pass, he does a guess at where the divisions might be.\nSpeaker G: And he overestimates.\nSpeaker G: And that's just a data reduction step so that you're not trying it every time interval.\nSpeaker G: And so those are the punitive places where he tries.\nSpeaker G: And right now he's doing that with silence.\nSpeaker G: And that doesn't work with the meeting recorder.\nSpeaker G: So if we use another method to get a first pass, I think it would probably work.\nSpeaker G: It's a good method.\nSpeaker G: As long as the segments are long enough.\nSpeaker G: That's the other problem.\nSpeaker E: Okay, so let me go back to what you had.\nSpeaker E: The other thing one to do is...\nSpeaker E: So you have two categories.\nSpeaker E: And the mark-off model is reached.\nSpeaker E: Good, you have a third category?\nSpeaker E: So you have non-speech, single-person speech, and multiple-person speech?\nSpeaker E: He has a non-speech board, actually.\nSpeaker B: Don't you have any other categories on the board?\nSpeaker E: And you have a mark-off model for each?\nSpeaker D: I'm not sure.\nSpeaker D: I'm not about adding another class to it.\nSpeaker D: It's not too easy, I think, the transition between a different class, in the system I have now.\nSpeaker D: But it could be possible, I think, in principle.\nSpeaker D: Yeah, I mean, this is all pretty gross.\nSpeaker E: I mean, the reason why I was suggesting originally to look at the features is because I thought, well, we're doing something we haven't done before.\nSpeaker E: We should at least look at the space, understand.\nSpeaker E: It seems like if two people, two or more people talk at once, should get louder.\nSpeaker E: Yeah.\nSpeaker E: And there should be some discontinuity and pitch contours.\nSpeaker E: And there should overall be a smaller proportion of the total energy that is explained by any particular harmonic sequence in the spectrum.\nSpeaker E: So those are all things that should be there.\nSpeaker E: So far, Jose has been, although I was told I should be calling you a paper, but anyway, has been exploring largely the energy issue.\nSpeaker E: As with a lot of things, it's not as simple as it sounds.\nSpeaker E: And there's a lot of energy, as it helps you see residual energy, is it delta of those things?\nSpeaker E: Obviously, just a simple number.\nSpeaker E: Absolute number isn't going to work, so it should be compared to watch.\nSpeaker E: There'd be a long window for the normalizing factor, a short window for what you're looking at, or how short should they be.\nSpeaker E: So that he's been playing around with a lot of these different things, and so far, at least has not come up with any combination that really gave you an indicator.\nSpeaker E: So I still have a hunch that it's in their someplace, but it may be given that you've limited time here, it just may not be the best thing to focus on some range.\nSpeaker E: So pitch related and harmonic related.\nSpeaker E: But it seems like if we just want to get something to work, that there's a suggestion of, they were suggesting going to mark up models, but in addition, there's an expansion of what have you idea of.\nSpeaker E: And one of those things, looking at the statistical component, even if the features that you give it are maybe not ideal for it, it's just a general filter bang or a faster one or something.\nSpeaker E: It's in there somewhere probably.\nSpeaker I: What did you just think about the possibility of using the Javier Sokhovur?\nSpeaker I: I mean, the bike criterion, to train the Gaussian using the mark by hand to train overlapping zone and SP zone.\nSpeaker I: I mean, I think that interesting experiment could be to prove that if we suppose that the first step, I mean the classified world world, classified from Javier or classified from Filo, what happened with the second step?\nSpeaker I: I mean, what happened with the clustering process using the Javier?\nSpeaker I: I mean, it's enough to operate or to distinguish between overlapping zone and SP zone.\nSpeaker I: If we develop and classify and the second step doesn't work, we have another problem.\nSpeaker G: I had tried doing it by hand at one point with a very short sample, and it worked pretty well, but I haven't worked with it a lot.\nSpeaker G: So I took a hand segmented sample and I added 10 times the amount of numbers at random, and it did pick out pretty good boundaries, but this was just very anecdotal sort of thing.\nSpeaker I: It's possible with my segmentation by Khan that we have information about the overlapping.\nSpeaker G: Right, so if we fed the hand segmentation to Javier's and it doesn't work, then we know something's wrong.\nSpeaker I: I think that's probably worth a while doing.\nSpeaker I: Do you know where our software is?\nSpeaker G: Do you use it?\nSpeaker G: I have as well, so if you need help, let me know.\nSpeaker G: Transcript 295129706030970801502.05884.\nSpeaker G: 1 6 2 8 5 8 3 2 3 3 0 3 1 5 4 5 0 9 9 7 1 1 2 8 4 0 0 9 4 0 7 1 0 1 2 4 1 5 3 1 2 6 7 2\nSpeaker E: correct that 6 7 2 1 0 8 6. Transfer 2 8 7 1-2 8 9 0 3 3 8 4 4 6 5 2 5 8 0 6 7 8 0 0 1 4 0 1 8 1 3 1 1 6 2 5 3 4 6 8 1 3 4 5 0 6 0 1 7 1 1 2 8 3 3 6 0 8 0 9 6 5 0 7 3 8 0 8 6 9 1 2 2\nSpeaker H: transcript 2 8 3 1-2 8 5 0 1 8 6 7 0 6 5 2 3 4 0 6 1 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 8 8 1 3 2 0 8 4 5 5 0 6 7 8 0 7 8 9 0 0 5 1 3 3 5 0 0 5\nSpeaker D: OK, transcript 2 7 9 1-2 8 1 0 0 0 0 2 3 7 3 3 2 4 7 5 3 0 2 6 7 8 0 8 0 1 0 7 0 3 1 5 0 3 6 2 8 0 0 0 3 4 5 0 2 0 0 7 2 0 3 3 1 7 8 9 5 1 2 2 0 9 8 5 3 9 7 8\nSpeaker I: 2 3 3 3 3 3 3 4 3 4 4 4 4 1 1 0 5 6 7 7 7 7 8 9 0 1 2 9 4 8 5 9 2 2 3 3 49567979 8350 02795108254\nSpeaker B: Transcript number 2771 2790 909 899 0150 1228 845 2500 3824 465608 819566 0502187 07029 1 2 3 0 0 5 3 6 3524 7764 862809\n"}, {"meeting_id": "TS3003b", "summary": "The whole meeting was focused on the functional design of the new remote control project. First of all, Marketing, User Interface, and Industrial Designer each gave a presentation about user requirements, technical function design, and working design respectively. Then, Project Manager announced some amendments to project requirements, which automatically rejected some of the previous assumptions and aroused further discussion about design and functionality with people above 40 as target customers.", "dialogue": "None: Big sitting, good, release.\nSpeaker D: Good session.\nNone: Take this in little bit.\nSpeaker D: Now my laptop is funny.\nSpeaker D: It does it again.\nNone: It's working.\nSpeaker D: Okay, good afternoon.\nSpeaker D: Okay, everybody found this place again?\nSpeaker D: Yeah, that's nice.\nSpeaker D: Okay, so this is our second meeting and still failing.\nSpeaker D: Yeah, and now we're going into the functional design.\nSpeaker D: The important thing of this phase is that we're going to try to get an agreement about the user requirements, technical function design and the working design.\nSpeaker D: So that we can move on to the second phase.\nSpeaker D: But first is phase.\nSpeaker D: First announcement, there's a little adaptation in the air conditioning system.\nSpeaker D: So there's our ghost mouse again.\nSpeaker D: But that means that you can have a little trouble with air conditioning.\nSpeaker D: That's because of this.\nSpeaker D: Okay, it's in wing C and E. So it should be over in a while, a couple of days.\nSpeaker D: It's going to be cold anyway, so I don't think you're going to need it.\nSpeaker D: Then our agenda, first the opening.\nSpeaker D: This time I will take the minutes.\nSpeaker D: You're going to have a presentation, all of you.\nSpeaker D: We've got 40 minutes for the whole presentations.\nSpeaker D: So I suggest we take about 7 minutes per presentation.\nSpeaker D: Then we can have a little discussion about the new project requirements, which have been sent to me.\nSpeaker D: Then the decision on the control functions, which we want to include.\nSpeaker D: So we've got 40 minutes for all of it.\nSpeaker D: So I suggest let's start with the first presentation.\nSpeaker D: Who wants to be first?\nSpeaker D: I think I'll go first.\nSpeaker D: Maybe it's easy if you will tell the presentation as well.\nSpeaker D: Which function you have and where you're going to talk about.\nSpeaker B: Okay, my name is Frank Vamponin.\nSpeaker B: I'm the market expert, but you already knew that.\nSpeaker B: I've done some research.\nSpeaker B: We have been doing research in a usability lab where we observe users operating remote controls.\nSpeaker B: We let them fill out a questionnaire.\nSpeaker B: We had 100 of these test subjects.\nSpeaker B: In addition, we did some market research.\nSpeaker B: See what the market consists of, what ages are involved.\nSpeaker B: Well, these are three quite astonishing results, I thought.\nSpeaker B: Remotes are being considered ugly.\nSpeaker B: They indicated that they thought they were ugly.\nSpeaker B: An additional 80% indicated that it would spend more money on a fancy-looking remote control.\nSpeaker B: In addition, remote were not very functional.\nSpeaker B: 50% indicated they only used about 10% of the buttons on a remote control.\nSpeaker B: 50% indicated that the remote tended to get lost in their room.\nSpeaker B: So, some things.\nSpeaker B: Then we did some research to do most relevant functions.\nSpeaker B: Channel selection and volume selection both got a 10 on a scale of 1 to 10 for relevancy.\nSpeaker B: The power button got a 9 and 10 attacks got a 6.5.\nSpeaker B: So, these are the most important functions of a remote control.\nSpeaker B: Then there are some one-time use function.\nSpeaker B: That's what I like to call them.\nSpeaker B: They're audio settings, video settings, and channel settings buttons, which are not really used very frequently, but are still considered to be of some importance.\nSpeaker B: Channel selection was also indicated to be used very frequently, 168 times per hour.\nSpeaker B: Then this is the market.\nSpeaker B: 60% of the market consists of users between the ages 16 and 46.\nSpeaker B: Main characteristic of this group is that they're very critical under remote control.\nSpeaker B: They like to use new functions, but they also are very critical.\nSpeaker B: They won't spend their money very easily.\nSpeaker B: The users of 46 to 65 years make up 40% of the market.\nSpeaker B: They're not really very interested in features, but they do tend to spend their money a lot easier.\nSpeaker B: What I think this indicates for our design, I think we should make a remote for the future.\nSpeaker B: This means we would have to focus on the ages 16 to 45.\nSpeaker B: This also makes the biggest part of the market, so it would also be where our main profit would be getable.\nSpeaker B: This would mean we would have to make a fancy design.\nSpeaker B: The results also indicated that about one quarter of the people questioned thought that remote control caused errors high.\nSpeaker B: This is certainly something to take into account, and 34% thought that it was hard to learn and how to operate the new remote control.\nSpeaker B: These are two factors that I think should be included in the design, besides of course that the remote must look very nice.\nSpeaker B: Functionality, as a lot of people indicated, they only use about 10% of the buttons.\nSpeaker B: I think we should make very few buttons.\nSpeaker B: This will also be beneficial to the design of the remote.\nSpeaker B: I think the most frequently used buttons should be emphasized, especially the channel selection and audio selection buttons, because they are used most, so they should be robust, they shouldn't break down easily.\nSpeaker B: Then, as a lot of people indicated that their remote got lost in the room, it might be, and I say it might be because it would certainly boost the production costs a lot, but it might be a good idea to make a docking station.\nSpeaker B: This would get a button in it which would send a signal to the remote which would then beep, so you know where it is in the room, and in addition to this it could recharge the batteries in the remote if you put it in.\nSpeaker B: Then, surprisingly, a great deal of people indicated that an LCD screen in the remote control would be preferred.\nSpeaker B: This was mostly people in the age of 16 to 25, but up to 45 it remains feasible.\nSpeaker B: This would also greatly increase the production costs, so I think these are just some small factors we could consider.\nSpeaker B: Okay. That would be all.\nSpeaker D: Thank you.\nSpeaker D: Anybody have any questions still now?\nSpeaker D: About functional requirements.\nSpeaker D: Okay, that's clear.\nSpeaker D: Add to the second.\nNone: Okay.\nSpeaker A: Okay.\nSpeaker A: You should interface.\nSpeaker A: Yeah.\nSpeaker A: For the technical functions of it.\nSpeaker D: Yeah, you can take your time.\nSpeaker D: We've got plenty of time.\nSpeaker D: That's pretty good.\nSpeaker A: Yeah, I think we must use the general functions of the remote control.\nSpeaker A: I've learned a little research on the internet.\nSpeaker A: Not much information about it.\nSpeaker A: About interface.\nSpeaker A: Yeah.\nSpeaker A: I've been thinking about simple manner to put a lot of functions in one remote control.\nSpeaker A: You've got a lot of devices like DVD, television, stereo.\nSpeaker A: But it must be user-friendly.\nSpeaker A: You can put a lot of functions in one remote control.\nSpeaker A: Yeah.\nSpeaker A: In one remote control.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Got many functions in one remote control.\nSpeaker A: You can see this is quite simple remote control.\nSpeaker A: A few buttons.\nSpeaker A: But this remote control.\nSpeaker A: Got a lot of buttons.\nSpeaker A: Yeah.\nSpeaker A: People don't like it.\nSpeaker A: So what I was thinking about was I was thinking about it like the on-off button.\nSpeaker A: I keep it like a red button.\nSpeaker A: Everybody knows it.\nSpeaker A: You don't have to change that.\nSpeaker A: My personal preferences.\nSpeaker A: Use a display for specific functions of the different device.\nSpeaker A: You've got the on-off button.\nSpeaker A: So you've got the functions of the device for a DVD player.\nSpeaker A: You've got real buttons for selecting a device.\nSpeaker A: So this button is for a DVD.\nSpeaker A: Every device you've got a part display of buttons.\nSpeaker A: So you've never got all the buttons on one device.\nSpeaker A: So that's my idea about it.\nSpeaker A: Yeah.\nSpeaker A: So that's me.\nSpeaker A: The buttons, the real buttons we have to use.\nSpeaker A: We better use quite large buttons for everybody after you see it.\nSpeaker A: So if all people, young people, so we must keep buttons quite simple and quite large.\nSpeaker A: Yeah.\nSpeaker A: That was my part of it.\nSpeaker D: So any other questions about technical functions?\nSpeaker C: Well, I think if we're going to use a touch screen, we're going to go way above the 12 and a half euros.\nSpeaker A: Yeah.\nSpeaker A: I don't think so.\nSpeaker A: You've got quite a cheap touch screen.\nSpeaker A: It's not in the corner or something.\nSpeaker A: It's just one call.\nSpeaker A: Yeah.\nSpeaker A: I've seen something on the Internet, not today, but a few weeks ago.\nSpeaker A: You've got quite a kind of touch screen.\nSpeaker A: It's for 20 euros or less.\nSpeaker A: So it's possible.\nSpeaker D: It's nice.\nSpeaker B: Well, we certainly make a fancy design.\nSpeaker C: It wouldn't be very robust.\nSpeaker C: It's very fragile and you can get scratches on it.\nSpeaker B: That is true.\nSpeaker B: Yeah.\nSpeaker D: Maybe we can first listen to your presentation.\nSpeaker D: Then we have a little discussion about the requirements.\nSpeaker D: Okay.\nSpeaker D: Design.\nSpeaker D: I think it's going to do.\nSpeaker D: It's not that much.\nSpeaker C: Okay.\nSpeaker C: I've got a presentation about the working design.\nSpeaker C: First about how it works.\nSpeaker C: It's really simple, of course, everybody knows how remote works.\nSpeaker C: The user presses a button.\nSpeaker C: The remote determines what button it is.\nSpeaker C: Uses the infrared to send the signal to the TV.\nSpeaker C: The TV switches to the frequency or what function it is.\nSpeaker C: So we've got the plate.\nSpeaker C: It's got conductive disks for every button.\nSpeaker C: When the user presses a button, a signal goes to the LED and transmits it to the TV.\nSpeaker C: It's a very simple device, technically speaking.\nSpeaker C: So this is a schematic overview.\nSpeaker C: You've got the buttons, the power source.\nSpeaker C: When a button gets pressed, it goes to the chip.\nSpeaker C: The chip controls the infrared bulb and perhaps a normal bulb.\nSpeaker C: When you press a button, you can actually see your pressed button.\nSpeaker C: Well, I think we should use default materials, simple plastics.\nSpeaker C: Keep the inner working simple so it's robust.\nSpeaker C: I think we should focus on aesthetics, the design and the user interface because if you're going to use high-tech materials, the price is going to go sky-high.\nSpeaker C: And you only have to design a remote once.\nSpeaker C: And if you use high-tech materials, it comes back in every product.\nSpeaker C: So in my idea, it's going to be smart to invest in design and not in the product itself.\nSpeaker C: That's it.\nSpeaker D: Okay, thank you.\nSpeaker D: Okay, now I hope everybody has a little bit more insight in the functions we all have and what we are doing right now.\nSpeaker D: I'm the project manager, so I'm here to mess things up and tell you some new requirements.\nSpeaker D: We've got to design a remote which is only suitable for TV.\nSpeaker D: That's because it will be too complex and the time to mark it will be too big if we want to have it for more functions.\nSpeaker D: So it has to be simple.\nSpeaker D: Another point is we have to skip the data text because in the world of upcoming Internet we think data text is going to be a thing of the past.\nSpeaker D: And it's a function we don't need in our remote control.\nSpeaker D: Internet is also mentioned just to go in a function we can use, maybe also on televisions that will be available as well.\nSpeaker D: Another one is the customer is 40 plus.\nSpeaker D: That's the market we have to target because we are going to develop a new product which is specially designed for the younger customers.\nSpeaker D: This is a bit pity for the marketing expert because he was aiming on the younger persons.\nSpeaker D: So we have to find a market which is above 40 plus but which will suit our remote control and the other way around.\nSpeaker D: And we have to be very attentive in putting the corporate image in our product.\nSpeaker D: So it has to be visible in our design in the way our device works.\nSpeaker D: And we have to be very clear on this point as well.\nSpeaker D: So I suggest let's have a discussion on the control functions.\nSpeaker B: So is there any discussion possible about the requirements?\nSpeaker D: We can see if we can find a way between the functions we want to use and the market we want to reach with our product.\nSpeaker B: You are saying that teletext is going to be an old feature and it's not going to be used anymore anyway.\nSpeaker B: Pretty soon new TVs will have internet access on them.\nSpeaker B: But I think if you are targeting people of 40 plus the chance that they will have a TV with internet access within the next 20 years is very slim.\nSpeaker B: In addition people indicated that teletext simply was an important feature for the remote control.\nSpeaker B: So I think it's pretty dumb to put no teletext feature on it.\nSpeaker B: I'm pretty much against it.\nSpeaker D: Against the no data text?\nSpeaker B: Yes.\nSpeaker B: Besides that I think the market for 40 plus is like pretty small.\nSpeaker B: But if I see this I think we are just going to go for another step.\nSpeaker B: Pretty much.\nSpeaker D: I think we can do a lot with the design and the simple buttons which are also mentioned.\nSpeaker D: If we put a lot of effort in those we can make remote control with just two or three buttons or just a remote which is suitable for the market.\nSpeaker D: It's suitable for the market you want to reach because it is 40% of the market and if you look in Holland the whole generation of 40 plus 50 plus it's the biggest share of the whole population now.\nSpeaker B: Yes but it's not the biggest part of the market.\nSpeaker B: Besides that they are not very critical so I mean they don't really care what the remote control is like.\nSpeaker B: They'll just pretty much take the first thing they see and which looks acceptable.\nSpeaker D: But don't you think that if we make a remote which is typically made for this market that people think from the people think that's the device I've looked for although I didn't realize it.\nSpeaker D: So let's try it.\nSpeaker B: No.\nSpeaker B: I think that would be the case in 16 to 45 age category because they are critical and they want to have a fancy remote control.\nSpeaker B: People are 40 plus I mean they want it to work but as soon as it works it's okay with them.\nSpeaker C: Yeah.\nSpeaker C: If we put our marketing right we can sell this just like I don't know if you've heard about it in the news the elderly mobile phone.\nSpeaker C: Yeah it's a big success.\nSpeaker C: If we make a remote control with that idea in mind we could make tons of money I think.\nSpeaker C: We don't have to focus on the design then but on functionality. We just change our focus on the project.\nSpeaker C: And I think we can sell this.\nSpeaker D: I simply think that the new products we're going to make specifically design are designed for younger people.\nSpeaker D: So maybe we can focus ourselves on the elderly people.\nSpeaker D: And I think we have to see what requirements we need for those remote controls because what you told is the channel selection is important.\nSpeaker D: The volume selection power and data text.\nSpeaker D: Yes.\nSpeaker D: Okay.\nSpeaker B: But obviously the board tends to disagree.\nSpeaker D: No we haven't voted yet.\nSpeaker D: I think that can be a function as well.\nSpeaker D: But only if it's going to hire the cost.\nSpeaker D: Because I don't know if it will be a lot more money to implement data text as well.\nSpeaker D: But I don't think that will be a problem.\nSpeaker D: Or is data text.\nSpeaker A: And deaf people need data text for subtitles.\nSpeaker D: Yeah.\nSpeaker B: So I suggest that.\nSpeaker B: Definitely be a bad idea not to include data text.\nSpeaker D: Is anybody really against data text?\nSpeaker D: No.\nSpeaker D: Just that.\nSpeaker D: Just keep the data text.\nSpeaker D: I think that's good idea as well.\nSpeaker D: Especially for the subtitles.\nSpeaker D: Maybe we can make that another point of advantage in our remote control.\nSpeaker D: If we make a button for example for big subtitles.\nSpeaker D: Which is instantly on the remote control for elderly people that they can think I want to have subtitles and push the button and they get the big subtitles.\nSpeaker D: That's a good idea.\nSpeaker D: So I think that can be very useful in our advantage.\nSpeaker D: Functionality should be few buttons you said.\nSpeaker D: Yes.\nSpeaker D: I think that's very important.\nSpeaker D: We have a few buttons.\nSpeaker D: So to keep it simple.\nSpeaker B: I don't think that's really an issue anymore.\nSpeaker B: If it's only for TV you're not going to need a lot of buttons anyway.\nSpeaker B: You need one to zero button.\nSpeaker B: Next channel, previous channel.\nSpeaker B: Pull him up, pull him down.\nSpeaker B: And some teletext buttons.\nSpeaker D: Yeah.\nSpeaker D: But do you need?\nSpeaker A: I think we can skip this place or we need it.\nSpeaker D: Do you need buttons for one to zero?\nSpeaker B: Maybe we can.\nSpeaker B: I think many people like to use that.\nSpeaker B: Maybe you can use it.\nSpeaker B: You want to switch from channel one to like 35.\nSpeaker B: You don't want to push the next channel button 35 times.\nSpeaker D: Maybe you can implement the scroll button.\nSpeaker D: Or joystick.\nSpeaker D: There are other ways to just look.\nSpeaker D: There are other phones.\nSpeaker B: There's only telephones in the scroll button which is very useful in searching names.\nSpeaker B: That's true.\nSpeaker B: I don't think there are many TVs that can switch channels that fast.\nSpeaker B: So you would need like the TV would need a function where you can actually view all channels and scroll through it.\nSpeaker B: If there are any TVs out there.\nSpeaker C: And besides that, if we're going to focus on elderly people, they'll have to adapt.\nSpeaker C: They're not used to using scroll buttons.\nSpeaker C: Perhaps we should stick to the basic layout.\nSpeaker D: Yes, the numbers.\nSpeaker D: Yeah.\nSpeaker D: We can see how much buttons there are going to be on the display.\nSpeaker D: And if it's too much, we can reconsider it.\nSpeaker D: But I think there won't be very much buttons or they're done.\nSpeaker B: I think if you're going to make remote control only to operate a TV, there's not much you can gain on having as few buttons as possible.\nSpeaker B: Because I think there are pretty many remote controls that can only operate a TV which already only have the minimum number of buttons.\nSpeaker B: I don't think there's much to be gained in that area.\nSpeaker D: The number of buttons.\nSpeaker D: I think it's very important in the design.\nSpeaker D: You can make a very fancy design with putting the buttons on the right places.\nSpeaker D: If you have less buttons, you can do a lot more with it.\nSpeaker B: That is true.\nSpeaker B: But I think there's simply not much to gain on the competition when you're making a remote control only for to operate only the TV.\nSpeaker D: To operate only the TV.\nSpeaker B: There's simply not a lot of buttons required.\nSpeaker B: There's not a lot of functions required.\nSpeaker B: So most existing remote controls simply don't have a lot of buttons either.\nSpeaker B: So I think it would be very hard to actually gain on the competition here.\nSpeaker B: So it would cost a big marketing expedition.\nSpeaker B: Which was one of the arguments to make it only for the TV because we didn't have the time to market a lot.\nSpeaker D: So you suggest we could better focus on, for example, the docking station.\nSpeaker D: Like other functions.\nSpeaker D: Instead of less buttons.\nSpeaker B: Well, I think we obviously need a good way to position all the buttons.\nSpeaker B: But I don't think we should spend very much time in that.\nSpeaker D: Do you think the docking station is allowed in the budget we have?\nSpeaker C: It should be possible, yes.\nSpeaker C: Because it's not too fancy.\nSpeaker C: And if the remote stays rather small, it should be possible.\nSpeaker D: Because I think that's a good advantage point as well if we have a fancy looking docking station.\nSpeaker D: Or very...\nSpeaker D: That's a nice requirement.\nSpeaker C: So we're just going to focus on the extras.\nSpeaker D: I think so.\nSpeaker B: Yeah, I think maybe we should do some research into what other people like to have in a...\nSpeaker B: We like to have X-raying in remote control.\nSpeaker D: That's a good point.\nSpeaker D: You said they easily get lost as well?\nSpeaker B: Yes, well, 50% of the people indicated that remote control tended to get lost.\nSpeaker D: So maybe we should implement the audio site or something.\nSpeaker B: Yeah, that was what I suggested.\nSpeaker B: Like with your keychain, if you were a solid girl, it makes us...\nSpeaker B: We have some phones too.\nSpeaker B: And you have a docking station.\nSpeaker B: You just press the button and the phone goes ringing.\nSpeaker D: Yeah, so the audio signal should be possible as well.\nSpeaker D: I think it's not too expensive.\nSpeaker D: Another point is the LCD screen.\nSpeaker D: I don't know if that will rise to cost too much.\nSpeaker C: I think we'll have to choose between the docking station or the screen, because...\nSpeaker D: Yeah, it's going to be too much as well.\nSpeaker B: I think since a lot of people indicated that a new remote control was hard to learn.\nSpeaker B: And we're focusing on elderly people here, which tend to have a hard time understanding new devices.\nSpeaker B: It might be a good idea to have just a little screen on it, which would explain a button if you press it, which would tell you what it does.\nSpeaker B: Yeah.\nSpeaker B: And it wouldn't have to be touchscreen or a very expensive screen.\nSpeaker D: Just the LCD or just the normal screen.\nSpeaker D: Just a small screen ID.\nSpeaker D: Some extra info.\nSpeaker D: Feedback.\nSpeaker D: Yeah, I think that's a good idea as well.\nSpeaker B: But I don't know if that would...\nSpeaker D: Just a small screen heading to the...\nSpeaker B: Cost.\nSpeaker D: Extra button info.\nSpeaker D: I think that should be possible as well.\nSpeaker D: Let's see, what did we say more?\nSpeaker D: Should be fancy design.\nSpeaker D: Easy to learn.\nSpeaker D: Few buttons.\nSpeaker D: Talking station LCD.\nSpeaker D: General functions.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And default materials.\nSpeaker D: I think that's a good idea as well because...\nSpeaker D: LDP people don't mind if it's a titanium cover or just a plastic one.\nSpeaker D: So it doesn't really matter.\nSpeaker B: I think...\nSpeaker B: Probably elderly people would be a little bit more careful with their remote controls and youngsters.\nSpeaker D: And let's specify the target group.\nSpeaker D: Are we talking about elderly people or people from 40 to 80?\nSpeaker D: Because I think what we're going to design now is for people above 60.\nSpeaker A: Maybe.\nSpeaker D: What do we want?\nSpeaker D: If we want to proflate us with, for example, the telephone for elderly people, we can target the real elderly people.\nSpeaker B: I think that would be... if we should do something like that, I think it would be really good for the image of the company.\nSpeaker D: Yeah.\nSpeaker B: And I think there would be a good market for it.\nSpeaker D: So that's the...\nSpeaker B: Really bringing enough of the product.\nSpeaker D: You're talking about 60 to 80, for example.\nSpeaker B: Yeah.\nSpeaker D: But 60.\nSpeaker B: I'd have to look into that a little more.\nSpeaker D: Okay.\nSpeaker D: And different cultures.\nSpeaker D: Why are we...\nSpeaker C: Well, I don't think they have different television sets in every country.\nSpeaker A: No.\nSpeaker A: Because...\nSpeaker D: We've got five minutes left.\nSpeaker D: It's a small word.\nSpeaker B: And with the little screen in it, which explains the buttons, I think it would be a lot easier to adapt to two different cultures.\nSpeaker D: Yeah.\nSpeaker D: In different languages.\nSpeaker D: Or you have to put a language button in it, but that will be a bit unnecessary, I think.\nSpeaker D: It's better to put it on different markets with...\nSpeaker D: Okay.\nSpeaker D: So that's the target.\nSpeaker D: Then a few small things.\nSpeaker D: Okay.\nSpeaker D: I will put the minutes in the project document folder.\nSpeaker D: What we're going to do for the next meeting is the industrial designer will do the components concept, user interface designer, user interface concept, and the trend watching.\nSpeaker D: So just keep in mind the things we've said about the target group requirements and the trends, the trends that are going on, and the specific instructions will be sent to you by the personal coach.\nSpeaker D: Okay.\nSpeaker D: So I thank you for this meeting.\nSpeaker D: And I think we have a lunch break now.\nSpeaker A: That's good.\nSpeaker D: So that's a good thing.\nSpeaker B: I'm not allowed to talk about it.\nSpeaker B: Thank you.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "Bed012", "summary": "The focus of the meeting was on a presentation of the work done already on the building of the Bayes-net. The complete prototype of the Bayes-net will be presented in the next meeting. After that, it will be possible to define interfaces and a dummy construction parser, in order to test and link modules together.", "dialogue": "None: Hello.\nSpeaker B: I just put it to the back and to the front.\nSpeaker B: It is, but it doesn't sit on the seat. I just put it to the back and to the front.\nSpeaker B: Is it picking me up? I should be five.\nSpeaker B: Hello. Test in.\nSpeaker B: Tested enough? Oh, I'm a testing machine.\nSpeaker C: I guess this is more or less known to get you up to day, Jono.\nSpeaker C: This is what is a meeting for me.\nSpeaker B: Did you add more stuff to it? Why? I don't know.\nSpeaker B: There were notes in the middle.\nSpeaker C: This is, so we thought that we can write up an element for each of the situation.\nSpeaker C: We observed it in the base. What is the situation like at the entity that is mentioned?\nSpeaker C: It is a stable. It is a forth going all the way through parking, location hotel, car, restroom, riots, fair strikes or disasters.\nSpeaker B: This is a situation where all the things can be happening right now.\nSpeaker B: Or what is the situation type?\nSpeaker C: That is just specifying the input for the...\nSpeaker B: Why are you specifying an XML?\nSpeaker C: It forces us to be specific about the values here.\nSpeaker C: This is what the input is going to be.\nSpeaker B: This is a schema.\nSpeaker B: Jono, this is what JavaBase takes.\nSpeaker C: We are sure going to interface to get an XML document from somewhere.\nSpeaker C: That XML document will say we were able to observe that the element of the location that the car is near.\nSpeaker B: This is a situation context. Is that what the situation is for?\nSpeaker C: This is just an XML document which defines a set of possible permissible XML structures which we view as input into the basenet.\nSpeaker B: We can possibly run one of them transformations to put in the JavaBase or whatever it wants.\nSpeaker C: Are you talking about the structure?\nSpeaker C: When you observe a node...\nSpeaker B: When you say the input to the JavaBase, it takes a certain format.\nSpeaker B: Which I don't think is this. Although I don't know.\nSpeaker C: No, it is certainly not this.\nSpeaker B: You can just run a XML.\nSpeaker B: Yeah, you can run it into the JavaBase format.\nSpeaker C: That's no problem.\nSpeaker C: I even think that once you have this running as a module, what you want is you want to say, give me the posterior probabilities of the go there, note when this is happening.\nSpeaker C: The person said this, the car is there, it's raining and this is happening.\nSpeaker C: With this you can specify what's happening in the situation and what's happening with the user.\nSpeaker C: After we've done through the situation, we get the user vector.\nSpeaker B: So this is...\nSpeaker B: So this is just a specification of all the possible inputs.\nSpeaker C: And all the possible outputs too.\nSpeaker C: So we have, for example, the go there decision node, which has two elements going there, and it's posterior probability and not going there, and it's posterior probability.\nSpeaker C: Because the output is always going to be all the decision notes and all the...\nSpeaker C: all the posterior probabilities for all the values.\nSpeaker B: And then we just look at the struct that we want to look at in terms of...\nSpeaker B: we're only asking about one of the...\nSpeaker B: So like, if I'm just interested in the going there node, I would just pull that information out of the struct that gets returned, that Java Bayes would output.\nSpeaker C: Pretty much yes, but I think it's a little bit more complex.\nSpeaker C: If I understand correctly, it always gives you all the posterior probabilities for all the values of all the decision notes.\nSpeaker C: So when we input something, we always get the posterior probabilities for all of these.\nSpeaker C: So there's no way of telling it not to tell us about the eva values.\nSpeaker B: Yeah, okay, that's...\nSpeaker C: Yeah, you're right.\nSpeaker C: So we get this whole list of things, and the question is, what to do with it?\nSpeaker C: What to hand on?\nSpeaker C: How to interpret it, in a sense?\nSpeaker C: So you said, if I'm only interested in whether he wants to go there or not, then I'd just look at that node...\nSpeaker C: Look at that struct in the output, right?\nSpeaker C: Yeah.\nSpeaker C: Look at that struct in the output, even though I wouldn't call it a struct, but...\nSpeaker B: Well, it's an XML structure that's being returned, right?\nSpeaker C: So every part of the structure is a struct?\nSpeaker B: Yeah, I just abbreviated it to struct in my head, starting going with that.\nSpeaker C: That element or struct?\nSpeaker C: Not a C struct.\nSpeaker B: That's not what I was talking about.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: And the reason is, I think it's a little bit more complex, so we can even think about it as an interesting problem in and of itself, is...\nSpeaker C: So the...\nSpeaker C: Let's look at an example.\nSpeaker B: Well, we're going to just take the structure that's outputted, and then run another transformation on it that would just dump the one that we wanted out.\nSpeaker C: Yeah, we'd need to prune, right?\nSpeaker C: Throw things away.\nSpeaker B: Well, actually, you don't even need to do that with XML.\nSpeaker B: Can't you just look at one specific...\nSpeaker B: Yeah, exactly.\nSpeaker C: The...\nSpeaker C: Circus allows you to say, just give me the value of that and that, and that.\nSpeaker C: But we don't really know what we're interested in before we look at the complete, at the overall result.\nSpeaker C: So the person said, um...\nSpeaker C: Where is x?\nSpeaker C: And so, we want to know, um, is...\nSpeaker C: does he want info on this, or know the location, or does he want to go there?\nSpeaker C: Let's assume this is our question.\nSpeaker C: Sure.\nSpeaker C: So...\nSpeaker C: Um...\nSpeaker C: Let's do this in parallel.\nNone: So we get...\nSpeaker C: So we get...\nSpeaker C: Okay.\nSpeaker C: Let's assume this is the output.\nSpeaker C: So we should be able to conclude from that that, I mean, it's always going to give us a value of how likely we think it is that he wants to go there and does want to go there.\nSpeaker C: Or how likely it is that he wants to get information, but maybe should just reverse this to make it a little bit more delicate.\nSpeaker C: So does he want to know where it is, or does he want to go there?\nSpeaker B: He wants to know where it is.\nSpeaker C: Right.\nSpeaker C: I tend to agree.\nSpeaker C: And if it's...\nSpeaker C: Well, I mean, you mean...\nSpeaker C: If there's sort of a clear winner here, and this is pretty...\nSpeaker C: indifferent, then we might conclude that he actually wants to just know where...\nSpeaker C: He does want to go there.\nSpeaker B: And I curiously, is there a reason why we wouldn't combine these three nodes into one smaller subnet that would just basically be the question for...\nSpeaker B: We have where as x is the question, right?\nSpeaker B: That would just be info on a location based upon...\nSpeaker C: Or go there.\nSpeaker C: A lot of people ask that if they actually just want to go there.\nSpeaker C: People come up to you on campus and say, well, it's the library.\nSpeaker C: You're going to say, go down that way.\nSpeaker C: You're not going to say it's 500 yards away from you, or it's north of you, or...\nSpeaker B: Well, I mean, so you just have three decisions for the final node that would link these three nodes in the...\nSpeaker B: in the...\nSpeaker B: Not together.\nSpeaker C: I don't know what to understand, but you mean...\nSpeaker C: But again, in this given this input, we also, in some situations, may want to postulate an opinion with a person who wants to go there now, the nicest way, use a cab, wants to know where it is because he wants something fixed there because he wants to visit it, or whatever.\nSpeaker C: So all I'm saying is, whatever our input is, we're always going to get the full output.\nSpeaker C: And some things will always be sort of...\nSpeaker C: too not significant enough.\nSpeaker B: Or it'll be tight.\nSpeaker B: It'll be hard inside.\nSpeaker B: But I guess the thing is...\nSpeaker B: This is another smaller case of reasoning in the case of uncertainty, which makes me think, Bayesnet should be the way to solve these things.\nSpeaker B: So if you had, for every construction, right?\nSpeaker B: Oh.\nSpeaker B: You could say, well, here's the where is construction.\nSpeaker B: And for the where is construction, we know we need to look at this node that merges these three things together as for to decide the response.\nSpeaker B: And since we have a finite number of constructions that we can deal with, we can have a finite number of nodes.\nSpeaker C: Okay.\nSpeaker B: So if we had to deal with arbitrary language, it wouldn't make any sense to do that because there'd be no way to generate the nodes for every possible sentence.\nSpeaker B: But since we can only deal with a finite amount of stuff.\nSpeaker C: So basically, the idea is to feed the output of that, believe in it, into another belief net.\nSpeaker B: Yeah. So basically, take these three things and then put them into another belief net.\nSpeaker C: But why only those three?\nSpeaker C: Well, I mean, for the where is question.\nSpeaker B: So we'd have a node for the where is question.\nSpeaker C: Yeah. But we believe that all the decision nodes can be relevant for the where is.\nSpeaker C: And how do I get to, or do I tell you something about?\nNone: You can come in if you want.\nSpeaker B: Is Putin online here?\nNone: Yes, it is allowed.\nSpeaker B: Is that actually you're not wearing your headphones?\nNone: All right, just say I'll be back.\nSpeaker B: Well, I see, I don't know if this is a good idea or not.\nSpeaker B: I'm just throwing it out.\nSpeaker B: But it seems like we could have, I mean, we could put all of the information that could possibly be relevant into the where is node answer.\nSpeaker B: Node thing stuff.\nSpeaker B: And, okay.\nSpeaker C: I mean, let's not forget we're going to get some very strong input from these, from these discourse things, right?\nSpeaker C: So tell me the location of X.\nSpeaker C: Or where is X located?\nSpeaker B: Wait.\nSpeaker B: Yeah, I know, but the base net would be able to, the way that's on the, on the nodes in the base net would be able to deal with that, wouldn't it?\nSpeaker B: Here's a, oh, I'll wait until you're plugged in.\nSpeaker B: Oh, don't sit there. Sit here.\nSpeaker B: You know how you don't like that one.\nSpeaker A: It's okay.\nSpeaker B: That's the weird one.\nSpeaker B: That's someone that's painful.\nSpeaker B: It hurts.\nSpeaker B: You're so bad.\nSpeaker B: I'm happy that they're recording that.\nSpeaker B: That headphone.\nSpeaker B: The headphone that you have to put on backwards with the little thing, and a little, a little phone block on it.\nSpeaker B: It's a painful, painful microphone.\nSpeaker C: I think it's called the crown.\nSpeaker C: The crown.\nSpeaker C: Yeah.\nSpeaker C: It was just the Sony.\nSpeaker A: The crown?\nSpeaker A: Is that the actual name?\nSpeaker A: Mm-hmm.\nSpeaker B: The manufacturer.\nSpeaker B: I don't see a manufacturer on it.\nSpeaker B: Oh, here it is.\nSpeaker B: This thingy.\nSpeaker B: Yeah, it's the crown.\nSpeaker B: The crown of pain.\nSpeaker B: You're on that?\nSpeaker B: Are you, are you my, is your mic on?\nSpeaker B: Okay.\nSpeaker B: So you've been working with these guys, you know what's going on.\nSpeaker B: Yes, I have.\nSpeaker B: I do.\nSpeaker A: No, I do.\nSpeaker A: It's a lot.\nSpeaker A: So where are we?\nSpeaker C: We're discussing this.\nSpeaker A: I don't think you can handle French.\nSpeaker C: So, we have something coming in.\nSpeaker C: Person says where is X, and we get a certain, we have a situation vector, and a user vector, and everything is fine.\nSpeaker C: And, and, and, and, and, or.\nSpeaker B: Did you just take the microphone actually in the T?\nSpeaker B: You know what?\nSpeaker A: And, I'm not drinking tea, what are you talking about?\nSpeaker C: Oh, yeah, sorry.\nSpeaker C: Let's just assume our base net just has three decision notes for the time being.\nSpeaker C: These three, he wants to know something about it.\nSpeaker C: He wants to know where it is, he wants to go there.\nSpeaker B: In terms of these would be how we would answer the question where is, right?\nSpeaker B: We, this is, this is what he's, it seemed like he explained it to me earlier.\nSpeaker B: We were, we want to know how to answer the question where is X.\nSpeaker C: No, I can, I can do the timing note in here too, and say okay.\nSpeaker B: Well, yeah, but in this, let's just deal with the simple case of, we're not worrying about timing or anything.\nSpeaker B: We just want to know how we should answer where is X.\nSpeaker B: Okay.\nSpeaker C: And, um, okay, and go there has two values, right?\nSpeaker C: Go there and not go there.\nSpeaker C: Let's assume those are the posterior probabilities of that.\nSpeaker C: InfoOn has two false and location.\nSpeaker C: So he wants to know something about it, and he wants to know something, he wants to know where it is, has these values.\nSpeaker C: And, um, Oh, I see why we can't do that.\nSpeaker C: And, um, in this case, we would probably all agree that he wants to go there.\nSpeaker C: I'll believe that things he wants to go there, right?\nSpeaker C: In the, whatever, if we have something like this here, and this, like that, and maybe here also some,\nSpeaker A: should probably make them happen. Yeah.\nSpeaker C: Something like that.\nSpeaker C: Then we would guess, aha, he, our belief net has stronger beliefs that he wants to know where it is, then actually wants to go there.\nSpeaker C: Right?\nSpeaker B: The, the, this is assumed though that they're evenly weighted.\nSpeaker B: Like, I guess they are evenly weighted.\nSpeaker A: The different decision nodes, you mean?\nSpeaker B: Yeah, they go there, the info on the location.\nSpeaker A: Well, yeah, this is making the assumption.\nSpeaker A: Yes.\nSpeaker C: I mean, by differently weighted, they don't fit into anything really anymore.\nSpeaker A: Or I mean, why do we, if we trusted the go there node more, much more than we trusted the other ones, then we would conclude even in the situation that he wanted to go there.\nSpeaker A: So in that sense, we weighed them equally.\nSpeaker A: Okay.\nSpeaker A: Makes sense.\nSpeaker B: So the, but I guess the, the question that I was at, or wondering, or maybe Robert was proposing to me, is how do we make the decision on as to which one to listen to?\nSpeaker A: Yeah, so the final decision is the combination of these three.\nSpeaker A: So again, it's, it's some kind of a, base net.\nSpeaker B: Yeah, actually.\nSpeaker B: Okay, so then the question, is that my question is to you then would be?\nSpeaker B: So the only reason we can make all these smaller base nets, because we know we can only deal with a finite set of constructions.\nSpeaker B: Because if we're just taking arbitrary language, then we couldn't have a node for every possible question, you know?\nSpeaker A: A decision on a February possible question, you mean?\nSpeaker B: Well, in the case of, yeah, in the case of any piece of language, we wouldn't be able to answer it with this system, if we just, because we wouldn't have the correct node, basically what you're proposing is a, where is node, right?\nSpeaker B: Yeah.\nSpeaker B: And if we, and if someone says, you know, something in Mandarin, yeah, to the system, we would know which node to look at to answer that question, right?\nSpeaker B: Yeah.\nSpeaker B: So, but if we have a finite, what?\nSpeaker C: I don't see a point.\nSpeaker C: What, what, what I am thinking of what we're about to propose here is, we're always going to get the whole list of values in their parts here, probabilities.\nSpeaker C: And now we need an expert system, or a belief net, or something that interprets that.\nSpeaker C: That looks at all the values and says, the winner is, timing now go there.\nSpeaker C: Go there timing now.\nSpeaker C: Or the winner is info on function off.\nSpeaker C: So, you want to know something about it and what it does.\nSpeaker C: Right?\nSpeaker C: Regardless of, yeah, but the input,\nSpeaker B: but how does the expert, how does the expert system know which one to declare the winner, if it doesn't know what question it is, and how that question should be answered?\nSpeaker C: Based on what the question was, so what the discourse, the autonomy, the situation, and the user model gave us, we came up with these values for these decisions.\nSpeaker B: Yeah, I know, but how do we wait what we get out?\nSpeaker B: As which ones are important?\nSpeaker B: So, if we were to do it with a base net, we'd have to have a node for every question that we knew how to deal with, that would take all of the inputs and wait them appropriately for that question.\nSpeaker B: Does that make sense?\nSpeaker B: Yeah, it may.\nSpeaker A: I mean, are you seeing that what happens if you try to scale this up to a situation where we're just dealing with arbitrary language?\nSpeaker A: Is that your point?\nSpeaker B: Well, no, I guess my question is, is the reason that we can make a node, or okay, so let me see if I'm confused.\nSpeaker B: Are we going to make a node for every question?\nSpeaker B: Does that make sense or not?\nSpeaker B: Every question?\nSpeaker A: Every construction.\nSpeaker A: I don't necessarily, I would think.\nSpeaker A: I mean, it's not based on constructions, it's based on things like, there's going to be a node for code, there are not, and there's going to be a node for attribute approach.\nSpeaker B: So someone asked a question.\nSpeaker B: How do we decide how to answer it?\nSpeaker C: Well, look at, look, face yourself with this question, you get this, this is what you get.\nSpeaker C: And now you have to make a decision, what do we think?\nSpeaker C: What does this tell us?\nSpeaker C: I'm not knowing what was asked and what happened, and whether the person was a tourist or a local, because all of these factors have presumably already gone into making these posterior probabilities.\nSpeaker C: Yeah.\nSpeaker C: What we need is a just a mechanism that says,\nSpeaker B: there is, I just don't think a winner take all type of thing is the,\nSpeaker A: I mean, in general, like, we won't just have those three, right? We'll have like many, many nodes.\nSpeaker A: So we have to like, so that it's no longer possible to just look at the nodes themselves and figure out what the person is trying to say.\nSpeaker C: Because there are interdependencies, right?\nSpeaker C: No, so if, for example, the go there, posterior probability is so high, if it has a risk of certain height, then all of this becomes relevant.\nSpeaker C: So even if the function or the history of something is scoring pretty good on the true node, true value.\nSpeaker B: I don't know about that, because I would suggest that, I mean, do they have to be mutual?\nSpeaker B: Do they have to be mutual exclusive?\nSpeaker C: I think to some extent, they are, or maybe they're not.\nSpeaker B: Because the way you describe what I meant, they weren't mutually exclusive to me.\nSpeaker C: Well, if he doesn't want to go there, even if the enter posterior probability, so go there is no, enter is high, and info on this.\nSpeaker B: Wait, I just added the other three that you had in the, those three nodes, they didn't seem like they were mutually exclusive.\nSpeaker B: No, there's no.\nSpeaker B: So yeah, but some things would drop out, and some things would still be important.\nSpeaker B: But I guess what's confusing me is if we have a base net to deal with, another base net to deal with this stuff, yeah, is the only reason, okay, so I guess if we have another base net to deal with this stuff, the only reason we can design it is because we know what each question is asking.\nSpeaker A: Yeah, that's true.\nSpeaker B: And then so with the only reason, the way we would know what question is asking is based upon, oh, so let's say I had a construction parser, and I would know what each construction, the communicative intent of the construction was, and so then I would know how to wait the nodes appropriately in response.\nSpeaker B: So no matter what they said, if I could map it onto a where is construction, I could say, ah, well, the intent here was where is, and I could look at those.\nSpeaker A: Yeah, yeah, I mean, sure.\nSpeaker A: You do need to know, do you need to have that kind of animation?\nSpeaker C: Yeah, I'm also agreeing that a simple, take the ones where we have a clear winner, forget about the ones where it's all sort of middle ground, prune those out and just hand over the ones where we have a winner.\nSpeaker C: Because that would be the easiest way.\nSpeaker C: We just compose as an output on X-Route-Best message that says, go there, now enter historical information, and not care whether that's consistent with anything.\nSpeaker C: Right? In this case, we say, definitely doesn't want to go there.\nSpeaker C: He just wants to know where it is, or let's call this, let's look at, he wants to know something about the history of, so he said, tell me something about the history of that.\nSpeaker C: Now, the, but for some reason, the endpoint approach gets a really high score, too.\nSpeaker C: We can't expect this to be sort of open, 3333, open, 3333, open, 3333, right?\nSpeaker C: Somebody needs to sap that, or no, there needs to be some knowledge that...\nSpeaker B: Well, yeah, but the BayesNet that would merge, and realize I had my hand in between my mouth and my, my, my, my, my, my, my, my, my, my, my, my, so then the BayesNet that would merge, there, that would make the decision between go there, info on location, would have a node to tell you, which one of those three you wanted, and based upon that node, then you would look at the other stuff.\nSpeaker B: I mean, does that make sense?\nSpeaker C: Sort of one of those, that's, it's more like a decision tree, if you want. You first look at the real ones, and then...\nSpeaker B: Yeah, I didn't intend to say that every possible, okay, there was confusion there, I didn't intend to say every possible thing should go into the BayesNet because some of the things aren't relevant in the BayesNet for a specific question, like the endpoint is not necessarily relevant in the BayesNet for where is until after you've decided whether you want to go there or not.\nSpeaker C: Right.\nSpeaker B: Show us the way, Bosch.\nSpeaker B: I just see other things that, yeah,\nSpeaker A: when you're asking specific questions, you don't even, like if you're asked a various question, you may not even look, like, ask for the posterior probability of the EVA node, right?\nSpeaker A: Because that's what, I mean, in the BayesNet, you always ask for the posterior probability of a specific node.\nSpeaker A: So, I mean, you may not even bother to compute things you don't need.\nSpeaker C: And we're always computing all?\nSpeaker A: No. You can compute the posterior probability of one subset of the nodes given some other nodes, but totally ignore some other nodes also.\nSpeaker A: Basically, things you ignore get marginalized over.\nSpeaker C: Yeah, but that's, that's a shifting the problem.\nSpeaker C: Then you would have to make a decision, okay?\nSpeaker C: Yeah, that's a various question, which is a node to our query?\nSpeaker A: Yes.\nSpeaker A: Well, I think that's what you want to do, right?\nSpeaker B: Well, eventually you still have to pick up which ones you're looking at.\nSpeaker B: So, it's pretty much the same problem.\nSpeaker C: Yeah, it's apples and oranges.\nSpeaker C: I mean, maybe it just makes a difference in terms of performance, computational times.\nSpeaker C: Either you always have to compute all the posterior probabilities for all the values, all nodes, and then prune the ones you think that are the same.\nSpeaker C: The ones you think that are irrelevant, or you just make a priori estimate of what you think might be relevant and query those.\nSpeaker A: Yeah.\nNone: So basically you'd have a decision tree query go there.\nNone: If that's false, query this one, if that's true, query that one, and just basically do a binary search through the...\nNone: I don't know if it would necessarily be that complicated, but...\nSpeaker B: Well, in the case of go there, it would be, in the case, because if you needed to...\nSpeaker B: If go there was true, you'd want to know what endpoint was, and if it was false, you'd want to look at either info on our history.\nSpeaker A: Yeah.\nSpeaker A: That's true, I guess.\nSpeaker A: Yeah, so in a way you would have that.\nNone: Awesome.\nNone: Some would be boggled by the hug and software.\nNone: Okay, why is that?\nSpeaker B: I can't figure out how to get the probabilities into it.\nSpeaker B: Like, I'd look at...\nSpeaker B: It's some way...\nSpeaker B: It's boggling me.\nSpeaker B: Okay.\nSpeaker B: All right.\nSpeaker B: But hopefully it's...\nSpeaker B: Oh yeah, I just think I haven't figured out what the terms in Huggin' Mean versus what Java-based terms are.\nSpeaker B: Okay.\nSpeaker C: By the way, do we know whether Jury and Nancy are coming?\nSpeaker A: Or...\nSpeaker A: They should come and they're done their stuff, basically, whenever that is.\nSpeaker A: So...\nSpeaker B: What do they need to do left?\nSpeaker A: I guess Jury needs to enter Marx, but I don't know if he's going to do that now or later, but if he's going to enter Marx, he's going to take him away, I guess, and he won't be here.\nSpeaker A: And what's Nancy doing?\nSpeaker A: Nancy...\nSpeaker A: She was sort of finishing up the calculation of Marx and his signing of grades, but I don't know if she should be here.\nSpeaker A: Well, or she should be free after that, so...\nSpeaker A: Assuming she's coming to this meeting.\nSpeaker A: I don't know if she knows about it.\nSpeaker B: She's on the email, let's try it.\nSpeaker A: Okay.\nSpeaker C: Okay.\nSpeaker C: Because...\nSpeaker C: Basically, what we also have decided prior to this meeting is that we would have a re-run of the three of us sitting together.\nSpeaker C: Okay.\nSpeaker C: So, we're going to come to this week, again, and finish up the values of this.\nSpeaker C: So we have...\nSpeaker C: Believe it or not, we have all the bottom ones here.\nSpeaker C: Well, either the bunch of notes or...\nSpeaker C: Yeah.\nSpeaker C: Actually, what we have is this line.\nSpeaker B: Right?\nSpeaker B: What do the stretchers do? So, for instance, this location now has two inputs.\nSpeaker A: Four.\nSpeaker A: Those are the bottom things are inputs also.\nSpeaker B: Oh, I see.\nSpeaker B: Okay, that makes a lot more sense to me now.\nSpeaker B: Because I thought it was like that one in Stuart's book about...\nSpeaker B: Alarm in the dog.\nSpeaker B: Yeah, or the earthquake in the alarm.\nSpeaker A: Sorry, yeah, I'm confusing too.\nSpeaker B: Yeah, there's a dog one too, but that's in Java Bay, isn't it?\nSpeaker B: Maybe.\nSpeaker B: Or there's something about bowel problems or something with the dog.\nSpeaker C: And we have all the top ones, all the ones to which no arrows are pointing.\nSpeaker C: What we're missing are the...\nSpeaker C: These arrows are pointing where we're combining top ones.\nSpeaker C: So we have to come up with values for this.\nSpeaker C: This, this, this, this, and so forth.\nSpeaker C: And maybe this fiddle around with it a little bit more.\nSpeaker C: And then it's just edges.\nSpeaker C: Many of edges.\nSpeaker C: And we won't meet next Monday.\nSpeaker B: So...\nSpeaker B: Just a memorial day.\nSpeaker A: Yeah, it would be next Tuesday, I guess.\nSpeaker B: When's Jerry leaving for Italy?\nSpeaker B: On Friday.\nSpeaker B: Which Friday?\nSpeaker B: This Friday.\nSpeaker A: Oh, this Friday?\nSpeaker B: As in four days?\nSpeaker B: Yeah.\nSpeaker B: Or three days?\nSpeaker A: How long has he gone for?\nSpeaker B: Two weeks.\nSpeaker A: Italy, huh?\nSpeaker A: What's there?\nSpeaker C: That's a country.\nSpeaker C: Billings.\nSpeaker C: People.\nSpeaker B: It does not account for anything.\nSpeaker A: He's just visiting.\nSpeaker A: Right.\nSpeaker A: Just visiting.\nSpeaker A: Vacation.\nSpeaker A: Let's be honest, please.\nSpeaker C: You can't really do that.\nSpeaker C: Do you guys...\nSpeaker C: Yeah.\nSpeaker C: So part of what we actually want to do is sort of sketch out what we want to surprise him with when he comes back.\nSpeaker C: I think we should disappoint him.\nSpeaker C: Or have a finished construction parser and working belief net.\nSpeaker B: That wouldn't be disappointing.\nSpeaker B: I think we should absolutely know work for the two weeks that he's gone.\nSpeaker C: Well, that's actually what I had planned.\nSpeaker C: Personally, I had sort of sketched it out in my mind that you guys do a lot of work and I do nothing.\nSpeaker C: And then I sort of...\nSpeaker C: Oh, that sounds good too.\nSpeaker C: Sort of bask in your glory.\nSpeaker C: But you guys have any vacation plans because I myself am going to be gone.\nSpeaker C: But this is actually not really important just this weekend.\nSpeaker C: So we're going to go and get this.\nSpeaker B: I want to be this guy this weekend too.\nSpeaker C: But we're all going to be here on Tuesday again.\nSpeaker C: Looks like it.\nSpeaker C: Okay, then let's meet again next Tuesday and finish up this base net.\nSpeaker C: And once we have finished it, I guess we can...\nSpeaker C: And that's going to be more...\nSpeaker C: Just you and me because Baskara is doing probabilistic, reclusive, structured, object oriented.\nSpeaker B: Killing machines.\nSpeaker C: Reasoning machines.\nSpeaker B: And...\nSpeaker B: Killing, reasoning.\nSpeaker B: What's the difference?\nSpeaker B: I think next Tuesday is it the whole group meeting or just working on it?\nSpeaker C: The whole group and we present our results.\nSpeaker C: A final definite.\nSpeaker B: So when you're saying we need to do a run of...\nSpeaker B: Like just working out the rest of the...\nSpeaker C: Yeah, we should do this the upcoming days.\nSpeaker C: So this weekend.\nSpeaker B: When you say the whole group, you mean the four of us and Keith?\nSpeaker C: And Ami might...\nSpeaker B: Be here and it's possible that Nancy will be here.\nSpeaker B: So...\nSpeaker B: Yeah.\nSpeaker C: Because once we have the...\nSpeaker B: You just have to explain it to me then on Tuesday how it's all going to work out.\nSpeaker B: Yeah.\nSpeaker C: We were.\nSpeaker C: Okay.\nSpeaker C: Because once we have it sort of up and running then we can start defining the interfaces and then feed stuff into it and get stuff out of it and then hook it up to some fake construction parser.\nSpeaker B: That you will have in about nine months or so, yeah.\nSpeaker B: And the first bad version will be done in nine months.\nSpeaker C: Yeah, I can worry about the ontology interface and you can...\nSpeaker C: Keith can worry about the discourse.\nSpeaker C: I mean this is pretty... I mean I hope everybody knows that these are just going to be dummy values, right?\nSpeaker C: With...\nSpeaker C: So if the endpoint...\nSpeaker C: If the go there is yes and no then go there discourse will just be 50-50, right?\nSpeaker A: What do you mean if the go there says no then the go there is...\nSpeaker A: I don't understand.\nSpeaker A: Like the go there depends on all those four things.\nSpeaker C: Yeah. But what are the values of the go there discourse?\nSpeaker A: Well it depends on this situation.\nSpeaker A: The discourse is strongly indicating that...\nSpeaker A: Yeah. But we have no discourse input.\nSpeaker A: Oh I see.\nSpeaker A: So you're specifically in our situation D and R are going to be...\nSpeaker A: Yeah.\nSpeaker A: Sure.\nSpeaker B: So far we have...\nSpeaker B: Is that what the Keith knows?\nSpeaker B: Yeah.\nSpeaker B: Okay. And you're taking it out for now.\nSpeaker C: Well this is D...\nSpeaker C: Okay.\nSpeaker C: This I can...\nSpeaker C: What the D's are.\nSpeaker C: I can get it in here.\nSpeaker C: So we have the...\nSpeaker C: Let's call it Keith, John O.\nNone: Note.\nSpeaker C: Note.\nSpeaker C: There is an H somewhere.\nSpeaker C: There you go.\nSpeaker A: People that have the same problem with my name.\nSpeaker C: And...\nSpeaker B: This is the H before the A or after the A.\nSpeaker B: Oh and my name before the A.\nSpeaker B: Okay good.\nSpeaker B: Because when you said people have the same problem with it.\nSpeaker B: Because my age goes after the A.\nSpeaker B: People have a worse problem with my name.\nSpeaker B: I always have to check every time I send you an email.\nSpeaker B: A past email if yours to make sure I'm spilling your name correctly.\nSpeaker B: That's good.\nSpeaker B: I worry about you.\nSpeaker C: I appreciate that.\nSpeaker C: But when you abbreviate yourself as the busman, you don't use any H.\nSpeaker A: Busman?\nSpeaker A: Yeah, it's because of the chest player name Michael Busman.\nSpeaker A: Busman, who is my hero?\nSpeaker A: Okay.\nSpeaker B: You're a geek.\nSpeaker B: It's okay.\nSpeaker B: How do you pronounce your name?\nSpeaker B: Eva.\nSpeaker B: Eva?\nSpeaker B: Yeah.\nSpeaker B: Not Eva.\nSpeaker B: What if I were to call you Eva?\nSpeaker B: I probably still respond to it.\nSpeaker B: I thought people would call me Eva but...\nSpeaker B: I don't know.\nSpeaker B: And I just Eva, Eva.\nSpeaker B: Like if I take the V and pronounce it like it was a German V.\nSpeaker C: Which is F.\nSpeaker B: Yeah.\nSpeaker B: No idea.\nSpeaker B: Loist.\nSpeaker B: What?\nSpeaker B: It sounds like an F.\nSpeaker C: There's also an F in German which is why...\nSpeaker C: It's just a difference between voice and unvoiced.\nNone: Okay.\nSpeaker B: As long as that's okay.\nSpeaker B: I mean I might slip out and say it accidentally.\nSpeaker A: That's all I'm saying.\nSpeaker A: It's fine.\nSpeaker A: Yeah, it doesn't matter what those nodes are anyway because we'll just make the weights here for now.\nSpeaker C: Yeah.\nSpeaker C: We'll make them 0 for now because who knows what they come up with.\nSpeaker C: What's going to come in there?\nSpeaker C: Okay.\nSpeaker C: And...\nSpeaker C: Then...\nSpeaker C: Should we start on Thursday?\nSpeaker C: And not meet tomorrow?\nSpeaker C: Sure.\nSpeaker C: I'll send an email.\nSpeaker C: Make a time suggestion.\nSpeaker B: Maybe it's okay so that we have one node per construction.\nSpeaker B: Because even in people, like they don't know what you're talking about if you're using some sort of strange construction.\nSpeaker C: Yeah, they would still sort of get the closest best fit.\nSpeaker B: Yeah, but I mean that's what the construction parts would do.\nSpeaker B: If you said something completely arbitrary, it would find the closest construction.\nSpeaker B: But if you said something that was completely...\nSpeaker B: Or theoretically the construction parts would do that.\nSpeaker B: If you said something for which there was no construction whatsoever, people wouldn't have any idea what you're talking about.\nSpeaker B: Like bust dog, fried egg.\nSpeaker B: Or if you've something Chinese for sure.\nSpeaker B: Or something in the internet.\nSpeaker B: Or Cantonese is the case maybe.\nSpeaker B: What do you think about that, boss?\nSpeaker A: I mean...\nSpeaker A: Well...\nSpeaker A: But how many constructions could we possibly have nodes for?\nSpeaker B: In this system or in...\nSpeaker A: No, we...\nSpeaker A: Like when people do this...\nSpeaker B: Oh, and how many constructions do people have?\nSpeaker A: Yeah.\nSpeaker A: I have no idea.\nSpeaker A: Is it considered to be like...\nSpeaker A: Are they considered to be like very...\nSpeaker A: Every now and then is the construction.\nSpeaker A: Okay, so it's like...\nSpeaker A: Thousands.\nSpeaker B: Any form meaning pair to my understanding is a construction.\nSpeaker B: And form starts at the level of...\nSpeaker B: Or actually maybe even sounds.\nSpeaker B: Yeah.\nSpeaker B: And goes upwards until you get the die transitive construction.\nSpeaker B: And then of course, I guess maybe there can be...\nSpeaker B: Can there be combinations of the die...\nSpeaker B: Yeah.\nSpeaker B: The giving a speech construction.\nSpeaker C: Retaric for construction, sir.\nSpeaker C: But I mean...\nSpeaker C: You know, you can probably count the ways.\nSpeaker C: It's probably...\nSpeaker B: I would definitely say it's finite.\nSpeaker B: Yeah.\nSpeaker B: And at least if you're a compiler, that's all that really matters.\nSpeaker B: As long as your analysis is finite.\nSpeaker A: How's this sound going to be finite again?\nSpeaker B: No, I can't think of a way it would be infinite.\nSpeaker B: Well, you can come up with new constructions.\nSpeaker B: Yeah.\nSpeaker B: If your brain is totally non-deterministic, then perhaps there's a way to get an infinite number of constructions.\nSpeaker B: You have to worry about...\nSpeaker A: What do you mean the fact that you can't say that it's impossible?\nSpeaker B: Right.\nSpeaker B: Because if you have a fixed number of neurons...\nSpeaker B: Yeah.\nSpeaker B: So the best case scenario would be the number of...\nSpeaker B: Or the worst case scenario is the number of constructions equals the number of neurons.\nSpeaker A: Well, two to the power of the number of neurons.\nSpeaker B: But still, finite.\nSpeaker C: Okay.\nSpeaker B: No, wait, not necessarily.\nSpeaker B: Is it...\nSpeaker B: We can end the meeting.\nSpeaker B: I just...\nSpeaker B: Can't you use different levels of activation across...\nSpeaker B: Hmm.\nSpeaker B:...lots of different neurons to specify different values?\nNone: Yeah.\nSpeaker A: There's a bandwidth issue, right?\nSpeaker A: Yeah.\nSpeaker A: You can't do better than...\nSpeaker C:...totally, by some other words, it gets really tough for a future.\nNone: Transcribers, too.\nNone: Transcribers, please.\n"}, {"meeting_id": "Bed006", "summary": "Meeting participants began by going over the logistics of recruiting participants, particularly through departmental mailing lists. They then moved onto discussing a new role, the instructor, who would help experimental participants figure out their intentions. This would eliminate a pre-written set of goals for the participants. Participants also discussed how the schema would incorporate other information, like choosing movies, and how more complex actions, like that of purchasing goods, could be incorporated into the source-path-goal schema. Finally, the participants concluded that an SPG schema of a different kind, one that can transfer previously acquired information, may be more helpful. The meeting concluded with the discussion of a new meeting time.", "dialogue": "None: Thanks.\nSpeaker B: Are you okay?\nSpeaker D: I am feeling high.\nSpeaker D: I remember talking about aspect.\nSpeaker D: You were my GSI briefly.\nSpeaker H: Some introductions are all there.\nSpeaker H: For those who don't know, everyone knows me.\nSpeaker H: Part of that, the old gang, John and Basgra, have been with us from day one.\nSpeaker H: They are engaged in various activities, some of which you will hear about today.\nSpeaker H: Ami is our counselor and spiritual guidance.\nSpeaker H: Also interested in problems concerning reference of the more complex type.\nSpeaker H: He sits in as an interested participant and helper.\nSpeaker H: Is that a good characterization?\nSpeaker H: Keith is not technically one of us yet.\nSpeaker H: But it's too late for him now.\nSpeaker H: I've got the headset on after all.\nSpeaker H: Officially, I guess you will be joining us in the summer.\nSpeaker H: Hopefully, it is by means of Keith that we will be able to get a better formal and a better semantic idea of what a construction is.\nSpeaker H: And how we can make it work for us.\nSpeaker H: Additionally, he is interested in surpassing English because it also includes German.\nSpeaker H: Next to a capability of speaking and writing and understanding and reading that language.\nSpeaker H: Is there anyone who doesn't know Nancy?\nSpeaker G: I made the joke already, Nancy.\nSpeaker G: What?\nSpeaker G: I don't know myself, joke.\nSpeaker G: Did you win?\nSpeaker G: Before you came in.\nSpeaker E: About me or you?\nSpeaker E: About me.\nSpeaker E: About you.\nSpeaker F: Well, I didn't know I didn't mean to be humor-copying, but I guess I know myself.\nSpeaker H: And Fei is with us six days ago, officially.\nSpeaker H: But in reality, he is already much, much longer.\nSpeaker H: Next to some more less bureaucratic stuff with the data collection.\nSpeaker H: She is also the wizard in the data collection.\nSpeaker H: We are sticking with the term wizard.\nSpeaker H: And...\nSpeaker H: Which?\nSpeaker H: Wizard that.\nSpeaker I: Wizard.\nSpeaker I: I think.\nSpeaker H: I think.\nSpeaker H: Wizard.\nSpeaker H: Wizard.\nSpeaker H: Okay.\nSpeaker H: By popular vote.\nSpeaker H: Take a vote.\nSpeaker H: Okay.\nSpeaker H: Why don't we get started on that subject anyways?\nSpeaker H: So we are about to collect data and the following things have happened since we last met.\nSpeaker H: One, we were three meters.\nSpeaker H: And...\nSpeaker H: What happened is that there was some confusion between you and Jerry, with the leading three you are talking to Catherine Snow.\nSpeaker H: And he was...\nSpeaker H: He agreed completely that something confusing happened.\nSpeaker H: His idea was to get sort of the list of mayors of the department, the students.\nSpeaker H: It's exactly how you interpreted it.\nSpeaker H: The list of mayors.\nSpeaker H: Major's, Major's.\nSpeaker H: Major's okay.\nSpeaker H: Major's.\nSpeaker H: Major's.\nSpeaker E: Major's.\nSpeaker E: Major's.\nSpeaker H: Major's.\nSpeaker H: And just sending the little right up that we did to those email lists.\nSpeaker D: So it was really Carol Snow who was confused not me and not Jerry.\nSpeaker H: Yeah.\nSpeaker H: Okay.\nSpeaker H: So that is...\nSpeaker D: So I should still do that.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And...\nSpeaker D: Using the thing that you wrote up.\nSpeaker H: Yeah.\nSpeaker H: Okay.\nSpeaker H: And we have a little description of asking people subjects to contact Fave for, you know, recruiting them for our thing.\nSpeaker H: And there was some confusion as to the consent form, which is basically that what you just signed.\nSpeaker H: And since we have one already.\nSpeaker E: Did Jerry talk to you about maybe using our class, the students in the undergrad class, like you're teaching?\nSpeaker H: Well, he said, we definitely, yes, however there is always more people in a department than are just taking his class or anybody else's class at the moment.\nSpeaker H: And we'll reach out and try to get them all.\nSpeaker H: Okay.\nSpeaker F: But I guess it's that people in his class cover a different set.\nSpeaker F: So then the...\nSpeaker F: Is it the college side department that you were talking about?\nSpeaker F: I guess.\nSpeaker F: You can't see.\nSpeaker D: Because we have suggested to him that there are various people like Jerry and George and etc.\nSpeaker D: And just advertising their classes as well.\nSpeaker D: Or even I could do the actual...\nSpeaker D: Because I mean, I know how to contact our students.\nSpeaker F: If there's something that you're sending out, you can also send me a copy, me your boss card, either of us could post it to.\nSpeaker F: If it's a general solicitation that, you know, just contact you, then we can totally post it to the new group.\nSpeaker F: Do it.\nSpeaker F: Okay.\nSpeaker F: So you'll send it or something?\nSpeaker F: Is it what effect...\nSpeaker F: You can send it to me.\nSpeaker F: No.\nSpeaker F: Don't worry.\nSpeaker H: I'll wrap it.\nSpeaker H: However, I suggest that if you look at your email carefully, you may think you may find that you already have it.\nSpeaker H: Really?\nSpeaker H: Maybe.\nSpeaker H: I don't remember.\nSpeaker H: Well, see.\nSpeaker H: Anyhow.\nSpeaker H: The...\nSpeaker H: Yeah, not only Cox, also we were talking about linguistics and of course computer science.\nSpeaker H: And then secondly, we had... you may remember the problem of the rephrasing that subjects always rephrase sort of the task that we gave them.\nSpeaker H: And so we had a meeting on Friday talking about how to avoid that.\nSpeaker H: And it proved finally fruitful in a sense that we came up with a new scenario for how to get the subject to really have intention and sort of to act upon those.\nSpeaker H: And they're the ideas now that next... actually we need to hire one more person to actually do that job because it's getting more complicated.\nSpeaker H: So if you know anyone interested in what I'm about to describe, tell that person to write a mail to me or very soon.\nSpeaker H: The idea now is to sort of come up with a high level of sort of abstract tasks, go shopping, take in a batch of art, visit... do some side-seeing.\nSpeaker H: Sort of analogous to what Faye has started in compiling here and already... she has already gone to the trouble of anchoring it with specific entities in real world places you will find in Heidelberg.\nSpeaker H: And so out of these high level categories, the subject can pick a couple, such as if there's a category in emptying your role of film, the person can then decide, okay, I want to do that at this place, sort of make up their own attinuery and tasks. And the person is not allowed to take sort of this high level category list with them but the person is able to take notes on a map that we will give him.\nSpeaker H: And the map will be a tourist sort of schematic representation with symbols for the objects.\nSpeaker H: And so the person can maybe make a mental note that I wanted to go shopping here and I wanted to maybe take a picture of that and maybe eat here and then goes in and solves the task with the system, i.e. Faye.\nSpeaker H: And we're going to try out any questions.\nSpeaker F: So you'll have them say somewhere what their intention was, so you still have the nice thing about having data where you know what the actual intention was.\nSpeaker F: But there's nothing that says, these are the things you want to do, so they'll say, well these are the things I want to do.\nSpeaker F: So they'll have a little bit more natural interaction.\nSpeaker F: Hopefully.\nSpeaker A: So they'll be given this map, which means that they won't have to ask the system for high level information, what do I think?\nSpeaker H: It'll still require that information.\nSpeaker F: It doesn't have streets on it.\nSpeaker H: Not really the street network.\nSpeaker H: So you're just saying like what part of town the things are in there?\nSpeaker H: The map is more means for them to have the buildings and their names and maybe some major streets and their names.\nSpeaker H: And we want to maybe ask them if you have get it sort of isolated street, whatever river street.\nSpeaker H: And we've decided that yes, that's where they want to do this kind of action that they have it with them and they can actually read them or sort of have the label for the object because it's too hard to memorize all these strange German names.\nSpeaker H: And we're going to have another trial run, either first with that new setup tomorrow at two.\nSpeaker H: And we have a real interesting subject, which is Ron Kay.\nSpeaker H: For those who know him, he's the founder of ICI.\nSpeaker H: So he's around seven, seven years old.\nSpeaker E: I didn't know he was the founder.\nSpeaker H: And he also approached me and he offered to help our project and he was more thinking about some high level thinking tasks.\nSpeaker H: So really help you can come in as a subject.\nSpeaker H: So that's what's going to happen tomorrow.\nSpeaker H: Using this new new new setup.\nSpeaker H: Which I'll hope police sort of scrap together.\nSpeaker H: But thanks to Fay, we already have some nice blueprint and I can work with that.\nSpeaker H: Questions, comments of that?\nSpeaker C: Not sure if we can move on.\nSpeaker H: No, no more questions.\nSpeaker H: I'm not sure I totally understand this.\nSpeaker I: I'm not sure I totally understand everything that's been talked about.\nSpeaker E: With the very rough setup of the data.\nSpeaker I: This is where they're supposed to talk to a machine and it breaks down and then the human comes on.\nSpeaker H: The question is just sort of how do we get the tasks in their head?\nSpeaker H: That they have an intention of doing something.\nSpeaker H: Have a need to ask the system for something.\nSpeaker H: Without giving them sort of a clear wording of phrasing of the task.\nSpeaker H: Because what will happen then is that people repeat errors much as they kind of that phrasing.\nSpeaker F: Are you worried about being able to identify the goals that you guys have been talking about?\nSpeaker F: Are these identifying which of three modes their question concerns?\nSpeaker F: So it's like the enter.\nSpeaker H: Yeah, we will sort of get a protocol of the prior interaction.\nSpeaker H: That's where the instructor, the person we are going to hire.\nSpeaker H: And the subject sit down together with these high level things.\nSpeaker H: The first question for the subject is, we thought the tours can do.\nSpeaker H: Is there anything that interests you?\nSpeaker H: This is something I would go shopping.\nSpeaker H: Then we can sort of the instructor can say, well, then you may want to find out how to get over here.\nSpeaker H: This is where the shopping district is.\nSpeaker F: So the interaction beforehand will give them hints about how specific or whatever.\nSpeaker F: The kinds of questions that they are going to ask during the actual session?\nSpeaker H: No, just sort of, okay, what would you like to buy?\nSpeaker H: Okay, you want to buy whatever Kukuk's clocks.\nSpeaker H: There is a store there.\nSpeaker H: So the task then for that person is finding out how to get there.\nSpeaker H: That's sort of what's left.\nSpeaker H: And we know that the intention is to enter because we know that the person wants to buy a Kukuk's clock.\nSpeaker F: That's what I mean.\nSpeaker F: So this task is all going to be unambiguous about which is the three.\nSpeaker F: Hopefully.\nSpeaker B: So the idea is to try to get the actual phrasing that they might use and try to interfere as little as possible with their choice of rules.\nSpeaker H: Yes, in a sense, that's exactly the idea, which is never possible in a lap situation.\nSpeaker B: There was one experiment that I've let somewhere use pictures to actually specify the task.\nSpeaker H: Yeah, but we had exactly that on our list of possible things.\nSpeaker H: So I even made a sort of silly thing how that could work, how you can show.\nSpeaker H: You are here.\nSpeaker H: You want to know how to get some place.\nSpeaker H: And this is the place and it's a museum and you want to do some, and there's a person looking at pictures.\nSpeaker H: So, you know, this is exactly getting someplace with the intention of entering and looking at pictures.\nSpeaker H: However, not only was the common census among all participants of Friday's meeting was, it's going to be very laborious to make these drawings for each different things, other different actions, if at all possible, and also people will get caught up in the pictures.\nSpeaker H: So all of a sudden we'll get descriptions of pictures in there.\nSpeaker H: Right.\nSpeaker H: And people talking about pictures and pictorial representations.\nSpeaker H: Right.\nSpeaker H: I would still be willing to try it.\nSpeaker B: I mean, I'm not saying it's necessary, but you might be able to combine text and some sort of picture.\nSpeaker B: And also, I think it would be a good idea to show them the text and kind of choose a task and then take the test away.\nSpeaker B: Yeah.\nSpeaker B: The test away.\nSpeaker B: They're not guided by what you wrote, but come up with your own.\nSpeaker H: They will have no more linguistic matter in front of them when they enter this room.\nSpeaker C: Okay.\nSpeaker H: Then I suggest we move on to the, we have the EDU project, let me make one more general remark, has sort of two side actions, it's action items that we're dealing with.\nSpeaker H: One is modifying the smart composer and the other one is modifying the smart, natural language generation, module.\nSpeaker H: And this is not too complicated, but I'm just mentioning it, put it in the framework because this is something we will talk about now.\nSpeaker H: I have some news from the generation.\nSpeaker H: Do you have news from the parser?\nSpeaker H: Oh, not.\nSpeaker H: But that look, I...\nSpeaker A: I have pretty, very, very, very, very, very, very talk about how I'm fighting.\nSpeaker H: Okay.\nSpeaker H: Yeah, wonderful.\nSpeaker H: Did you run into problems or did you run into not having time?\nNone: Yeah.\nSpeaker H: Okay, so that's good.\nSpeaker H: That's better than running into problems.\nSpeaker H: And I do have some good news for the natural language generation, however.\nSpeaker H: And the good news is, I guess it's done.\nSpeaker H: Meaning that Tillmann Becker, who does the German one, actually took out some time and already did it in English for us.\nSpeaker H: And so the version he's sending us is already producing the English that's needed to get by in version 1.1.\nSpeaker A: So I think it's not as similar to what he did for the parsing.\nSpeaker H: Yeah, even though the generator is a little bit more complex and it would have been...\nSpeaker H: not changing 100 words, but maybe 400 words, but it would have been pre-emergerous.\nSpeaker H: So this is, I guess, good news.\nSpeaker H: And the time, especially Basquira and...\nSpeaker H: What do I have here?\nSpeaker H: No.\nSpeaker H: The time is now pretty much fixed.\nSpeaker H: It's the last week of April until the 4th of May.\nSpeaker H: So it's 26th through 4th.\nSpeaker H: Will they be here?\nSpeaker H: That they'll be here.\nSpeaker H: So it's extremely important that the two of you are also present this town through that time.\nSpeaker G: Wait, what are the days?\nSpeaker G: April 26th through the May 4th?\nSpeaker G: Yeah, something like that.\nSpeaker G: I'll probably be here.\nSpeaker I: You will be here.\nSpeaker H: Isn't finals coming up then pretty much after that?\nSpeaker A: Finals was that.\nSpeaker F: It doesn't really have much meaning to grad students, but final projects might.\nSpeaker G: Anyway, so this...\nSpeaker G: Well, I'll be here working on something.\nSpeaker G: Guaranteed, it's just...\nSpeaker G: Well, I'll be here, you know.\nSpeaker G: I'll be here too, actually.\nSpeaker H: No, it's just...\nSpeaker H: They're coming for us so that we can bug them and ask them more questions.\nSpeaker H: And it's a done-together and right sensible code.\nSpeaker H: So they can give some nice talks and stuff, but...\nSpeaker G: So it's not like we need to be with them 24 hours a day.\nSpeaker G: It's seven days that they're here.\nSpeaker H: Not unless you really want to.\nSpeaker H: They're very dependent.\nSpeaker H: Not unless you really want to.\nSpeaker H: And they're both nice guys, so you really make one.\nSpeaker H: Okay, that much from the pass through and generator side.\nSpeaker H: And there's more questions on that.\nSpeaker E: So no sample generator output yet?\nSpeaker E: No.\nSpeaker H: Okay.\nSpeaker H: Just a mail that, you know, he sent me the stuff soon.\nSpeaker H: And I was completely flammogasted.\nSpeaker H: And it's also...\nSpeaker H: It's going to produce the concept to speech, blah, blah, blah, information for necessary for 1.1 in English, based on English.\nSpeaker H: In English, so.\nSpeaker H: I was like, okay.\nSpeaker I: We're done.\nSpeaker F: So that was like one of the first, you know, the first task was getting it working for.\nSpeaker F: So that's basically over now, is that right?\nSpeaker F: Yeah.\nSpeaker F: So the basic requirement fulfilled.\nSpeaker H: The basic requirement is fulfilled almost when Andreas Stolke and his gang, when they have changed the language model of the regularizer and dictionary, then we can actually put it all together and you can speak into it and ask for TV and movie information.\nSpeaker H: And then when, if something actually happens and some answers come out, then we're done.\nSpeaker F: So it's not done, please.\nSpeaker F: And they kind of are correct.\nSpeaker I: Perhaps the answer says something to do with the questions for us then.\nSpeaker E: Is it using the database, the German TV movie?\nSpeaker E: Yeah.\nSpeaker F: Okay. So all the actual data might be German.\nSpeaker H: Well, actually...\nSpeaker F: You know, they're all like American TV programs.\nSpeaker I: I want to see Dukes von Hezzar in the middle of the day.\nSpeaker H: Dukes.\nSpeaker H: Okay. So you don't know how the German dialogue, the German demo dialogue actually works.\nSpeaker H: The first thing is what's showing on TV and then the person is presented with what's running on TV in Germany on that day, on that evening.\nSpeaker H: And so you take one look at it and you say, well, that's really nothing.\nSpeaker H: There's nothing for me there.\nSpeaker H: What's running in the cinema?\nSpeaker H: So maybe there's something better happening there.\nSpeaker H: And then you get, you know, shown what movies play which films.\nSpeaker H: And it's going to be, of course, all the highly burnt movies and what films there are actually showing.\nSpeaker H: And most of them are going to be Hollywood movies.\nSpeaker H: So American beauty is American beauty.\nSpeaker H: Right?\nSpeaker H: And...\nSpeaker F: But they're showing like on a screen.\nSpeaker F: I mean, so what the generator, like the English language sentence of is, these are the following films are being shown.\nSpeaker H: Yeah, but in that sense, it doesn't make sense to read them out loud.\nSpeaker F: Right. So it'll just display...\nSpeaker F: Okay.\nSpeaker H: I don't tell you that this is what's showing in Heidelberg. There you go.\nSpeaker H: And the presentation agent will go like that of our time.\nSpeaker H: And then you pick a movie and it shows you the times and you pick a time and you pick seeds and all of this.\nNone: Okay.\nSpeaker H: Pretty straightforward.\nSpeaker H: Okay.\nSpeaker H: But it's... So this time we are at an advantage because there was a problem for the German system to incorporate all these English movie titles.\nSpeaker H: But in English, that's not really a problem.\nSpeaker H: And as we get some topical German movies that have just come out and they're in their database.\nSpeaker H: So the person may select H\u00fchnerinnen.\nSpeaker H: Okay. Then onto the modeling.\nSpeaker H: Right?\nSpeaker H: Yeah, I guess.\nSpeaker H: The modeling.\nSpeaker H: Here it is.\nSpeaker I: Okay. Yeah, the next thing.\nSpeaker H: This is very rough but this is sort of what John O'Neil managed to come up with.\nSpeaker H: The idea here is that...\nSpeaker G: This is the schema of the XML here, not an example.\nSpeaker H: This is not an XML. This is sort of towards an schema definition.\nSpeaker H: The idea is, so imagine we have a library of schema such as the search path goal and then we have forced motion, we have cost action.\nSpeaker H: We have a whole library of schemas.\nSpeaker H: And they're going to be, you know, fleshed out in their real ugly detail.\nSpeaker H: So path goal and there's going to be a lot of stuff on the goal and blah, blah, blah.\nSpeaker H: That a goal can be and so forth.\nSpeaker H: What we think is, and all the names should be taken cum granozales.\nSpeaker H: So this is the fact that we're calling this action schema right now, should not entail that we are going to continue calling this action schema.\nSpeaker H: But what it means is we have here, first of all, and the first iteration, a stupid list of search path goal actions.\nSpeaker G: Actions that can be categorized with or that are related to source path goal.\nSpeaker H: Okay, true that schema.\nSpeaker H: And we will have, you know, forced motion and cost action actions.\nSpeaker G: And then those actions can be in multiple categories at the same time if necessary.\nSpeaker H: So a push may be in both, you know, push in this or this.\nSpeaker H: Force motion and cost action.\nSpeaker H: Exactly.\nSpeaker H: Also, these things may or may not get their own structure in the future.\nSpeaker H: So this is something that, you know, may also be a result of your work in the future.\nSpeaker H: We may find out that, you know, they're really subtle, these subtle differences between, even within the domain of entering in the light of a sort of path goal schema, we need to put in fill in additional structure up there.\nSpeaker H: But it gives us a nice handle.\nSpeaker H: So with this, we can basically, you know, slaughter the cow any way we want.\nSpeaker H: It is, it was sort of a, gave us some head out.\nSpeaker H: How do we avoid writing down that we have sort of the enter, sort of path goal, but this sort of gets the job done in that respect.\nSpeaker H: And it is even conceptually somewhat adequate in the sense that we're talking about two different things.\nSpeaker H: We're talking more on the sort of intention level up there and more on the, this is your basic bone schema down there.\nSpeaker G: One question, Robert, when you point at the screen, is it your shadow that I'm supposed to look at?\nSpeaker G: Okay, we're looking where your hand is.\nSpeaker G: Well, that wouldn't have helped you at all.\nSpeaker G: Basically, what this is is that there's an interface between what we are doing and the action planner.\nSpeaker G: And right now, the way the interface is action, go, and then they have the, what the person claimed is the source and the person claimed is the goal passed on.\nSpeaker G: And the problem is that the current system does not distinguish between, goes of type, going into, goes of type, want to go to a place where I can take a picture of, etc.\nSpeaker H: So this is sort of what it looks like. Now, some simple go action from an object named Peter's curve here of the type church, to an object named Paul Dattower of the type tower.\nSpeaker F: This is what the action planner uses.\nSpeaker F: Very currently.\nSpeaker F: And is that, and that's changeable or not?\nSpeaker F: Yeah, well, are we adapting to it?\nSpeaker H: No, this is the output of the natural language understanding, right?\nSpeaker H: The input into the action planning, as it is now.\nSpeaker H: And what we are going to do, we're going to, and you can see here, and again, for John O'Plee's focus on the shadow, we're going to, here you have the action and the domain object.\nSpeaker H: What did you think he was doing?\nSpeaker I: A laser pointer would be most appropriate here.\nSpeaker G: I remember it likes to be abstract, and that's what I just thought he was doing.\nSpeaker H: So you look up here.\nSpeaker H: Instead of between here and here, so as you can see, this is on one level.\nSpeaker H: And we are going to add another struct, if you want, i.e. our rich action description on that level.\nSpeaker H: So the future.\nSpeaker F: So it's additional information.\nSpeaker H: Exactly. The future, the content of a hypothesis will not only be an object and an action and a domain object, but an action, a domain object, and a rich action description.\nSpeaker G: Which we're abbreviating as rad.\nSpeaker I: Rad.\nSpeaker A: So you have like an action schema and a source path goal schema, right?\nSpeaker A: So how does a source path goal schema fit into the action schema?\nSpeaker A: Like, is it one of the factors?\nSpeaker C: I think I'll back to that one.\nSpeaker G: So the source path goal schema, in this case, if I understand how we describe, we set this up.\nSpeaker G: Because we've been arguing about it all week.\nSpeaker G: But we'll hold the, in this case, it will hold the features, I guess.\nSpeaker G: It's hard for me to exactly see.\nSpeaker G: So basically that will store the object that is, the source will store the object that we're going from.\nSpeaker G: The goal will store the, the full those in, full those roles in, right?\nSpeaker G: These action schemas basically have extra, so those are schemas exist because in case we need extra information, instead of just making it an attribute in which, which is just one thing, we decided to make it its own entity so that we could explode it out later on, in case there is some structure that we need to exploit.\nSpeaker F: Okay, so that, sorry, I just don't know, this is just XML notation, but the fact that it's action schema and then sort of slash action schema, that's the whole...\nSpeaker F: That's a block, yeah.\nSpeaker F: Whereas source is just an attribute, is that?\nSpeaker H: No, no, no, no, source is just that spelled out here.\nSpeaker H: Okay.\nSpeaker H: Source meaning, source will be, we'll have a name, a type, maybe a dimensionality, maybe conanical orientation.\nSpeaker H: It could also be blocked out.\nSpeaker H: Yeah, so, yeah.\nSpeaker H: So it will be, we know a lot about sources, so we'll put all of that in source.\nSpeaker H: But it's independent whether you're using the SPG schema in an interview or approach mode, right?\nSpeaker H: This is just properties of the SPG schema.\nSpeaker H: We can talk about paths being the fastest, the quickest, the nicest, and so forth, or the trajectory should be coming in there as well, and then the same about goals.\nSpeaker F: Okay, so I guess the question is when you actually fill one of these out, it'll be under action schema.\nSpeaker F: You'll pick one of those for what?\nSpeaker F: Okay, this is just a layout of the possible things.\nSpeaker G: So the roles will be filled in with the schema, and then what actual action is chosen will be in the action schema section.\nSpeaker F: So one question, in this case it's all clear, sort of obvious, but you can think of the interview and approach as each having their roles, right?\nSpeaker F: I mean, it's implicit that the person that's moving is doing, entering, viewing, and approaching.\nSpeaker F: But the usual thing is we have bindings between sort of like action specific roles and the more general source path goal specific roles.\nSpeaker F: So are we worrying about that or not?\nSpeaker H: Yes, yes.\nSpeaker H: Since you bring it up now, we will worry about it.\nSpeaker H: We're about it.\nSpeaker F: Oh, I guess I may be just reading this and interpreting it into my head in the way that I've always viewed things, and that may or may not be what you guys intended, but if it is, then the top block is sort of like, you know, you have to list exactly what X schema or it's action schema, there'll be a certain one that has its own structure, and maybe it has stuff about that specific to entering, or viewing, or approaching.\nSpeaker F: But those could include roles like the thing that you're viewing, the thing that you're entering, the thing that you're,\nSpeaker I: or very specific roles, viewed thing, enter a thing.\nSpeaker F: So you can enter view and approach as frames, and they have frame specific parameters and roles. And you can also describe them in a general way as source path goals schema, and maybe there's other image schemas that you could add after this, that, you know, how do they work in terms of, you know, a force dynamics, or how do they work in terms of other things.\nSpeaker F: So all of those have basically either specific frame specific roles or more general frame specific roles that might have bindings.\nSpeaker F: The question is how to represent when things are linked in a certain way.\nSpeaker F: So we know for enter that there's container potentially involved, and it's not, I don't know if you want to have in the same level as the action schema and the SPG schema.\nSpeaker F: It's somewhere in there that you need to represent that there is some container, and the interior of it corresponds to some part of the source path goal, you know, goal, goal, like us in this case.\nSpeaker F: So is there an easy way in this notation to show when there's identity basically between things?\nSpeaker F: And I don't know if that's something we need to invent or, you know.\nSpeaker G: But wasn't there supposed to be a link in the, I don't know if this answers the question, I was just staring at this while you were talking. Sorry.\nSpeaker G: Okay.\nSpeaker G: A link between the action schema, a field in the schema for the image schemas that would link us to which action schema we were supposed to use.\nSpeaker H: Yeah.\nSpeaker H: Well, that's one thing is that we can link up, I think also that we can have one or as many as we want links from the schema up to the action description of it.\nSpeaker H: But the notion I got from Nancy's idea was that we may find sort of concepts flowing around in the action description of the action, the action for the entire frame up there that are when you talk about the real world, actually identical to the goal of the, so the goal schema.\nSpeaker H: And do we have means of telling it within that and the answer is absolutely.\nSpeaker H: Yeah.\nSpeaker H: We absolutely have those means there even part of the M3L API meaning we can reference.\nSpeaker H: Great. That's exactly what it is.\nSpeaker H: And this referencing thing however is of temporary nature because sooner or later the W3C will be finished with their X path.\nSpeaker H: Specification and then it's going to be even much nicer than we have real means of pointing at an individual insensation of one of our elements here and link it to another one.\nSpeaker H: And this not only within a document but also via documents and and all in a very easy, which you know, happened to know how what what sooner or later means like practice.\nSpeaker H: Let's look at it on the back.\nSpeaker H: So it's the spec is there and it's going to be part of the M3L API by the end of this year.\nSpeaker H: So this means we can start using it basically now.\nSpeaker H: But this is a technical detail.\nSpeaker G: So it's going to wait a few minutes.\nSpeaker G: Basically references from the roles in this game at the bottom team is to the actions game is.\nSpeaker H: Okay.\nSpeaker H: I'm assuming.\nSpeaker H: I mean personally I'm looking even more forward to the day where we're going to have X forms which is a form of notation where it allows you to say that if the SPG action up there is enter.\nSpeaker H: Then the goal type can never be a statue.\nSpeaker F: So you have constraints that are dependent on the actual specific filler.\nSpeaker H: Exactly.\nSpeaker H: You know this of course does not make sense in light of the set of liberty or ever.\nSpeaker H: It is sort of these sort of things are imaginable.\nSpeaker F: Or the gateway arch and statement.\nSpeaker A: So like are you going to have similar schemas for FM like force motion and caused actions of like you have for SPG.\nSpeaker A: Kind of so like can are you able to enforce that you know if it's if it's a SPG action then you have that schema if it's a force motion that you have the other schema.\nSpeaker H: We have absolute no we have absolutely no means of enforcing that.\nSpeaker H: So it would be considered valid if we have an SPG action entry and no SPG schema.\nSpeaker F: But a force action schema could happen which is not bad because I mean that this multiple.\nSpeaker F: I mean that particular case there's more there's a force side of.\nSpeaker H: Of that.\nSpeaker H: It may be means we had nothing to say about the sort that goal.\nSpeaker H: What is also nice and for me in my mind it's it's crucially necessary is that we can have multiple schemas and multiple action schemas in parallel.\nSpeaker H: And we started thinking about going through our bakery questions so when I say is there a bakery here you know I do ultimately want our module to be able to first of all tell.\nSpeaker H: The rest of the system hey this person actually wants to go there and be that person actually wants to buy something to eat there.\nSpeaker H: And if these are two different schemas I eat a so called schema of getting there and then the buying snacks schema.\nSpeaker F: Would they both be listed here and under so our under action schema there's a list that can include both.\nSpeaker H: Right they would both schemas would appear so what is the is there buying a snack schema what is the.\nSpeaker H: The bikes next schema I'm sure there's a commercial event scheme commercial event or something.\nSpeaker H: So we would we would instantiate the SPG schema with a sort of goal blah blah blah and the buying event.\nSpeaker H: So you know at which however that looks like the place.\nSpeaker F: Would you say that the like I mean could have flat structure and just say these are two independent things but there's also this sort of like causal well so one is really facilitating the other and it's part of a compound action.\nSpeaker F: There's some kind of structure.\nSpeaker H: So it's technically possible that you can fit schema within schema and schema nicer for a lot of reasons but might be a pain so well for me it seems that.\nSpeaker F: I mean there are truly times when you have two totally independent goals that they might express at once yes in this case it's really like there's a purpose means you know for achieving some other purpose well if I'm\nSpeaker H: the recipient of such a message and I get a sort of goal where the goal is a bakery and then I get a commercial action which takes place in a bakery right and and they are obviously we are the first identify to be the same thing here yeah see that that bothers me that they're the same thing no just yeah yeah because there are two different things one of\nSpeaker F: which is you could think of one is a sub you know whatever the condition for the second yeah yeah yeah so so okay so there's like levels of granularity so there's there's a single event of which they are both apart and then they're independently they are events which have very different characters as far as source pathical whatever so when you identify source path goal and whatever there's going to be a desire or whatever eating hungry whatever other other frames you have involved they have to match up in in nice ways so it seems like each of them has its own internal structure and mapping to these schemas you know from the\nSpeaker H: but you know that's just that's just me well I think we're going to hit a lot of interesting problems and as I preface that this is the result of one week of arguing about it\nSpeaker I: yeah and and so I still am not entirely sure that I really fully grasp this in tax of this you know it's not it's not actually very\nSpeaker H: actually that's actually that I mean the intended interpretation of this well I should have we should have added an an XML example or some XML examples yeah that would be and and this is on on on on my list of things until next next week it's also a question of the recursiveness and and a hierarchy in there do we want the schemas just plump plump plump plump I mean it's if we can actually you know get it so that we can out of one utterance activate more than one schema I mean then we're already pretty good right well you have to be careful with that I think because I mean many actions\nSpeaker B: suppose some almost infinitely many other actions so if you go to your bakery you have a general intention of not being hungry there was specific intention to cross the traffic light yeah you have a further specific intentions to lift your so I mean you really have to focus on on on decide the level of of abstraction that you am at the kind of zero in on that and more or less ignore the rest unless there is some implications that you want to come to draw from from\nSpeaker F: sub-tasks that are relevant I mean but what are the other thing that I just thought you could want to go to the bakery because you're supposed to meet your friend there or some you so you like being able to infer the second thing is very useful and probably often right well the the utterance is our bakery around here not well I want to go to a bakery meet them in a bakery around the area yeah I'm okay but it's still the case that you could you could override that default giving extra information which is to me a reason why you would keep the inference of that separate from the knowledge of okay they really want to know if\nSpeaker H: it's a bakery around here which is direct well there there should never be a hard-coded shortcut from the bakery question to the double schema thing how and as a matter of fact when I travel with my friends we make these exactly these kinds of appointments\nSpeaker F: yeah exactly that's one of the bakery you know in the Victoria station train train station London before I have a question about this lot of the SPG action\nSpeaker B: so the interview approach the ever of fixed slots in this particular action every action of this kind would have a choice or it would just every SPG action either is an enter or a view or a product right right so I mean for each particular action that you may want to characterize you would have some number of slots that define you know in some way what this action is all about can be either a B or C so is it a fixed number or do you live it open it could be between one and 15 it's flexible\nSpeaker H: the well it sort of depends on if you actually write down the schema then you have to say it's either one of them or it can be none or it can be any of them however the it seems to be sensible to me to to view them as mutually exclusive maybe even not\nSpeaker B: do you mean within the sort of action yeah and how where's the end so there's no no actually but I think my question is simple that is is okay so you have an SPG action and it has three different aspects because you can either enter a building or view it or approach it now you define another action for section course motion SPG one action that has to do with writing a letter I mean not even within this context but a different action and this action to have various variable possibilities of interpreting what you would like to do in a way similar to either enter view approach you may want to send a letter or big take a letter let's say okay maybe the these actions I don't know if I'm going to answer your question or not\nSpeaker G: with this but the categories inside of action schema so SPG action is a category I think what we're specifying here is this is a category where the actions enter view and approach fall into because they have a related source path goal schema in our tourist domain because viewing in a tourist domain is going up to it and we're actually going from one place to another to take a picture in this\nSpeaker B: sort of automatic derived from the structure that is built elsewhere\nSpeaker I: derived I don't know if this is a category structure here right action schema what are some types of actions well one of the types of action schemas is source path goal action and what are some types of that and enter view and approach\nSpeaker G: those are all source path goal inside of enter there will be roles that can be filled basically so if I want to go from outside to inside then you have the roles that need to be able to have a source path goal set of roles so you have the source to be outside and path to the door whatever right so if you wanted to have a new type of action you create a new type of category then this category would be put it or not necessarily we would put a new action in the in the categories that in which it has the well every action has a set of related schemas like source path goal or force whatever right so we put right a letter in the categories that in which it had it would have there could be a scheme of action or exactly schemas that type and then later you know that we have a communication event action where we define it down there as\nSpeaker F: so there's a bit of redundancy right in which the things that go into a particular you have categories at the top under action schema and the things that go into particular category are supposed to have a corresponding schema definition for that type so I guess what's the function of having it up there too I mean I guess I'm wondering whether you could just have at an action schema you could just sort of say whatever you know it's going to be interview or approach or whatever number things and possibly because you need to know somewhere that those things fall into some categories and it may be multiple categories you say which is the reason why it gets a little messy but if it has if it's supposed to be categorized in category X then the corresponding schema X will be among the structures and follow\nSpeaker H: this is one of the things we are okay this is for this is probably the way that that's the way they see more to to Toronto I guess also for a while but now you guys have seen the light\nSpeaker G: no no no no we have not seen the light because the reason why we're doing it this way is in case there's extra structure that's in the enter action that's not captured by these\nSpeaker F: schemas right why I would think you would say enter and to say all the things that are relevant specifically enter and then the things that are abstract will be in the abstract things as well and that's why the bindings become useful right but like so you're saying you can practically turn the structure inside out or something no basically I see which you mean by that but I don't know if I would I would\nSpeaker H: get rid of the sort of sbg slash something right or these sub actions category because what does it tell us yeah and I agree that you know in fact\nSpeaker F: what you could say is for enter you could say here list all the kinds of schemas that on the category that list all the parent categories is just like frame hierarchy right you have he's blinded so you say enter and you say my parent frames are such a such and then those are the ones that actually you then actually define and say how the roles bind to your specific roles which will probably be richer and fuller it sounds like a paper I've read around here yeah like I said I'm sure I'm just hitting everything with the hammer that I developed I mean you know it's I'm just telling you\nSpeaker I: what I think I'm just hitting the button but there's a good question here like I mean do you when do you need this when you need this I just don't see why you would I mean this who uses this this data structure you know like do you say all right I'm going to do an SPG action and then you know somebody neither the computer or the user says all right well I know I want to do a source path goal action so what are my choices among that and oh okay so I can do an interview approach it's not like that right it's more like you say I want to well I want to enter and then you're more interested in knowing what the parent categories are of that right so that the sort of representation that you were just talking about seems more relevant\nSpeaker G: I think I'm not sure if I understand your question only one of those things are going to be lit up when you pass this on okay so only enter will be if we if our\nSpeaker H: modules have the enters the case view and approach will not be there okay well it's it sort of came into my mind that sometimes even two could be on\nSpeaker G: or would be interesting um nevertheless well in that case we can we can't we can't if it goes to the thing is that the case we are don't think our system can handle that currently no not at all but\nSpeaker H: so the I think the in some sense we got the task done extremely well because this is exactly the discussion we need need period no more qualifiers and this is useful yeah and and and I hope let's make a sharper claim we will not end this discussion anytime soon yeah they could guarantee you that and it's going to get more and more complex the complex that are not our domain sketch and and I think we will have all of our points in writing pretty soon so this is nice about being being recorded also the\nSpeaker G: the in terms of why is it laid out like this versus some other yeah um that's kind of a contentious point between the two of us but this is one what so this is a way to link uh the way these rules are filled out to the action because if we know that enter is a as an SPG action right we know to look for an SPG schema and put the\nSpeaker F: appropriate fill in the appropriate roles later on and you could have also indicated that by saying enter what are the kinds of action right yeah right to just like sort of reverse organization I can let's I mean we are the reasons why one is better than the other\nSpeaker H: I mean again from other sources yes because no but no the modules don't this is this is a schema that defines XML messages that are passed from one module to another yeah mainly meaning from the natural language understanding or the deep natural language understanding to the action planner now the the reason for for not using this approach is because you always will have to go back each module will try to have to go back to look up which uh you know entity can have which parents and then so you always need the whole body of of your your model um to figure out what belongs to what or you always send it along with it mm-hmm so you always send up here I'm I'm this person and I can have these parents in every message okay so it's just like a pain to have to send it it may or may not be just a pain it's I'm completely okay willing to to throw all of this away and completely redo it you know and and and after some iterations we\nSpeaker I: may just do that I would just like to ask like if it could happen the next time I mean just because because I'm new and I don't really just I just don't know what to make of this and what this is for and stuff like that you know if there if someone could make an example yeah of what would actually be in it like first of all what modules are talking to each other using this right and I will promise\nSpeaker H: for the next time to have fleshed out and XML examples for a run through and see how this this then translates and how this can come about including the sort of miracle occurs here right part and is there more to be said I think in principle what I think that this approach does and whether or not we take the interim view and we all throw up up the letter um well how do how does professor Peter call that the uh silent sublimination throwing somebody up the stairs have you never read the\nSpeaker A: Peters principle no oh uh people reach their level if uh Max their level if maximum incompetence\nSpeaker H: yeah then you throw them up the stairs um promotion yeah okay so we can promote the interview in all all upper bit and get rid of the uh blah blah x-plub uh asterisk subaction item all together no no problem with that and we we'll play around with all of them but the principle distinction between having the the pure schema and their insunciations on the one hand and adding some whatever more intention oriented specification um parallel to that this approach seems to be uh workable to me I don't know in this view I'll share that opinion then\nSpeaker G: that made my day much happier this is a simple way to basically link uh rolls to actions\nSpeaker F: yeah yeah that's fine that's that that was the intent of it basically that's true although um\nSpeaker G: rolls so I don't I don't I don't I don't happen when he uses the word yeah I think I made\nSpeaker I: the ROLs so I'm in pace for you Sam yeah this is a big example got it all right\nSpeaker H: help I'll agree to that then okay that's all I have oh no there's one more issue basket I brought that one up meeting time rescheduling and did you say something about Friday or yeah so it looks like you have not been particularly the Monday three o'clock time has turned out to be not good anymore so people have been thinking about an alternative time and the one we came up with is Friday to 33 what was it you have class until two right so if we don't want to if we don't want him to run over here to 30 years or 30 years or 30 or something around that time um how how are you uh Friday uh and I know that you have until three you're busy two three sounds good\nSpeaker I: yeah I'll be I could do that yeah I mean earlier on Friday is better but three you know I mean if it were a three or three thirty time then I would take the three or whatever but yeah sure\nSpeaker H: three is fine yeah and you can always make it shortly after three probably yeah and I don't need to be here particularly off me no but uh whenever you are more than welcome if you think that this kind of gets you anywhere in your life then you're free I'm just glad that I don't have to work it out\nSpeaker D: because I'm just glad that I don't have to work it out myself I'm not involved at all and they're working out of it but you're a linguist but yeah that's why I'm glad that I'm not\nSpeaker H: in the next week so Friday is free and um so already again this week how diligent do we feel yeah do we feel that we've done our chores for this week or so I mean clearly that is I can talk about\nSpeaker A: the uh by searching changes on Friday okay bus car will do the big show on Friday and you guys\nSpeaker G: will argue some more yeah and bring now and then yeah promise we will don't worry we'll get the\nSpeaker F: summary like the short version and I would like to second kiss request to an example would be nice\nSpeaker H: yes yes I guess I'm on the record for promising that so how have it in writing\nSpeaker G: better speech is it and the other good thing about it is Jerry can be on here on Friday and he can\nSpeaker H: weigh in as well yeah and um if you can get that binding point also maybe with a nice exception that would be helpful for John oh and yeah give us that one on hand yeah I have\nSpeaker D: several in my head yeah always thinking about binding well the the binding is technically no problem\nSpeaker H: but it's for me it seems to be conceptually important that we find out if we can if there if there are things in there that are sort of of a general nature we should just still them out and put them where the schemas are if there are things that you know our intention specific then we should put them up somewhere so in general they'll be bindings across both\nSpeaker F: intentions and the actions so that's wonderful yeah so it's general it's general across all of these things it's like I mean I think Shostra would say you know finding is like your essential process so okay so I don't think it'll be isolated to one or the two but you can definitely figure out where yeah sometimes things belong so actually I'm not sure I would be curious to see how separate the intention part and the action part are in the system like I know the whole thing is like intention ladders or something like that right so is the right right now the idea is the rich rich the rad whatever is one you know potential block inside intention it's still it's still mainly intention hypothesis and then that's just one way to describe the action\nSpeaker G: part of it okay it's in a temporary find it basically and yeah it's in it it's sort of not\nSpeaker F: just that you want to go from here to here it's that yeah the action is what you intend in this action contistas all these complicated modules and image schemas and whatever yeah and there will be\nSpeaker H: a relatively high level of redundancy in the sense that ultimately one so if we want to get really cocky we we will say well if you really look at it you just need our rad you can throw the rest away right because you're not going to get any more information out of the action as you find it there and the domain object but then again in this case the domain object may contain information that we don't really care about either so but we'll see that how it sort of evolves I mean if if people really like or or rad I mean what might happen is that they will get rid of that action thing completely you know and leave it up for us to get the parser input we know the\nSpeaker F: things that make use of this thing so that we can just change them so that they make use of rad\nSpeaker E: okay okay\nSpeaker G: rads a great term it's rad then why it happened like well it just has to be the acronym\nSpeaker F: yeah that doesn't make it a great term it's just like those jokes\nSpeaker H: just leave it as work as we know it's just easy to see it but if you work in that XML community it is a great acronym because it evokes whatever RDF is the biggest thing right that's the rich resource description framework and and also so description having the word this term description and there's wonderful rich is also great everybody likes action plus the kids alike it\nSpeaker F: it's rad too and tensions will be rid okay are the sample data that you guys show some time ago like the things maybe maybe you're gonna run a trial tomorrow I mean I'm just wondering whether they act some of the actual sentences from this domain will be available because be nice for me to like look if I'm thinking about examples I'm mostly looking at child language which you know we'll have somewhere we'll laugh at not total with the kinds of things that you guys are getting so you showed some in this here before and maybe you posted before but where would I look if I\nSpeaker H: want to see oh you want audio or do you want transcript well just transcript is just not a way to because nobody has transcribed it yet okay I can eat I can I transcribe it though okay well\nSpeaker F: don't don't bake it a high priority yet in fact if you just tell me like you know two examples that the representational problems are I'm sure we'll be there okay like enough for me to think about\nSpeaker H: okay so Friday whoever wants and comes can just Friday it's a big pass or show so you can all turn off your\n"}, {"meeting_id": "ES2011b", "summary": "The meeting started by discussing the functional design of the remote control. Marketing gave the study on user needs and target groups. User Interface gave the presentation on technical effect design and Industrial Designer gave the presentation on working design. Then Project Manager initiated new project requirements brainstorming. The group had agreed on the on-call button design, DVD capability button design, power button design, and denied speech recognition design as well as mouse pad design. They would further discuss the function switch button and its compatibility in the next meeting.", "dialogue": "Speaker A: Okay, so we'll try to zip through this since we're short on time.\nSpeaker A: Welcome everybody.\nSpeaker A: Hope your sessions went well.\nSpeaker A: So this is our functional design meeting.\nSpeaker A: We're going to consider user needs, technical effects, and the working design of our remote control.\nSpeaker A: I've been taking meetings on the minutes, and I'll be putting them in the shared documents folder so if there's anything you need to refer to, you can find them in there.\nSpeaker A: I'll get the ones up next time.\nSpeaker A: They're not finished yet.\nSpeaker A: So can we have updates from everyone from what you've worked on?\nSpeaker A: Just kind of a quick summary of anything interesting that you'd like to share or discuss\nSpeaker C: in the list. Is that an order?\nSpeaker C: No, I decided on an order.\nSpeaker A: No, any order is fine.\nSpeaker C: How do I put this?\nSpeaker C: I'll just put the cable in.\nSpeaker A: Oh yeah, sorry.\nNone: Is that it?\nSpeaker C: Can you see?\nSpeaker C: It's all up.\nSpeaker B: It takes a few seconds I think.\nSpeaker B: You may need to.\nSpeaker D: Who's that?\nSpeaker D: Look.\nSpeaker B: It's like a three-setting cycle, so press it a couple of times.\nSpeaker C: Hold down function and then press function.\nSpeaker A: Yeah.\nNone: And then F8.\nSpeaker D: Could you just plug it back into hers?\nSpeaker C: Yeah.\nSpeaker B: Okay.\nSpeaker C: In order to see what the functional requirements were to be found, 100 people were tested in usability laboratory through just their habits.\nSpeaker C: They're habits were observed and questionnaires were given out.\nSpeaker C: The findings in some cases matched what we were talking about.\nSpeaker C: Customers and users don't like the way remote controls look.\nSpeaker C: They don't like the way they feel.\nSpeaker C: They don't think they match their operating behavior.\nSpeaker C: And an example is what we were talking about, the buttons.\nSpeaker C: They only use 10% of the buttons.\nSpeaker C: And later on there's a study of the buttons that they use most.\nSpeaker C: And I think we should design according to these buttons.\nSpeaker C: Easy to lose and RSI.\nSpeaker C: I don't know what RSI means.\nSpeaker C: The person that did the questionnaire, obviously did.\nSpeaker C: I don't know.\nSpeaker C: According to the frequency of use and the relevance of each buttons, I have made a list of the buttons that we should focus on in order of importance.\nSpeaker C: The most important buttons in those do with channel selection.\nSpeaker C: They're used about 160 times per hour.\nSpeaker C: And people like to zap a lot, apparently.\nSpeaker C: So this is the order, translation, teletext, volume and power.\nSpeaker C: The other ones are the settings and they're used less than 0.8 to 0.5 times per hour.\nSpeaker C: This means that I think we could have a button for all the settings.\nSpeaker C: And then just one and then from there go on to the audio on the screen, either on the remote or on the television.\nSpeaker C: About the screen and speech recognition, some people are more willing than others to actually pay for that.\nSpeaker C: And if we look at the market, people from 15 to 35 year olds, I don't really know how to describe this.\nSpeaker C: So that's the number of people that are above 35 years old.\nSpeaker C: Go from 35% to 8%.\nSpeaker C: So people that are 65, for example, wouldn't actually pay for this sort of thing.\nSpeaker C: I don't know what the decision to be made is, but I think that the people that actually do buy remotes more are those like teenagers and young professionals and most likely, but we should discuss this together.\nSpeaker C: And that's what I have to say about the matter.\nSpeaker C: Shall I? What do I do? Do I give this to someone else?\nSpeaker C: Yeah, just move right on.\nSpeaker D: Let's do this.\nSpeaker D: Okay, so now I need to press F8. What is it?\nSpeaker D: A function of it.\nSpeaker D: What's function?\nSpeaker D: It's the little blue.\nSpeaker A: It's the function.\nSpeaker A: That should be cool.\nSpeaker D: Okay.\nSpeaker D: This is my presentation about the technical functions design.\nSpeaker D: I basically just tried to focus on what functions we need and how to make that the best function for the user.\nSpeaker D: I thought about it myself and looked on the internet.\nSpeaker D: I think it is really important that we sort of get this done in a user-friendly and fashionable way.\nSpeaker D: Keeping buttons together, close together, that are used in the same way or maybe making the same color, keeping the number of buttons to a minimum.\nSpeaker D: Can you, small enough, large enough, I'm not sure.\nSpeaker D: I guess we would need to do some research about what size is appropriate and that sort of thing.\nSpeaker D: But basically we need to make sure that it turns on and off the TV.\nSpeaker D: Does it have capacity to change the channels?\nSpeaker D: Do we need to have functions for cable or VCR and then is it findable and how do we want to do that?\nSpeaker D: I just thought that these two remotes were pretty boring.\nSpeaker D: I think we can find something that is more fun to look at and use than either of those.\nSpeaker D: I like the one on the right better just because it does have fewer buttons.\nSpeaker D: I think we can sort of think about things like color and size shapes and that sort of thing to best fit the user.\nSpeaker D: That concludes my presentation.\nSpeaker A: Thank you.\nSpeaker B: This is on the working design, which is one of the mechanical functions of the remote.\nSpeaker B: The method I used was to basically look at existing designs and incorporate ideas from our last meeting.\nSpeaker B: I think we need two basic functions, which is the basic remote functions.\nSpeaker B: User can input a channel or a volume setting and it would be output to the TV.\nSpeaker B: Also we talked about a location function where maybe you could press a button on the TV and it would send a signal to the remote where it would be for flash or vibrate or whatever to tell you where the remote is.\nSpeaker B: The components we need are an energy source to power through a node.\nSpeaker B: Processor to take the information.\nSpeaker B: Something to transmit it to the TV.\nSpeaker B: We also need something on the remote that would receive the location signal and have an output, like possibly a beep or a vibration.\nSpeaker B: You need a sender for location signal, which would probably be a separate thing that we would have to sell with the remote and if you could stick it on the TV or stick it on the wall.\nSpeaker B: This is an overview of how the remote works.\nSpeaker B: Power comes from the battery, it goes to the chip, and then it is sent from there to an infrared bulb, which would probably be the easiest way to send to the TV.\nSpeaker B: Then the location function would have a sender on the TV, which would output some sort of signal.\nSpeaker B: We could use IR, but we probably want to use radio instead.\nSpeaker B: That signal would go to a receiver, it would process it, and it would be output in the form of a buzzer or a light lighting up.\nSpeaker B: My personal preference is for how to build the remote would probably be a battery for the energy source, that way you wouldn't have to plug it in.\nSpeaker B: A button pad, input, we can purchase a pre-made chip, which will handle all the processing stuff.\nSpeaker B: IR transmitter to communicate to the TV, that is standard.\nSpeaker B: Most TVs have an IR receiver.\nSpeaker B: It is a radio receiver to send out a location function and to receive it.\nSpeaker B: That includes my presentation.\nSpeaker D: Do you know about, I don't know, do you seem like you know about?\nSpeaker B: Yeah, I was an engineer.\nSpeaker B: Okay.\nSpeaker A: Cool.\nSpeaker A: So, we have, we'll discuss that in a minute, I just want to mention some new project requirements that came in.\nSpeaker A: Teletext is apparently outdated, so due to internet popularity, so that's off the list.\nSpeaker A: Also, our remote should be used only for television.\nSpeaker A: No, extra internet, kind of fancy things, just throw out in the television.\nSpeaker A: And also, we need to incorporate our corporate image onto this.\nSpeaker A: So, the phase is we put fashion in electronics, so let's be fashionable with the bullet, I guess.\nSpeaker A: If we have something, I mean silver and gold or yellow are colors, so if we have like a kind of silver on like you saw.\nSpeaker A: And yellow writing, so now watch that.\nSpeaker A: Okay.\nSpeaker A: So, we need to make some decisions on the remote control functions.\nSpeaker A: Yeah.\nSpeaker A: Do, I guess we should, yeah, make some kind of brainstorming.\nSpeaker D: Like in terms of how it looks or like what it does?\nSpeaker A: What, well, probably our target group and how it's going to appeal to our target group in, I don't know, the buttons and what it does.\nSpeaker A: Yeah.\nSpeaker A: That sort of thing.\nSpeaker D: So, it's our target group, then people, so do you want to go ahead and design this thing with the finder button?\nSpeaker A: So, I think that seems to, yeah.\nSpeaker C: Okay.\nSpeaker C: So, yeah, yeah.\nSpeaker D: So then our target age group would be the 19 to 35 range.\nSpeaker A: Yeah, teenagers and young professionals.\nSpeaker C: Well, that's the speech recognition.\nSpeaker C: Oh, well, that's the screen.\nSpeaker B: I was thinking about that, but, I mean, speech recognition is really hard to program.\nSpeaker B: And also, if the TV is on, it's making sound and the people on the TV are talking, you know, somebody says like one and the TV is going to switch itself to channel one.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: It seems like a silly.\nSpeaker B: Yeah.\nSpeaker B: I'm not sure how you would implement it.\nSpeaker C: I just put the button.\nSpeaker A: I mean, if you consider a budget, if you consider a budget to have speech recognition programmed in every single remote might be a little crazy.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, okay.\nSpeaker D: So, are we going to have just some kind of a, like, we'll have the buzzer on the, like, on the TV itself?\nSpeaker B: Well, you would have to have a button on a TV or on your wall or someplace, but it's a TV already has power.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: You click the button.\nSpeaker B: It's going to send out a signal.\nSpeaker B: You know, I was thinking IR is line of sight.\nSpeaker B: So, unless the remote is like, actually in front of the TV, it's not going to work.\nSpeaker B: So probably like a radio signal.\nSpeaker B: Okay.\nSpeaker B: So, you'd need like a separate base for that or like something.\nSpeaker B: It would have to be sold separately because if the sender is on the remote, then you'd have to find the remote first to click the button.\nSpeaker B: So, do you plug it in the TV?\nSpeaker C: Yeah, probably just take it on your TV.\nSpeaker B: Okay.\nSpeaker D: So it's now like a two part.\nSpeaker D: Yeah.\nSpeaker B: Okay.\nSpeaker D: So we get to design that too.\nSpeaker D: Thank you.\nSpeaker D: Fashionable.\nSpeaker D: Okay.\nSpeaker A: So do you think even though we're not talking about speech recognition, our target group should still be teenagers and young professionals?\nSpeaker C: Yeah.\nSpeaker C: Just, yeah.\nSpeaker C: Yeah.\nSpeaker C: We should that thing be on the thing to put the, you're talking about the home for it.\nSpeaker C: Do you still want to build a little thing next to the telly or to pan on the wall or shall we leave that for now?\nSpeaker A: We probably leave that.\nSpeaker A: I mean, I guess one takes care of the other.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: If you can, yeah, if you can call it, then it's.\nSpeaker A: Then it can live anywhere.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: So you have that button that so this is our light or shall we leave just have a radio.\nSpeaker C: Oh, we just have a TV on the phone on the phone.\nSpeaker C: Um, seems like a need to lie.\nSpeaker B: I think that's what the phone is.\nSpeaker B: Yeah.\nSpeaker B: We only need to find your phone.\nSpeaker B: You just have to call it and it starts ringing somewhere and then you can figure out.\nSpeaker A: And like if the phone's under the couch, you're right.\nSpeaker A: That's the light.\nSpeaker A: Yeah.\nSpeaker C: Um, so I need the other buttons.\nSpeaker C: So you have this.\nSpeaker A: So I mean, the two remotes that you had shown.\nSpeaker A: I don't remember who shown me.\nSpeaker A: Yeah, you need to.\nSpeaker A: Um, there.\nSpeaker A: I mean, one look like it was for VCR type thing and the other looked like just television.\nSpeaker D: I think they're both sort of just like general.\nSpeaker D: They're both general.\nSpeaker D: They're both general.\nSpeaker A: That is something we have to decide is whether we want to have VCR capabilities.\nSpeaker A: Yeah.\nSpeaker B: Does anyone know VCRs are the same across international?\nSpeaker C: They're not.\nSpeaker C: They're not.\nSpeaker B: Okay.\nSpeaker B: So you need like a whole difference.\nSpeaker A: Yeah.\nSpeaker A: It's not VCRs.\nSpeaker A: But DVD probably is.\nSpeaker B: Yeah, other than that region including thing.\nSpeaker B: Um, but VCRs.\nSpeaker A: And if we're targeting young professionals and teenagers, I mean, it's going to be DVD type.\nSpeaker D: That's the technology these days.\nSpeaker D: So, okay.\nSpeaker D: I think still it shouldn't be that hard to type like just reduce the number of buttons, you know?\nSpeaker D: You just have like one menu button that works like with, you know, where you can just kind of scroll through the options.\nSpeaker C: Well, for sure we need the, I think we can just design the channels.\nSpeaker C: I mean, power is just about.\nSpeaker C: Yeah.\nSpeaker C: And it's not used that much.\nSpeaker C: So it's usually that red.\nSpeaker C: Yeah.\nSpeaker C: And I think it's quite nice to keep it like.\nSpeaker A: You know, I've seen some remotes that were you just hold one, like if you hold one down, it's a different color than the other buttons, but that turns it on.\nSpeaker A: So you don't actually have a separate power button.\nSpeaker A: Oh, okay.\nSpeaker D: It seems like that would be hard though.\nSpeaker D: I mean, because unless you know.\nSpeaker D: Am I pecan confusing?\nSpeaker D: Yeah.\nSpeaker D: Just because I wouldn't, I would probably pick it up and just be like, uh, why is there no\nSpeaker A: audience? Besides you like to be able to go power.\nSpeaker A: Yeah.\nSpeaker A: I have the power.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So we definitely want a power button and numbers.\nSpeaker C: Yeah.\nSpeaker C: Or even, um, I don't know if people like this, but if you want to reduce the number of buttons, instead of having like one to nine, have a sort of, that sort of like joystick flat.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Because people like seem, now the eye puts up.\nSpeaker C: And I like this thing that there's no, no, you don't have one to.\nSpeaker D: Yeah.\nSpeaker D: I think that's an interesting idea because it's cool.\nSpeaker D: It's, it's funny.\nSpeaker D: Like you've, like, I just, I don't have an iPod, but like, you know, I just like started messing around with one of my friends the other day and you just sort of, and it's funny how you pick it up and you just figure out how to use it quite easily.\nSpeaker D: Like it's not that hard.\nSpeaker D: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It's just, and it's one thing.\nSpeaker C: Yeah.\nSpeaker D: And it is.\nSpeaker D: Yeah.\nSpeaker D: It is really, but do you need a screen then?\nSpeaker C: Do you have to have a screen then?\nSpeaker C: Well, can't it tell the, like, can't you?\nSpeaker C: Yeah.\nSpeaker C: If you have the number, you can have the number.\nSpeaker B: Okay.\nSpeaker B: The telly go.\nSpeaker C: Yeah.\nSpeaker C: One, two, three, four, five.\nSpeaker D: Once you scroll.\nSpeaker D: Okay.\nSpeaker D: So we have this like scrolling sort of button.\nSpeaker A: Is that like an on a mouse pad where you're kind of, yeah.\nSpeaker D: It's like, it's just like the same.\nSpeaker C: It's like, analogy as a mouse pad.\nSpeaker C: Like that.\nSpeaker C: And then you do that.\nSpeaker C: Okay.\nSpeaker C: And then you can have, if you actually just want to zap, you can have like a thing.\nSpeaker C: Like that and that and then it can just be plus and minus.\nSpeaker D: Okay.\nSpeaker D: So like it's like a little part of the circle that.\nSpeaker D: Or yeah.\nSpeaker D: Oh, so it's just a region of the circle that you can.\nSpeaker D: Yeah.\nSpeaker C: Click.\nSpeaker A: Okay.\nSpeaker A: We could even have four buttons.\nSpeaker A: Like if that's the, if that's the mouse, you could have the volume.\nSpeaker A: Yeah.\nSpeaker A: The channel changes.\nSpeaker A: Yeah.\nSpeaker D: So volume could be like the top and the bottom.\nSpeaker D: So do you need to.\nSpeaker B: Okay.\nSpeaker B: So it'll be.\nSpeaker D: Well, you have to like be able to change the function of it to like.\nSpeaker C: What do you mean the function?\nSpeaker D: I mean, like, okay, cause so.\nNone: I don't know.\nSpeaker D: I guess.\nSpeaker D: Okay.\nSpeaker D: So when you scroll your thumb like around it, it'll like, let's say you're going, you're going to clockwise that.\nSpeaker D: That means you're going to go up the channels and then scroll the other way and it'll go down.\nSpeaker D: Yeah.\nSpeaker D: But then so if you want to switch to the, do you have to switch to a function where.\nSpeaker D: Like, so either in that mode or you're in the mode where like.\nSpeaker D: It just has like the four, like, you know, this is channel that way.\nSpeaker D: That's that way and volume is up and down.\nSpeaker C: Yeah, but it knows for some reason.\nSpeaker C: It just, the iPod knows.\nSpeaker C: It just knows.\nSpeaker C: The iPod knows.\nSpeaker B: If it works on an iPod, then it works.\nNone: Yeah.\nSpeaker D: So you just, you just can either do this or like you can just touch it if you want.\nSpeaker C: Well, for the volume, you have to press the middle.\nSpeaker C: Okay.\nSpeaker D: That's what I mean.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: So you have to like press this middle region and then you can scroll up.\nSpeaker D: Go up and then it's like holding.\nSpeaker C: So it's like holding.\nSpeaker C: But if you like that makes more sense because there's already ones with up and down here.\nSpeaker A: Okay.\nSpeaker A: That I've seen.\nSpeaker A: And you, is there an extra, actual button or you actually, you're just using the mass to go up and down?\nSpeaker C: Well, what do you, for the, I can put you press and, right, if you're on the channel, let's say, then you press on the middle and then if you do that again, the volume goes up and if you do that, it goes down.\nSpeaker C: But if you want to keep it with volume here and here, I'm pretty sure.\nSpeaker A: Well, I mean, if we are going to use this for channels, right, and you, wouldn't the volume need to be separate somehow?\nSpeaker A: Like you could just have, I don't know, you can then have it up and down.\nSpeaker D: Or you could actually.\nSpeaker D: I think we can go on the fact that it does just work with you.\nSpeaker D: But the only thing is like, I've had, they're so expensive.\nSpeaker D: Like it has to be.\nSpeaker D: Is that part of it?\nSpeaker D: Is that what makes me expensive?\nSpeaker D: Yeah.\nSpeaker A: I don't know.\nSpeaker A: They have so much memory, though.\nSpeaker A: I think so.\nSpeaker A: I think it's the.\nSpeaker A: I think it's the capability.\nSpeaker A: I mean, they can hold what?\nSpeaker A: Like 5,000.\nSpeaker D: I'm thinking we could, if we have, so, I mean, but an iPad just has that circle thing.\nSpeaker B: Yeah.\nSpeaker B: So that's really why they're expensive.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But like, since it just has this circle thing, you could make it a kind of cool shape.\nSpeaker D: Like it could be a cool sort of, you know, because it could be circular.\nSpeaker D: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Or something weird like that.\nSpeaker C: It could just be simple instead of being a mess.\nSpeaker C: Because the other thing I didn't tell you on my presentation is that people find it, find that it's a big waste of time to have to learn how to use your remote.\nSpeaker C: Yeah.\nSpeaker C: That's another thing they complained about.\nSpeaker C: Yeah.\nSpeaker C: What are the buttons, whether?\nSpeaker C: Volume.\nSpeaker C: I don't know if it's just said that.\nSpeaker D: So, okay.\nSpeaker D: This is just for TV.\nSpeaker D: It's not for, or it is, does need to be compatible with.\nSpeaker B: DVD is simple.\nSpeaker B: You just have play, pause.\nSpeaker A: So how do you know, actually, our new project requirements, I'm not sure if they meant, oh, use only for television as a not for DVD or just not internet type thing.\nSpeaker A: So I'll check that and update you on the next.\nSpeaker A: So like if we have that, we'll hold off on that.\nSpeaker A: Yeah.\nSpeaker A: So yeah.\nSpeaker C: But DVD players do usually have their own remote.\nSpeaker D: That's true.\nSpeaker D: Yeah.\nSpeaker D: So I know I'm not clear.\nSpeaker B: I'm clear I don't have it all on one because you know, you turn it on, then you want to turn off the volume, and then you want to go to the menu.\nSpeaker D: So you'd have to have like, I think you would have to have like a function switch button, you know, somewhere.\nSpeaker D: So like you're either on TV or on DVD or you're on VCR or you're like, so.\nSpeaker B: Well, the DVD is only like four buttons.\nSpeaker D: Yeah, but I mean like to switch the, so like, to switch the function of the little circle disk, the touchpad.\nSpeaker B: Yeah.\nSpeaker B: The circle only does channel is an applicable to DVD really because you don't want to.\nSpeaker B: Yeah, but it would be.\nSpeaker B: Volume is, and volume is actually controlled on the TV so you don't have to switch.\nSpeaker D: So, but I'm saying like, does it make sense to have like some kind of a button?\nSpeaker D: So like you're, if you're on TV, like you can switch channels, but then if, if you're on DVD, then like the channel button, like the, the region of the disk that was for channels is for like switching to different tracks or, you know, to different.\nSpeaker D: I mean, do we need to think about that?\nSpeaker A: That like, yeah, let's think about it because we need to wrap up.\nSpeaker A: Let's see.\nSpeaker A: So everyone's going to go finish their questionnaire.\nSpeaker A: Then when we come back after lunch, we'll have 30 minutes of individual work.\nSpeaker A: Yeah.\nSpeaker A: So think about the things we've discussed and bring some ideas to our next meeting.\nSpeaker A: See you soon.\nSpeaker D: Okay.\nSpeaker D: So I feel like I want to spaceship.\nNone: We're done.\n"}, {"meeting_id": "Bed002", "summary": "The meeting was about a computer based navigation system that will be used to conduct linguistic experiments on navigational terminology and intentionality, and it was simultaneously being recorded for another project. The navigational interface allows people to walk around a place and perform basic actions, like entering a building. Its goal is to understand what people intend to do based on granular linguistic features. The experimental set up is that people talk to a human assistant who controls the program in an attempt to complete a certain task. To study intentionality, the team will determine important linguistic features and use a Bayes model to see if they can predict the intention based on utterances.", "dialogue": "None: Yes.\nSpeaker F: Okay, we're on.\nSpeaker F: So just make sure that the wireless mic is on.\nSpeaker F: If you're wearing wireless.\nSpeaker F: Check one.\nSpeaker F: And you should be able to see which one.\nSpeaker F: Which one you're on by watching the little bars change.\nSpeaker D: So which is my bar?\nSpeaker D: Number one.\nSpeaker F: So actually if you guys want to go ahead and read digits now as long as you've signed the consent form, that's all right.\nSpeaker F: Are we supposed to read digits at the same time?\nSpeaker F: No.\nSpeaker F: No, each individually.\nSpeaker F: We're talking about doing all at the same time, but I think cognitively that would be really difficult to try to read them while everyone else is.\nSpeaker F: So when you're reading the digits strings, the first thing to do is just say which transcript\nSpeaker E: you're on.\nSpeaker F: So you can see the transcript. There's two large number strings on the digits.\nSpeaker F: So you would just read that one.\nSpeaker F: And then you read each line with a small pause between lines.\nSpeaker F: And the pause is just so that the person transcribing it can tell where one line ends and the other begins.\nSpeaker F: And I'll read the digits strings first so you can see how that goes.\nSpeaker F: Again, I'm not sure how much I should talk about stuff before everyone's here.\nSpeaker E: Well, one more coming.\nSpeaker F: Well, I'm going to go ahead and read digit strings and then we can go on from there.\nSpeaker F: So this is transcript.\nSpeaker F: 3, 6, 5, 1, 3, 6, 7, 0, 3, 3, 5, 1, 9, 0, 3, 6, 4, 0, 2, 3, 6, 5, 1, 3, 6, 5, 1, 0, 5, 1, 0, 5, 2, 1, 5, 6, 4, 7, 6, 7, 8, 0, 2, 1, 4, 0, 4, 6, 0, 5, 5, 8, 0, 2, 8, 1, 6, 8, 2, 8, 2, 9, 3.\nSpeaker E: Okay, well, we can start doing it.\nSpeaker D: Okay, this is transcript 3, 6, 3, 1, 3, 6, 5, 1, 3, 1, 5, 4, 2, 9, 7, 6, 3, 9, 5, 7, 1, 6, 7, 4, 4, 6, 7, 8, 9, 0, 4, 3, 0, 2, 4, 9, 2, 6, 9, 1, 7, 4, 1, 2, 3, 4, 0, 4, 7, 9, 3, 7, 4, 5, 6, 8, 3, 9, 9, 3, 4, 7, 0, 7, 0, 9, 1, 7, 1, 2,\nSpeaker E: 3, 6, 7, 4, 7, 6, 7, 6, 6, 7, 8, 7, 8, 9, 1, 8, 7, 8, 8, 8, 8, 8, 9, 0-0-3129-889-0.\nSpeaker E: 0-0-3129-8235-33049-490507-670531-80620-015-03832.\nSpeaker E: 0-0-3309-490507-670531-80620-015-03832.\nSpeaker E: 0-0-3309-015-030830620-015-03832.\nSpeaker C: 0-0-3309-99900.\nSpeaker B: 0-0-3351-357-0. 0-9-0802-0. 0-1-0-0-7-4-3401-4218-5478-8682-7.\nSpeaker B: 0-0-7-4401-4218-5478-8682-7.\nSpeaker B: 0-0-66450-901-152-6-0-3.\nSpeaker B: 0-0-7-6-3100-5-6-827-92959-5904-2-3.\nSpeaker B: 0-0-7-6-6-6-6-6-6-6-6-5708-5902-0.\nSpeaker F: 0-0-7-6-6-6-7-6-7.\nSpeaker F: 0-0-7-6-6-6-6-6-6.\nSpeaker F: only once the speaker form and the consent form.\nSpeaker F: And the short form, I mean you should read the consent form, but the thing to notice is that we will give you an opportunity to edit all the transcripts.\nSpeaker F: So if you say things and you don't want them to be released to the general public, which these will be available at some point to anyone who wants them, you'll be given an opportunity by email to bleep out any portions you don't like.\nSpeaker F: On the speaker form, just to tell you as much of the information as you can, if you're not exactly sure about the region, we're not exactly sure either.\nSpeaker F: So don't worry too much about it.\nSpeaker F: It's just self-rating.\nSpeaker F: I think that's about it.\nSpeaker F: And should I do want me to talk it all about why we're doing this and what this project is?\nSpeaker E: Yeah, no, let's see.\nSpeaker C: Does that see no, that we're meeting in here?\nSpeaker E: She's got an even, she was notified whether she knows.\nSpeaker E: It's another question.\nSpeaker E: So are the people going to be identified by name?\nSpeaker F: Well, we'll anonymize it in the transcript, but not in the audio.\nSpeaker E: OK, so in terms of people worrying about excising things from the transcript that's unlikely since it isn't attributed.\nSpeaker E: Oh, I see, but the other thing.\nSpeaker F: Right, so if I said, oh, hi, Jerry, how are you?\nSpeaker F: We're not going to go through and cancel out the jerry.\nSpeaker F: Sure.\nSpeaker F: So we will go through.\nSpeaker F: And in the speaker ID tags, there'll be M1 or 7M108.\nSpeaker F: But I don't know a good way of doing it on the audio and still have people who are doing discourse research to be able to use the data.\nSpeaker E: No, I was in the complain here.\nSpeaker E: I just wanted to understand.\nSpeaker D: Right, OK.\nSpeaker D: We can make up LES for each of us.\nSpeaker F: OK, I mean, whatever you want to do is fine.\nSpeaker F: But we find that we want the meeting to be as natural as possible.\nSpeaker F: We're trying to do real meetings.\nSpeaker F: And so we don't want to have to do aliases and we don't want people to be editing what they say.\nSpeaker F: So I think it's better just as a post-process to edit out every time you bash Microsoft.\nSpeaker E: OK, so why don't you tell us briefly?\nSpeaker E: Give your normal spiel.\nSpeaker F: So this is a project that's called Meeting Recorder.\nSpeaker F: And there are lots of different aspects of the project.\nSpeaker F: So my particular interest is in the pda of the future.\nSpeaker F: This is a mock-up of one.\nSpeaker F: Yes, we do believe the pda of the future will be made of wood.\nSpeaker F: The idea is that you'd be able to put a pda at the table at an impromptu meeting and record it and then be able to do querying and retrieval later on on the meeting.\nSpeaker F: So that's my particular interest is a portable device to do information retrieval on meetings.\nSpeaker F: Other people are interested in other aspects of meetings.\nSpeaker F: So the first step on that in any of these is to collect some data.\nSpeaker F: And so what we wanted is a room that's instrumented with both the tabletop microphones.\nSpeaker F: And these are very high quality pressure zone mics, as well as the close talking mics.\nSpeaker F: What the close talking mics gives us is some ground truth gives us high quality audio, especially for people who aren't interested in the acoustic parts of this corpus.\nSpeaker F: So for people who are more interested in language, we didn't want to penalize them by having only the far-field mics available.\nSpeaker F: And then also, it's a very, very hard task in terms of speech recognition.\nSpeaker F: And so on the far-field mics, we can expect very low recognition results.\nSpeaker F: So we wanted the near-field mics to at least isolate the difference between the two.\nSpeaker F: So that's why we're recording in parallel with the close talking and the far-field at the same time.\nSpeaker F: And then all these channels are recorded simultaneously and framed synchronously so that you can also do things like beam forming on all the microphones and degree search like that.\nSpeaker F: Our intention is to release this data to the public, probably through a body like the LDC, and just make it as a generally available corpus.\nSpeaker F: There's other work going on in meeting recordings.\nSpeaker F: So we're working with SRI with UW, NIST has started an effort, which will include video.\nSpeaker F: We're not including video, obviously.\nSpeaker F: And then also a small amount of assistance from IBM.\nSpeaker F: So I'm so involved.\nSpeaker F: Oh, and the digit strings.\nSpeaker F: This is just a more constrained task.\nSpeaker F: So because the general environment is so challenging, we decided to do at least one set of digit strings to give ourselves something easier.\nSpeaker F: And it's exactly the same digit strings as in TI digits, which is a common connected digits corpus.\nSpeaker F: So we'll have some comparison to be able to be made.\nSpeaker F: OK.\nSpeaker F: Anything else?\nSpeaker F: Nope.\nSpeaker F: OK, so when the last person comes in, just have him wear a wireless.\nSpeaker F: It should be on already, either one of those.\nSpeaker F: And read the digit strings and fill out the forms.\nSpeaker F: So the most important form is the consent form.\nSpeaker F: So be sure everyone signs that if they consent.\nSpeaker F: It should be usual for meeting certain people come late.\nSpeaker F: So you don't have to use these.\nSpeaker F: And just give me a call, which my team is up there when your meeting is over.\nSpeaker F: And I'm going to leave the mic here, but I'm not going to be on.\nSpeaker F: So don't have them use this one.\nNone: Thank you.\nNone: Thank you.\nSpeaker D: Any further?\nSpeaker D: Yeah, there we go.\nSpeaker E: Anyway, Adam, we will be using the screen as well.\nSpeaker E: So yeah.\nSpeaker E: Wow.\nSpeaker E: Organization.\nSpeaker E: So you guys got an email about this Friday or something about what we're up to.\nSpeaker C: No.\nSpeaker D: I got it.\nSpeaker C: What was the nature of the email?\nSpeaker E: Oh, this was about inferring intentions from pictures and context and the word.\nSpeaker E: It's like go to sea or visit.\nSpeaker C: Well, I didn't get it.\nSpeaker C: I don't think I did it.\nSpeaker E: I guess these have got better filters.\nSpeaker E: Because I said it to everybody.\nSpeaker E: You just blew it off.\nSpeaker E: OK.\nSpeaker D: It's only simple.\nSpeaker D: So this is the idea.\nSpeaker D: We could pursue, if we thought it's worth it, but I think we will agree on that, to come up with a very, very first crew prototype and do some implementation work and do some research and some modeling.\nSpeaker D: So the idea is if you want to go somewhere and focus on that object down, oh, I can actually walk around this.\nSpeaker D: Down here.\nSpeaker D: That's the powder tower.\nSpeaker D: Now, we found in our data experiments that there are three things you can do.\nSpeaker D: You can walk this way and come really, really close to it and touch it.\nSpeaker D: But you cannot enter or do anything else unless you're interested in rock climbing.\nSpeaker D: It would do you no good standing there.\nSpeaker D: It's just a dark alley.\nSpeaker D: You can touch it.\nSpeaker D: If you want to actually go up or into the tower, you have to go this way and then through some buildings and upstairs and so forth.\nSpeaker D: If you actually want to see the tower, and that's what actually most people want to do, is just have a good look of it, take a picture for the family.\nSpeaker D: You have to go this way and go up here.\nSpeaker D: There you have a really view.\nSpeaker D: It exploded during the 30 years war, really interesting side.\nSpeaker D: And these lines are paths, that's the street network of our geographic information system.\nSpeaker D: You can tell that we deliberately cut out this part, because otherwise we couldn't get our GIS system to lead people this way.\nSpeaker D: It would always use the closest point to the object and then the tourists would be faced in front of a wall, but would do them absolutely no good.\nSpeaker D: So what we found interesting is, first of all, intentions differ.\nSpeaker D: Maybe you want to enter a building.\nSpeaker D: Maybe you want to see it, take a picture of it.\nSpeaker D: Or maybe you actually want to come as close as possible to the building for whatever reason.\nSpeaker C: That's it.\nSpeaker C: What's it made out of?\nSpeaker C: Red limestone.\nSpeaker C: So maybe you would want to touch it.\nSpeaker D: Yeah, maybe you would want to.\nSpeaker D: OK, these intentions we could, if you want to call it the Vista mode, really just want to get the overview or look at it.\nSpeaker D: The Enter mode and the Tango mode, I always call it, with silly names.\nSpeaker D: So this Tango means literally translated to touch.\nSpeaker D: But sometimes the Tango mode is really relevant in the sense that if you don't have the intention of entering a building, but you know that something is really close to it, you just want to approach it or get to that building.\nSpeaker D: Consider, for example, the post office in Chicago building, so large that it has its own zip code.\nSpeaker D: So the entrance could be miles away from the closest point.\nSpeaker D: So sometimes it makes sense maybe to distinguish there.\nSpeaker D: So I've looked through 20 some.\nSpeaker D: I didn't look through all the data.\nSpeaker D: And there is a lot more different ways in people, the ways people phrase how to get, if they want to get to a certain place.\nSpeaker D: And sometimes here it's a little bit more obvious.\nSpeaker D: Maybe I should go back, come on steps.\nSpeaker E: OK, come in, sit down.\nSpeaker E: Grab yourself a microphone.\nSpeaker D: You need to sign some stuff.\nSpeaker E: Well, you can sign afterwards.\nSpeaker C: Yeah, absolutely.\nSpeaker C: I'll have to read some digits afterwards.\nSpeaker D: There are two.\nSpeaker A: Maybe small?\nSpeaker A: OK, I see.\nSpeaker A: OK.\nSpeaker E: Thank you.\nSpeaker E: OK, that was our idea.\nSpeaker E: It also has to be switched on.\nSpeaker E: No, it's already on it.\nSpeaker E: That's all.\nSpeaker D: OK, good.\nSpeaker D: Thank you.\nSpeaker D: OK.\nSpeaker D: There was a year people, when they want to go to a building, sometimes they just want to look at it.\nSpeaker D: Sometimes they want to enter it.\nSpeaker D: Sometimes they want to get really close to it.\nSpeaker D: That's something we found.\nSpeaker D: It's just a truth.\nSpeaker D: And the place is where you will lead them for these intentions is sometimes incredibly different.\nSpeaker D: I gave an example where at the point where you end up, if you want to look at it, it's completely different.\nSpeaker D: You want to enter it.\nSpeaker D: So this is sort of how people may phrase those requests to a mock-up system at least, the way they did it.\nSpeaker D: And we get tons of these.\nSpeaker D: How do I get to?\nSpeaker D: I want to go to.\nSpeaker D: But also, give me directions to, and I would like to see.\nSpeaker D: And what we can sort of do if we look closer at the data that was the wrong one, we can look at some factors that may make a difference.\nSpeaker D: First of all, very important.\nSpeaker D: And that I've completely forgot that when we talked.\nSpeaker D: This is of course a crucial factor.\nSpeaker D: What type of object is it?\nSpeaker D: So some buildings you just don't want to take pictures of or very rarely, but you usually want to enter them.\nSpeaker D: Some objects are more picturesque, more highly photographed.\nSpeaker D: Then of course, the actual phrases may give us some idea of what the person wants.\nSpeaker D: Sometimes I found in looking at the data in a superficial way I found some sort of modifiers that may also give us hint.\nSpeaker D: I'm trying to get to a need to get to sort of instead of the fact that you're not really sightseeing and just for their full pleasure and so forth.\nSpeaker D: And this is straight to the context, which ought to should be considered.\nSpeaker D: That whatever it is you're doing at the moment may also influence the interpretation of a phrase.\nSpeaker D: So this is really my suggestion.\nSpeaker D: It's really simple.\nSpeaker D: We start with now, let me say one more thing.\nSpeaker D: What we do know is that the parser reuse in the smart group system will never differentiate between any of these.\nSpeaker D: So basically all of these things will result in the same XMLM3L structure, sort of action go and then an object and a source.\nSpeaker D: So it's way to capture those differences in intentions.\nSpeaker D: So I thought maybe for a deep understanding task, that's a nice sort of playground or first little thing we can start it and sort of look, OK, we need, we're going to get those M3L structures, the crude, undifferentiated parsed, interpreted input.\nSpeaker D: We may need additional part of speech or maybe just some information on the verb and water fires, auxiliaries, we'll see.\nSpeaker D: And I will try to sort of come up with a list of factors that we need to get out of there.\nSpeaker D: And maybe we want to get a switch for the context.\nSpeaker D: So this is not something we can actually monitor now, but just something we can set.\nSpeaker D: And then you can all imagine sort of a constraint, satisfaction program, depending on what comes out.\nSpeaker D: We want to have a structure resulting if we repeat it through belief net or something along those lines.\nSpeaker D: We get an inferred intention.\nSpeaker D: We produce a structure that differentiates between the vista of the enter and the tango mode, which I think we may want to ignore, but that's my idea.\nSpeaker D: It's up for discussion.\nSpeaker D: We can change all of it any bit of it.\nSpeaker D: Go it all the way.\nSpeaker B: Now, this email you sent actually.\nSpeaker B: What?\nSpeaker B: Now I remember the email.\nSpeaker C: OK.\nSpeaker C: Huh.\nSpeaker C: Still, I have no recollection whatsoever of the email.\nSpeaker C: I'll have to go back and check.\nSpeaker E: Not important.\nSpeaker E: So what is important is that we understand what the proposed task is.\nSpeaker E: And Robert and I talked about this some on Friday.\nSpeaker E: And we think it's well-formed.\nSpeaker E: So we think it's a well-formed starter task for this deep understanding in the tourist domain.\nSpeaker B: So where exactly is the deep understanding being done?\nSpeaker B: Like, I mean, is it before the base net?\nSpeaker B: Is it?\nSpeaker E: Well, it's always all of it.\nSpeaker E: So in general, it's always going to be the answer is everywhere.\nSpeaker E: So the notion is that this isn't real deep, but it's deep enough that you can distinguish between these three quite different kinds of going to see some tourist thing.\nSpeaker E: And so that's the quote deep that we're trying to get at.\nSpeaker E: And Robert's point is that the current front end doesn't give you any way to not only doesn't it do it, but it also doesn't give you enough information to do it.\nSpeaker E: It isn't like if you just took what the front end gives you and use some clever inference algorithm on it, you would be able to figure out which of these is going on.\nSpeaker E: So in general, it's going to be true of any kind of deep understanding.\nSpeaker E: There's going to be contextual things.\nSpeaker E: There's going to be linguistic things.\nSpeaker E: There's going to be discourse things.\nSpeaker E: And they've got to be combined.\nSpeaker E: And my idea on how to combine them is with a belief net.\nSpeaker E: Well, it may turn out that some totally different thing is going to work better.\nSpeaker E: The idea would be that you take your editing your slide.\nSpeaker D: Yeah.\nSpeaker D: So as I get ideas, so discourse, I thought about the course that needs to be.\nSpeaker D: I'm sorry.\nSpeaker E: OK.\nSpeaker E: So this is taking minutes as we go in his own way.\nSpeaker E: But anyway, so the thing is, naively speaking, you've got for this little task a belief net, which is going to have as output the conditional probability of one of three things that the person wants to view it, to enter it, or to tango with it.\nSpeaker E: So the output of the belief net is pretty well formed.\nSpeaker E: And then the inputs are going to be these kinds of things.\nSpeaker E: And then the question is, the two questions is, one, where do you get this information from?\nSpeaker E: And two, what's the structure of the belief net?\nSpeaker E: So what are the conditional probabilities of this, that, and the other, given these things?\nSpeaker E: And you probably need intermediate notes.\nSpeaker E: I don't know what they are yet.\nSpeaker E: So it may well be that, for example, that knowing whether, oh, another thing you want is some information about the time of day.\nSpeaker E: Now, let me want to call that part of context.\nSpeaker E: But the time of day matters a lot.\nSpeaker E: And if things are obviously closed, then people don't want to enter them.\nSpeaker E: And if it's not obvious, you may want to actually point out the people that it's closed, what they're going to is closed, and they don't have the option of entering it.\nSpeaker E: So another thing that can come up, and will come up as soon as you get serious about this, is that another option, of course, is to have it more of a dialogue.\nSpeaker E: So if someone says something, you could ask them.\nSpeaker E: And now, one thing you could do is always ask them.\nSpeaker E: That's boring.\nSpeaker E: And it will also be a pain for the person using it.\nSpeaker E: So one thing you could do is build a little system that say, whenever you've got a question like that, I've got one of three answers, ask them which one you want.\nSpeaker E: But that's not what we're going to do.\nSpeaker D: But maybe that's a false state of the system that it's too close to call.\nSpeaker E: Oh, yeah.\nSpeaker E: You want the ability to ask.\nSpeaker E: You want the ability to ask.\nSpeaker E: But what you don't want to do is build a system that always asks every time.\nSpeaker E: And that's not getting out the scientific problem.\nSpeaker E: And in general, it's going to be much more complex than that.\nSpeaker E: This is purposely a really simple case.\nSpeaker D: So I have one more point to the bus question.\nSpeaker D: I think also the deep understanding part of it is going to be in the extent that we wanted in terms of modeling.\nSpeaker D: We can start, you know, basic from human beings, model that, it's motions, going, walking, seeing.\nSpeaker D: We can model all of that and then compose whatever inferences we make out of these really conceptual primitives.\nSpeaker D: That will be extremely deep in my understanding.\nSpeaker E: So the way that might come up, if you want to do that, you might say, as an intermediate step in your belief net, is there a source path goal schema involved?\nSpeaker E: And if so, is there a focus on the goal, or is there a focus on the path or something?\nSpeaker E: And that could be one of the condition, you know, in some piece of the belief net, that could be the appropriate thing to enter.\nSpeaker B: So when would we extract that information from the M3L?\nSpeaker E: No.\nSpeaker E: No.\nSpeaker E: So the M3L is not going to give you, what he was saying is, the M3L does not have any of that.\nSpeaker E: All it has is some really crude stuff saying, person wants to go to a place.\nSpeaker C: The M3L is the old smart com.\nSpeaker E: Right.\nSpeaker E: Well, M3L itself refers to modeling and markup language.\nSpeaker E: So we have to have a better way of referring to the parts that we're going to put.\nSpeaker D: Yeah, that implies speech.\nSpeaker D: I think you can actually actually intentionally let us, is what we're going to do.\nSpeaker E: But they call it intention lattice.\nSpeaker D: But in the intention lattice, they call it intention hypotheses.\nSpeaker E: So they're going to give us some, or we can assume that you get this crude information about intention.\nSpeaker E: And that's all they're going to provide.\nSpeaker E: And they don't give you the kind of object.\nSpeaker E: They don't give you any discourse history.\nSpeaker E: If you want to keep that, you have to keep it somewhere else.\nSpeaker D: Well, they keep it where they have to request it.\nSpeaker D: Right.\nSpeaker E: Well, they keep it by their lights.\nSpeaker E: It may or may not be what we want.\nSpeaker C: So if someone says I want to touch the side of the powder tower, that would basically we need to pop up Tango mode in the direction.\nSpeaker C: If you've got a simple as that, yeah.\nSpeaker E: But it wouldn't.\nSpeaker C: But that isn't necessarily, we have to infer a source path goal to something really good, touching the side, right?\nSpeaker D: There is a point there if I understand you correct me.\nSpeaker D: Because sometimes people just say, you find very often, where is the city hall?\nSpeaker D: And they don't want to see it on a map, or they don't want to know it's 500 yards away from you, or it's to the north.\nSpeaker D: They want to go there.\nSpeaker D: That's what they say is, where is it?\nSpeaker C: Where is it, that thing?\nSpeaker C: And the parser would output.\nSpeaker D: Well, that's a question mark.\nSpeaker D: A lot of parses just way beyond their scope of interpreting that.\nSpeaker D: But still outcome, the outcome will be some form of structure with a town hall, and maybe it's a WH focus on the town hall.\nSpeaker D: But to interpret it, somebody else has to do that later.\nSpeaker C: I'm trying to figure out what this mark is just without, but depending on these things.\nSpeaker D: It will probably tell you how far away it is.\nSpeaker D: That's even what you've mapped as it has to how far away it is and shows it to your own map.\nSpeaker D: Because we cannot differentiate at the moment between the intention of wanting to go there, or the intention of just knowing or wanting to know where it is.\nSpeaker A: People might not be able to infer that either.\nSpeaker A: I could imagine if someone came up to me and asked where is the city hall, are you trying to get there?\nSpeaker A: Because how I describe this location, probably kind of whether I should give them directions now or say whatever, a half a mile away or something like that.\nSpeaker D: granularity factor because we're people ask you.\nSpeaker D: Where is New York?\nSpeaker D: We'll tell them it's on the East Coast.\nSpeaker D: You won't tell them how to get there.\nSpeaker D: Check that bus to the airport in blah, blah, blah.\nSpeaker D: But if it's the post office, we will tell them how to get there.\nSpeaker D: So they have done some interesting experiments of that in Hamburg.\nSpeaker E: But go back to the, yeah,\nSpeaker D: this is on tour is knowledge about buildings, they're opening times, and then coupled with time of day.\nSpeaker A: So that context was like their presumed purpose context, like business or travel, as well as the utterance context.\nSpeaker A: Like I'm now standing at this point.\nSpeaker E: Yeah, I think we have all along.\nSpeaker E: We've been distinguishing between situational context, which is what you have is context in discourse context, which you have is D H, I don't know what the H means.\nSpeaker E: History, discourse.\nSpeaker E: OK, whatever.\nSpeaker E: So we can work out terminology later.\nSpeaker E: So they're quite distinct.\nSpeaker E: I mean, you need them both, but they're quite distinct.\nSpeaker E: And so what we're talking about doing as a first shot is not doing any of the linguistics except to find out what seems to be useful.\nSpeaker E: So the reason the belief net is in blue is the notion would be, this may be a bad idea, but the idea is to take us a first goal, see if we could actually build a belief net that would make this three-way distinction in a plausible way, given all these transcripts, and we're able to by hand extract the features to put into belief net, saying, here are the things which if you could get them out of the language and discourse and put them into the belief net, it would tell you which of these three intentions is most likely.\nSpeaker E: And to actually do that, build it, run it on the data where you hand transcribe the parameters and see how that goes.\nSpeaker E: If that goes well, then we can start worrying about how we would extract them.\nSpeaker E: So where would you get this information and expand it to other things like this?\nSpeaker E: But if we can't do that, then we're in trouble.\nSpeaker D: And if you can't do this task, we need a different engine.\nSpeaker E: Or something.\nSpeaker E: Well, if it's the belief net, we'll switch to logic or some terrible thing.\nSpeaker E: But I don't think that's going to be the case.\nSpeaker E: I think that if we can get the information, belief net is a perfectly good way of doing the inferential combination of it.\nSpeaker E: The real issue is what are the factors involved in determining this?\nSpeaker E: How many seconds?\nSpeaker E: So I know.\nSpeaker E: Is it clear what's going on here?\nSpeaker A: I missed the beginning.\nSpeaker A: But I guess could you go back to the slide, the previous one?\nSpeaker A: So is it that it's these are all factors that, these are the ones that you said that we are going to ignore now or that we want to take into account.\nSpeaker A: You're saying you're taking them into account.\nSpeaker E: But you don't worry about how to extract them.\nSpeaker E: So let's find out which ones we need first.\nSpeaker A: OK.\nSpeaker A: And it's clear from the data, the correct answer in each case.\nSpeaker E: But let's go back to the slide of data.\nSpeaker A: Like do we know from the data?\nSpeaker A: Which?\nSpeaker A: OK.\nSpeaker A: Not from that data.\nSpeaker D: But since we are designing, compared to this bigger data collection effort, we will definitely take care to put it in the form of the other.\nSpeaker D: To see whether we can get sort of empirically validated data.\nSpeaker D: From this, we can sometimes, you know, and that's not what we need for leave that anyhow.\nSpeaker D: It's sort of sometimes what people want to see.\nSpeaker D: They phrase it more like this.\nSpeaker D: But it doesn't exclude anybody from praising it.\nSpeaker D: Totally different.\nSpeaker D: They still.\nSpeaker D: But then other factors may come into play, the changey outcome of the leave that.\nSpeaker D: So this is exactly what, because you can never be sure.\nSpeaker D: And I'm sure even the most deliberate data collection experiment will never give you data that say, well, if it's phrased like that, be intentional.\nSpeaker A: I mean, the only way you could get those if you were to give the subjects a task, you're a current goal is to.\nSpeaker A: Yeah, that's what we're doing.\nSpeaker D: So that's what you want.\nSpeaker D: We will still get the phrasing all over the place.\nSpeaker D: I'm sure that.\nSpeaker A: No, that's fine.\nSpeaker A: I guess it's just knowing the intention from.\nSpeaker A: From that task, that's the experiment.\nSpeaker E: So I think you all know this.\nSpeaker E: But we are going to actually use this little room and start recording subjects, probably within a month or something.\nSpeaker E: So this is not any, and you guys worry, except that we may want to push that effort to get information we need.\nSpeaker E: So our job is to figure out how to solve these problems if it turns out that we need data of a certain sort, then the sort of data collection branch can be asked to do that.\nSpeaker E: And one of the reasons why we're recording the meeting for these guys is because we want their help when we start doing recording of subjects.\nSpeaker E: So yeah, you're absolutely right, though.\nSpeaker E: No, you will not have.\nSpeaker E: And there it is.\nSpeaker E: And the.\nSpeaker A: And I think the other concern that has come up before too is if it's, I don't know if this was collected.\nSpeaker A: What situation the data was collected?\nSpeaker A: It was the one that you showed in your talk.\nSpeaker A: Like people.\nSpeaker A: No, no.\nSpeaker A: OK.\nSpeaker A: So it was just like someone actually mobile, using a device?\nSpeaker D: No, not.\nSpeaker D: It was mobile, but not really with the system.\nSpeaker A: So there were never answers.\nSpeaker A: OK.\nSpeaker A: OK.\nSpeaker A: But it was it.\nSpeaker A: I guess I don't know the situation of collecting the data.\nSpeaker A: Here, you can imagine them being walking around the city.\nSpeaker A: It's like one situation.\nSpeaker A: And then you have all sorts of other situational context factors that would influence how to interpret, like you said, the scope and things like that.\nSpeaker A: If they're doing it in a, you know, I'm just going to hear with a map and ask questions.\nSpeaker A: I would imagine that the data would be really different.\nSpeaker A: So it's just.\nSpeaker D: Yeah.\nSpeaker D: But it was never the goal of that data connection to surface at a certain purpose.\nSpeaker D: So that's why, for example, the tasks were not differentiated by attentionality.\nSpeaker D: There was no label, you know, attention A, attention B, attention C, or task ABC.\nSpeaker D: I'm sure we can produce some if we needed that will help us in other times.\nSpeaker D: But you've got to leave something for other people to model, so to finding out what the context of the situation really is.\nSpeaker D: It's an interesting thing.\nSpeaker D: So if I'm at the moment curious and I want to approach it from the end where we can sort of start with this toy system that we can play around with.\nSpeaker D: So that we get a clearer notion of what input we need for that, what suffices and what doesn't.\nSpeaker D: And then we can start worrying about where to get this input.\nSpeaker D: What do we need, you know, ultimately, once we are all experts in changing that parser, for example, maybe there's just a couple of really things we need to do.\nSpeaker D: And then we get more whatever part of speech, more construction type, like stuff out of it.\nSpeaker D: Procmedic approach for the moment.\nSpeaker C: How exactly does the data collection, do they have a map?\nSpeaker C: And then you give them a scenario of some sort?\nSpeaker D: OK.\nSpeaker D: You mentioned you're the subject you're going to be in here.\nSpeaker D: And you see either the 3D model or a quick time animation of standing in a square in Heidelberg.\nSpeaker D: So you actually see that.\nSpeaker D: The first thing is you have to read a text about Heidelberg.\nSpeaker D: So just offer textbook, the tourist guide, to familiarize your sample, the sort of all-sound in German street names like Lerkerkasse.\nSpeaker D: So that's part one.\nSpeaker D: Part two is you're told that this new, wonderful computer system exists.\nSpeaker D: You can tell you everything you want to know.\nSpeaker D: And it understands your completely.\nSpeaker D: And so you're going to pick up that phone dialer number, and you get a certain amount of tasks that you have to solve.\nSpeaker D: First you have to know, find out how to get to that place.\nSpeaker D: Maybe with the intention of buying stamps in there.\nSpeaker D: Maybe some next task is to get to a certain place, take a picture for your grandchild.\nSpeaker D: The third one is to get inspiration\nNone: of the history of the object. The fourth one, and then the system breaks down of crashes.\nSpeaker D: At the third, right then?\nNone: After the third task.\nSpeaker D: And then, or after the fourth, sometimes you give that file.\nSpeaker D: And then a human operator comes on and apologizes that the system has crashed.\nSpeaker D: Or just you to continue.\nSpeaker D: Now we're the human operator.\nSpeaker D: And so you have basically the same tasks again, just with different objects.\nSpeaker D: And you go through it again, and that was it.\nNone: Oh, and one little bit.\nNone: The computer, you've been told the computer system knows exactly where you are via GPS.\nSpeaker D: When the human operator comes on, that person does not know.\nSpeaker D: So the GPS is crashed as well.\nSpeaker D: So the person first has to ask you, where are you?\nSpeaker D: And so you have to do some tell the person sort of where you are, depending on what you see there.\nSpeaker D: This is a bit that I don't think we did.\nSpeaker D: We discussed that bit.\nNone: Just sort of squeezed that in now.\nNone: But it's something that we would provide some very interesting data for some people, I know.\nSpeaker D: So.\nSpeaker A: So in the display, you said you might have a display that shows.\nSpeaker D: Yeah.\nSpeaker D: Additionally, you have a sort of a map type.\nSpeaker A: Your perspective.\nSpeaker A: And so as you, OK, so as you move through it, that's they just track it on the for themselves.\nSpeaker D: You don't, I don't know.\nSpeaker D: I don't think you really move.\nSpeaker D: OK.\nSpeaker D: Yeah.\nSpeaker D: I mean, that would be an enormous technical effort.\nSpeaker D: Unless we can show it, walks through, we can have movies of walking, you walking through Hydeburg and ultimately arriving there, maybe we want to do that.\nSpeaker A: I was just trying to figure out how.\nSpeaker D: The map was sort of a bit.\nSpeaker D: You want to go to that place.\nSpeaker D: And then sort of there, you see the label of the name.\nSpeaker D: So we get those labels, pronunciation stuff.\nSpeaker D: And so we can change that.\nSpeaker A: So your tasks don't require you to, I mean, you're told.\nSpeaker A: So when your task is, I don't know, a goodbye stamp or something like that.\nSpeaker A: So do you have to respond or do you, what are you supposed to be telling the system?\nSpeaker A: What you're doing now or?\nSpeaker A: Well, we'll see what people do.\nSpeaker A: OK.\nSpeaker A: So it's just like this figure out what they would say.\nSpeaker A: Yeah.\nSpeaker D: And we will record both sides.\nSpeaker D: I mean, we'll record the wizard.\nSpeaker D: In both cases, it's going to be human and the future and the operating limit.\nSpeaker D: And there will be some dialogue.\nSpeaker D: So you first have to listen to that and see what they say.\nSpeaker D: We can instruct the wizard and how expressive and talkative should be.\nSpeaker D: But maybe what you're suggesting is what you're suggesting is it might be to pour the data if you sort of limit it to this ping pong one.\nSpeaker D: Your task results in a question.\nSpeaker D: And then there's an answer.\nSpeaker D: And that's the end of the task.\nSpeaker D: Do you want to have it more steps sort of?\nSpeaker A: I don't know how much direction is given to the subject about what their interaction.\nSpeaker A: I mean, they're unfamiliar with interacting with the system.\nSpeaker A: All they know is it's this great system that can do stuff.\nSpeaker E: So some extent, this is a different discussion.\nSpeaker E: OK.\nSpeaker E: So we have to have this discussion of the experiment, the data collection, and all that sort of stuff.\nSpeaker E: And we do have a student who is a candidate for wizard.\nSpeaker E: She's going to get in touch with me.\nSpeaker E: It's a student of Eves, F-E-Y-F-A spelled F-E-Y-D.\nSpeaker E: Oh, PayPal.\nSpeaker E: You know her?\nSpeaker E: OK.\nSpeaker A: She started taking the class last year and then didn't continue.\nSpeaker A: She's a graduate.\nSpeaker A: She's a graduate.\nSpeaker A: Yeah, she is a good.\nSpeaker A: OK.\nSpeaker A: Yeah.\nSpeaker A: I know her very, very briefly.\nSpeaker A: I know she's interested in.\nSpeaker A: OK.\nSpeaker E: So anyways, she's looking for some more part-time work while she's waiting, actually, for graduate school.\nSpeaker E: And she'll be in touch.\nSpeaker E: So we may have someone to do this.\nSpeaker E: And she's got some background in all this stuff, and there's a linguist.\nSpeaker E: So Nancy, at some point, we'll have another discussion on exactly how that's going to go.\nSpeaker E: And Jane, but also Liz, have offered to help us do this data collection and design and stuff.\nSpeaker E: So when we get to that, we'll have some people doing it to know what to do.\nSpeaker A: I guess the reason I was asking the details of this kind of thing is that it's one thing to collect data for.\nSpeaker A: And speech recognition are various other tests that have pretty clear correct answers.\nSpeaker A: But with intention, obviously, as you point out, there's a lot of other factors.\nSpeaker A: I'm not really sure how the question of how to make it a appropriate toy version of that.\nSpeaker A: It's just hard.\nSpeaker A: So I mean, I guess that was my question.\nSpeaker C: Is the intention implicit in the scenario that's given?\nSpeaker A: It is if they have these tasks that they're supposed to.\nSpeaker C: Yeah, it just wasn't sure what level of detail it has.\nSpeaker D: Yeah.\nSpeaker D: Right.\nSpeaker D: No one is a moment.\nSpeaker E: So that's part of what we'll have to figure out.\nSpeaker E: But the problem that I was going to try to focus on today was let's suppose by magic you could collect dialogues in which one way or the other, you were able to figure out both the intention and set the context and know what language was used.\nSpeaker E: So let's suppose that we can get that kind of data.\nSpeaker E: The issue is can we find a way to basically featureize it so that we get some discrete number of features so that when we know the values to all those features or as many as possible, we can come up with the best estimate of which of the, in this case, three little intentions are most likely.\nSpeaker A: But where did the three intentions go there to see it?\nSpeaker A: And to come this close to?\nSpeaker E: The terminology we're using is to go back.\nSpeaker E: To view it, okay, to enter it.\nSpeaker E: Now, it seems to me, you have no trouble with those being distinct.\nSpeaker E: Take a picture of it.\nSpeaker E: You might well want to be really rather different place than entering it.\nSpeaker E: And for objects, it's at all big, sort of getting to the nearest part of it could be quite different than either of those.\nSpeaker A: Just sort of.\nSpeaker A: Okay, so now I understand the referendum.\nSpeaker A: Tango mode.\nSpeaker C: So I would have thought it was more of a wall.\nSpeaker C: To wall suits?\nSpeaker A: Yeah, like how close are you?\nSpeaker A: Yeah, because it's really close.\nSpeaker A: Yeah, well anyway.\nSpeaker B: So, so like what features do you want to try to extract from either parts or whatever?\nSpeaker B: Like the presence of a word or the presence of a certain stem?\nSpeaker E: Right.\nSpeaker E: Is there a construction or the kind of object or anything else that's in the, it's either in the discourse itself or in the context.\nSpeaker E: So if it turns out that whatever it is, you want to know whether the person's a tourist or not.\nSpeaker E: Okay, that becomes a feature.\nSpeaker E: And then how you determine that is another issue.\nSpeaker E: But for the current problem, it would just be, okay, if you can be sure that it's a tourist versus a businessman versus a native or something, that would give you a lot of discriminatory power and then you just have a little section in your belief net that said, though it's in the short run, you'd set them and see how it worked.\nSpeaker E: And then in the longer run, you would figure out how you could derive them from previous discourse or anything else you knew.\nSpeaker B: So what's the, like how should we go about it?\nSpeaker E: Okay, so first of all, do we, either of you guys, you got a favorite belief net that you've played with Java Bayes or something?\nSpeaker B: No, I'm getting it.\nSpeaker E: Okay, anyway, get one.\nSpeaker E: Okay, so one of the things we want to do is actually pick a package, it doesn't matter which one.\nSpeaker E: Presumably one that's got good interactive abilities because a lot of what we're gonna be, we don't need the one that'll solve massive belief nets quickly, these are not gonna get big in the foreseeable future.\nSpeaker E: And we do want one in which it's easy to interact with and modify it because a lot of what it's gonna be is playing with this.\nSpeaker E: And probably one in which it's easy to have what amounts to transcript files so that if we have all these cases, okay, so we make up cases that have these features, okay, and then you'd like to be able to say, okay, here's a bunch of cases.\nSpeaker E: There are even ones that you can do learning.\nSpeaker E: Okay, so you have all their cases and their results and you have algorithms go through and run around trying to accept the probabilities for you.\nSpeaker E: Probably that's not worth it.\nSpeaker E: I mean, my guess is we aren't gonna have enough data that's good enough to make these data fitting ones worth it that I don't know.\nSpeaker E: So I would say the first task for you too, guys, is to pick a package.\nSpeaker E: You know, the standard thing, you want it stable, you want it.\nSpeaker E: And as soon as we have one, we can start trying to make a first cut at what's going on.\nSpeaker E: But what I like about it is it's very concrete.\nSpeaker E: We know what the outcomes are gonna be and we have some data that's loose.\nSpeaker E: We can use our own intuition and see how hard it is.\nSpeaker E: And importantly, what intermediate notes we think we need.\nSpeaker E: So it turns out that just thinking about the problem you call things that you really need to, this is the kind of thing that is an intermediate little piece in your belief net.\nSpeaker E: That'd be really interesting.\nSpeaker D: It may self as a platform for a person, maybe me or whoever who is interested in doing some linguistic analysis, we have a whole frame that grew up here.\nSpeaker D: We can see what they have found out about those concepts already that are contained in the data.\nSpeaker D: You know, to come up with a nice set of features and maybe even means abstracting them.\nSpeaker D: And that altogether could also be a common nice paper that's going to be published somewhere between the two of them, right?\nSpeaker D: And when you said Java-based belief net, you were talking about when said Rano Coffee or that are in programming.\nSpeaker E: No, it turns out that there is a, the no end of Java libraries.\nSpeaker E: Okay, and it turns out one called Java-based, which is one that people around here use a fair amount.\nSpeaker E: I have no idea whether that's, the obvious advantage of that is that you can then relatively easily get all the other Java packages for GUIs or whatever else you might want to do.\nSpeaker E: So that's, I think, why a lot of people are doing research use that.\nSpeaker E: But that may not be, I have no idea whether that's the best choice.\nSpeaker E: And there are plenty of people around students in the department who live and breathe based nets.\nSpeaker E: So.\nSpeaker A: There's the toolkit that Kevin Murphy has developed which might be useful to you.\nSpeaker E: So yeah, Kevin would be a good person to start with.\nSpeaker E: Nancy knows them well.\nSpeaker E: I don't know whether you guys have met Kevin yet or not, but yeah, this is really probably a pretty sure\nNone: that the, for example, this, the dialogue history is producing XML documents everywhere.\nSpeaker D: Of course, XML and the ontology that a student is constructing for me back in, in email is in oil.\nSpeaker D: It's also an XML.\nSpeaker D: So that's where a lot of knowledge about bakery is about what else about the process and stuff that is going to come from.\nSpeaker D: So it has, I owe capability and it's a job of engineering and then it'd be able to.\nSpeaker E: So yeah, we're sort of committed to XML as the kind of interchange, but that's not a big deal.\nSpeaker E: So in terms of interchange in and out of any module we build, it'll be XML.\nSpeaker E: And if you're going off the queries to the ontology, for example, you'll have to deal with its interface, but that's fine.\nSpeaker E: And all of these things have been built with much bigger projects than this in mind.\nSpeaker E: So they've worked very hard.\nSpeaker E: It's kind of blackboards and multi-way blackboards and ways of interchanging and registering.\nSpeaker E: And so that, I don't think, is even worth us worrying about just yet.\nSpeaker E: I mean, if we can get the core of the thing to work in a way that we're comparable with, and we can even out of it XML little descriptors.\nSpeaker D: I believe.\nSpeaker D: I don't see that.\nSpeaker D: I like for some of the, what you said about the getting input from just Pius about where you have the data is specified for features and so forth.\nSpeaker D: That's, of course, easy also to do with.\nSpeaker E: You can make an XML format for that, sure.\nSpeaker E: You know, feature value XML format.\nSpeaker E: Probably as good a way as any.\nSpeaker E: So it's all, yet, I guess it's also worth while you're proking around for XML packages, the things you'd like.\nSpeaker D: Does my comps just have set it back?\nSpeaker D: Yeah, sure.\nSpeaker D: The library does that.\nSpeaker E: And the question is, you have to look.\nSpeaker E: That should be, we should be able to look at that.\nSpeaker E: No, the, what I sort of came to my mind\nSpeaker D: was the notion of an idea that if there are nets that can actually try to set their own probability factors based on input, just in file format, if we get really wild on this, we may actually want to use some copper that other people made.\nSpeaker D: And for example, if they are in mate, then we get XML documents with discourse annotations.\nSpeaker D: From the discourse act down to the phonetic level, Michael has a project about recognizing discourse acts and he does it all in mate.\nSpeaker D: And so they're actually annotating data and data and data.\nSpeaker D: So if we think it's worth it one of these days, not with this first prototype, maybe with a second.\nSpeaker D: And we have the possibility of taking input that's generated elsewhere and learning from that.\nSpeaker D: That'd be nice.\nSpeaker E: It'd be nice, but I don't want to come.\nSpeaker E: I mean, you can't run your project based on the speculation that the data will come and you don't have to actually design the nets.\nSpeaker E: Just a backdoor.\nSpeaker E: It could happen.\nSpeaker E: So in terms of what the smart com gives us for M3L packages, it could be that they're fine.\nSpeaker E: It could be, you know, you don't really like it.\nSpeaker E: So we're not required to use their packages.\nSpeaker E: We are required at the end to give them stuff in their format, but hey, it doesn't control what you do in internal language.\nNone: What was the time frame for this?\nNone: Today?\nNone: I have to do this.\nSpeaker E: Yeah, but this week, to have you guys pick a package and tell us what it is and give us a point where you can play with it or something.\nSpeaker E: And then as soon as we have it, I think we should start trying to populate it for this problem, make a first cut, you know, what's going on.\nSpeaker E: And probably the easiest way to do that is some online way.\nSpeaker E: I mean, you can figure out what you want.\nSpeaker E: I'm going to website her.\nNone: You know, I was actually more joking with the 2.3.\nNone: It was a usual joke.\nNone: It was a big as long as you guys need for that.\nSpeaker D: Maybe it might be interesting if the two of you can agree on who's going to be the speaker next, but to tell us something about it, you should know what it does\nNone: and how it does that. Well, or about the speaker.\nSpeaker D: Yeah, or you split it up.\nSpeaker D: So there will be sort of the assignment for that.\nSpeaker D: Force lines or whatever.\nNone: I think that what it can do and how far you can.\nSpeaker E: Well, I like the also that I have a first cut at what the bleaching at looks like.\nSpeaker E: It's really crude.\nSpeaker C: So, you know, here are your features and whatnot.\nNone: Right.\nNone: Yeah.\nSpeaker E: And as I said, what I like to do is, I mean, what would be really great is bring it in if we could in the meeting, say, you know, here's the package here is the current one we have, what other ideas you have.\nSpeaker E: And then we can think about this idea of making up the data file of get a tentative format for it.\nSpeaker E: Let's say XML that says, you know, these are the various scenarios we can just add that.\nSpeaker E: And then we just this file of them.\nSpeaker E: And when you think you've got a better belief in it, use run it against this data file.\nSpeaker B: So we'd be like, yeah.\nSpeaker B: Oh, yeah.\nNone: Until we know more.\nSpeaker C: And what's the relation to this with the table so that the system works out with English?\nSpeaker D: So this is why what you are doing this, I received two lovely emails, the full and T in the whole Linux version.\nSpeaker D: There, I uploaded the most.\nSpeaker D: And I started to unpack the Linux one, the int1 work kind of has a package.\nSpeaker D: Linux, when it told me that you can't really unpack it because it's a future date.\nNone: So this is the tentative.\nNone: It's between Germany.\nNone: I had to wait until one of the talk was actually a new two package.\nSpeaker D: Now, then it will be my job to get this whole thing running both on suite and on this machine.\nSpeaker D: And so that we have it.\nSpeaker D: And then hopefully that hoping that my urgent message will now come through to Ralph and Tillman that it will send some more documentation along.\nSpeaker D: We, I can show, maybe that's what I will do next Monday.\nSpeaker D: I'll show the state of the system and show that.\nSpeaker D: Yeah.\nSpeaker E: So the answer to John O. is that these are at the moment separate.\nSpeaker E: One hopes is that when we understand how the analyzer works, we can both worry about converting it to English and worry about how it could extract the parameters we need for the belief net.\nSpeaker C: I guess my question is more about time frame.\nSpeaker C: So we're going to do belief nets this week.\nSpeaker C: And then, oh, yeah, I don't know.\nSpeaker E: None of this is neither of these projects has got a real tight timeline in the sense that over the next month, there's a deliverable.\nSpeaker E: So it's opportunity.\nSpeaker E: In that sense, it's opportunistic.\nSpeaker E: If we don't get any information for these guys for several weeks, then we aren't going to sit around wasting time trying to do the problem.\nSpeaker E: Or guess what?\nSpeaker E: I'm just going to do other things.\nSpeaker D: Yeah, but this point is really, I think, very valid.\nSpeaker D: But ultimately, we hope that both will merge into harmonious the wonderful state where we cannot only do very necessities, i.e. changing the tables.\nSpeaker D: That does exactly in English where it doesn't German.\nSpeaker D: But also that we can sort of have the system where we can say, OK, this is what it usually does.\nSpeaker D: And now we add this little thing to it, whatever John knows of musk or us, great belief net.\nSpeaker D: We plug it in.\nSpeaker D: And then for these certain tasks, and we know that navigational tasks are going to be core domain of the new system, it all of a sudden it does much better.\nSpeaker D: And because it produced better answers, tell the person, as I showed you on this map, produce either a red line that goes to the vista point or a red line that goes to the tango point or a red line that goes to the door, which would be great.\nSpeaker D: So you don't only can you show that you know something sensible, but ultimately, if you produce a system like this, it takes the person world points to go, rather than taking them always to the geometric center of a building, which is what they do now.\nSpeaker D: And we even had to take out, the men's you missed that part.\nSpeaker D: We had to take out a bit of the road work so that it doesn't take you to the wall.\nSpeaker D: Every time.\nSpeaker D: Yeah, really?\nSpeaker D: So this was actually an actual problem that we encountered, which is your way up.\nSpeaker D: Because current navigation systems don't really care.\nSpeaker D: They get you to the beginning of the street, somehow do the house number.\nSpeaker D: But even that is problematic.\nSpeaker D: If you go, if you want to drive to the SAP in Balthorff, I'm sure the same is true with Microsoft.\nSpeaker D: It takes you to the address, whatever street number blah, blah, blah.\nSpeaker D: I'm miles away from the entrance.\nSpeaker D: Because this postal address is maybe a mailbox somewhere.\nSpeaker D: But the entrance where you actually want to go is something completely different.\nSpeaker D: So unless you're a mail person, you really don't want to go.\nSpeaker E: Probably not then, because you probably can't drop the mail there anyway.\nSpeaker E: I don't even know if you're better.\nNone: Clear?\nNone: OK.\nNone: That's it.\nSpeaker C: Outer towers made of red limestone.\nSpeaker A: I was wondering.\nSpeaker A: Do you want to see a picture?\nSpeaker D: Sure.\nSpeaker D: After we boot for that, though.\nSpeaker A: So you two who will be working on this, I mean, are you supposed to just do it by thinking about the station?\nSpeaker A: Can you use the sample data?\nSpeaker A: Of course, I do just sample data.\nSpeaker A: Is there more than there are lots of sample data that\nSpeaker D: is beyond what you have there? I think this is an in part my job to look at that and to see whether there are features in there that can be extracted.\nSpeaker D: And to come up with some features that are not purely based on a real experiment or on reality, but sort of on pure intuition of real, this is maybe a sign for that.\nSpeaker D: And this is maybe a sign for this.\nSpeaker B: So later this being should get together.\nSpeaker B: Talked down.\nSpeaker B: Let's see what we look at.\nSpeaker E: OK, we can end the meeting and call Adam.\nSpeaker E: And then we want to look at some filthy pictures of high version.\nSpeaker E: We can do that as well.\nSpeaker D: Is that OK?\nSpeaker D: They used the ammunition.\nSpeaker D: They start the ammunition in that tower.\nSpeaker D: And that's why when it was hit by a cannon ball.\nSpeaker C: That's what they call it, the powder tower.\nSpeaker C: At this time, I had so many material.\nSpeaker C: That's why I asked.\nSpeaker A: That's right.\nSpeaker A: OK.\nNone: The little synthrodotter, powder metal.\nNone: Hi there, we're done.\nSpeaker A: Is there a lot of ontological information available about the very time marks then?\nSpeaker A: And it's presumably what title we're doing.\nSpeaker A: I mean, we have this example of post office behind that.\nSpeaker A: Well, I think whether post office is going to be\nNone: building a contextual feature, which we determined would have wanted to look at it.\nSpeaker A: Yep.\nSpeaker A: So going to those facts about that.\nSpeaker A: No.\nSpeaker E: Well, Robert knows exactly what they're going to be.\nSpeaker E: But the point is that, again, for our purposes, the probability to be six or seven buildings that will drop in the dialogue and we will put as much in as we need.\nSpeaker E: And if that turns out to be critical, that's part of what we learned is, OK, we're going to mean that kind of information in order to do this.\nSpeaker E: Oh, Nancy didn't read her numbers yet.\nSpeaker E: Sorry.\nSpeaker E: You love me, Andy?\nSpeaker F: Yeah.\nSpeaker F: So it's been a little bit more since then.\nSpeaker F: I think every time.\nSpeaker E: Let's see.\nSpeaker E: So you turned off your rhythm for two seconds.\nSpeaker F: I turned it off for two seconds.\nSpeaker F: So I have to turn off the strings like this.\nSpeaker E: Right.\nSpeaker E: OK.\n"}, {"meeting_id": "Bed016", "summary": "The meeting was taken up by discussion about a thesis proposal and a talk about to take place at EML. It was advised that more emphasis should be put on the role of construal in the understanding of metaphor and metonymy. Base constructions deal with the norm, while further general domain mechanisms determine how the constructions are invoked depending on the context. Several potential examples of polysemy were discussed in detail: \"walk/run into\", \"on the bus\", \"out of film\", \"where is X?\". However, none of them was an example of lexical polysemy resolved by construal straightforward enough to include in the proposal; the tourist domain is not metaphor rich.", "dialogue": "Speaker A: I am so comfortable.\nSpeaker B: I'm going to like Taiwan and other places too.\nSpeaker A: The first place I can have a meal with is the best restaurant in my career.\nSpeaker E: I'm just one of the most frustrating meetings of my career.\nSpeaker C: It's definitely not the most frustrating meeting I've ever had.\nSpeaker B: Remember you're being recorded at this point.\nSpeaker B: Yeah.\nSpeaker B: We need to specify it with whom.\nSpeaker F: So that's why Keith and I are going to be a little bit used for the first half of the meeting.\nSpeaker F: I avoided that as long as I couldn't.\nSpeaker E: I'm very appreciative of that.\nSpeaker E: I know you were doing that.\nSpeaker C: How did it exactly that paperly Danny like breaks?\nSpeaker E: I have to tell you what they're up to.\nSpeaker E: I love that story.\nSpeaker E: It's a great story.\nSpeaker E: Okay.\nSpeaker E: Oh my goodness.\nSpeaker D: Oh yeah.\nSpeaker D: This is just that we could start off by doing the digits all at the same time.\nSpeaker C: What?\nSpeaker C: All at the same time.\nSpeaker C: I don't know if I would get distracted and confused probably.\nSpeaker F: Are you being silly?\nSpeaker C: Everybody's got different digits right?\nSpeaker C: Yeah.\nSpeaker F: You're kidding.\nSpeaker F: Any rate?\nSpeaker F: Yes.\nSpeaker F: Okay.\nSpeaker D: This is what I want to hear.\nSpeaker D: This is transcript L6-4-3-7-4-1.\nNone: This is a 6.\nNone: This is a 6.\nNone: This is a 6.\nNone: This is a 6.\nNone: This is a 6.\nNone: This is a 6.\nNone: This is a 6.\nNone: 85\u306f\u3044\u3044\u308f 3All 6-0center 5All 6-1 That's a great meeting.\nNone: Now, why?\nSpeaker D: Just, just a few times.\nSpeaker D: Okay.\nSpeaker D: It doesn't matter for them.\nSpeaker B: Are we gonna start all our meetings out that way for no one?\nSpeaker B: No.\nSpeaker B: No.\nSpeaker B: I kind of like it.\nSpeaker B: Could we?\nSpeaker B: It's a satisfy.\nSpeaker B: Yeah.\nSpeaker B: It's so cool.\nSpeaker C: Are we there just to make sure I know what's going on.\nSpeaker C: We're talking about Robert's thesis proposal today.\nSpeaker C: Is that we?\nSpeaker D: We are.\nSpeaker D: We might.\nSpeaker D: Okay.\nSpeaker F: Well, you said there were two things that you might want to do.\nSpeaker F: One was rehearse your...\nSpeaker C: Oh, yes.\nSpeaker D: Not rehearse.\nSpeaker D: I mean, I have just not spent any time on it.\nSpeaker D: So I can show you what I've got, get your input on it, and maybe some suggestions that would be great.\nSpeaker D: And the same is true for the proposal.\nSpeaker D: I will have time to do some revision and some additional stuff on various airplanes and trains.\nSpeaker D: So, I don't know how much of a chance you had to actually read it.\nSpeaker D: Because...\nSpeaker D: But you could always send me comments for an electronic mail.\nSpeaker D: And they will be incorporated.\nSpeaker D: The...\nSpeaker D: It basically says, well, this is Construle, and then it continues to say that one could potentially build a probabilistic relational model that has some general domain general rules, how things are construed, and then the idea is to use a entology situation user and discourse model to instantiate elements in the classes of the probabilistic relational model to do some inferences in terms of what is being construed as what?\nSpeaker D: In our beloved tourism domain.\nSpeaker D: But with the focus on this, this is good.\nSpeaker E: I need a copy of this, yes.\nSpeaker E: Can I get a copy around?\nSpeaker C: Yes, I can pass my extra copy around.\nSpeaker C: Okay, you can read it.\nSpeaker C: You can read it.\nSpeaker D: I don't know.\nSpeaker D: Actually, this is the newest version after your comments.\nSpeaker F: Yeah, no, I see this has got the castle in it and stuff like that.\nSpeaker C: Oh, maybe the version I didn't have.\nSpeaker C: Did the one you sent on the email?\nSpeaker C: Have the...\nSpeaker C: Yeah.\nSpeaker C: That was the most recent one?\nSpeaker F: Yeah, I think so.\nSpeaker C: Yeah.\nSpeaker C: Because I read halfway, but I didn't see a castle thing.\nSpeaker B: I'm changing this.\nSpeaker D: Yeah.\nSpeaker D: If you would have checked your email, you may have received a note from you, as you to send me the...\nSpeaker D: Okay.\nSpeaker D: Sorry.\nSpeaker D: Current formulas.\nSpeaker B: I will.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker D: But for this, it doesn't matter.\nSpeaker B: We can talk about it later.\nSpeaker B: That's not even ready, so.\nSpeaker B: Okay, go on to it, whatever.\nSpeaker B: I'm making changes to not worry about that.\nSpeaker B: Oh, okay, sorry.\nSpeaker B: Come on.\nSpeaker D: In any type of comment, whether it's a spelling or a syntax or...\nSpeaker D: I don't really want to see it.\nSpeaker E: I just want to ask an interesting...\nSpeaker E: I'm page five.\nSpeaker E: Interesting.\nSpeaker B: Anyway.\nSpeaker B: Email anytime, but most useful before...\nSpeaker C: The 21st, I'm assuming.\nSpeaker D: The 21st?\nSpeaker D: The 21st.\nSpeaker C: Today's the 21st.\nSpeaker C: I'll better hurry up.\nSpeaker A: Oh, man.\nSpeaker A: The 29th.\nSpeaker A: The 29th.\nSpeaker A: Okay.\nNone: That's when I'm meeting with Wafeng Walsh, or to sell him this idea.\nSpeaker D: Okay.\nSpeaker A: Okay.\nNone: Okay.\nSpeaker D: Then I'm also going to present a little talk at email.\nSpeaker D: Okay.\nSpeaker D: About what we have done here.\nSpeaker D: And so, of course, I'm going to start out with this slide.\nSpeaker D: So, the most relevant aspects of our stay here.\nSpeaker D: And then I'm asking them to imagine that they're standing somewhere in Hydeburg and someone asked them in the morning.\nSpeaker D: The cave 45 is a well-known discotheque, which is certainly not open at that time.\nSpeaker D: And so, they're supposed to imagine that, you know, do they think the person wants to go there, or just know where it is?\nSpeaker D: Which is probably not the case in that discotheque example.\nSpeaker D: In the Bavaria example, it just want to know where it is.\nSpeaker D: And so, basically, we can make a point that here is where we can make a point that here is ontological knowledge.\nSpeaker D: But if it's 9 p.m. in the evening, then the discotheque question would be, for example, one that might ask for directions, sort of just location.\nSpeaker D: And so forth.\nSpeaker D: And so forth.\nSpeaker D: That's sort of motivating it.\nSpeaker D: Then what have we done?\nSpeaker D: So, while we had our little bit of smart com stuff that we did.\nSpeaker D: I would have done the parser done.\nSpeaker D: Sorry.\nSpeaker D: That's not the construction parser.\nSpeaker D: That's the template paste.\nSpeaker D: Okay.\nSpeaker D: Parser and the generation.\nSpeaker D: Halfway done?\nSpeaker C: Yeah.\nSpeaker D: That's done.\nSpeaker D: You actually change the strategies, right?\nSpeaker D: Yeah.\nSpeaker D: That's 10 words.\nSpeaker C: Well, it, you know, maybe 12.\nSpeaker D: 12?\nSpeaker D: Okay.\nSpeaker D: And, um, and phase doing the synthesis stuff as we speak.\nSpeaker D: That's all about that.\nSpeaker D: Then I'm going to talk about the data.\nSpeaker D: You know these things about, actually, have an example.\nSpeaker D: Probably.\nSpeaker D: Two.\nSpeaker D: Can you hear that?\nSpeaker D: We should have turned the volume on.\nSpeaker C: I can hear it.\nSpeaker C: They might not hear it in the...\nSpeaker C: Or maybe it will.\nSpeaker B: Where are you going to take my club?\nSpeaker B: This was the natural subjects.\nSpeaker B: It sounds like fake.\nSpeaker D: Yeah.\nSpeaker D: But they're mimicking the synthesis when they speak to the computer.\nSpeaker D: You can observe that all the time.\nSpeaker D: They're trying to match their prosody onto the machine.\nNone: The positive thing is, who is the young cancer?\nSpeaker E: Oh, it's pretty slow.\nNone: Where is it?\nSpeaker D: I wonder who\u00e6rr 2gsan the average 22kL.\nSpeaker B: What is this? Oh\nSpeaker D: Okay, answer for, answer for. I will talk about problems with the refraising and how we solved it and some preliminary observations also. I'm not going to put in the figures from this, but I thought it would be interesting to point out that it's basically the same.\nSpeaker D: It is an every human-human telephone conversation and the human computer telephone conversation is of course quite different from some first observations.\nSpeaker D: Then sort of feature back to our original problem, because how to get there, but actually is happening there today.\nSpeaker D: And then maybe talk about the big picture here. Tell a little bit as much as I can about the ETL story.\nSpeaker D: I'm not quite sure about this, whether I should put this in, that you have these two sort of different ideas that are different camps of people envisioning how language understanding works.\nSpeaker D: And then talk a little bit about the embodied simulation approach, pivot here, and as a prelude, talk about monkeys in Italy.\nSpeaker D: And Thrinny was going to send me some slides, but he didn't do it. So from that I have to paper, I can mic a resume of that.\nSpeaker D: I stole an X-Skima from one of your talks, I think.\nSpeaker A: I was like, where'd you get that? It looks familiar.\nSpeaker D: I think that's Bergen Chang, something with the other.\nSpeaker D: I'm not going to bring that. So that's basically what I have so far, the rest is for airplanes.\nSpeaker D: So X-Skimas, then I would like to talk about the construction aspect and then at the end about our basenet.\nSpeaker D: End of story.\nSpeaker D: Anything I forgot that we should mention?\nSpeaker D: Now maybe the FMRI stuff. Should I mention the fact that we're also actually starting to start to look at people's brains in a more direct way?\nSpeaker B: I just want to like tack that on as a comment.\nSpeaker D: Future activities, something.\nSpeaker F: It's hard to mention that you mentioned that when you talk about mirror neurons, then you should talk about the recent stuff about the Tiki.\nSpeaker E: Yeah.\nSpeaker F: And the man has seen how to explain the same phenomenon that we're talking about.\nSpeaker F: So that's the end of the work that I'm going to do.\nNone: So that's one thing.\nNone: It depends. There is a language learning story.\nSpeaker D: Yeah.\nSpeaker D: Actually, even on your black face, you've got a role where the picture is in the bottom.\nSpeaker D: Yeah. I do have it here.\nSpeaker D: And of course, the big picture is this bit.\nSpeaker D: But I don't think I am capable of putting this off and doing justice to the matter.\nSpeaker D: I mean, there is interesting stuff in terms of how language works.\nSpeaker D: So the emergentism story would be nice to tell people what's happening there.\nSpeaker D: Plus how the language learning stuff works.\nSpeaker D: But...\nNone: I mean, I think that's not...\nNone: What is it?\nNone: It may not be that you might want to...\nNone:...and the slides are...\nNone:...simulation.\nSpeaker D: Yeah. That comes up to the ex-schema slide.\nSpeaker D: Okay.\nSpeaker D: So basically, I'm going to steal that from Nancy.\nSpeaker B: Okay. I can give you a more recent...\nSpeaker B:...if you want.\nSpeaker D: Well, that might help.\nSpeaker D: But I also have stuff you trash you left over, your quals and your triple AI.\nSpeaker B: Yeah.\nSpeaker B: Which I can even email you that, you know, like the probably was a few changes.\nSpeaker B: Not big deal.\nSpeaker B: Yeah, you could steal anything you want, actually.\nSpeaker B: You already done, obviously.\nSpeaker D: I don't feel bad about it at all because you are on the title.\nSpeaker B: That's great. That's great.\nSpeaker D: And glad to see the complication.\nSpeaker D: Provocated?\nSpeaker D: Yeah, that's great.\nSpeaker D: I mean, you might even mention that this work you're doing is sort of also with the MPI in Liptsick.\nSpeaker B: It's certainly related.\nSpeaker D: Because email is building up a huge thing in Liptsick.\nSpeaker D: Is it?\nSpeaker D: It's a bio-computation.\nNone: This is the DNA building.\nSpeaker B: How do you look at building?\nSpeaker B: Kind of a different level of analysis.\nSpeaker F: You have multiple buildings at the moment.\nSpeaker C: All sorts of clear things.\nSpeaker C: They're building in the shape of DNA.\nSpeaker E: So you said...\nNone: What?\nSpeaker E: That's really annoying.\nSpeaker F: I mean, I spent a lot of money there.\nSpeaker F: I spent a few hours doing this.\nSpeaker C: Oh, you definitely want to...\nSpeaker C: What's the building?\nSpeaker C: Don't want to waste that money on research.\nSpeaker C: It's horrible.\nSpeaker F: You didn't build with research.\nSpeaker B: I think you did more money.\nSpeaker B: They just want a fun place for them to...\nSpeaker E: And everybody gets a trampoline in there often.\nSpeaker D: The offices are actually a little...\nSpeaker D: You think of ramps coming out of the double headaches.\nSpeaker D: And then you have these half-doms.\nSpeaker D: Class half-doms.\nSpeaker D: And the offices are in the class half-dom.\nSpeaker E: All right, let's start talking about this.\nSpeaker E: Does it exist yet?\nSpeaker D: Is it a model?\nSpeaker D: I think it's...\nNone: It's a lot of...\nNone: It's interactive.\nNone: It's a little bit of a system.\nNone: And...\nSpeaker F: The combination of the biology and the likes of the invention.\nNone: And that's...\nSpeaker D: I mean, another of that was talking about your thesis.\nSpeaker D: Yeah, if somebody has something to say.\nSpeaker E: You might want to double check the spellings of the author's names on your references.\nSpeaker E: You had a few misspells in your slides there.\nSpeaker E: Like, I believe you had Jack and Dorf.\nNone: Unless there's a person called Jack and Dorf.\nNone: On that one.\nSpeaker E: That's the only thing I noticed in there.\nSpeaker B: In the presentation?\nSpeaker B: I'll probably have comments for you separately.\nSpeaker B: Oh, in the presentation?\nSpeaker D: Yeah, that's actually worried about Biptech.\nSpeaker D: No, that's quite possible.\nSpeaker D: It's copy and paste from something.\nSpeaker F: So I think it looks like...\nSpeaker D: It did.\nSpeaker D: There is a reference to Srini.\nSpeaker F: Oh, did you put in something about...\nSpeaker F: Metalline metaphor here, right?\nSpeaker F: The individual...\nSpeaker F: We talked about putting in something about people and...\nNone: Oh, yeah, okay.\nNone: Good.\nNone: I see where you have it.\nSpeaker F: I think that's the problem.\nNone: What I think is...\nSpeaker F: I think even if you put this...\nSpeaker F: It's a little bit of a false verb.\nNone: Unless you put it there, it's a verb.\nNone: I think we talked about scalability yet by...\nNone: Binding and destruction.\nNone: General Constructural mechanism.\nSpeaker D: Is that in there?\nSpeaker D: Yeah.\nSpeaker D: It's good for us.\nSpeaker D: But I did not focus on that aspect, but...\nSpeaker D: It's just underneath that reference to metaphor.\nSpeaker D: So it's last paragraph before, so on page two, a main focus.\nSpeaker D: But it's pretty...\nSpeaker D: It's not about that, yeah.\nSpeaker D: No, it says it, but it doesn't say it does.\nSpeaker D: Why?\nSpeaker D: It doesn't give much of a...\nSpeaker F: Okay, let me tell you what.\nNone: I think the punchline is because it's actually working.\nSpeaker F: Which is that the constructions that...\nNone:...are doing...\nNone:...are in a way quite...\nSpeaker F:...but cover only base cases.\nNone: And to make them a fine...\nSpeaker F:...to work with cases, an inmate cases, and all those things.\nSpeaker F: It requires additional mechanism of...\nSpeaker F:...and the punchline is...\nNone:...then...\nNone:...right, you can get essentially your flogginality...\nNone:...that you have to do some new construction...\nNone:...at the base level.\nNone: It should interact with all the metatonnes and metaphors...\nSpeaker F:...and all of the projections of it also should work.\nSpeaker F: And similarly, it's the new metaphor...\nNone:...and it should...\nSpeaker E:...compose with all the constructions.\nNone: That's true, they're...\nNone:...it's a big way of...\nSpeaker C: So does that mean instead of having tens and tens of rules in your...\nSpeaker C:...contextually grammar, you just have...\nSpeaker C:...these base constructions and then general mechanism for co-wersing in them.\nSpeaker C: So that, for example, in the metaphor case...\nNone:...the every time direct idea of source path and goal...\nNone:...and any metaphor will be tensed, right?\nSpeaker A: And you can say grammar is the same thing.\nNone: But the trick is...\nNone:...the way the constructions...\nSpeaker F:...it requires the logic of the position...\nNone:...for example, the heat as well, trouble is in the containers...\nSpeaker C:...which is extremely stressful.\nNone: So that's where...\nSpeaker C: So with constriction, you don't have to have a construction...\nSpeaker C:...for every possible thing to can fill the rule.\nNone: It's a very big deal if you have to string work here.\nNone: This is probably the same as the other.\nNone: I don't think so.\nSpeaker D: That doesn't say it at all, but...\nNone:...even though...\nSpeaker D:...even though...\nSpeaker D:...if there are basic cases, even...\nSpeaker D:...it seems like nothing is context free.\nSpeaker F: But walked into the cafe and ordered a drink...\nNone:...and walked into the cafe and broke his nose.\nSpeaker D: It doesn't mean there are any reasons.\nSpeaker D: It can't be explained.\nSpeaker D: It's a normal word, it's a physical object.\nSpeaker F: So it's multiple of the rules.\nSpeaker F: And if I take this part...\nSpeaker F:...it's a very simple thing.\nSpeaker F: It's a normal rule.\nSpeaker F: And if I take part of what we've done...\nSpeaker A:...it's just interaction we can leave the analysis...\nSpeaker A:...and it can stroll.\nNone: So yeah, it doesn't mean that you can't do it a go-play.\nNone: It does say that...\nSpeaker E:...if you walk into a cafe and you walk into a cafe...\nNone:...it's not consistent with other things.\nNone: You got to reject that reading.\nSpeaker C: You can't...\nSpeaker C:...dition me with your first sentence.\nSpeaker C: And so I thought, why would he walk into the cafe and then...\nSpeaker C:...somehow break his nose?\nSpeaker E: Just flip it on the way forward.\nSpeaker D: You don't find that usage...\nSpeaker D:...I checked for it in the Brown National Corpus.\nSpeaker D: So walking into never really means...\nSpeaker D:...what smack.\nSpeaker D: But if you walk smack into the cafe or slam into the wall...\nNone:...you will find running to...\nSpeaker C:...car's running the telephone balls all the time.\nSpeaker D: Or you can run into an old friend or...\nSpeaker A: No!\nSpeaker B: Yeah, running to might even be more impact sense...\nSpeaker B:...than, you know, container sense.\nNone: Like running to an old friend...\nSpeaker F:...crow, if someone is writing...\nSpeaker F:...you have George, we have George, we have something...\nSpeaker F:...complicative, reason-wondering...\nNone:...something else, and maybe it's...\nNone:...car-vide-event.\nSpeaker B: Oh, contact.\nSpeaker B: I mean, there's contact that's social contact, whatever.\nSpeaker E: Sutton's surprising contact.\nSpeaker E: First of all.\nSpeaker E: But of course, no, I mean, it has a life of its own...\nSpeaker E:...and sort of partially inspired by the spatial...\nSpeaker E:...yeah.\nSpeaker F: Oh yeah, but for sure.\nSpeaker F: Yeah.\nSpeaker C: Do you have a teacher that says that?\nSpeaker B: There's lots of things we can make teachers out of.\nSpeaker B: But, uh...\nSpeaker B:...we don't need to be...\nSpeaker D:...we don't need to be- Probably not your marks in the kitchen today.\nSpeaker B: No, no, no, no, no, no, no, no, no, no.\nSpeaker B: Okay.\nSpeaker E: In other news?\nSpeaker D: Well, well, the...\nSpeaker D: Actually, the...\nSpeaker D: What would have been really nice is to find an example for...\nSpeaker D:...all of this from our domain.\nSpeaker D: So maybe if we can make one up...\nSpeaker D:...now that would be incredibly helpful.\nSpeaker B: So where it should illustrate...\nSpeaker B:...how do we say all of this, do you mean like...\nSpeaker B:...I don't know, the related work stuff as well as...\nSpeaker D:...mapping?\nSpeaker D: Well, we have, for example, a canonical use of something.\nSpeaker D: And we have some constructions and then it's construed as something.\nSpeaker D: And then we may get the same constructions with a metaphorical use...\nSpeaker D:...that's also relevant to the domain.\nSpeaker F: Okay, so it should be published here.\nSpeaker F: What's the first one?\nSpeaker F: It's where the buzz and on the buzz are.\nNone: That actually a little is really getting used because...\nNone:...there's some incident in this.\nSpeaker D: Okay.\nSpeaker D: I had two hours with George on this.\nSpeaker D: What did he do?\nSpeaker D: Join the call.\nSpeaker D: The buzz is a metaphorical metonym that relates some metapath...\nSpeaker D:...metaphorically and you're on that path.\nSpeaker D: I mean, there's a platform notion, right?\nSpeaker D: He's standing on the buzz waving to me.\nSpeaker D: But the regular, as we speak, John O. is on the buzz to New York.\nSpeaker D: That's what did I call it here?\nSpeaker D: The transportation schema, something...\nSpeaker D:...you can be on the first flight, on the second flight, and you can be...\nSpeaker D:...on the wagon.\nSpeaker F: So that may or may not be what you want to do.\nSpeaker F: It could do something much simpler.\nSpeaker D: But it's unfortunate.\nSpeaker D: This is not really something a tourist would ever say.\nSpeaker D: Well, as it was repairing it, there was something...\nSpeaker D:...but...\nSpeaker D:...we had initially started discussing the audio film.\nSpeaker D: And there's a lot of audio analysis.\nSpeaker D: Could we capture that?\nSpeaker D: With a different control of...\nSpeaker B: Yeah, we've thought about it before.\nSpeaker B: It's used as examples in other papers.\nSpeaker B: It's a little complicated.\nSpeaker B: I would have felt like it's a state of...\nSpeaker B:...there's resource, right?\nSpeaker B: What is film?\nSpeaker B: You're out of the state of having film, right?\nSpeaker B: And somehow film is standing for the resource.\nSpeaker B: But the state of having some resource is just labeled as that resource.\nSpeaker E: Yeah, I mean, plus the fact that there's also...\nSpeaker E:...I mean, can you say, like, the film ran out, you know, or...\nSpeaker E: Yeah, it's film with a projector.\nSpeaker E: So like the film went away from where it should be, namely with you, or something like, you know...\nSpeaker E:...the film is gone, right?\nSpeaker E: I never really knew what was going on.\nSpeaker E: I mean, I find it sort of a little bit far-fetched to say that...\nSpeaker E:...that I'm out of film means that I have left the state of having film or something like that.\nSpeaker B: Or having is also associated with location, right?\nSpeaker B: Yeah.\nSpeaker B: So if the film left, you know, the...\nSpeaker D: So running out of something is different from being out of somewhere.\nNone: Well, being out of something is...\nSpeaker B: So running out of it is just a process.\nSpeaker B: But that's some run, yeah.\nSpeaker B: That's okay.\nSpeaker D: I mean, the final state of running out of something is being out of it.\nSpeaker D: Yeah.\nSpeaker E: You got there.\nSpeaker E: That part is fine.\nSpeaker E: You got that out of it.\nNone: So, no one is...\nSpeaker F:...the professional linguists they have.\nSpeaker B: This whole piece is on that.\nSpeaker B: There was...\nSpeaker A: Out.\nSpeaker E: There was one on...\nSpeaker E:...an out or out of.\nSpeaker E: Maybe it just out.\nSpeaker E: Okay.\nSpeaker F: It was over, but there was also a big one out.\nSpeaker E: Yeah, Susan, one right?\nSpeaker E: Oh, yeah.\nSpeaker E: The syrup spread out, anything.\nSpeaker B: Yeah.\nSpeaker B: And, undoubtedly, there's been...\nSpeaker B:...remes of work about it in...\nSpeaker B:...tyre linguistics, but...\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker B: It's not one of the...\nSpeaker B:...more straightforward ones to defend.\nSpeaker B: So, you probably don't want to use it for purposes.\nSpeaker B: These are...\nSpeaker B:...you're just like computational linguists, right?\nSpeaker D: Or...\nSpeaker D:...are you?\nSpeaker D: There is...\nSpeaker B:...it's going to be four computational linguists.\nSpeaker B: Okay.\nSpeaker B: But more...\nSpeaker B:...compices on the computational...\nSpeaker D:...for emphasis on the linguists.\nSpeaker D: More...\nSpeaker D:...there's going to be...\nSpeaker D:...just for computational linguists,...by coincidence,...for the rest is whatever bio-computing people and...\nSpeaker D: Okay.\nSpeaker B: So, I mean, this, you know, like...\nSpeaker D: The computer should be very computational.\nSpeaker B: Okay.\nSpeaker B: So, I would stay away from one that involves weird...\nSpeaker B:...construal stuff.\nSpeaker B: You know, it's an obvious one.\nSpeaker D: I mean, the...\nSpeaker D:...our bakery example might be nice.\nSpeaker D: Is there a bakery around here?\nSpeaker D: Yeah.\nSpeaker D: So, if you could really just...\nSpeaker D:...construit as a...\nSpeaker D:...no, it's the bakery itself.\nSpeaker D: Is it a building that you want to go to?\nSpeaker D: Oh, yeah.\nSpeaker D: Or is it something to eat that you want to buy?\nSpeaker D: And then...\nSpeaker B: No.\nSpeaker B: What?\nSpeaker B: The bakery can't be something you're going to eat.\nSpeaker C: This is a speech act.\nSpeaker C: It's true.\nNone: Exactly.\nSpeaker F: You want to...\nSpeaker F:...you...\nSpeaker F:...be green as a place that...\nSpeaker F:...you can get baked goods.\nSpeaker F: Well, that's one.\nSpeaker F: You want to buy something.\nSpeaker F: But the other is...\nSpeaker F:...you might smell the smell or just curious about what it would be...\nSpeaker E:...the bakery and the Naglewood.\nSpeaker E: You know, wonder how people here right there living...\nNone:...if they're living in such a reason why you might...\nSpeaker F:...pick up the eggs.\nSpeaker F: It doesn't mean I want to buy some baked goods.\nSpeaker A: Okay.\nNone: Those are interesting examples.\nSpeaker B: It's not clear that there's a lot of pragmatics there.\nSpeaker B: There's a lot of stuff.\nSpeaker A: Maybe beyond what you want to use.\nSpeaker A: So, let's first do...\nSpeaker F:...the first thing that's fun of eating is probably easiest than that.\nNone: And actually...\nSpeaker B:...the one you have...\nSpeaker B:...is the...\nSpeaker B:...you mean the steak wants to pay?\nSpeaker B: No, not that one.\nSpeaker F: That's a sort of background.\nSpeaker F: This is the page line.\nSpeaker C: The Play-Toe in the book.\nSpeaker C: No.\nSpeaker D: How much does it cost?\nSpeaker D: Where's the castle?\nSpeaker D: How old is it?\nSpeaker D: How much does it cost?\nSpeaker B: To go in.\nSpeaker E: 200 million dollars.\nSpeaker F: So...\nSpeaker E: Yeah, I think that's gonna be good.\nSpeaker E: Yeah, that's good.\nSpeaker D: But as Nancy just suggested, it's probably elliptic.\nSpeaker B: Like, it doesn't refer to thing.\nSpeaker B: It refers to...\nSpeaker B:...things downing for active, most relevant activity for a tourist.\nSpeaker B: Good to give it that way.\nSpeaker E: Yeah.\nSpeaker E: Well, shoot.\nSpeaker E: Isn't that...\nSpeaker E: I mean, that's fun.\nSpeaker D: I mean, my argument here is...\nSpeaker D: It's the same thing as Play-Toe's on the top shelf.\nSpeaker D: You can refer to...\nSpeaker D:...a book of Play-Toe by you, you should play it over here.\nSpeaker D: No, no, I'm agreeing that this is a good...\nSpeaker D:...back to it.\nSpeaker D: So, you can castles have...\nSpeaker D:...as truth, I'd say, have admission fees.\nSpeaker D: Where is the castle?\nSpeaker D: How much has it cost?\nSpeaker D: Um...\nSpeaker D: How far is it?\nSpeaker D: From here.\nNone: So...\nSpeaker D: You're also not...\nSpeaker D:...refereg to the...\nSpeaker D:...wets of the objectors.\nNone: Okay, do I think that a nice...\nNone:...can afford what you're aware of in a tourist building.\nNone: Oh, again, I was...\nNone:...you think that she's aware of...\nNone:...the amount of land.\nNone: Where was...\nNone:...there was height of earth...\nNone:...in 34 or something.\nSpeaker F: Yeah.\nSpeaker F: Yeah, or some such thing.\nSpeaker E: Like, what side were they on?\nSpeaker B: Okay, it's like this.\nSpeaker B: It's here.\nSpeaker B: Um...\nSpeaker B: I'm like, it's developmental state or something like that.\nSpeaker B: You could...\nSpeaker B:...make it as you could get that.\nSpeaker E: I mean, there's also things like...\nSpeaker E:...I guess I could ask something like...\nSpeaker E:...where can I find out about blah blah blah...\nSpeaker E:...and the sort of...\nSpeaker E:...doesn't necessarily...\nSpeaker E:...I don't necessarily have to care about the spatial location.\nSpeaker E: Just give me a phone number and I'll call them or something like that.\nNone: We're starting to get that, yeah.\nSpeaker F: Where could I find it so open?\nSpeaker F: Yeah.\nNone: The dams.\nNone: And the traffic is horrible.\nNone: But there's...\nNone:...there's things about it.\nNone: You can also think about...\nSpeaker D:...how could I get out of the way?\nSpeaker A: State.\nNone: Is this the issue?\nSpeaker F: Is that maybe just a message?\nSpeaker E: That it's not the thing that's forthcoming.\nSpeaker B: Alright.\nSpeaker B: It's a more exotic one, version of that.\nSpeaker B: I'm really into...\nSpeaker B:...do you really say that?\nSpeaker F: Which you really say that?\nSpeaker C: I'm really into art.\nSpeaker B: Yeah, as I can say.\nSpeaker B: Fix expression, yeah.\nSpeaker B: There are all sorts of fixed expressions.\nSpeaker B: Out of sorts now.\nSpeaker B: I mean trouble.\nSpeaker D: Well, I mean, the data that I've looked at so far, there's tons of cases for polysamy.\nSpeaker D: So making reference to buildings as institutions, as containers, as...\nSpeaker D:...whatever.\nSpeaker D: So in music, for example, museums, as a building, or as something where pictures hang, or something that puts on exhibits, so forth.\nSpeaker B: Why don't you want to use any of this?\nSpeaker D: You know, one of these ones.\nSpeaker D: That's what I've started doing.\nSpeaker D: Metronomy, publicity.\nSpeaker B: How did you see the van go?\nSpeaker A: Oh, geez.\nSpeaker D: But I think the argument should be...\nSpeaker D:...could be made that despite the fact that this is not the most metaphorical domain, because people interacting with HEI systems try to be straightforward and less lyrical.\nSpeaker D: Constructural still is, you know, completely key in terms of finding out any of these things.\nSpeaker F: That's a reasonable point.\nSpeaker D: This domain, you're going to get less metaphor in both sides.\nSpeaker D: With a student, I looked at the entire database that we have on Heidelberg for cases of autonomy, hardly anything.\nSpeaker D: So not even in descriptions, did we find anything relevant?\nNone: I have to go.\nSpeaker D: But okay, this is just something we'll see.\nSpeaker D: And deal with.\nSpeaker D: I mean, maybe the where is something questioned as a whole?\nSpeaker D: It can be construed as a location versus instructional request.\nSpeaker D: So if we not talk about the direction.\nSpeaker D: Location versus what?\nSpeaker B: Instruction.\nSpeaker B: Oh, directions.\nSpeaker B: Yeah.\nSpeaker B: Oh, I thought that was...\nSpeaker B: Dimmalintry did as an example of construal.\nSpeaker D: Yeah, but then, you know, on the lexical level.\nSpeaker B: This sort of one level higher.\nSpeaker D: Oh, you want a lexical example.\nSpeaker F: But I don't need it.\nSpeaker A: Yeah.\nSpeaker D: Also, it would be nice to get ultimately to get a nice mental space example.\nSpeaker D: So, even temporal references are just the spatial domain of rare.\nSpeaker F: But it's easy to make up plausible ones.\nSpeaker D: When you're getting information on objects.\nSpeaker D: Right.\nNone: Yeah.\nSpeaker D: Yeah.\nNone: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker D: Maybe we can include that also in our second data run.\nSpeaker D: We can show people pictures of objects and then have them ask the system about the objects and engage a conversation on the history and the art and the architecture and so forth.\nNone: Okay.\nSpeaker F: So, why don't we give you feedback on what you're talking about?\nNone: I wish you were a true person.\nNone: I'm so sorry.\nSpeaker C: For some reason when you said feedback electronically, I thought of that. You ever see the symptoms where the family's got the buzzers and they buzz each other when they don't like what the other one is saying.\nSpeaker B: That's the first one, I think.\nSpeaker C: It was a very early one.\nSpeaker C: I don't know.\nSpeaker C: Okay.\nSpeaker D: First one.\nSpeaker D: I'm trying to play break the road record for different occasions, for being at Milan Airport.\nSpeaker F: Yeah.\nNone: I didn't see that.\nNone: Yeah.\nSpeaker D: I don't know.\nSpeaker F: I don't know.\nSpeaker B: I don't know.\nSpeaker C: I don't know.\nSpeaker B: I don't know.\nSpeaker F: I don't know.\nSpeaker A: I don't know.\nNone: I don't know.\nSpeaker F: I don't know.\nNone: I don't know.\n"}, {"meeting_id": "ES2010b", "summary": "This was the functional design meeting, where the team discussed what functions should be included in the remote and how they should be carried out. User Interface gave a presentation on the design of the remote, offering a complex version and a simple one. Marketing followed with a presentation on the functional requirements of the remote, pointing out that the remote should be both high-tech looking and user-friendly, preferably with fewer buttons, a lost-and-found function and speech recognition. Industrial Designer then presented the working design for the remote, including its power source, chips and the component layout. The team then went into a further in-depth discussion, dealing with the specific points raised in the above presentations. After everyone had shared their ideas, Project Manager closed the meeting by distributing new tasks among the team members for them to work on until their next meeting.", "dialogue": "None: I don't know.\nSpeaker B: I'm going to go.\nSpeaker D: Was the questioner supposed to be filled in before or after the matter?\nSpeaker D: We just got a third one.\nSpeaker D: Do you want us to do it really fast or wait?\nNone: Okay.\nNone: Oh, yeah.\nNone: That was quick.\nNone: Okay.\nNone: Okay.\nNone: Oh, there we go.\nSpeaker D: Okay.\nSpeaker D: Welcome to our second meeting.\nSpeaker D: This is the functional design meeting.\nSpeaker D: And I hope you all had a good individual working time.\nSpeaker D: Okay.\nSpeaker D: Let's get started.\nSpeaker D: Okay.\nSpeaker D: Here's the agenda for the meeting.\nSpeaker D: After the opening, I am going to fulfill the role of secretary.\nSpeaker D: Take the meeting minutes.\nSpeaker D: Then we're going to have three presentations, one from each of you.\nSpeaker D: Then we're going to discuss some new project requirements.\nSpeaker D: Going to come to a decision on the functions of their mode control.\nSpeaker D: And then we're going to close up the meeting.\nSpeaker D: And we're going to do this all in about 40 minutes.\nSpeaker D: Okay.\nSpeaker D: First, I want to discuss the goals of this meeting.\nSpeaker D: First, we need to determine the user requirements.\nSpeaker D: And the question that we can ask ourselves is what needs and desires are to be fulfilled by this remote control.\nSpeaker D: And then we're going to determine the technical functions.\nSpeaker D: What is the effect of the apparatus?\nSpeaker D: What actually is it supposed to do?\nSpeaker D: What do people pick up a remote and use it for?\nSpeaker D: And then lastly, we're going to determine its working design.\nSpeaker D: How exactly will it perform its functions?\nSpeaker D: That's the whole technical side of.\nSpeaker D: I'll just give you a minute because it looks like you're making some notes.\nSpeaker D: Okay.\nSpeaker D: Oh, well let's go ahead and...\nSpeaker D: I'm going to go back.\nSpeaker D: So what I want to do right now is hear from all three of you on your research that you just did.\nSpeaker D: Who would like to start us off?\nSpeaker B: On the line going first.\nSpeaker D: Okay.\nSpeaker D: Do you have a PowerPoint or no?\nSpeaker B: Yeah, it should be in the...\nSpeaker B: I'm going to do our PowerPoint now.\nSpeaker B: You can do it yourself actually.\nSpeaker E: Oh, did you send it?\nSpeaker E: Save it in the project documents?\nSpeaker A: Yeah.\nNone: Okay.\nNone: This one?\nNone: Sure.\nNone: Okay.\nNone: Yeah.\nNone: Okay.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: Great.\nSpeaker A: Okay.\nSpeaker D: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: Okay.\nNone: Okay.\nSpeaker E: Okay.\nNone: Great.\nSpeaker B: Oh, the function of a remote control as we've been informed is basically to send messages to the television side.\nSpeaker B: For example, switch it on, switch it off, go to this channel, go to channel nine, turn the volume up, etc.\nSpeaker B: So, some of the considerations is just, for example, what it needs to include is the numbers, you know, zero to nine, so you can move to channel.\nSpeaker B: The power button on slash off.\nSpeaker B: The channel going up and down, volume going up and down, and then mute.\nSpeaker B: And mute function.\nSpeaker B: And then functions for VHS DVD, for example, play, rewind, fast forward stop pause, enter, and enter would be for like, you know, the menu.\nSpeaker B: And then other menus for DB as well as TV, whether that means like we can go and decide the brightness of the screen, things like that, all the more complicated functions of menus.\nSpeaker B: And we can decide if that's what we want.\nSpeaker B: If we want to include that on the remote, or if that's something that would stay on the TV itself, for example.\nSpeaker B: Okay.\nSpeaker B: These are two examples, and you can see on the left, it's got a lot more buttons, and I don't know if you can read it, but it says step, go to freeze, slow repeat program mute.\nSpeaker B: And so there's just some of the buttons, and so it gives you an idea, one example, and then on the right, it's a lot more simpler, it's got volume, it's got the play, like circle set, which is play rewind, but it's also what is fast forward is also like next on a menu.\nSpeaker B: So you have functions that are duplicating.\nSpeaker B: And you have a mute button in the numbers, and the eject and the power button, so that gives you two different kinds of more complex and more simple version.\nSpeaker B: And then lastly, it's just the questions that we want to consider, like what functions do we want it to include, and how simple complex it should be, and what functions it needs to complete.\nSpeaker B: What are needed to complete the installation process, because it's something that also has to be considered, and it's going to be hopefully a one time thing when you set it up should be set to go, but we have to include the functions that can allow it to set up in the first place.\nSpeaker B: Okay, so that's it.\nSpeaker D: All right.\nSpeaker D: Very good presentation.\nSpeaker D: Thank you.\nSpeaker D: Speak with such authority on the menu.\nSpeaker D: Okay, who would like to follow that one up?\nSpeaker D: Yeah, that we just go. Okay.\nSpeaker D: Do you want me to run it or you want to run it? Okay.\nSpeaker D: Do you do functional requirements?\nSpeaker D: Yes.\nSpeaker D: All right.\nSpeaker D: Now we have Courtney with the functional requirements.\nSpeaker C: Yes. Okay, so we tested 100 subjects in our lab, and we just we watched them, and we also made them fill out a questionnaire and we found that the users are not typically happy with current remote controls.\nSpeaker C: 75% think they're ugly.\nSpeaker C: 80% want they are willing to spend more, which is good news for us.\nSpeaker C: If we make it look fancier.\nSpeaker C: And basically we just need something that really I mean there's some other points up there, but they need to be snazzy and but yet simple. So that's really what we need to do.\nSpeaker C: And we need we need it to be simple yet.\nSpeaker C: It needs to be high-tech looking.\nSpeaker C: So that like they like I guess use the buttons a lot.\nSpeaker C: Oh, yeah.\nSpeaker C: I don't know. It's from my research.\nSpeaker C: Okay, my team wasn't very clear. Oh, sorry.\nSpeaker D: What do you mean by the current remote controls not match well the operating behavior of the user like I think it's like the engineering versus user.\nSpeaker C: Okay.\nSpeaker C: Whereas like the engineering she showed that the engineering ones are more complex and users don't really need all of the buttons that are contained over because they only use 10% of the buttons.\nSpeaker C: Yeah.\nSpeaker E: Okay, we use 10% of our brains.\nSpeaker E: Good point.\nSpeaker C: It works.\nSpeaker C: It's a necessary tool.\nSpeaker C: Yeah.\nSpeaker C: Ready for the next five.\nSpeaker C: And so people say that they typically lose it as you yourself know because you probably lose your remote control the time much like any small planes like cell phone.\nSpeaker C: And they we need something simple because most people well 34% say that it's just too much time to learn how to use a new one and we don't want to go.\nSpeaker C: We don't want to vary too far from the normal standard remote.\nSpeaker C: I mean they do need to be able to identify it and or I am not very sure what that is.\nSpeaker C: It's okay.\nSpeaker C: It's very important.\nSpeaker C: It is important for the remote control world.\nSpeaker B: Is that like ergonomics like your hand movement system?\nSpeaker C: It could be.\nSpeaker C: Possibly.\nSpeaker E: Maybe really need to provide more information on what RS I am.\nSpeaker C: Yeah, that's what my website said.\nSpeaker D: I think that's a pretty good guess though.\nSpeaker C: Yeah, I would assume so.\nSpeaker C: I think we're supposed to know it as remote control experts.\nSpeaker C: But also so the channel the volume and the power buttons are most important on our company website.\nSpeaker C: You can find like the specific statistics concerning to how much each button is used.\nSpeaker C: But those are the definitely the top ones.\nSpeaker C: Okay.\nSpeaker C: Next slide.\nSpeaker C: Yes.\nSpeaker C: And so personally I think that we need a modern eye catching design but it really needs to be simple.\nSpeaker C: So saying from your site, your presentation, the engineering versus the user specified or modes.\nSpeaker C: I think that we should go with something that's more user friendly where the engineering ones, the boxes tend to make it look more complicated than it really is.\nSpeaker C: The functionality of the product really needs to be considered as to like what type of buttons do we really need on it.\nSpeaker C: And it needs to be open to a wide range of consumers.\nSpeaker C: So even though we need a small number of buttons, we also need to take in like our most people going to be using it for a DVD player or a TVO.\nSpeaker C: So what exactly are we using it for as well as the age range.\nSpeaker C: So we need a hip but not a corny marketing scheme for promoting our product.\nSpeaker C: And also we found our team found that speech recognition is it's like an up and coming thing.\nSpeaker C: They really consumers are really interested in it.\nSpeaker C: And since our findings found that people are willing to pay more money for a remote for it to be more high class.\nSpeaker D: Just to clarify by speech recognition you mean they would say channel 5 and I guess so.\nSpeaker B: Okay.\nSpeaker B: I had to just say where are you?\nSpeaker B: I guess we can interpret it.\nSpeaker C: We can just try out different types of speech recognition with an R.\nSpeaker D: Didn't our rival companies manufacture a remote that you would press the button on TV and it would promote with beeps so that if you ever lost it.\nSpeaker D: We can just use a remote phone.\nSpeaker B: Oh yeah.\nSpeaker D: That's true.\nSpeaker D: We could definitely include that if we wanted to.\nSpeaker D: If it's within our race.\nSpeaker D: Okay.\nSpeaker D: Are we ready for our last presentation?\nSpeaker E: Yeah, I'm just trying to move it.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: I think it should be there.\nSpeaker E: Working design.\nSpeaker E: Okay.\nSpeaker E: I didn't get a chance to complete this one because some of the tools that I was giving more frustrate.\nSpeaker E: Oh, that's fine.\nSpeaker E: Okay.\nSpeaker E: So.\nSpeaker E: Method.\nSpeaker E: Method of our design.\nSpeaker E: I think I just started listing some of the things that we actually need to put into this.\nSpeaker E: We need a power source.\nSpeaker E: We're going to need a smart chip if we're going to make it multifunctional.\nSpeaker E: Extra functions will probably need an additional chip either that or this smart chip will have to be extremely smart.\nSpeaker C: What exactly is a smart chip?\nSpeaker E: Usually a smart chip is just a chip that's been programmed and designed so that it can complete a fair range of functions.\nSpeaker B: Well, how much extra would the additional chip be?\nSpeaker B: Is that going to push us over our production costs?\nSpeaker E: I wouldn't think so because we could probably get it from bulk from a newer company and they tend to sell their chips pretty cheap.\nSpeaker E: Okay.\nSpeaker E: Yep.\nSpeaker E: Nothing here.\nSpeaker E: That's okay.\nSpeaker E: Power source.\nSpeaker E: I figured batteries because they're easily available.\nSpeaker E: Typically, remote has either two double A's or four triple A's, sometimes three.\nSpeaker E: It really kind of depends on the size of the actual remote itself.\nSpeaker E: A large on off button demographically, we're moving towards an older generation of people, so a large on off button would probably be good.\nSpeaker E: Selection button for various entertainment devices, so you want something that will permit you to select the DVD player or the TV or the stereo system.\nSpeaker E: A smart chip that permits, sorry, universal application.\nSpeaker E: Again, something that will allow us to skip over between devices.\nSpeaker E: That's kind of it.\nSpeaker E: This is my 52nd design.\nSpeaker E: Power source over here.\nSpeaker E: We're going to have a switch, obviously, between the power source and the rest of it.\nSpeaker E: You're going to need the switch.\nSpeaker E: Extra bulb could just be for flashiness.\nSpeaker E: Subcomponent, which would be like a way of diverting the power to different parts of the device.\nSpeaker E: The chip and, of course, the infrared bulb, so it can communicate with the various devices that it needs to talk to you.\nSpeaker C: So what exactly are you looking at?\nSpeaker C: Is this like the front of the...\nSpeaker E: This is just like a rough schematic.\nSpeaker D: So this is the internal workings.\nSpeaker D: So the red would be the front of the remote, though.\nSpeaker B: Yeah.\nSpeaker B: That's going to be what's communicating with the TV, but the other bulb, I think, is good just to indicate I'm doing something sort of like a...\nSpeaker E: We don't have to stare at that infrared because, you know, when the battery starts dying on your remote, currently you have to actually stare at that bulb and go, okay, when I push this button, is it working?\nSpeaker E: We can skip that whole thing, right?\nSpeaker E: Yeah.\nSpeaker E: So you can put it in the dark.\nSpeaker E: Yeah, that's good.\nSpeaker D: We should make a glow in the dark.\nSpeaker D: Yeah, definitely.\nSpeaker D: Okay, next.\nSpeaker D: Yeah, that's it.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Anything you want to add for personal preferences, though?\nSpeaker D: You said already that we need a large on-off button, you think.\nSpeaker E: I think that that's a good idea because, you know, that's one of the most important buttons.\nSpeaker B: Well, should it be larger buttons in general?\nSpeaker B: You know, like the examples that I had, they're quite small, so should we try and go for something that has larger buttons?\nSpeaker C: I think we should.\nSpeaker C: Like, I think that would be an, like, for the design, sorry.\nSpeaker C: I think we should definitely go with buttons that don't look like a normal remote because most remotes have small square buttons.\nSpeaker C: I think we should do something like bigger and round.\nSpeaker C: Like, almost.\nSpeaker D: Okay.\nSpeaker D: Okay, let's talk about all of our.\nSpeaker D: We'll come to a decision later about all the components that we need to include.\nSpeaker D: Let's wrap up this one and I'm going to go back to my PowerPoint because we need to discuss the new project requirements, which you might have already seen flashed up on the screen a bit earlier.\nSpeaker D: Wait.\nSpeaker D: Come back.\nSpeaker D: All right.\nSpeaker D: Sorry.\nSpeaker D: Let's go through this.\nSpeaker D: All right.\nSpeaker D: Here we go.\nSpeaker D: New product requirements.\nSpeaker D: First, it's only going to be a TV remote.\nSpeaker D: We're trying not to over-complicate things.\nSpeaker D: So no DVD, no Tivo, no stereo.\nSpeaker D: It's not going to be multifunctional.\nSpeaker D: Hey, and we need to promote our company more.\nSpeaker D: So we need to somehow include our color and our company slogan on the remote.\nSpeaker D: We're trying to get our name out there in the world.\nSpeaker D: Okay.\nSpeaker D: And you know what teletext is?\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: In the States, we don't have it, but it's like they just have this channel where it just has news and weather kind of sports.\nSpeaker D: It's very bland looking.\nSpeaker D: It's just text on the screen.\nSpeaker A: Not even black with text.\nSpeaker D: Like running along the bottom.\nSpeaker E: Yeah.\nSpeaker E: You can also get the kind of the TV guide.\nSpeaker C: Wait, is it like the weather channel where it's got like the ticker running on the bottom?\nSpeaker B: It's the entire screen.\nSpeaker E: It's the entire screen.\nSpeaker E: It's just running information.\nSpeaker E: You can see me.\nSpeaker B: So anyway, you can pick the news, entertainment.\nSpeaker D: So it's like a separate channel from like what I'm watching.\nSpeaker D: But it's becoming outdated now because of the internet.\nSpeaker D: Nobody needs to go to the teletext channel to check the news.\nSpeaker D: And we have 24 hour news channels now too.\nSpeaker D: So those are our new product requirements.\nSpeaker D: Okay.\nSpeaker E: So do we have to include the company color within that?\nSpeaker D: Yes.\nSpeaker D: It's part of the logo.\nSpeaker D: Okay.\nSpeaker B: Company color being yellow.\nSpeaker D: What we're going to do right now is come to some decisions definitive that we can all agree on about the target group and the functions and just definite things that we need to do.\nSpeaker D: And then we'll close up the meeting.\nSpeaker D: So.\nSpeaker D: All right.\nSpeaker A: Okay.\nSpeaker D: So our target group is you mentioned older people.\nSpeaker D: Would it just be universal for everyone you think because I think even if something has large buttons as long as they're not childishly large, like even technically non technically challenged people are going to use it.\nSpeaker D: I mean, they want something user friendly.\nSpeaker E: So well, even if we kept the regular standard size of remote, if we reduce the buttons down to the ones that people are saying that they use the most often and a couple extra, because they're saying they only use 10% of them.\nSpeaker E: Yeah.\nSpeaker E: Then we should be able to accommodate fairly decent sized buttons.\nSpeaker D: Okay.\nSpeaker D: So we want for our target group, would we say, I mean, young and old, all age ranges, all not kids, obviously, right?\nSpeaker C: No, kids need to know how to use a remote.\nSpeaker C: Most of them will into a picture between Disney channel, Cartoon Network.\nSpeaker D: Okay.\nSpeaker D: So we're going to go anywhere.\nSpeaker D: Yeah, I think we need to adult in the age range.\nSpeaker D: What about technical specifications, like how, how technically literate are these people who are going to be using our vote.\nSpeaker C: I would say we should say dumber than the average person.\nSpeaker B: Go for the lowest denominator.\nSpeaker D: Yeah.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: High school.\nSpeaker D: So they need no technical experience to operate.\nSpeaker E: How about little to no, because there's no way that you're going to be able to make it no.\nSpeaker D: Okay.\nSpeaker D: And we also need to determine the specific functions of this just to get it all out on paper.\nSpeaker D: So we said it needs to send messages to the TV, needs to change the channel, turn on and off, just basic simple stuff like this.\nSpeaker B: So if you have something, just say it and we'll add it to my meeting minutes.\nSpeaker C: Yeah.\nSpeaker D: Those are the most important ones.\nSpeaker D: Right.\nSpeaker D: And we want to keep, I'll make a note here that we want to keep the number buttons down.\nSpeaker D: Correct?\nSpeaker D: Because people only use 10%.\nNone: Yeah.\nSpeaker D: Hey, what else?\nSpeaker E: Do we want this thing to be able to be found easily?\nSpeaker D: I think so.\nSpeaker D: What do you.\nSpeaker B: Sure.\nSpeaker B: Yeah.\nSpeaker B: So what?\nSpeaker B: Finding.\nSpeaker D: We need to like,\nSpeaker E: like, underneath the couch.\nSpeaker E: How are we going to accommodate the quick ability to finding?\nSpeaker C: Because people really are looking for our mother's more high tech.\nSpeaker E: What if we gave it a charger and on the charger, just like a phone, like you get a portable phone and it's got a charger.\nSpeaker E: Leave your phone somewhere.\nSpeaker E: You push the button to find it.\nSpeaker E: It finds.\nSpeaker E: Do you think the phone really goes for that though?\nSpeaker E: Because useful for the remote phone.\nSpeaker D: Yeah.\nSpeaker D: Would that add to our cost though?\nSpeaker D: I wonder.\nSpeaker C: I would think so because you'd have to develop a base.\nSpeaker C: Right.\nSpeaker B: Well, if you have a base, you can start putting in a charger and then you have a different kind of battery.\nSpeaker E: Yeah.\nNone: Mmm.\nSpeaker E: Returgical batteries are cheaper usually.\nSpeaker E: Yeah.\nSpeaker D: Okay.\nSpeaker D: I think we can make a decision about that later.\nSpeaker D: We'll still put that as a point that we need to discuss.\nSpeaker D: So that would include battery source.\nSpeaker D: Power source rather.\nSpeaker D: Is it going to have a charger?\nSpeaker D: Or is it going to be run strictly off batteries?\nSpeaker D: And we asked the need to deal with the issue that you mentioned of speech recognition if we want.\nSpeaker B: If we have the speech recognition, then we can start aiming at it like another kind of more handicapped.\nSpeaker B: Disabled demographic.\nSpeaker C: Well, there's the people who desire speech recognition.\nSpeaker C: The different demographics have different desires.\nSpeaker C: I don't know if you guys can.\nSpeaker C: You could, it wouldn't copy onto the things.\nSpeaker C: It's black.\nSpeaker C: But all the different age groups have different desires for speech recognition.\nSpeaker C: So basically older people don't really care.\nSpeaker C: It's really the people 25 to 35.\nSpeaker C: I feel those are the people that really watch a lot of TV though.\nSpeaker C: They're the ones that get addicted to soap operas and sitcoms.\nSpeaker D: And if we introduce it when they're at this stage, they're going to probably always buy a remote that has.\nSpeaker B: Then do you put the voice recognition?\nSpeaker B: Do you put the receiver on the actual television in the base or in the actual remote?\nSpeaker B: Because then you've already got the remote in the hand.\nSpeaker B: Why are you just going to speak to the remote?\nSpeaker B: Right.\nSpeaker B: Whereas if you just speak in general and you don't have to have the remote in your hand and talk at it.\nSpeaker D: Yeah.\nSpeaker D: Well, and the speech recognition could be part of the lost and found device too.\nSpeaker D: If we said find remote, locate remote or something.\nSpeaker D: A certain phrase, then it could be just thrown it out there.\nSpeaker A: Yeah.\nSpeaker D: Well.\nSpeaker B: 15 minutes.\nSpeaker D: Okay.\nSpeaker B: Anything else you want to discuss?\nSpeaker B: Well, do we want to include the numbers like 0-3-9?\nSpeaker B: Can we conceive of leaving them out?\nSpeaker B: Wait on the remote itself?\nSpeaker B: Yeah.\nSpeaker B: Like you have 1-2-3-4-5-6-7-8-9-0.\nSpeaker B: Well, we definitely need those.\nSpeaker B: How do you leave those out?\nSpeaker B: Well, I don't know.\nSpeaker B: I mean, if you can conceive, it's like, well, I don't know.\nSpeaker E: It's just a way of leaving them out.\nSpeaker E: I think people would find that too foreign.\nSpeaker D: Yeah.\nSpeaker D: Yeah, that's true.\nSpeaker D: And also remember that in this day and age, we need, you know, like a hundred button too.\nSpeaker D: I used to have a remote that's not even go up past like 50.\nSpeaker D: So I couldn't, whenever I got cable, I had to get a new TV.\nSpeaker E: So we need to get to satellite.\nSpeaker D: Yes, I get your own remote or digital cable.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: I guess we're going to discuss the project financing later, making sure that we can fit all of the stuff that we want to on our budget.\nSpeaker E: Yeah, because I don't have any material pricing information available to me at the moment.\nSpeaker D: So I don't forget we need to include the color of our company and the logo.\nSpeaker B: The color being yellow?\nSpeaker D: I'm guessing.\nSpeaker D: I feel like the double R would be too garish.\nSpeaker C: We could just have the logo in yellow or maybe a yellow light for the keys.\nSpeaker C: I'll put like stripes.\nSpeaker C: Oh, yeah, yellow.\nSpeaker E: Yeah, yellow could be and it doesn't have to be huge.\nSpeaker B: Yeah, well, if you have like, you have this sort of strip kind of down at the bottom, the base of it, it's just like yellow with the R.\nSpeaker A: Right.\nSpeaker D: So we've simplified, we don't need all those play fast forward, we're wind.\nSpeaker D: Right.\nSpeaker D: You know, no menu buttons.\nSpeaker D: So we've pretty much pared it down to on off volume mute channel up down the numbers.\nSpeaker A: Yeah.\nSpeaker D: Can we go back to, I'm going to look really quickly back at those two examples and see if there's anything, which one was yours technical functions for functional requirement?\nSpeaker C: Yeah.\nSpeaker C: Audio settings and screen settings.\nSpeaker C: We need those like audio settings, mono, stereo, pitch, screen settings like brightness color.\nSpeaker C: Or do we just want that access from television itself?\nSpeaker D: I think that that's fine just through the TV.\nSpeaker D: I mean, how often does the average user need to do that kind of stuff?\nSpeaker B: Well, the other options sort of like down at the bottom, like farther away, you just have this sort of box in set where it's like the buttons that you don't use as much, but occasionally you will use.\nSpeaker C: Yeah, because we definitely need to have buttons for like subtitles and things like that.\nSpeaker C: Because the foreign film market is expanding and stuff.\nSpeaker C: And like on television, like I know living in Los Angeles, it's tons of Spanish network television.\nSpeaker C: And if it has English subtitles, it's definitely helpful.\nSpeaker D: Can we do that off through wind button?\nSpeaker D: There's a menu button that pops up with a menu on the TV that says audio, video, whatever language.\nSpeaker C: So we need up, down and side to side buttons for the menu.\nSpeaker B: Well, that's what you know you can just double up with like the channel or the volume buttons.\nSpeaker B: That's true.\nSpeaker B: Yeah.\nSpeaker B: The channel's just up and down.\nSpeaker C: Okay, yeah.\nSpeaker D: And then something that such as, yeah, the one, you know, the one over there on the left, the engineering center one.\nSpeaker C: Right.\nSpeaker C: Right, right, right.\nSpeaker B: So we just have like add a menu button then for the various things needed, including the middle of the mission.\nSpeaker B: If we have any like settings for voice recognition, though.\nSpeaker B: Included menu.\nSpeaker B: Yep.\nSpeaker C: We just got an idea for design.\nSpeaker D: Okay.\nSpeaker D: Anybody have anything else I'd like to bring up in this meeting?\nSpeaker E: I had something, but I forgot.\nSpeaker D: Okay.\nSpeaker D: Get out of here.\nSpeaker D: Let's go back to the meeting closure then and see what we need to do next.\nSpeaker D: All right.\nSpeaker D: After this meeting, we're going to be sent a questionnaire and summary again, which we need to reply to that email.\nSpeaker D: And then we're going to have lunch break.\nSpeaker D: And after lunch, 30 minutes of individual work time.\nSpeaker D: I'm going to put the minutes, I put the minutes for the first meeting already in the product documents folder.\nSpeaker D: If you'd like to review them and I'm going to type up the minutes for this one as well.\nSpeaker D: Here's what we're each going to do.\nSpeaker D: The ID is going to work on the components concept.\nSpeaker D: UID, the user interface concept, and you're going to do some try and watch it.\nSpeaker D: Hey, specific instructions will be sent to you by your personal coach.\nSpeaker D: And if anybody has anything they would like to add?\nSpeaker D: No?\nSpeaker D: Okay.\nSpeaker D: Well, this meeting is officially over.\nSpeaker D: Thank you all.\n"}, {"meeting_id": "IS1006b", "summary": "The meeting was basically on functional design. First, the marketing talked about the user requirements based on a market survey and suggested making their new product good-looking and equipped with less buttons and a speech recognition system. Then the user interface designer stressed that they should design nothing but a remote control because their principle was to make users' simple desires into simple actions. Therefore, the proposed concept for design included just a few buttons, a screen with a back light and titanium, which was accepted by the group. Finally, when it came to the working design of the remote control, the industrial designer described the product as a portable device with some energy source, an interface, a processor, some infrared source, different encryption codes for different TVs, and speech recogniser. All these components would be within budget. At the end of the meeting, the group also talked about how to enable the users to find the controller.", "dialogue": "Speaker D: So I hope you're ready.\nSpeaker D: For this functional design team, of course.\nSpeaker D: So I will take the minutes.\nSpeaker D: You three are going to do presentation.\nSpeaker D: And we want to know, at the end, to know the new projects requirement.\nSpeaker D: So we need to know the user needs that we want to fulfill.\nSpeaker D: From the technical part, we want to know how it's going to work.\nSpeaker D: And the part...\nSpeaker D: I don't remember.\nSpeaker D: This is not very good.\nSpeaker D: Of course, how to design this stuff.\nSpeaker D: So let's go for the presentation.\nSpeaker F: So first, marketing expert.\nSpeaker D: Okay.\nSpeaker F: So I don't know if I can do that like this.\nSpeaker F: So is being modified, do you want...\nSpeaker F: Yeah, open it only.\nSpeaker F: I hope I saved it.\nSpeaker F: So...\nSpeaker F: Benjung.\nSpeaker F: Yeah.\nSpeaker F: I know, just my name.\nSpeaker F: So as you know, I think you already know me, Sammy Benjio.\nSpeaker F: I'm the expert in marketing and I want to tell you about what people want and like and dislike in remote controls.\nSpeaker F: And I hope this is going to help you to design it correctly.\nSpeaker F: So next, please.\nSpeaker F: Yeah, it is good.\nSpeaker A: The full page presentation.\nSpeaker F: Yeah, maybe in the full page because I spend lots of time doing this presentation.\nSpeaker F: Okay.\nSpeaker F: So basically what I suggest is that instead of deciding ourselves what...\nSpeaker F: That could be and what should be good, the remote control.\nSpeaker F: Let's ask people who are users of remote controls how they feel about the current remote controls, what they like, what they don't like and what they do with them, by the way, because they are supposed to be useful.\nSpeaker F: Don't forget about that.\nSpeaker F: So we've conducted a survey on the use of remote controls and I'd like to show you some of the results we found on this survey.\nSpeaker F: Next, please.\nSpeaker F: Yeah.\nSpeaker F: So basically what we found was that there are several things that the user don't like in remote controls.\nSpeaker F: First of all, they find it very ugly.\nSpeaker F: Current remote controls, as you know, they are the same as this one.\nSpeaker F: They are not nice color, not nice shape, all the same.\nSpeaker F: And they are not good looking.\nSpeaker F: What is interesting is that in fact it seems that people are ready to pay for a nice and fancy looking remote control.\nSpeaker F: So I think we should pretty spend lots of time and effort in that.\nSpeaker F: And the other thing is that the current remote controls are not so easy to use and the current facilities that they offer do not match what people really want to use their remote controls.\nSpeaker F: For instance, we see that the ZAP very often.\nSpeaker F: So I think this is a very important functionality.\nSpeaker F: It should be easy for them to ZAP in one way or another.\nSpeaker F: And most of the buttons on the current remote controls are not used.\nSpeaker F: So I think we should design something where some of the buttons which are those that are used should be easier to see and use than others that only a couple of people are using.\nSpeaker F: Next, please.\nSpeaker F: Now people are very frustrated with their remote controls.\nSpeaker F: For instance, they don't even find it.\nSpeaker F: It's often lost somewhere in your home and nobody knows where it is.\nSpeaker F: Maybe if we have something where we could ask the remote control, please, where are you?\nSpeaker F: Something to like I think phones, some of the phones have some of these kind of functionality.\nSpeaker F: Of course phone you can always phone your phone.\nSpeaker F: You can't phone your remote.\nSpeaker F: But why not?\nSpeaker F: And because of the fact that there are so many buttons in these remote controls that nobody use, in fact they don't even know how to use them.\nSpeaker F: So most of the people say they don't know how to use properly their remote controls.\nSpeaker F: And they are bad for AirAssy but I don't remember what is AirAssy.\nSpeaker F: So I think they are bad.\nSpeaker C: That's not it.\nSpeaker F: AirAssy.\nSpeaker F: Nobody has any idea about that?\nSpeaker F: Well, I'll check with my...\nSpeaker A: Electromagnetic waves or something.\nSpeaker F: I think it's a technical thing which...\nSpeaker A: Because infrared uses some electromagnetic...\nSpeaker A: Electromagnetic.\nSpeaker A: So it seems like a 6%.\nSpeaker F: So a lot of people call concept that we don't know.\nSpeaker F: But we have to take this into account.\nSpeaker F: Every force, you know.\nSpeaker E: Every people really, some of us knows.\nSpeaker F: So anyway, that's for what the biggest frustration of the user and what else do I have?\nSpeaker F: Next slide.\nNone: Oh, yeah.\nSpeaker F: So we've listed a couple of functions that may be used by the user in the current available remote controls.\nSpeaker F: And well, the table has been very nice to read.\nSpeaker F: But what is important is to understand that the power button is not used often because in general you use it only once per session.\nSpeaker F: But it is very relevant.\nSpeaker F: People want to have a power button.\nSpeaker F: Visual selection is often used, very often used, and indeed very relevant.\nSpeaker F: Now I remember what is RSI.\nSpeaker F: It's rapid TBT stress injury.\nSpeaker F: We have to be careful with that word.\nSpeaker F: Anyway, I got into my presentation.\nSpeaker F: So yeah, channel selection is very important, very often used.\nSpeaker F: Volume is not often used, but people want to have control on volume and that makes sense, of course.\nSpeaker F: And then you have things which are very much less often used, like the settings, audio settings, screen settings, even teletext and channel settings.\nSpeaker F: All of them, they are not often used and they are more or less relevant.\nSpeaker F: It seems that people find teletext relevant, even if I personally never use it, but it seems that it's average relevant at least.\nSpeaker D: I have been told that we don't consider teletext that it's out of date because of internet.\nSpeaker F: I can tell you that in a scale between 1 and 10 relevant, not relevant to relevant people score the 6 on this, which is not as these two ones were had I think 10.\nSpeaker F: But if you compare with these ones, I think they scored a 1 or 2, not very relevant.\nSpeaker F: So if there are good reasons not to put teletext, it's okay, but just know that people find it somehow relevant.\nSpeaker F: That's for the main functions I think.\nSpeaker F: And then we can ask ourselves what people don't have that may be useful, for instance, I think next slide.\nSpeaker F: One of the things, the trend that you are probably aware of is the possibility, the eventual possibility of having special recognition in your remote control.\nSpeaker F: So you wouldn't have to tap in your buttons, but just tell your remote control or whatever unit you have, what you want.\nSpeaker F: So we've conducted a survey about whether people would like or not to have this kind of functionality in their remote control.\nSpeaker F: And as we can see, it really depends on the age.\nSpeaker F: Young people, probably because it's a buzzword, find it very relevant.\nSpeaker F: And as the age goes up, the relevance goes down.\nSpeaker F: So now it really depends on the kind of targeting, who are we targeting with this remote control.\nSpeaker F: I think if we are targeting young people, then it's probably something we have to consider.\nSpeaker F: If we are targeting young, very old people, this is something that we don't know why it should be.\nSpeaker F: So now this is of course, depends on that.\nSpeaker F: And I don't have any conclusion, I didn't have time, the meeting was very tight.\nSpeaker F: So that's basically my findings.\nSpeaker F: If you have any question?\nSpeaker D: I think it's good, okay.\nSpeaker D: I can rely on one question.\nSpeaker D: One question.\nSpeaker F: Yeah, one question.\nSpeaker E: Yeah.\nSpeaker E: You're a market expert.\nSpeaker E: So should we aim at young people or not?\nSpeaker F: I think we should aim at the young people.\nSpeaker F: But I think there are those who might be more interested in a new device.\nSpeaker F: In general, the early adopters of a new device are young people less than...\nSpeaker F: Okay.\nSpeaker F: Zental text is useless.\nSpeaker F: Zental text is useless for them, I think.\nSpeaker F: Because they have other means of finding their information.\nSpeaker F: But that's a good point.\nSpeaker F: Okay.\nSpeaker D: Thank you.\nSpeaker D: So now I think it's the turn of the...\nSpeaker D: I'm not sure.\nSpeaker D: Of the technical function?\nSpeaker D: So...\nSpeaker A: Do I think it's you?\nSpeaker A: No?\nSpeaker A: That's my question.\nSpeaker F: What if...\nSpeaker F: So you're a requirement now?\nNone: Yeah.\nSpeaker D: Thank you.\nSpeaker A: I have to do working design too.\nSpeaker E: So you...\nSpeaker E: That's just what number?\nNone: Three.\nSpeaker A: Yes.\nSpeaker E: So my name is Mark Twight and I'm responsible for user interface design.\nSpeaker E: However, the product manager asked me to give you some presentation about technical functions design.\nSpeaker E: I assume more an artist.\nSpeaker E: That's going to be less technical functions, but more user interface and current intentions and everything which is linked with this.\nSpeaker E: So next slide please.\nSpeaker E: And the general method which seems to be very useful for our task is not to forget about the outcome razor.\nSpeaker E: We should never complicate things too much.\nSpeaker E: We should only make a remote control.\nSpeaker E: Nothing more.\nSpeaker E: Nothing more than this.\nSpeaker E: Just remote control.\nSpeaker E: Because current remote controls, they are never easy enough to use.\nSpeaker E: So make it click this.\nSpeaker E: So here is user mode control.\nSpeaker E: It's quite a standard one, but it's not from a TV.\nSpeaker E: It's from a much easier device like air conditioning or something.\nSpeaker E: But you know, we can use it for a TV easily.\nSpeaker E: Any buttons we need is on off volume channels and maybe some options or some snails.\nSpeaker E: And please make a click compared to this one.\nSpeaker E: Which one would you prefer?\nNone: I guess this.\nSpeaker F: I would say the simplest one as long as there are the, I find the buttons that I need every time I need a button.\nSpeaker E: Sure.\nSpeaker A: Sure.\nSpeaker A: And maybe a middle of like between those two.\nSpeaker E: And our method is going to be provide simple, simple desires into simple actions.\nSpeaker F: Nice.\nSpeaker F: Nice sentence.\nSpeaker E: Fighting.\nSpeaker D: Oh, sorry.\nSpeaker E: Our passion is a style.\nSpeaker E: We should remember that our company boots fashion into electronics.\nSpeaker E: And we should never forget about it.\nSpeaker E: Concept.\nSpeaker E: Simple.\nSpeaker A: Simple.\nSpeaker E: And you lean on this market.\nSpeaker E: Market is off remote controls.\nSpeaker E: You know it better.\nSpeaker E: It's very.\nNone: Wow.\nSpeaker E: It's not an easy field to play, you know.\nSpeaker E: So be simple.\nSpeaker E: For personal preferences, I think that to make a baby proof remote control got to be a titanium.\nSpeaker E: It's a really good style.\nSpeaker E: It's going to look like this.\nSpeaker E: It is unbreakable and it is very universal.\nSpeaker E: We'll have a screen with a backlight.\nSpeaker E: We can change colors and we can put all the options into this screen.\nSpeaker E: We'll need only a few buttons.\nSpeaker E: All the other things can be controlled through the screen.\nSpeaker E: And all these buttons should be easy to find and to click because when you watch a movie and you want to change something, you always try to find a good button and click it.\nSpeaker E: But you should do it by touching it and finding it easily just by touching.\nSpeaker E: So.\nSpeaker E: Press.\nSpeaker E: I would propose this concept for design.\nSpeaker E: Just a few buttons.\nSpeaker E: A screen with a backlight which can change colors.\nSpeaker E: The titanium, I think and what else.\nSpeaker E: I got just very few good ideas.\nSpeaker E: We need power and volume.\nSpeaker E: And let us include two nice features into this device.\nSpeaker E: First, power on and off can be made fully automatic.\nSpeaker E: When you go to the sofa, take your control and point it to the TV.\nSpeaker E: The TV turns on.\nSpeaker F: When does it turn off?\nSpeaker E: When you don't touch the control but you go out of it for enough time.\nSpeaker E: Like you.\nSpeaker F: Sensing sensor machine.\nSpeaker E: That's a question to our technical designer, to engineers.\nSpeaker E: And another nice feature that I would like to implement is the volume control.\nSpeaker E: Suppose you set up some volume and then you move to the other corner of the room and take your control with you.\nSpeaker E: Like you want to change the chair or you want to move to the armchair from the sofa.\nSpeaker E: What do you want to do?\nSpeaker E: Something.\nSpeaker E: And then the volume changes.\nSpeaker E: It's easy to do.\nSpeaker E: You just control the.\nSpeaker E: According to your distance.\nSpeaker F: And the angle maybe if you have a stereo.\nSpeaker F: I'm not sure about the screen.\nSpeaker F: What is the usefulness of the screen?\nSpeaker F: Is it a touch screen by the way?\nSpeaker E: I think it can be just a menu which can be controlled with a left right up down and enter.\nSpeaker F: So it gives instructions but it has to be the backlight somehow.\nSpeaker E: So its main purpose in fact is the backlight.\nSpeaker E: Which change colors which makes it easier to find.\nSpeaker E: And it can respond for your voice.\nSpeaker E: Like it can turn on the light for you just to find it easily.\nSpeaker E: Okay.\nSpeaker E: So basically that's it.\nSpeaker D: I see that you target several applications not only TV but like we talk about universal remote control.\nSpeaker E: It can be easily done.\nSpeaker E: Because you got simple designs.\nSpeaker E: We should put it to simple actions.\nSpeaker E: Let it be universal.\nSpeaker E: So you want to use it for your high-fives system.\nSpeaker E: You want to change tracks and you want to adjust volume.\nSpeaker E: Just a few actions.\nSpeaker E: Few actions for everything.\nSpeaker E: Also rest we will put it into this menu on the screen.\nSpeaker D: Since we are targeting a really soon date for the issue of this remote control.\nSpeaker D: I think we will only concentrate on TV for the moment.\nSpeaker D: Okay.\nSpeaker D: And maybe.\nSpeaker D: Okay.\nSpeaker E: But it's quite universal.\nSpeaker E: You can just extend it to any device.\nSpeaker F: For instance if I want to go directly to channel 25.\nSpeaker F: How would I do that with this?\nSpeaker F: On 25.\nSpeaker F: Yeah.\nSpeaker F: Let's say I'm on channel 8 now.\nSpeaker F: You know these days we have hundreds of channels.\nSpeaker F: That's not so easy to go just next, next, next when you have hundreds of channels.\nSpeaker E: In fact I would propose another solution.\nSpeaker E: Basically you use just four or five channels.\nSpeaker E: Most people yeah.\nSpeaker E: So set up your TV set like channels that you use.\nSpeaker E: They are one, two, three and five.\nSpeaker E: And you will never have to go to a 25 channel.\nSpeaker F: In fact in one remote control that I've seen instead of doing that you could just say these are the channel 3, 28, 48 and 64 are those that I want to be cycled with my next button.\nSpeaker D: Yes.\nSpeaker F: It's the same solution.\nNone: Yeah.\nSpeaker A: But even we can have some LCD display that you can do.\nSpeaker A: You can do this with the five button the number and then go.\nSpeaker F: The thing is that as I said in my presentation people really do like to zap.\nSpeaker F: So even if they are only watching four or five channels I think they want to zap out of the 100 channels just because this is one kind of thing they do zap in.\nSpeaker D: And zap it's only next.\nSpeaker D: And it's only next.\nSpeaker F: So but you have to go to this button here.\nSpeaker E: Okay.\nSpeaker E: Next, next.\nSpeaker E: Or say this can be back.\nSpeaker A: But otherwise like we can put some display on numbers and then they can just press.\nSpeaker A: So it was two five they just press two and five and then.\nSpeaker D: Maybe we can make different modes for each button.\nSpeaker D: You can change mode zapping mode or.\nSpeaker D: All right.\nSpeaker D: Curant.\nSpeaker A: We were since we are focusing only on TV remote control.\nSpeaker A: So we can have more functions for TV.\nSpeaker A: If you want to go for universal then we have to limit for functions for each of our devices.\nSpeaker A: Okay.\nSpeaker E: Could we carry out some research if you really need this?\nSpeaker E: Like how many people really need to go to channel number 25 and then to 64?\nSpeaker F: Oh I could have a look at that.\nSpeaker F: Maybe.\nSpeaker F: I'll check in my department if there's someone special is in them.\nSpeaker F: Although I don't know.\nSpeaker E: Thanks for your attention.\nSpeaker D: So now the technical aspects of this new device.\nNone: Sorry.\nSpeaker A: I see you know I'm Mr. Ramaro.\nSpeaker A: I'm an expert in industrial design of all electronic devices.\nSpeaker A: I previously devised many like digital calculators and electronic calculators.\nSpeaker A: So now I briefly describe the working design of our remote control.\nSpeaker A: Well as you know the basic function of remote control is sending some message to the device like TV or VCR or DVD player etc.\nSpeaker A: So we will have a portable device which will send message to the main device like TV.\nSpeaker A: So we need to have some energy source to do what to do the functions what we want on this portable device and usually these so to do these functions we need an interface.\nSpeaker A: It's basically some kind of pressing buttons or keys or like moving jack or something like that and then these messages these key buttons can be transferred into some kind of message and then it will processed by the chip and then it will generate some information to the main device.\nSpeaker A: It's generally in the form of infrared or some kind of sensor information.\nSpeaker A: Then we will have the main control in the main device to do the particular action what we want.\nSpeaker A: So basically we need since we are focusing on our interface device that's our remote control we need few components mainly the energy source like the battery and then we have user interface like the keypad and new buttons we want and then we have some chip it's mainly digital signal processing chip because since we are involved doing mostly digital devices we have to have some kind of processor which take care of all these functions and put it in some digital format and then we will have the infrared LED source which sends the information to the main device then we will have switch in our main device to do particular operations and we have to do different codes for different TVs so some TVs will have different encryption codes for doing channel changing and these things.\nSpeaker D: To make it quite a universal device.\nSpeaker A: Yeah because people don't use one particular band so at least we have more than five brands which are really good so we need to check their specifications and do our encryption that's passing information to the TV device so we need to have particular encryption codes.\nSpeaker A: Then components so we have the main energy source and then we will have some buttons and then we will have infrared source and then we have some inside some chip in the device since I don't have much time so I didn't put the connection to all these components and since I also want to know feedback from our marketing expert and user interface so if you want to add some more components we can incorporate them.\nSpeaker D: From the discussion we had can you make it on the whiteboard?\nSpeaker A: Yeah I'm sure because since our user interface engineers suggesting speech recognition and also marketing expert found that speech recognition is really handy so we can have another simple speech recognition on our DSP chip since we have some kind of energy this is our just normal battery so this battery once you switch on it will take power and we can have some speech recognition and in our generally small digital signal processing chip so that and we will put the small simple speech recognition and we can also try and this speech recognition for particular user so you just yeah such we just use simple to contract.\nSpeaker A: But very good to sell yeah even if you can find even a simple mobile device like any mobile brand you can have this voice delers or they got all these young people who would\nSpeaker F: love to say that this remote control only works for them. You cannot use my remote control because it's targeted to me.\nSpeaker D: Whatever.\nSpeaker D: What about the price of this component?\nSpeaker A: Maybe we can make within five euros and even less than that because we want to have billions millions and in bulk so we can make really simple and we want to make really simple device because we have only very few words like like powers which on or some like then we will have something like this we'll have volume and then we'll have particular channel so users can.\nSpeaker F: So you will just be able to say please can you pop up the volume or it will be something like volume.\nSpeaker A: Yeah up.\nSpeaker A: The user can use any kind of center but they should have this prompted words volume and\nSpeaker D: then. With keywords.\nSpeaker A: The volume decrease or increase so we try to only recognize both words and because we can't really say user to say same word and then it become more mechanical.\nSpeaker A: And then we can have channel they can say okay I want it because we don't know like users have different programs I mean they don't really follow same channel structure so we just want channel number we don't want like BBC or CNN or of course it has to be complicated.\nSpeaker A: So we'll have only this three main basic.\nSpeaker A: Anyway volume is not really speech recognition problem it will be take care of our main.\nSpeaker E: No it's a conceptual question because now I see the picture in front of my eyes like a user taking his remote control and shouting into it volume up volume up.\nSpeaker E: But I think he's coming you know he's really annoyed with this.\nSpeaker E: Down up down.\nSpeaker F: First of all I think this is not a functionality that is going to be instead of using the buttons.\nSpeaker F: No not only speech.\nSpeaker F: It's on top of using the button.\nSpeaker E: Okay because it's budget like 12 euros.\nSpeaker E: Well I don't know.\nSpeaker A: Actually we can have one switch to like switch on and off this DSP processor and this really parallel to the suppose like here we have our main chip which controls power volume and this buttons and this DSPs again these two have some interaction like suppose people use DSP then it particularly sends some information to the chip like in some form like volume increase or like this key.\nSpeaker A: It need not be like very expensive because since we are only focusing on TV remote controls\nSpeaker D: and we have only few things here. Okay.\nSpeaker E: What you consider is a gesture recognition like if I want to put volume up I like to I take my remote control to do something like volume up and volume down.\nSpeaker A: At this point we didn't consider because it's very expensive.\nSpeaker A: It's very expensive because our target is only like 12.5 euros.\nSpeaker D: What about the idea of automatic on off on?\nSpeaker A: And what do you think?\nSpeaker A: Yeah even an automatic on off is also a bit problematic because it's different criteria for different people like so suppose people are really they just they don't touch the remote and you don't know how much time you need to switch on or switch off and should\nSpeaker F: we target the user personalize the remote control so in a given room there might be more than one remote control.\nSpeaker F: We would have each one.\nSpeaker F: Yeah and with our own personal settings.\nSpeaker A: Yeah that can be possible especially for power settings so user can say okay suppose they're watching a tennis match or something then they can say okay after one hour.\nSpeaker A: Wouldn't that make a...\nSpeaker D: They can make arguments.\nSpeaker C: Yeah of course.\nSpeaker F: I want to do the trick.\nSpeaker F: That's the problem we will sell more.\nSpeaker F: Yeah we can have...\nSpeaker C: And we can increase this.\nSpeaker C: Yeah exactly.\nSpeaker C: You can buy.\nSpeaker C: Well that's more.\nSpeaker E: Yeah let's sell more.\nSpeaker E: Okay.\nSpeaker D: Okay you have something else to say.\nSpeaker A: Ah not very much.\nSpeaker A: No.\nSpeaker D: Okay thanks.\nNone: Okay thanks.\nNone: Thank you.\nNone: So.\nSpeaker C: Can you just...\nSpeaker C: I love it.\nSpeaker C: Yeah I think.\nSpeaker C: Okay we're just on time.\nSpeaker D: So we are now going to have the lunch break.\nSpeaker D: Great.\nSpeaker D: Then we will work again for 30 minutes in the video work.\nSpeaker D: Then we will meet again for the next meeting.\nSpeaker D: And in this one we want to be more focused on the individual actions.\nSpeaker D: No no.\nSpeaker D: On the components.\nSpeaker D: So you will focus on the component concept.\nSpeaker D: Of course the UID mark will be focused on the user interface concept and our market expert Sami Bencho on the train watching.\nSpeaker D: So of course like before specific instruction will be sent to you by your personal coach.\nSpeaker D: Oh I think that's all.\nSpeaker D: And we have maybe we have to say only for TV.\nSpeaker D: Not a red text.\nSpeaker D: I think automatic on off control it's not possible.\nSpeaker A: It's a current phrase.\nSpeaker D: Yeah maybe in the next step if we make it work.\nSpeaker A: But speech recognizer can be possible.\nSpeaker D: Yeah you can think about that.\nSpeaker D: And do you see something else?\nSpeaker E: No.\nSpeaker E: Should it be equipped with speakers?\nSpeaker F: Speakers.\nSpeaker E: Like you want to find it.\nSpeaker D: You shout.\nSpeaker E: Yeah.\nSpeaker E: And the answers I'm here.\nSpeaker F: It just beeps.\nSpeaker F: That would be a good fit.\nSpeaker D: Or maybe you want to find him.\nSpeaker F: Yeah, that's expensive.\nSpeaker A: Especially the power it really consumes because it should be all the time on.\nSpeaker F: I heard of devices where you just whistle them.\nSpeaker F: And because of the frequency they just answered to that.\nSpeaker F: You can't whistle.\nSpeaker F: Or a clap.\nSpeaker F: Clap.\nSpeaker C: Clap.\nSpeaker F: Clap is good.\nSpeaker C: I think.\nSpeaker F: It's a good fit.\nSpeaker F: We have only one hand.\nSpeaker A: I think it's good.\nSpeaker F: He's another target.\nSpeaker A: Oh, okay.\nSpeaker A: But it's a good feature I guess we need to think about how to incorporate.\nSpeaker E: Just don't interfere with other devices.\nSpeaker E: Because always people do that their lights are turning on with clapping.\nSpeaker F: That already exists.\nSpeaker F: Yeah, I thought it was my home.\nSpeaker F: Yeah, you do have.\nSpeaker F: Wow.\nSpeaker D: You're trendy.\nSpeaker D: So let's think about it.\nSpeaker D: I think that could be in the component concept.\nSpeaker F: Yes.\nSpeaker D: Okay.\nSpeaker D: Good.\nSpeaker D: Right.\nSpeaker D: Yeah.\nSpeaker A: Thank you.\nSpeaker A: Thank you very much.\nSpeaker D: Thank you.\nNone: Thank you.\n"}, {"meeting_id": "Bmr014", "summary": "The group discussed recording equipment issues, including the purchase of two additional headsets and the prospect of getting a new base station and a set of wireless microphones to replace those wired microphones currently in use. Speaker fe008 presented the current status on transcriptions, and explained procedures for cleaning up transcripts and ensuring they conform with set conventions. Speaker mn014 briefly described his efforts to normalize loudness levels across speech channels to distinguish between foreground and background speech. Finally, the group discussed legal and procedural issues concerning the provision of transcripts to meeting participants for 'bleeping out' any sections of speech they want excluded from the Meeting Recorder database.", "dialogue": "None: All right.\nSpeaker E: Are we going?\nSpeaker E: It must be February 15th.\nSpeaker H: I think the date's written in there.\nSpeaker H: And actually if everyone could cross out the R9 next session and write MR11.\nSpeaker E: Yeah, we didn't have a front-end meeting today.\nSpeaker H: And let's remember also to make sure that once you get smart as unread, unused.\nSpeaker H: MR11.\nSpeaker E: Okay, so.\nSpeaker H: Lots of clicking on sure as I'm trying to get this to work.\nSpeaker E: Agenda.\nSpeaker E: Agenda items today.\nSpeaker H: I want to talk a little bit about getting how we're going to get people to edit bleeps parts of the meeting that they don't want to include.\nSpeaker H: What I've done so far and I want to get some opinions on how to finish it up.\nSpeaker H: Okay.\nSpeaker C: I'm just going to ask about you monitoring some of the particular topics.\nSpeaker C: In listening to some of these meetings for our recorded, there are sometimes big spikes\nSpeaker G: on particular things. I don't know what the electronics is.\nSpeaker H: Yeah, well I think it could be a number of things.\nSpeaker H: It could be touching and fiddling and the other thing is that it could, the fact that it's on a wired mic is suspicious, it might be a connector.\nSpeaker C: Well maybe then we don't really have to talk about that.\nSpeaker G: You could try and experiment and say, okay, I'm about to test for spikes and then wiggle the thing there.\nSpeaker G: Right, and when they go to transcribe it, it could ask them to come get you.\nSpeaker G: Come get me when you transcribe this and see if there's any.\nSpeaker E: I'm just...\nSpeaker E: I mean, were this a professional audio recording?\nSpeaker E: Well what you would do is, in testing it is you would actually do all this wood letting them make sure that things are not giving that kind of performance that they are than to campus use.\nSpeaker E: Right, so let's see, I guess I would like to have a discussion about where we are on recording transcription, you know, basically where we are on the corpus.\nSpeaker E: And then the other thing which I would like to talk about, which is a real matter question, is maybe I'll start with that actually, Andreas brought up the fact that he would kind of like to know if possible what we're going to be talking about because he's sort of peripherally involved at this point and if there's going to be a topic about, a discussion about something that he strongly cares about then he would come.\nSpeaker E: And I think part of his motivation of this is he's trying to help us out because of the fact that the meetings are tending to become reasonably large now and days when everybody shows up and so he figures he could help that out by not showing up if it's a meeting that he's...\nSpeaker E: So in order, I think that this is a wish on his part, it's actually going to be hard because it seems like a lot of times things come up that are unanticipated.\nSpeaker E: But we could try anyway, do another try at coming up with the agenda at some point before the meeting.\nSpeaker H: Well, maybe it would be a good idea for one of us to, like on Wednesday or Tuesday, send out a reminder for people to send in agenda items.\nSpeaker E: Okay, you want to alter it to that? Sure.\nSpeaker E: Okay, so we'll send out agenda requests.\nSpeaker H: That'll be... but that'll help air-brainer or will not help a lot.\nSpeaker E: Yeah, I have to tell you for the admin meeting that we have, Lyle, does that every time before the admin meeting and she ends up getting the agenda requests 10 minutes before the meeting.\nSpeaker E: But we can try. Maybe it'll work.\nSpeaker H: Yeah, maybe. Where are things have happened?\nSpeaker C: I'm wondering if he were to specify particular topics that we'd be able to meet that requested business.\nSpeaker G: I would also guess that as we get more into processing the data and things like that, there'll be more things of interest to him.\nSpeaker E: Yeah, actually, this maybe brings up another topic which is...\nSpeaker E: So we're talking about that topic. The other topic I was thinking of was the status on microphones and channels.\nSpeaker H: Yeah, actually, I was going to say we need to talk about that too.\nSpeaker H: What line do we do that?\nSpeaker H: Okay, the new microphones, the two new ones are in.\nSpeaker H: And they are being assembled as we speak, I hope.\nSpeaker H: And I didn't bring my card today, so I'm going to pick them up tomorrow.\nSpeaker H: And then the other question I was thinking about is, well, a couple of things.\nSpeaker H: First of all, if the other headsets are a lot more comfortable, we should probably just go ahead and get them.\nSpeaker H: So we'll have to evaluate that when they come in and get people's opinions on what they think of them.\nSpeaker H: Then the other question I had is maybe we should get another wireless setup.\nSpeaker H: It's expensive, but it does seem to be better than the wired.\nSpeaker E: So how many channels do you get to have?\nSpeaker E: You know, wireless setup?\nSpeaker H: Well, I'm pretty sure that you can daisy chain them together.\nSpeaker H: So what we would do is replace the wired mics with wireless.\nSpeaker H: So we currently have one base station with six wireless mic, possibility of six wireless receivers.\nSpeaker H: And apparently you can chain those together.\nSpeaker H: And so we could replace our wired mics with wireless if we bought another base station and more wireless mics.\nSpeaker H: So it's still at 15 minus 6.\nSpeaker H: So we could have up to 9.\nSpeaker E: And right now we can know at the 6.\nSpeaker H: And we have 5, we're getting one more.\nSpeaker H: And it's about $900 for the base station and then 800 per channel.\nSpeaker E: Oh, so yeah, so the only, beyond the mic cost of the mics, the only thing is the base station is $900.\nSpeaker E: Oh, we should do it.\nSpeaker H: Okay.\nSpeaker H: So I'll look into how you daisy chain them and then just go ahead and understand how that works.\nSpeaker H: So we're not increasing the number of channels. We're just replacing the wired, the two wired that are still working, along with a couple of the wired that aren't working, one of the wired that's not working with a wireless.\nSpeaker G: Yeah, basically three wired's work, right?\nSpeaker G: I guess three wired's work.\nSpeaker E: But we've had more problems with that.\nSpeaker E: And this sort of bypasses the whole gym box thing.\nSpeaker E: And so we seemed to have a reliable way of getting the data in which is through the Sony Radio Max, as long as we're conscious about the batteries.\nSpeaker F: Right.\nSpeaker F: Everyone's battery, okay?\nSpeaker G: I'm sure.\nSpeaker G: I checked them this morning.\nSpeaker G: I should be.\nSpeaker E: Yeah.\nSpeaker E: That's the only thing with them.\nSpeaker E: But the quality seems really good.\nSpeaker E: I heard from UW that they're very close to getting their set up purchased.\nSpeaker E: They're buying something that you just sort of buy off the shelf.\nSpeaker H: Well, we should talk to them about because I know that SRI is also in the process of looking at stuff.\nSpeaker H: And so, you know, we should try to keep everyone on the same page with that.\nSpeaker H: So, are I really?\nSpeaker H: Yeah.\nSpeaker H: They got a parent.\nSpeaker H: No.\nSpeaker H: Maybe this needs to be bleeped out.\nSpeaker E: I don't know how much of it's probably working.\nSpeaker E: I don't know how much of it's probably working.\nSpeaker E: I'm talking about funding stuff.\nSpeaker E: Right.\nSpeaker E: But anyway, there's other activities that are going on there.\nSpeaker E: And NIST and UW.\nSpeaker E: But I think that at least the message we can tell other people that our experience is quite positive with the Sony Radio Max.\nSpeaker E: Right.\nSpeaker E: Now, the one thing that you have said that actually concerns me a little is you're talking about changing headsets, changing the connector, which means some hands-out or something.\nSpeaker H: No, we're having them do it.\nSpeaker H: So, it's hand-sortering it, but I'm not doing it.\nSpeaker H: Okay.\nSpeaker H: So, they charge.\nSpeaker E: No, I guess you're doing your hands-out.\nSpeaker H: You've never seen my hands-out or anything.\nSpeaker H: But, as I've said, they're coming in.\nSpeaker H: I mean, yeah.\nSpeaker H: As professionally as I guess you can get it done.\nSpeaker H: I mean, if they do it...\nSpeaker H: I mean, it's just their repair shop.\nSpeaker E: Right.\nSpeaker E: They're maintenance people.\nSpeaker E: We'll see what it's like.\nSpeaker E: That can be quite good.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: Good.\nSpeaker H: Yes, let's go with that.\nSpeaker H: And then you will see tomorrow.\nSpeaker E: It looks like.\nSpeaker E: So, Dave isn't here, but he was going to start working on some things with the digits.\nSpeaker E: So, he'll be interested in what's going on with that.\nSpeaker E: I guess the decision last time was that the transcribers were going to be doing stuff with the digits as well.\nSpeaker E: Is that Starters?\nSpeaker C: It would be to use his interface now as going to meet with him today about that.\nSpeaker H: Right.\nSpeaker H: So, the decision was that Jane did not want the transcribers to be doing any of the paperwork.\nSpeaker H: So, I did all that last week.\nSpeaker H: So, all the forms are now on the computer.\nSpeaker H: And then I have a bunch of scripts that will read those and let the transcribers use different tools.\nSpeaker H: And I just wanted to talk to Jane about how we transition to using those.\nSpeaker D: So, he hasn't set up that they could be efficient than I just did.\nSpeaker H: Okay.\nSpeaker H: So, I don't think it'll take too long.\nSpeaker H: So, you know, just a matter of a few days, I suspect.\nSpeaker E: So, anyway, I think we have at least one user for the digits once they get done, which will be Dave.\nSpeaker H: Right.\nSpeaker H: I've already done five or six sets.\nSpeaker H: So, if you wanted to, you know, just have a few to start with, he could.\nSpeaker H: And I also have a bunch of scripts that will generate key files and run recognition on them also.\nSpeaker D: Right.\nNone: Okay.\nSpeaker E: It's Dave.\nSpeaker E: I don't know if Dave is on the list.\nSpeaker E: He's invited to these meetings.\nSpeaker C: I don't tend to get an invitation myself.\nSpeaker H: We don't have an active one, but I'll make sure he's on the list.\nNone: Should we call him?\nSpeaker C: I mean, you see, you see, he definitely not available today.\nSpeaker H: I don't know.\nSpeaker H: Should I call him?\nSpeaker H: He's still taking classes.\nSpeaker C: So, he may well have classes.\nSpeaker E: He wasn't there at college.\nSpeaker E: Yeah, so this might be a conflict.\nNone: Okay.\nSpeaker F: Okay.\nSpeaker F: So, did Indy say his signal processing class was like, I think he has Tuesdays and Thursdays?\nSpeaker F: Yeah.\nSpeaker F: He might have.\nSpeaker F: Okay.\nSpeaker A: Talking about David Gilbert.\nSpeaker A: Yeah.\nSpeaker A: I think he's taking 225A, which is now.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: So, that's why we're doing team.\nSpeaker E: Okay.\nSpeaker E: Transcription is beyond the digits where we are.\nSpeaker E: Okay.\nSpeaker E: And the recording is also just where we are.\nSpeaker C: So, should we do a little recording?\nSpeaker H: It's the first turn.\nSpeaker H: Oh, we have about 32 hours.\nSpeaker H: I guess a week and a half ago.\nSpeaker H: So, you probably now have about 35 hours.\nSpeaker E: And that's how much of those digits?\nSpeaker E: That's including digits.\nSpeaker E: That's including digits.\nSpeaker H: So, I haven't separated out, so I've no clue how much of that is digits.\nSpeaker E: So, anyway, there's at least probably 30 hours.\nSpeaker E: Got to be more than 30 hours.\nSpeaker E: Of non-digits?\nSpeaker E: Yeah, absolutely.\nSpeaker H: I mean, the digits don't take up that much time.\nSpeaker H: Okay.\nSpeaker C: Okay.\nNone: And the transcribers, I don't have the exact numbers.\nNone: I think it would come to about 11 hours that are finished transcribing from them right now.\nNone: The next step is to do the recording.\nNone: And then to ensure that the data are clean first.\nNone: And then to channelize.\nSpeaker D: The only clean is that there's a check.\nSpeaker C: And the markup is not only throughout, but also that we now incorporate these additional conventions that, uh, Liz requested in terms of, uh, in terms of having us, uh, systematic handling of numbers and acronyms, which I hadn't been specific about.\nSpeaker C: Uh, for example, they'll say 92.\nSpeaker C: And, you know, so how you could...\nSpeaker C: 92.\nSpeaker C: Exactly.\nSpeaker C: So, if you just say 92, then there are many ways that could have been expressed.\nSpeaker C: And I just had them.\nSpeaker C: I mean, a certain number of them did put the words down.\nSpeaker C: But now we have a convention which also involves having it followed by a gloss.\nSpeaker C: You know, Jay?\nSpeaker G: Um, one suggestion that you may already be doing this.\nSpeaker G: But I've noticed in the past that when I've gone through transcriptions and, you know, in order to build lexiconcant things, if you, um, just take all the transcriptions and separate them into words and then alphabetize them, a lot of times just scanning down that list you'll find a lot of misspelled inconsistency.\nSpeaker C: Just talking about the tech token frequency listings.\nSpeaker C: And I use those two.\nSpeaker C: You mean, just a, on each line, there's one word, right?\nSpeaker C: It's one token from the corpus.\nSpeaker C: Yeah, those are extremely efficient.\nSpeaker C: And I agree that's a good decision.\nSpeaker C: So you already have that, okay.\nSpeaker C: Well, that's a way, you know, the spell check basically does that.\nSpeaker C: But, but in addition, yes, that's exactly the strategy.\nSpeaker C: I want to do in terms of locating these things, which are, you know, locally spoken forms, which aren't the likes of them.\nSpeaker C: Exactly.\nSpeaker G: So a lot of times they'll appear next to each other and, uh, in alphabetize list, they'll appear next to each other.\nSpeaker G: Exactly.\nSpeaker C: I agree.\nSpeaker C: That's a very good suggestion.\nSpeaker C: And that was, that's my strategy, praneling a lot of these things in terms of things that need to be glossed.\nSpeaker C: I didn't get to that point.\nSpeaker C: But so there are numbers, then there are acronyms.\nSpeaker C: And then, um, there's a, she wants to be, uh, actually, an explicit marker of what type of comment this is.\nSpeaker C: So it curly inside the curly brackets, I kind of put either vote for vocalize, like cough or laugh or whatever, non-vogue for door slam and gloss for things that have to do with, if they set us, uh, spoken form with this, maybe this pronunciation, or I already had that convention that I've been asking these people to do this,\nNone: because I think it can be most efficiently handled by, by, uh, a filter, and that's what I was always planning on. So, you know, you get a whole long list, exactly what you're saying.\nSpeaker C: You get a whole list of things that say curly bracket, laugh curly bracket.\nSpeaker C: Then, you know, it's, it's, you risk less error if you handle it by a filter.\nSpeaker C: Then if you have the transcribed, laboriously typing in, it's going to be a voc space.\nSpeaker C: So many ways that can be error.\nSpeaker C: Right.\nSpeaker C: So, um, I'm, I'm going to convert that via a filter into these tagged, uh, subcategorized comments.\nSpeaker C: And same thing with, you know, you see, you get a subset, when you do what you're saying, you end up with, with, uh, your claps can cross the frequency, you just have to open, so you can, uh, have a filter which more efficiently makes those changes.\nSpeaker C: But the numbers and acronyms have to be handled by hand, because, you know,\nSpeaker H: you don't know what they could be.\nSpeaker C: Yeah, no timents clear, um, and PLP is clear. But, uh, there are things that are not so well known, or, or have variant, you use like the numbers, you could say 9 to, you could say 92.\nSpeaker H: So, how are you doing the, I handle the numbers individually.\nSpeaker H: How are you doing the, uh, acronyms?\nSpeaker H: So if I say PZM, what would it be here on the transcript?\nSpeaker C: The letters would be separated.\nSpeaker C: Okay.\nSpeaker C: Space.\nSpeaker C: And potentially they'll have, uh, curly bracket thing afterwards.\nSpeaker C: But I'm not sure if that's necessarily clarifying what it is, so gloss up, whatever.\nSpeaker C: Right.\nSpeaker C: I don't know if that's really necessary to do that.\nSpeaker C: It's a nice thing to do because of, it then indicating this is, uh, a step away from, indicating that it really is intentional, those spaces are there, indicating why they're there, indicating that it's, uh, the, uh, enumerated, or, I mean, that's not a good way of saying it.\nSpeaker C: But it's, it's the, it's a thick, uh, way of stating these, these letters.\nSpeaker C: So it sounds good.\nSpeaker C: And so anyway, the clean, those are those things, and then, channelized is to then, uh, get it into this multi-channel format.\nSpeaker C: And at that point, then it's ready for use by Liz and Dawn.\nSpeaker C: But that's been my top priority.\nSpeaker C: Beyond getting a channel channelized, the next step is to work on tightening up the boundaries of the time.\nSpeaker C: Right.\nSpeaker C: And, uh, Tila had a, uh, I've right through this, this last week in terms of getting the channel based, uh, speech-non-speech segmentation, uh, up and running, and I haven't, I haven't been able to use that yet, because I'm working through it. This is my top priority. Get the data clean.\nSpeaker H: Have you also been doing spot checks, Jane?\nSpeaker C: Oh, yes. Well, you see, that's part of the cleaning process.\nSpeaker C: I spent, um, actually, um, I have a segment of 10 minutes that was transcribed by two of our transcribers.\nSpeaker C: Okay. Good.\nSpeaker C: And I went through it last night. It's, it's almost spooky how similar these are.\nSpeaker C: Word for word. And there are some differences in commas, because commas, I left in discretion at commas.\nSpeaker C: Right.\nSpeaker C: And so, because it's not part of our, our, our needed conventions, and, um, and, um, and it still will be different from commas, but it's, it's word by word the same, in, in huge patches of the data.\nSpeaker C: And I have 10 minutes stretch where I can, where I can show that.\nSpeaker C: And, and sometimes it turns out that one of these transcribers has a better ear for technical jargon, and the other one has a better ear for co-ocleal speech.\nSpeaker C: So, um, the one, the, the co-ocleal speech person picked up gobble-dee book, and the other one did, and on this side, this one's picking up things like neural nets, and the one that's good on this vocabulary.\nSpeaker C: Right.\nSpeaker G: Yeah, co-ocleal speech.\nSpeaker G: For the person who missed gobble-dee book, what did they put?\nSpeaker C: It was, uh, interesting approximation put in parentheses, because I had this convention that if they're not sure what it was, they put it in parentheses, so I tried to approximate it, but it was spelled GA, BB, L.\nSpeaker C: Sort of how it sounds.\nSpeaker C: Yes, more of an attempt to, I mean, apparently it was very clear to her, these, these, this, this, this, this, this, this, this, It was a technical term that she didn't recognize.\nSpeaker C: Yeah, but she knew she didn't know it, maybe it was a technical term exactly, but she even know her technical perception is just really, you know, I'm tempted to ask her, is she's taking any courses in this area, or is she's taking cognitive science, or is that right?\nSpeaker C: Because neural nets, and oh, she has some things that are, oh, down sample, she got that right, and some of these are rather, uh, I'm excited.\nSpeaker C: That's obscure, yeah.\nSpeaker C: But ten solid, she's trying to have ten solid minutes where they both coded the same data, and, uh,\nSpeaker E: and again, the main track that you're working with is 11 hours, is that right?\nSpeaker C: Yes, exactly.\nSpeaker C: Okay.\nSpeaker C: And that's probably, uh, 11 hours.\nSpeaker E: Is that, is that, that including digits?\nSpeaker E: Yes it is.\nSpeaker E: Yeah.\nSpeaker E: So let's say roughly 10 hours or so.\nSpeaker E: I mean it's probably more than that.\nSpeaker C: It ain't been more than that because my recollection is a minute, and that's not, that, that, that, that just don't take more than half a minute for a person.\nSpeaker C: Okay.\nSpeaker C: But, uh, the, uh, the total set that I gave them is 12 hours of tape, But they haven't gotten to the end of that yet, so there's still working on some of them.\nSpeaker C: Two of them are still working on completing that.\nSpeaker G: Quite him, moving right along.\nSpeaker C: Yeah, they are.\nSpeaker C: There's some that have more hours that they devote to it than others.\nSpeaker E: So what's the deal with your talent thing?\nSpeaker B: Oh, it's just I ran the Rekinaza, other speech and speech detector on different channels.\nSpeaker B: It's just in this new multi-channel format and output.\nSpeaker B: I just gave one meaning to Liz who wanted to try it for the Rekinaza.\nSpeaker B: As apparently the Rekinaza had problems with those long channels of speech, which took too much memory or whatever.\nSpeaker B: And so she will try that.\nSpeaker B: I'm raking on it.\nSpeaker H: Is there anything different than the HMM system you were using before?\nSpeaker B: I used some different features, but not the basic thing is this.\nSpeaker H: So there's still no knowledge using different channels at the same time?\nSpeaker B: Yeah, it's as the energy is normalized across all of them.\nSpeaker B: That's one of the main changes.\nSpeaker E: First of the other features, besides the energy.\nSpeaker E: You said you're trying some different features.\nSpeaker B: Oh, I just used our loudness based things.\nSpeaker B: I used to stay before the area where some in the locked domain.\nSpeaker B: I changed just to the cube root.\nSpeaker B: No, I changed just to the loudness thing you had.\nSpeaker B: What do you call it?\nSpeaker E: The Trim.\nSpeaker B: I'm not sure about the Trim.\nSpeaker B: I'll look it up.\nSpeaker B: That's basically the thing.\nSpeaker B: I tried to normalize the features, loudness and modified loudness within one channel.\nSpeaker B: Because they're to be able to distinguish between foreground and background speech.\nSpeaker B: And it works quite well, but not always.\nSpeaker E: I think the re-basic redone with the transcription.\nSpeaker E: So I guess the next thing is this.\nSpeaker E: We better.\nSpeaker H: Right, so the idea is that we need to provide the transcripts to every participant of every meeting.\nSpeaker H: To give them an opportunity to leave out sections they don't want.\nSpeaker H: So I've written a bunch of tools that will generate web pages with the transcription in it so that they can click on them and listen to pieces and they can scroll through and read them.\nSpeaker H: And then they can check on each one if they wanted excluded.\nSpeaker H: And then it's a form, a single form so they can submit it and it will end up sending me email with the times that they want excluded.\nSpeaker H: And so some of the questions on this is what do we do about the privacy issue?\nSpeaker H: And so I thought about this a little bit and I think the best way to do it is every participant will have a password, a single password.\nSpeaker H: Each person will have a single password, username and password.\nSpeaker H: And then each meeting will only allow the participants who are at that meeting to look at it.\nSpeaker H: And that way each person only has to remember one password.\nSpeaker E: I can't help but wonder if this is maybe a little more elaborate than is needed.\nSpeaker E: I mean, for me I would actually want to have some pieces of paper that have a transcription and sort of flip through it.\nSpeaker E: And then if I thought it was okay, I'd say it's okay.\nSpeaker E: And I mean, it depends how this really ends up working out. But I guess my thought was that the occasion of somebody wondering whether something was okay or not, the meeting to listen to it was going to be extremely rare.\nSpeaker H: Right. I mean, so the fact that you could listen to it over the web is a minor thing that I already had done for other reasons.\nSpeaker H: And so that's a minor part of it. I just wanted some web interface so that people, you didn't actually have to send everyone the text.\nSpeaker H: So what my intention to do is that as the transcripts become ready, I would take them and generate the web pages and send email to every participant or contact them using the contact method they wanted.\nSpeaker H: And just tell them, here's the web page. You need a password. So question number one is how do we distribute the passwords.\nSpeaker H: And question number two is how else do we want to provide this information if they want it?\nSpeaker E: Let's think what I was sort of saying is that if you just say here is it, here is, maybe it sounds peveolithic, but I thought if you handed them some sheets of paper that said, here's what was said in this transcription, is it okay with you?\nSpeaker E: And if it is, here's the other sheet of paper that you sign that says that it's okay.\nSpeaker H: I think that there are a subset of people who want printouts that we can certainly provide, but certainly I wouldn't want to print out. These are big. And I would much rather be able to just sit and read through it.\nSpeaker E: I mean, how do you read books?\nSpeaker H: Well, I certainly read books by hand, but for something like this, I think it's easier to do it on the web.\nSpeaker H: Does your going to get, you know, I mean a bunch of meetings and I don't want to get a stack of these, I want to just be able to go to the website and visit it as I want.\nSpeaker E: Going to a website is easy, but flipping through 100 pages of stuff is not easy on the web.\nSpeaker H: Well, I don't think it's that much harder than paper.\nSpeaker C: So, I have a question. So, are you thinking that the person would have a transcription go strictly through the transcript? Because I do think that there's a benefit to being able to hear the tone of voice.\nSpeaker E: So, here's where I was imagining it, maybe I'm wrong, but the way I mentioned it was that the largest set of people is going to go, oh yeah, I didn't say any funny that meaning just go ahead, where's the release?\nSpeaker E: There will be a subset of people, right? Because think of who it is we've been recording. There will be a subset of people who will say, well yeah, I really would like to see that.\nSpeaker E: And for them, the easiest way to flip through, if it's a really large document, I mean unless you're searching, searching, of course, should be electronic.\nSpeaker E: But if you're not, so if you provide some search mechanism, go to every place they said something or something like that, see them or get more elaborate with this thing.\nSpeaker E: If you don't have search mechanisms, you just sort of have this really, really long document. I mean whenever I've had a really, really long document that is sitting on the web, I've always ended up printing it out.\nSpeaker E: I mean, so it's, I mean, you're not necessarily going to be sitting at the desk all the time, you want to figure you have a train ride, there's all these situations where, I mean, this how it's imagining it anyway.\nSpeaker E: And then I figured that out of that group, there would be a subset who would go, you know, I'm really not sure about this section here. And then that group would need it.\nSpeaker E: It seems like, if I'm right in that, it seems like you're setting it up for the most infrequent case rather than for the most frequent case.\nSpeaker E: So that now we have to worry about, we'll see if we're going to talk about all these passwords for different people.\nSpeaker H: For the most frequent case, they just say, it's okay. And then they're done. And I think almost everyone would rather do that by email than any other method.\nSpeaker C: The other thing too is, it seems like.\nSpeaker H: Because you don't have to visit the web page if you don't want to.\nSpeaker E: I guess, yeah, I guess we don't need their signature. I guess an email.\nSpeaker H: Oh, that was another thing. I had assumed that we didn't need their signature, that an email approval was sufficient.\nSpeaker G: Are people going to be allowed to leap out sections of a meeting where they weren't speaking?\nSpeaker G: Yes.\nSpeaker H: If someone feels strongly enough about it, then I think they should be allowed to do that.\nSpeaker G: So that means other people are editing what you say.\nSpeaker G: I don't know about that.\nSpeaker G: I don't know if I like that.\nSpeaker H: Well, the only other choice is that the person would say, no, don't distribute this meeting at all. And I would rather they were able to edit out other people than just said, don't distribute it at all.\nSpeaker E: But what they signed in the consent form was something that said, you can use my voice.\nSpeaker H: Well, but if someone is having a conversation and you only bleep out one side of it, that's not sufficient.\nSpeaker E: Yeah, but that's our decision, right?\nSpeaker H: I don't think so. Because if I object to the conversation, if I say we were having a conversation, and I consider that conversation private, and I consider that your side of it is enough for other people to infer.\nSpeaker H: I want to be able to bleep out your side.\nSpeaker C: I agree that the consent forms were, I agree with what Adam's saying, that the consent form did leave open this possibility that they could edit things which they found offensive.\nSpeaker C: And the other thing is from the standpoint of the lawyer, but it strikes me that we wouldn't want someone to say, oh yes, I was a little concerned about it, but it was too hard to access.\nSpeaker C: So I think it's kind of nice to have this facility to listen to it, not in terms of like editing it by hand.\nSpeaker C: I think it's, some people would find that easier to specify the bleep part by having a document they edited, but it seems to me that sometimes, you know, if the person had a bad day and they had a tone in their voice that they didn't really like, you know, it's nice to be able to listen to it, be sure that that was okay.\nSpeaker H: Yeah, I can certainly provide a printable version if people want it.\nSpeaker E: And it's also a mixture of people. I mean, some people are, do their work primarily by sitting at the computer and flipping around the web and others do not.\nSpeaker E: Others would consider it a set of skills that they would have to gain.\nSpeaker H: Well, I think most of the people in the meetings are the former so far.\nSpeaker E: Yeah, the meeting so far. Yeah, but we're trying to expand this, right?\nSpeaker E: So I absolutely think that papers, the more universal thing. Well, but if they want to print it out, that's all right. I think everyone in the meeting can access the web.\nSpeaker E: No, I think we have to be able to print it out. It's not just if they want to print it out.\nSpeaker H: Okay, so does that mean that I can't use email? Or what?\nSpeaker C: Because you can send a through email, I think.\nSpeaker H: Well, I don't think I, well, I don't think we can send the text through email because of the privacy issues.\nSpeaker H: So giving them, you think a website to say if you want to print it out here, it is not sufficient.\nSpeaker E: Certainly for everybody who's been in the meeting so far would be sufficient.\nSpeaker H: Yeah, I'm just thinking for people that that's not sufficient for what the only sufficient thing would be for me to walk up to them and hand it to them.\nSpeaker C: It may not be. It's equivalent.\nSpeaker C: I think it's easier to put the box back.\nSpeaker B: Just put the button on the web page, which say please send me the scripts.\nSpeaker B: That's interesting.\nSpeaker G: When you display it on the web page, what are you showing them utterances?\nSpeaker G: And so can they bleep within an utterance?\nSpeaker H: No, whole utterances.\nSpeaker H: And that was just convenience for my sake that it's, it would end up being fairly difficult to edit the transcripts if we want to do it at the sub-eitorance level.\nSpeaker H: Because this way I can just delete an entire line out of a transcript file rather than have to do it by hand.\nSpeaker E: There's other aspect to this, which maybe is part of why it's bothering me.\nSpeaker E: I think you're really trying very hard to make this as convenient as possible for people to do this.\nSpeaker H: I mean that's why I did the web form because for me that would be my most convenient.\nSpeaker H: I know where you go.\nSpeaker E: See, because you're going to end up with all these little patchy things, where really what we want to do is have the bias towards letting it go.\nSpeaker E: Because there was one or twice in the meetings we've heard where somebody said something that they might be embarrassed by.\nSpeaker E: But overall, people are talking about technical topics.\nSpeaker E: Nobody's going to get hurt. Nobody's being liable.\nSpeaker E: We're playing the lawyers game and we're looking for the extreme case.\nSpeaker E: If we really orient it towards that extreme case, make it really easy.\nSpeaker E: We're going to end up encouraging the headache.\nSpeaker E: I think that's sort of psyching myself out here.\nSpeaker H: I guess I don't see having a few phrases here and there in a meeting being that much of a headache.\nSpeaker H: Well, so.\nSpeaker G: I think what Morgan's saying is that easier it is the more it's going to be bleeped.\nSpeaker E: And it really depends on what kind of research you're doing.\nSpeaker E: I think some researchers who are going to be working with this group as years from now are really going to be cursing the fact that there's a bunch of stuff in there that's missing from the dialogue.\nSpeaker E: It depends on the kind of research they're doing, but it might be really a pain.\nSpeaker E: And where it's really going to hurt somebody in some way, the one who said it or someone who's being spoken about, we definitely want to allow the option of being bleeped out.\nSpeaker E: But I really think we want to make it the rare incidence.\nSpeaker E: And I'm just a little worried about making it so easy for people to do and so much fun.\nSpeaker E: But they're going to go through and bleep that stuff.\nSpeaker E: I think I'm bleeped out stuff they don't like too, right?\nSpeaker E: Somebody else, as you say, you know?\nSpeaker H: Well, I don't see any way of avoiding that.\nSpeaker H: We have promised that we will try them to transcript them that they can remove parts that they don't like.\nSpeaker H: Yeah, so that the...\nSpeaker E: No, no, no. The only question is...\nSpeaker E: It doesn't mean to that, but I just think that we should make it harder to do.\nSpeaker H: The problem is if it's harder for them, it's also harder for me.\nSpeaker H: Whereas this web interface, I just get email, it's all formatted, it's all ready to go.\nSpeaker E: So maybe you don't get the access to the web interface unless they really need it.\nSpeaker E: Oh, I can...\nSpeaker E: So, I'm sorry. So maybe this is a way out of it.\nSpeaker E: You've provided something that's useful for you to handle and useful for someone else if they need it.\nSpeaker E: But I think the issue of privacy and ease and so forth should be that they get access to this if they really need it.\nSpeaker G: So you're saying the sequence would be more like...\nSpeaker G: First Adam goes to the contact lists, contacts them, whatever their preferred method is, to see if they want to review the meeting.\nSpeaker G: Right.\nSpeaker G: And then if they don't, you're done.\nSpeaker G: If they do, then he provides them access to the website.\nSpeaker H: Well, to some extent, I have to do it anyway because as I said, we have to distribute passwords.\nSpeaker E: But you don't necessarily have to distribute passwords, that's what I'm saying.\nSpeaker H: Well, but only if they...\nSpeaker H: What I'm saying is that I can't just email them the password because that's not secure.\nSpeaker E: No, no, no, but you have to call me and ask.\nSpeaker E: Right.\nSpeaker E: But we don't even necessarily need to end up distributing passwords at all.\nSpeaker H: Well, we do because of privacy. We can't just make it openly available.\nSpeaker E: No, no, you're missing the point.\nSpeaker E: We're trying to make it less of an obvious, just fall off a log to do this.\nSpeaker C: Not everyone gets a password.\nSpeaker E: Right.\nSpeaker E: So what I would see is that first you contact them and ask them if they would like to review it, or to check for it, not just for fun.\nSpeaker E: Okay.\nSpeaker E: But to check this for things that they're worried about having said, or if they're willing to just send approval from their memory.\nSpeaker E: And we should think carefully, actually, we should review, go through how that's worded.\nSpeaker E: Okay.\nSpeaker E: Then if someone wants to review it, and I know you don't like this, but I'm offering this as a suggestion, is that we then give them a printout.\nSpeaker E: And then if they say that I have a potential problem with these things, then you say, okay, well, you might want to hear this in context.\nSpeaker E: I think if you need that, you issue them a password.\nSpeaker H: The problem with what you're suggesting is it's not just inconvenient for them.\nSpeaker H: It's inconvenient for me, because that means multiple contacts every time, for every single meeting, every time anyone wants anything.\nSpeaker H: I would much prefer to have it all be automatic.\nSpeaker H: They visit the website if they want to.\nSpeaker H: Obviously they don't have to.\nSpeaker E: I know you prefer it, but the problem is there's a problem.\nSpeaker H: So I think you're thinking people are going to arbitrarily start bleeping, and I just don't think that's going to happen.\nSpeaker C: I'm also concerned about the spirit of the event.\nSpeaker C: I think it's important to want something, because I think if they feel that it's...\nSpeaker C: If it turns out that something gets published in this corpus, that someone really should have eliminated and didn't detect, then it could have been because of their own negligence that they didn't pursue that next level and get the password and do that.\nSpeaker C: But they might be able to argue, oh well, it was cumbersome, and I was busy, and it was going to take me too much time to trace it down, so it could be that the burden would come back on to a song a little bit worried about making it harder.\nSpeaker E: Where you can go to far in that direction, you need to find somewhere between, I think.\nSpeaker H: It seems to me that sending the email, saying, if you have an okay, reply to this email and say, okay, if you have a problem with it, contact me, and I'll give you a password.\nSpeaker H: It seems like it's a perfectly reasonable compromise, and if they want to print out, they can print it out themselves.\nSpeaker H: Well, we can print it up for them, let me give you off of that.\nSpeaker C: But there's another aspect of that, and that is that in the informed consent form, my impression is that we offered them, at the very least, that they definitely would have access to the transcript.\nSpeaker C: And I know that there's a chance of really skipping that stage.\nSpeaker C: I thought that maybe I'm a sort of turpid, and what you said about this.\nSpeaker E: Having access to it doesn't necessarily mean having it.\nSpeaker E: Having it.\nSpeaker E: Having the correct answer.\nSpeaker H: Well, the consent form is right in there, if anyone wants to look at it.\nSpeaker H: You want me to grab it?\nSpeaker H: Yeah, but you're wired.\nSpeaker E: I asked you.\nSpeaker E: Yeah, I mean, I don't want to fool them.\nSpeaker E: I just meant that every, anytime you say anything to anyone, there is, in fact, a bias that is presented.\nSpeaker H: If you agree to participate, you'll have the opportunity to have anything x-anything excised, which you would prefer not to have included in the data set.\nSpeaker H: Once the transcript is available, we will ask your permission to include the data in the corpus for the larger research community.\nSpeaker H: There, again, you will be allowed to indicate any sections that you'd prefer to have x-ays from the database, and they will remove both from the transcript and the recording.\nNone: Well, that's more open than I realized.\nSpeaker H: Well, I mean, the one question is definitely clear, anything, as opposed to just what you said.\nSpeaker C: Yeah, I know that.\nSpeaker C: That's more severe, but the next one says the transcript will be around, and it's really, say, we'll send it to you, or it'll be available to you on the web.\nSpeaker G: Thank you.\nSpeaker G: I hope it leaves it open.\nSpeaker C: How do we get it to that?\nSpeaker C: It leaves it open.\nSpeaker C: It means also we don't have to give it to him.\nSpeaker H: They just have to make sure that it is available to them.\nSpeaker C: Is it able to, if they ask?\nSpeaker E: Yeah.\nSpeaker E: Okay, so I think I have an idea that maybe you satisfy both you and me in this, which is, we just go over carefully how these notes to people are worded.\nSpeaker E: So I just wanted to be worded in such a way where it gives the strong impression, gives very, I mean, nothing hidden, very strongly, the bias, that we would really like to use all of these data.\nSpeaker E: Right. That we really would rather it wasn't a patchwork of things tossed out, that it would be better for our field if that is the case.\nSpeaker E: But if you really think something is going to, and I don't think there's anything in legal aspects that is hurt by our expressing that bias.\nSpeaker E: Great, great.\nSpeaker E: And then my concern about, which, you know, you might be right, it maybe was just paranoia in my part, but people just, it's, I'm not worried about you making this interface so much fun, and people start bleeping stuff out just because they can't.\nSpeaker H: It's just a checkbox, that's to the text, it's not any fun at all.\nSpeaker E: Why do I kind of had fun when you played me something was bleeped out, you know?\nSpeaker H: Well, but they won't get that feedback.\nSpeaker H: No, because it does not automatically bleep it at the time.\nSpeaker H: It just sends me, right?\nSpeaker H: Oh, good.\nSpeaker H: It just sends me the time intervals, and then at some point, I'm going to break them all and put bleeps.\nSpeaker H: I mean, I don't want to do that yet until we actually release the data because then we have to have two copies of every meeting, and we're already short on disk space.\nSpeaker H: Yeah.\nSpeaker H: So I want to, I just keep the times until we actually want to release the data, and then we bleep it.\nSpeaker E: Okay.\nSpeaker E: Alright, so I think, yeah, so if we have, again, let's, you know, sort of circulate the wording on each of these things and get it right.\nSpeaker H: Well, since you seem to feel strong as to about it, would you like to do the first pass?\nSpeaker H: Okay.\nSpeaker E: Fair enough.\nSpeaker E: Fair enough.\nSpeaker E: Fair enough.\nSpeaker C: There's this other question, a legal question that I answered about whether we need a concrete signature or, you know, it suffices or whatever.\nNone: I don't know how that works.\nSpeaker C: And there's something down there about if you...\nSpeaker E: I don't think so.\nSpeaker E: I thought about it with one of my background processes, and it's fine to do the email.\nSpeaker H: Okay.\nSpeaker H: Because they're signing here that they're agreeing to the paragraph, which says, you'll be given an opportunity.\nSpeaker H: Yeah.\nSpeaker H: And so I don't think they need another signature.\nSpeaker E: Well, and furthermore, I, it's now fairly routine in a lot of arrangements that I do with people on contracts and so forth, that, that, if it's, if it's that sort of thing where you're, you're saying, okay, I agree, we want 80 hours of this person that's such a mouth, and I agree that's okay.\nSpeaker E: If it's a follow-up to some other agreement where there's a signature, it's often done in email now.\nSpeaker E: So it's okay.\nSpeaker H: So I guess I probably should at the minimum think about how to present it in a printed form.\nSpeaker H: I'm not really sure what's best with that.\nSpeaker H: The problem is a lot of them are really short, and so I don't necessarily want to do one per line, but I don't know how else to do it.\nSpeaker C: I also have this.\nSpeaker C: I think it's nice you have a view of her here on the web for those who might wonder about non-briple sign.\nSpeaker C: I mean, I agree that our bias should be, as I expressed here.\nSpeaker C: But I think it's nice to a person could check, because sometimes, you know, the words on the page come out sounding different terms of social dynamics at the end of the year.\nSpeaker C: And I realize we shouldn't emphasize that people shouldn't borrow travel.\nSpeaker H: I think actually, my opinion probably is that the only time someone who will need to listen to it is if the transcript is not good, you know, if there are lots of mumbles in parentheses and things like that.\nSpeaker C: Or what if there was an error in the transcript?\nSpeaker C: Right.\nSpeaker C: Didn't get detected, and there was a whole segment against some personal.\nSpeaker H: That was all mumbled.\nSpeaker H: I think Microsoft is really.\nSpeaker H: Exactly.\nNone: Sorry, transcripters.\nSpeaker C: Or even there was a line, you know, about how Bill Gates and I know.\nSpeaker C: But it was all, the words were all visible, but they didn't end up.\nSpeaker H: They're going to hate this meeting.\nSpeaker H: Yeah.\nSpeaker H: Actually, Liz will like it, you know.\nSpeaker H: But this will like it.\nSpeaker E: We had pretty strong agreement going on.\nSpeaker H: Yeah, that's right.\nSpeaker C: So I don't know.\nSpeaker C: I guess we're assuming that the transcript is close to an approximation and that my double checking will be so close to absolutely perfect that nothing will slip behind.\nSpeaker E: But something might sometime.\nSpeaker E: If it's something that they said, they might, I mean, you might be very accurate in putting down what they actually said.\nSpeaker E: But when they hear it themselves, they may hear something different because they know what they're meant.\nSpeaker E: I don't know how to note it.\nNone: Sarcasm.\nSpeaker G: How do you indicate Sarcasm?\nSpeaker G: No, I'm serious.\nSpeaker E: So we might get some feedback from people that such and such was, not really what I said.\nSpeaker H: Well, that would be good to get that.\nSpeaker H: Yeah, sure.\nSpeaker H: Just for corrections.\nSpeaker H: Yeah.\nSpeaker H: So in terms of password distribution, I think phone is really the only way to do it, phone an in person.\nSpeaker H: Yeah.\nSpeaker H: Or mail, physical mail.\nSpeaker H: We can have their voice mail.\nSpeaker H: Any sub word level.\nSpeaker H: Any sub word?\nSpeaker H: Yeah.\nSpeaker H: Okay.\nSpeaker H: I mean, you could do it with PGP or things like that, but it's too complex.\nSpeaker C: I just realized something, which is this question about the possible mismatch of, I mean, well, and actually also the lawyer is saying that we shouldn't really have them, have the people believing that they will be cleared by our checks.\nSpeaker C: So it's like, in a way, it's nice to have the responsibility still on them to listen to the tape and hear the transcript to have that be.\nSpeaker E: Well, yeah, but you can't, I mean, most people will not want to take the time to do that.\nSpeaker E: Yeah.\nSpeaker C: Okay, sure.\nSpeaker C: And they're observing the responsibility.\nSpeaker C: And they have to.\nSpeaker C: So it's not, yeah, good.\nSpeaker E: But I mean, if you write a meeting and you don't think at least that you said anything funny and the meeting was about, you know, some funny thing about semantics or something, or you probably won't listen to it.\nSpeaker C: It is true that the content is technical, and so we're not having these discussions, which I mean, when I listen to these things, I don't find things that are questionable.\nSpeaker C: Yeah.\nSpeaker C: People's speech are in my own way.\nSpeaker E: You think of Bebe Rare?\nSpeaker E: I mean, we're not talking about the energy crisis or something.\nSpeaker E: Man.\nSpeaker H: How about them energy crises?\nSpeaker H: Yeah.\nSpeaker E: I think we're done.\nSpeaker E: Kind of done.\nSpeaker E: Actually, I was kind of, did you have anything going on?\nSpeaker A: Not really.\nSpeaker A: No, my project is going along, but I really just shared to follow the project, the overall progress.\nSpeaker A: I don't really have anything specific to talk about.\nSpeaker E: Is there more to talk about?\nSpeaker E: No, okay.\nSpeaker E: You know, I've made things.\nSpeaker E: Transcribers, he was rattling in the barbells in his brain.\nSpeaker H: So we did digits?\nSpeaker H: Oh, yeah.\nSpeaker H: Oh, by the way, I did find a bunch.\nSpeaker H: We should count out how many more digits to forms do we have back there?\nSpeaker H: Quite a few.\nSpeaker H: That's what I thought.\nSpeaker H: I was going through them all, and I found actually a lot filed in with them that were blanks that no one had actually read.\nSpeaker H: And so we still have more than I thought we did.\nSpeaker H: Okay.\nSpeaker H: So we have a few more digits before we're done.\nSpeaker H: So transcript 30113030.\nSpeaker H: 9858-001-207-4241866-54622-63781168889869-0090-1194-254-34689266.\nSpeaker H: 4901 correction 4905708467075793.\nSpeaker G: Transcript 3531-355090111914700226351839556.\nSpeaker G: 6, 7, 9253-05564-0549014234-6287256-8709873450.\nSpeaker E: Transcript 2851-2870.\nSpeaker E: 2390392-377-485-6282864-950-0712-0972-192-305-5165-7769-88409-00.\nSpeaker E: 0, 0, 0, 2, 4, 2, 1.\nSpeaker C: Transcript 2631-265-0.\nSpeaker C: 300-514-066-225-6773-89181-910-227-003245-02426-599-67.\nSpeaker C: 7, 8, 0, 2, 5, 0135655-04310-1756-2949578-3990467.\nSpeaker B: Transcript 2411-243057844-69621-78901-1286-235-2027-3951-45656-81396204-0801407-0166.\nSpeaker B: 2667-1123-5263.\nSpeaker A: Transcript 3211-3230-60811-95103060502020231-22810308-207.\nSpeaker A: 32308-228-5215-6274-3248-9880010-0252036-4833-00157076.\nSpeaker G: loop wasn't\n"}, {"meeting_id": "ES2006b", "summary": "This meeting was about functional design. Firstly, the group had three presentations on working design, technical functions and functional requirements. They suggested basic components of the remote control including the size, the colour, the ease of using and main buttons. Then group mates had a discussion on general requirements. They agreed to target disposable income groups between 15 to 35 years of age group. Also, they decided to have a simple, small-sized remote with key buttons and a menu button. Lastly, they agreed to integrate the voice recognition technology because the group mates thought they should not only focus on the target age group from 15 to 25, but also the age group from 25 to 35.", "dialogue": "Speaker B: Oh, hooked up.\nSpeaker B: Okay, so now we're here at the functional design meeting.\nSpeaker B: Hopefully, this meeting I'll be doing a little bit less talking than I did last time because this is when you get to show us what you've been doing individually.\nSpeaker B: The agenda for the meeting, I put it in the shared documents folder, I don't know if that meant that you could see it or not.\nSpeaker B: No, no.\nSpeaker B: Oh well, I'll try and do that for the next meeting as well.\nSpeaker B: So if you check in there, there's a shared project documents folder.\nSpeaker B: And it should be in there.\nSpeaker D: And I'm under what?\nSpeaker D: Oh, project documents, yeah.\nSpeaker C: Yeah, okay.\nSpeaker D: Yeah.\nSpeaker B: So I'll put it in there.\nSpeaker B: Oh, okay.\nSpeaker B: Is it best if I send you an email maybe to let you know it's there?\nSpeaker B: Yes.\nSpeaker B: Yeah, thanks.\nSpeaker B: I'll do that next time.\nSpeaker B: I'll act as secretary for this meeting in just take minutes as we go through and then I'll send them to you after the meeting.\nSpeaker B: The main focus of this meeting is your presentations that you've been preparing during the time.\nSpeaker B: So we'll go through each of you one by one.\nSpeaker B: Then we need to briefly discuss the new project requirements that were sent to us.\nSpeaker B: Yeah, the last minute.\nSpeaker B: Yeah.\nSpeaker B: The last minute, I'm sorry about that.\nSpeaker B: But we can see how that affects what you were doing.\nSpeaker B: And then we need to, by the end of the meeting, come to some kind of decision on who our target group is going to be and what the functions of the remote control.\nSpeaker B: That's the main goal is to come up with those two things, target group and functions of the remote control.\nSpeaker B: And we've got 40 minutes to do that in.\nSpeaker B: So I would say target groups.\nSpeaker A: What does that mean?\nSpeaker B: So who it is that we're going to be trying to sell this thing.\nSpeaker B: So we need to have a fairly defined group that we want to focus on and then look at the functions of the remote control itself.\nSpeaker B: So with that, I think it's best if I hand over to you.\nSpeaker B: Does anyone have a preference for going first?\nSpeaker A: I can go first.\nSpeaker B: Okay.\nSpeaker B: So we need to unplug my laptop and plug in yours.\nSpeaker B: So we'll just pull it out.\nSpeaker A: So from the.\nSpeaker B: Just before you start to make it easier, would you three mind emailing me your presentations once we you don't have to go back.\nSpeaker B: Just so that I don't have to scribble everything down.\nSpeaker A: So with regard to the working design of this remote control, I've identified a few basic components of the remote.\nSpeaker A: And from the design functional design perspective, we can now know what exactly the components are and how they work together with each other.\nSpeaker A: So this is the method that I'll mostly be following in my in my row, the identification of the components.\nSpeaker A: And since since I'm dealing only with the technical aspects, I would need feedback from the marketing person and from the user interface person.\nSpeaker A: We'll then integrate this into the product design at a technical level and basically update and come up with a new design.\nSpeaker A: So it's a cyclical process.\nSpeaker A: Okay.\nSpeaker A: So these were the basic findings from today.\nSpeaker A: The last three bullets have been integrated from the last minute email.\nSpeaker A: I just quickly jotted them down.\nSpeaker A: So basically, as I told you, the identification of how the remote control works and what are the various parts to it and what are the different processes and how the parts communicate with each other.\nSpeaker A: Okay, so the email said that teletext is now outdated.\nSpeaker A: So we need to do away with that functionality of the remote control.\nSpeaker A: Also the remote control should be used only for television because incorporating other features makes it more complex.\nSpeaker A: And the reason why teletext is outdated because of internet and the variability of internet or television, however, our remote control will only be dealing with the use for television in order to keep things simple.\nSpeaker A: Also the management wants that our design should be unique.\nSpeaker A: So it should incorporate color and the slogan that our company has it standard.\nSpeaker A: Okay, so here is a functional overview of the remote control.\nSpeaker A: There's basically an energy source at the heart which feeds into the chip and the user interface.\nSpeaker A: The user interface communicates with the chip.\nSpeaker A: So I'll basically go over to the, okay.\nSpeaker A: So if this is our energy source, this is a cell, it communicates, it feeds energy into the chip which basically finds out how to do everything.\nSpeaker A: There is a user interface here.\nSpeaker A: So when the user presses a button, it feeds into the chip and the chip then generates a response and takes the response to an infrared terminal which then, so the output of the chip is an infrared bit code which is then communicated to the remote site which has an infrared receiver.\nSpeaker A: There can be a bulb here or something to indicate whether the remote is on or communicating.\nSpeaker A: So these are the essential.\nSpeaker A: So all the functionality of the remote control, whatever new functions that we need to do, make the chip more complicated and bigger basically.\nSpeaker A: So in my personal preferences, I'm hoping that we can keep the design as simple and clear as possible.\nSpeaker A: This would help us to upgrade our technology at a future point of time.\nSpeaker A: And also if we can incorporate the latest features in our chip design so that our remote control does not become outdated soon and is compatible with most televisions, that's about it.\nSpeaker B: So anything that you would like to know or do you have any idea about costs at this point?\nSpeaker A: No, I don't have any idea about what each component costs.\nSpeaker B: Because that's something to consider I guess if we're using more advanced technology it might.\nSpeaker A: Certainly, yeah.\nSpeaker A: So that, yeah, we definitely need to operate within our constraints.\nSpeaker A: But unfortunately I do not have any data so I just have identified the functional components.\nSpeaker B: That's fine.\nSpeaker B: Are there any more questions or should we just skip straight to the next one and then we can discuss all of them together?\nSpeaker D: I think we need some general discussion at the end.\nSpeaker B: Yeah, I think that will do.\nSpeaker D: Okay, so do you want to...\nSpeaker D: Yeah, I think since we were discussing some design issues then I would like to continue.\nSpeaker B: Okay.\nSpeaker B: I think that has to come out of here.\nSpeaker B: Yeah.\nSpeaker B: Thanks.\nSpeaker B: I thought those last minute things are going to hit us at worst.\nSpeaker B: Yeah.\nSpeaker D: Okay, can we help?\nSpeaker D: Right.\nSpeaker D: Should it just...\nSpeaker A: Yeah, it'll take some time.\nSpeaker D: It'll take some time.\nSpeaker B: It's just not an interview.\nSpeaker B: You need to then also press one.\nSpeaker B: Alright, right, right.\nSpeaker B: And f8.\nSpeaker B: So the blue function key at the bottom.\nSpeaker B: Okay.\nSpeaker B: And f8.\nSpeaker B: Nothing.\nSpeaker B: Okay.\nSpeaker D: Okay, something is coming out.\nSpeaker D: It'll come up with...\nSpeaker A: No signal.\nSpeaker C: No signal.\nSpeaker C: Why?\nSpeaker C: Maybe again?\nSpeaker A: Yeah, yeah, it says something now.\nSpeaker A: Oh.\nSpeaker D: My computer won blank now.\nSpeaker D: Okay.\nSpeaker D: Adjusting.\nSpeaker D: Adjusting.\nSpeaker D: But I don't see anything.\nSpeaker D: Okay.\nSpeaker D: I don't see anything on my computer now.\nSpeaker D: This is the problem.\nSpeaker B: Oh, if you press function and that again, there's usually three modes.\nSpeaker B: One where it's only here.\nSpeaker B: One where it's only there and one where it's...\nSpeaker B: Okay, so one more time.\nSpeaker D: And now it's okay.\nSpeaker D: No?\nSpeaker B: No.\nSpeaker B: Yeah, just wait for a moment.\nSpeaker D: Adjusting.\nNone: Okay.\nSpeaker D: Okay, that's fine.\nSpeaker D: That's good.\nSpeaker D: Okay, let's start from the beginning.\nSpeaker D: So I'm going to speak about technical functions design.\nSpeaker D: Just like some first issues that came up.\nSpeaker D: Okay, so the method I was adopting at this point, it's not for the whole period of the project.\nSpeaker D: It's just at this very moment.\nSpeaker D: My method was to look at other remote controls mostly just by searching on the web and to see what functionality they used.\nSpeaker D: And then after having got this inspiration and having compared what I found on the web, just to think about what they usually really need and what the user might desire as additional functionalities.\nSpeaker D: And then just to put the main function of the remote control in words.\nSpeaker D: So the findings where the main function of the remote control is just sending messages to the television set.\nSpeaker D: So this is quite straightforward.\nSpeaker D: And some of the main functions would be switching on, switching off.\nSpeaker D: Then the user would like to switch the channel.\nSpeaker D: For example, just changing to the next channel to flip through all of the possible channels.\nSpeaker D: Or the other possibility would be that she might just want to choose one particular channel.\nSpeaker D: So we would need the numbers.\nSpeaker D: And also the volume is very important.\nSpeaker A: So could you go back for a second?\nSpeaker A: Okay.\nSpeaker A: Switching on off channel.\nSpeaker A: Okay.\nSpeaker D: That's good.\nSpeaker D: Okay.\nSpeaker D: Among the findings, I found that most of the presently available remote controls also include other functionalities in their design, like operating a VCR, but they don't seem to be able to deal with DVD players.\nSpeaker D: But then there are, surely there are many other functions that could possibly be added to them.\nSpeaker D: But according to the last minute update, actually we do not want to have all these complicated functions added to our design.\nSpeaker D: So my personal preference would be to keep the whole remote control small, just like the physical size.\nSpeaker D: And then it must be easy to use.\nSpeaker D: So it must follow some conventions.\nSpeaker D: Like we're about to find the on-off button and maybe the color tends to be red or something.\nSpeaker D: Then, yeah, the most-have buttons would be on-off and then the channel numbers and then the one that allows us to go to the next of the previous channel.\nSpeaker D: And then volume has to be there.\nSpeaker D: But then other functionalities could be just, there could be a menu button and you could change things on the screen.\nSpeaker D: Then, for example, brightness and similar functions could be just done through the menu.\nSpeaker D: And yeah, the last question I had about whether we wanted to incorporate more functionalities, the answer was already no because of the last minute update.\nSpeaker D: So for the time being, that's all if you have questions.\nSpeaker B: I mean, that was the directive that came through from management.\nSpeaker B: But if we had a decent case for that we really think it's important to include video on DVD, I could get back to them and see if it's just whether it's worth arguing about.\nSpeaker D: Yeah, and also it's-the question is because there are so many different-and there are so many different things that could possibly be included.\nSpeaker D: Besides video and DVD, there are video CDs and whatever.\nSpeaker D: So yeah, it might be problematic to choose between all these possible things.\nNone: Okay.\nSpeaker B: Are there any questions for clarification of Marika before we go on to the next one?\nSpeaker A: So in the user interface requirements, we have been able to identify what are the basic buttons that we do want.\nSpeaker A: But so at this stage, how we go about implementing those buttons, we will not identify or I mean, we can completely do away with buttons and have some kind of fancy user interface or something like that.\nSpeaker A: But is there any thoughts on that?\nSpeaker D: Well, I think the buttons are still kind of the most easy for the user to use.\nSpeaker D: I mean, what other options would you have, a little screen or something, but this would be really kind of I think a lot of learning for the user.\nSpeaker D: And I mean, the user just wants to get a result quickly, not to spend time in like giving several orders.\nSpeaker D: I don't know, I think I would think the buttons, but if you have other proposals.\nSpeaker A: I think the costs will also play a big role when we come to know about them.\nSpeaker A: So we can probably wait until we have more knowledge on that.\nSpeaker A: If the costs allow, we can have like an LCD display and because we do want something fancy and fashionable as well.\nSpeaker A: So, yeah.\nSpeaker B: Sure, we can discuss that maybe after the next one.\nSpeaker E: Cool. Do you want to give me a little cable thing?\nSpeaker D: Alright, am I going in the right direction? No.\nSpeaker D: Wait here.\nSpeaker C: Oh, I'm getting hungry.\nSpeaker D: Okay, here you are.\nSpeaker E: That's why we meet.\nSpeaker E: Okay, cool.\nSpeaker E: You said?\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: Okay.\nSpeaker E: Thanks for your requirements.\nSpeaker B: We need to do the function key thing so that it comes up on the right.\nSpeaker C: Hello.\nSpeaker A: We try to transfer.\nSpeaker A: Okay.\nSpeaker A: It's working.\nNone: Okay.\nSpeaker E: Excellent.\nSpeaker E: Okay.\nSpeaker E: So, what I have, like where I've got my information from is a survey where the usability lab observed remote control use with 100 subjects and then they gave them a questionnaire.\nSpeaker E: So it was all about how people feel about the look and feel of the remote control.\nSpeaker E: We'll see most annoying things about remote controls and the possibility of speech recognition and LCD screens in remote control.\nSpeaker E: They actually gave me any answers on the LCD screen, so I should have taken that back.\nSpeaker E: Anyway.\nSpeaker E: Okay.\nSpeaker E: So, what they found is that people don't like how common remote controls are.\nSpeaker E: So, you know, definitely we should be looking at something quite different.\nSpeaker E: 75% of users find remote controls ugly.\nSpeaker E: The other 25% are no fashion sense.\nSpeaker E: 80% of users would spend more to get, you know, a nice looking remote control.\nSpeaker E: Current mode controls.\nSpeaker E: They don't match the user behavior well, as you'll see on the next slide.\nSpeaker B: I don't know what zapping is, but switching between channels sort of randomly going through.\nSpeaker E: Oh, right.\nSpeaker E: But you have that little thing that comes up at the bottom and tells you what's on.\nSpeaker E: Okay. 50% of users say they only use 10% of the buttons, so that's going back to what, you know, we were saying earlier about, you know, do you need all the buttons on the remote control?\nSpeaker E: It is make it look ugly.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker B: So, this is one little bit difficult to see.\nSpeaker B: If you explain it to us, it'll be fine.\nSpeaker E: Okay. Well, I can send it to all of you.\nSpeaker E: Yeah.\nSpeaker E: What it is is it's cones because I thought they'd be more exciting.\nSpeaker B: I like that.\nSpeaker E: Oh, come back.\nSpeaker E: All right.\nSpeaker E: Cool.\nSpeaker E: Okay.\nSpeaker E: I'm going to stop playing with the little thing.\nSpeaker E: Okay.\nSpeaker E: So, like, what it shows is how much things are used relatively.\nSpeaker E: And what you can clearly see from that is the thing that's used most is the channel selection.\nSpeaker E: What you can't see is volume selection is a little bit higher than all the others.\nSpeaker B: That's the next one along.\nSpeaker B: Yeah.\nSpeaker E: So, what the graph shows is that, you know, power, channel selection and volume selection are important.\nSpeaker E: And the rest of them, you know, nobody really uses.\nSpeaker E: And so, the numbers are not on the top, but present there, like, they're important, you know, so on a scale of one to ten, how important is that?\nSpeaker E: You know, channel selection and volume selection are absolutely essential and the power, while it's not quite so essential, apparently, although I didn't understand that it could fit.\nSpeaker E: And everything else, I think, you know, you can forget about having those buttons on the remote control because they're just not needed and they're not used.\nSpeaker E: Okay.\nSpeaker E: This is the bit that the email messed up for me.\nSpeaker E: That's what I was fiddling about with the beginning of the thing.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: Say.\nSpeaker E: Okay.\nSpeaker E: So, this is what people find annoying about remote controls.\nSpeaker E: And that they get lost, that they're not intuitive and that they're bad for repetitive strain injury.\nSpeaker E: I think if you're watching enough TV to get repetitive strain injury from, you know, watching TV, then that's the least of your problem, but, you know, it's up there.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: So, I mean, the RSI thing would be that, like, when you have the computer keyboards and you need the rest up, it would be something that encourages, you want something with an ergonomic design that encourages good use of the remote control and, you know, not strain ink, you're at rest watching TV.\nNone: Yes.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: Right.\nSpeaker E: Sorry.\nSpeaker E: This is pink because I was copying and pasting the table.\nSpeaker E: I didn't have time to write it out again.\nSpeaker E: That's all right.\nSpeaker E: Okay.\nSpeaker E: But that shows how people, whether they would pay more for voice recognition software.\nSpeaker E: So you can see from that that, you know, younger people, the age of 35, are quite likely to pay quite a lot more, well, quite likely to pay more for voice recognition software, whereas as people get older, they're a bit more skeptical about it and they're less willing to try it.\nSpeaker E: So clearly voice recognition is something to think about, but, you know, I do wonder how well that would work given that a TV, you know, tends to be people talking and, you know, how you're going to stop it from just flipping channels or watching TV.\nSpeaker E: Okay.\nSpeaker E: Cool.\nNone: Okay.\nSpeaker E: So these are my personal preferences.\nSpeaker E: So yeah, sleek, stylish, sophisticated.\nSpeaker E: You know, so something that's a bit cool, you know, functional.\nSpeaker E: So it's useful, but minimalist.\nSpeaker E: There's an important thing that, you know, people use when, you know, when you're filling up your home, you know, a lot of people fill up their home with bits of crap, basically, you know, and you've got this stuff and you're just like, what the hell is that?\nSpeaker E: Who is ever going to use it?\nSpeaker E: You know, so things should either be functional or beautiful or preferably both.\nSpeaker E: So I think we need to aim for both.\nSpeaker E: Okay.\nSpeaker E: Then a long battery life, like you were talking about earlier, and, you know, I was thinking that solar power would be quite cool because, you know, your remote control just sits there and you could just sit it in the sunshine and save the environment a bit.\nSpeaker E: And then like a locator.\nSpeaker E: So, you know, kind of like you have a mobile phone or not a mobile phone.\nSpeaker E: Some kind of a ring.\nSpeaker E: Yeah, that's it, you know.\nSpeaker E: Whistle on it.\nSpeaker E: And it's weird, my flatmate and I were talking about this on the way into you this morning and I was like, I need to get off everything.\nSpeaker E: So yeah, so maybe something where you clap and then it beeps, something, a kind of sound that you don't often hear on the TV, you know, because you don't want you remote control beeping every five minutes because you, you're deliberately losing by throwing out the window or something.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: That's me.\nSpeaker B: That's you.\nSpeaker B: Excellent.\nSpeaker B: Very good.\nSpeaker B: I'm just going to take this.\nSpeaker B: So, we've got about 10, 15 minutes to discuss.\nSpeaker A: I think one of the very interesting things that came up in Cape Cat Cat's presentation was this issue of like voice recognition being more popular with younger people.\nSpeaker A: So if you need to have a target group, then I think as far as the motto of our company is concerned, if you want to have something sleek and, you know, good looking, we are better off targeting a younger audience than, you know, people who are comparatively elderly.\nSpeaker A: Yeah.\nSpeaker E: I mean, that's the thing is that it didn't say in the survey, you know, whether, you know, these are the people that will pay more for more stylish remote control, but I'm assuming, you know, yes.\nSpeaker A: But the survey did say that things like voice recognition are more popular with them.\nSpeaker A: So if you want to put in something stylish, then it will certainly be more popular with the younger people as compared to older people.\nSpeaker B: Yeah.\nSpeaker B: Then again, I guess where it was most popular was the 15 to 25 brackets.\nSpeaker B: Right.\nSpeaker B: And I don't know how often they are buying television.\nSpeaker B: Right.\nSpeaker E: Well, that's when you go to uni, isn't it?\nSpeaker E: Yeah.\nSpeaker B: But you don't have much money generally.\nSpeaker B: I would share a television or something.\nSpeaker B: It's more that 25 to 35 when people are really moving out and they've got their first job and they want their nice toys.\nSpeaker A: Yeah.\nSpeaker A: But still, if you can go back to that slide and how popular was it?\nSpeaker A: Oh, okay.\nSpeaker A: That's all right.\nSpeaker A: If you can just look it up on your computer.\nSpeaker A: Yeah.\nSpeaker A: People between 25 to 35, how popular was it?\nSpeaker A: It was still quite popular amongst them.\nSpeaker A: So even they are 76% of their height.\nSpeaker E: Yeah.\nSpeaker E: I know what you're saying about the 15 to 25 yards, but it has been proven that people of that age group have a higher disposable income because they don't have like, I mean, if you're at university, you're paying your rent, but you don't have a mortgage.\nSpeaker E: You don't have a life insurance policy.\nSpeaker B: Right.\nSpeaker E: Usually not a car.\nSpeaker E: Normally you have a car.\nSpeaker C: Yeah.\nSpeaker C: So you're more likely to move.\nSpeaker E: You're more likely to see that.\nSpeaker E: It's cost more than a car.\nSpeaker E: Yeah.\nSpeaker E: So I mean, like it is an age group to target really.\nSpeaker B: Yeah.\nSpeaker B: And if we're talking 25 euros, it's a price that's not unaffordable even for...\nSpeaker B: No, I mean, that's what it's like.\nSpeaker E: 15 pounds, you know?\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: This is not unaffordable, but the problem is whether people need it, whether they do have a TV.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker E: I don't know many people without a TV.\nSpeaker E: Yeah.\nSpeaker E: But today, the TV last year, and everyone thought we were off our heads.\nSpeaker B: But the TVs are often kind of somewhat old TV that's well-balanced.\nSpeaker C: Listen, yeah.\nSpeaker C: And the strange stuff.\nSpeaker D: It's more control, like, not even remote.\nSpeaker D: It might not even function whether it's TV.\nSpeaker E: Yeah.\nSpeaker E: Well, we've...\nSpeaker A: But even in the case of 25 to 35, it's quite popular, right?\nSpeaker A: Yeah, we're still, yeah.\nSpeaker A: So...\nSpeaker D: Or maybe we can just kind of...\nSpeaker E: I think the fact that, you know, 91.2% of 15 to 25-year-olds are saying, yes, I would pay more for a voice recognition.\nSpeaker E: Yeah.\nSpeaker E: And the world is saying quite a lot, really, you know.\nSpeaker E: So, I mean, that and the disposal of Bullencom, I don't think it's something to ignore, you know.\nSpeaker D: Yeah, but at the same time, I think maybe we can just decide to have both of these groups as our target, because actually, I mean, they are all still good young people.\nSpeaker E: It's not a massive difference, yeah.\nSpeaker B: Yeah, if we take 15 to 35, but that one does imply that we should try and incorporate voice recognition.\nSpeaker B: Is that going to happen in an implication?\nSpeaker A: I was having a general outlook on most sophisticated features, but voice recognition itself, I'm not very sure about, because one of the things that Kat pointed out was, how do we go about implementing it?\nSpeaker E: You do have any mobile phone they don't need, because you have, like, I mean, every mobile phone now has, like, call this person, and it calls it there.\nSpeaker A: How frequently do we use it anywhere, and how good is it, you know, voice recognition software is still quite...\nSpeaker B: But with a TV remote, it's going to be quite limited if we're saying the main things people want to do is on, off, channel 5.\nSpeaker B: Yeah.\nSpeaker A: Right.\nSpeaker B: Louder.\nSpeaker B: So, that should be relatively simple.\nSpeaker E: You would maybe need a code word, do you know what I mean? So, like, when you say change, so that's been said quite a lot on TV, so maybe, like, you know, remote.\nSpeaker E: I mean, how often do people say remote on TV? Although I only watched Charms, so really, I wouldn't let it be.\nSpeaker E: But, like, say, just say remote 5, you know, remote 10, remote 1.29.\nSpeaker E: Okay.\nSpeaker A: So, it seems like a feasible thing to implement. Yeah, but maybe if you want to look into that just to check.\nSpeaker B: So, if we go for the 15 to 35 age group.\nSpeaker B: Yeah, but then of course we're going to get anyone who's older than 35 who wants to look young and hip and trendy and have money, then they'll still go for the same advertising.\nSpeaker E: I think there's a lot of voice recognition.\nSpeaker D: Well, now the voice recognition, if it works wonderfully, we could possibly deal with all buttons, but I think this is not really the right moment yet because people are just so used to buttons and, yeah, it's kind of safe.\nSpeaker C: I think we need safer. So, we need both.\nSpeaker D: So, the voice recognition would be just an extra. It wouldn't really reduce the size of the remote.\nSpeaker A: What I was thinking is that there is this separation between what the channels are on TV and how they are numbered on the remote control.\nSpeaker A: If we can do with away with that, our product can be really popular in the sense that a person can say, I want to watch it one instead of saying that I want to go on channel number 45.\nSpeaker A: Yeah.\nSpeaker A: So, if something like that can be incorporated.\nSpeaker E: So, if that was in the voice recognition, that would be great.\nSpeaker E: Yeah, but then the code word would be even more important because, I mean, sky advertised on every channel.\nSpeaker E: So, then it would be watching charm and then the sky added for it. Come on, and it would change the sky.\nSpeaker E: Yeah, yeah, and that would be really annoying.\nSpeaker A: Right.\nSpeaker D: But that's definitely it. But on the other hand, remote control isn't as close to you.\nSpeaker D: You probably might just speak into it and the TV would be already farther away.\nSpeaker D: So, it might not pick up the other things coming from the...\nSpeaker E: Do you not think that defeats the objects of having voice recognition on a remote control?\nSpeaker B: So, that you can yell at it.\nSpeaker E: Yeah, you know, so you have to have the remote control. It's more like if you lost it and it's down the surface and something can yell at it.\nSpeaker E: So, it's changing, you can look for it later.\nSpeaker E: Yeah.\nSpeaker B: All right.\nSpeaker D: Yeah, but then the remote control, I think, I mean, the idea is kind of, it's not that it's sitting there on top of the television because then you could already yell at the television and you wouldn't need the remote control.\nSpeaker D: So, the remote control is still something you keep near yourself.\nSpeaker E: Yeah, I suppose near it to you.\nSpeaker E: Like, if you have some downtown, then...\nSpeaker A: Yeah, and it might become very difficult from a distance for the television to understand what you're saying because of the noise.\nSpeaker A: Yeah, yeah.\nSpeaker A: So, the remote control being...\nSpeaker D: No, but I was just defending the fact why we want to keep the remote control close to us and not to yell at it from the distance.\nSpeaker A: So, another thing that can be used is that there can be a beeper button on the TV.\nSpeaker A: You can go and press that button and the remote control wherever it is, it will be.\nSpeaker B: But then if you're buying the remote separately, but you could have something that you could stick on to the TV or something.\nSpeaker B: Right, yeah.\nSpeaker B: If you bought it in a two-part pack, so one part attaches to the TV.\nSpeaker E: Yeah, because it's quite important that you don't lose the bit to locate the remote control.\nSpeaker B: Well, that's right.\nSpeaker B: But it solves the problem of having different noises.\nSpeaker E: Yeah, definitely.\nSpeaker E: Yeah.\nSpeaker B: Okay, I think we're going to have to wrap this up.\nSpeaker B: But if we go away with that kind of general specification in mind that we're looking at 15 to 35-year-olds, we want it to look simple, but still have the button so it's easy to use.\nSpeaker B: But only those key buttons.\nSpeaker B: The major buttons and then one sort of menu one, and then voice recognition included as an option.\nSpeaker B: Right.\nSpeaker B: But that obviously needs a little bit more working out as to whether it's really feasible and some of those problems we were mentioning.\nSpeaker B: What we have to do now is to go back to our little places, complete our questionnaire and some sort of summarization, which you'll get immediately by email.\nSpeaker B: Send me your presentations so that I can use them to make the minutes.\nSpeaker B: And then we've got a lunch break.\nSpeaker B: And after lunch, we go back to our own little stations and have 30 minutes more work.\nSpeaker B: I'll put the minutes in that project documents folder, but I'll send you an email when I do it so that you know.\nSpeaker B: It should be on your desktop.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: So I'll put them there as soon as I've written that.\nSpeaker B: Did you find it?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: And email them round.\nSpeaker E: Oh, so you want our PowerPoint presentations and that?\nSpeaker B: Yeah, that would be great.\nSpeaker D: So we'll just put them there.\nSpeaker D: Yeah.\nSpeaker D: We don't have to email them.\nSpeaker E: But is everyone's called functional requirements?\nSpeaker E: No.\nSpeaker B: They're all called something slightly different.\nSpeaker B: Technical requirements.\nSpeaker B: Okay.\nSpeaker B: So that's good.\nSpeaker B: So if you put them in there, we'll all be able to see them and refer to them if we need to.\nSpeaker B: Okay.\nSpeaker B: So as to where we're going from here, you're going to look at the components concept.\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker B: Whatever that means.\nSpeaker B: Yeah.\nSpeaker A: I guess I'll find out.\nSpeaker C: You'll be looking at the user interface.\nSpeaker C: Something conceptual.\nSpeaker B: Yeah.\nSpeaker B: Something conceptual.\nSpeaker B: And you're watching trends to see how we go and surely voice recognition or form of the math or something that will keep our up to go.\nSpeaker A: And I was supposed to look into components.\nSpeaker D: Sorry.\nSpeaker D: But the next meeting, are we going to have it right after lunch or shall we prepare?\nSpeaker B: We have after lunch, we have 30 minutes to prepare.\nSpeaker B: Okay.\nSpeaker B: That's fine.\nSpeaker B: But before lunch, we just have to complete the questionnaire and some sort of summary.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Well, right on time.\nSpeaker B: Okay. So you can, I guess we'll see you for lunch.\nSpeaker C: Okay.\nSpeaker C: See you.\n"}, {"meeting_id": "ES2006c", "summary": "This meeting was about technical design. Firstly, the group had three presentations on remote components, user interface and market trends. They presented remote control features including the circuit board, the chip, the kinetic energy, the interface mechanism of operation, the scroll wheel to be added as well as the LCD and spinning innovation. Then, the group had a discussion about the key remote concepts. They decided to have a LCD display with a spinning wheel. They would not have hand dynamos, instead, they would have a wind dynamo. And they would keep the kinetic energy as an option and look at more spongy material preferences with rubber or plastic.", "dialogue": "Speaker C: Excellent. So I sent you the agenda that was in the project document, I don't know if you've got a chance to just have a look at it.\nSpeaker C: Anyway, the meeting is going to follow more or less the same structure as last time, so we'll go around each of you in turn and you can give your presentations on what you've been up to.\nSpeaker C: And at the end of that we need to discuss what you've come up with so that we can make a decision on the key remote control concepts.\nSpeaker C: So that's we need to know about the components, properties, materials, the user interface and any trends that the marketing expert has been watching.\nSpeaker C: Okay, do you understand again? Okay. I've got 40 minutes.\nSpeaker A: So I haven't played a PowerPoint. We have a bit of a passion.\nSpeaker A: I thought I'll use the whiteboard instead.\nSpeaker A: I'll put that back in and go down.\nSpeaker A: Okay, so basically I'll start off by, I thought I'll use the whiteboard because we have so many different options.\nSpeaker A: And what we can do is that we can start rubbing off the options that we do not require.\nSpeaker A: I'm putting in the options that are more highlighted and aligning them or something like that.\nSpeaker A: Okay, so I'll start again with a brief introduction to that anyway.\nSpeaker A: Brief introduction to the insides of a mode control and then we can probably discuss the various components.\nSpeaker A: Okay, so what you see here is this is the outside of the remote.\nSpeaker A: If you open it, you have a circuit board here.\nSpeaker A: And this is the chip that I was talking about last time.\nSpeaker A: This basically sends information to a transistor here, which then sends the information to an LED device here.\nSpeaker A: If you flip the printed circuit board, and this is the most important point here, everything else is kind of.\nSpeaker A: Okay, so if you flip the circuit board, this is what it looks like.\nSpeaker A: So you see, for example, a particular button attaches to a particular place on the PCB.\nSpeaker A: And on pressing this button, a circuit completes.\nSpeaker A: The information goes to the chip, which is somewhere here, and the chip then translates the code into an infrared radiation, which goes out through there.\nSpeaker A: So the important point that I read over the website was that the configurations of these printed circuit boards are quite too many.\nSpeaker A: You can get them printed as you want to.\nSpeaker A: So we can have a configuration irrespective of the cost the way we want to have it.\nSpeaker A: So that's the important point here.\nSpeaker A: So these are the different options that we have.\nSpeaker A: Okay, so the batteries, I'll start with the battery.\nSpeaker A: So they can be simple, which is like the normal batteries in our results.\nSpeaker A: Yeah.\nSpeaker A: These are the kind of different kinds of batteries that the company makes.\nSpeaker A: So hand dynamos.\nSpeaker A: Does that mean like a wind up one?\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: I don't know if you want to consider this, but these are the different things that the company makes.\nSpeaker A: So they'll come internally from the company.\nSpeaker A: They'll be cheaper.\nSpeaker A: All these options.\nSpeaker A: So the third one is the kinetic energy one.\nSpeaker E: You can make the hand dynamo into an exercise bike and then people could exercise walls, watching TV and charging the remote.\nSpeaker C: Yeah.\nSpeaker E: And still worrying about the whole RSI.\nSpeaker A: It's a good option.\nSpeaker A: So what was the kinetic energy one?\nSpeaker A: The kinetic energy one is that they're usually in modern watches since our hand keeps moving.\nSpeaker A: It keeps the watch ticking.\nSpeaker A: But I don't know if it is a good idea for a remote control because it'll just lie there for a long while sometimes.\nSpeaker A: As soon as you pick it up, it moves and then again it recharges or something.\nSpeaker A: And the fourth option is the solar cells, which are also made by the company.\nSpeaker A: Environment friendly.\nSpeaker A: Okay.\nSpeaker A: So I'll enlist things and then we can come back and discuss what we think from everybody's perspective.\nSpeaker A: There are different cases that can be provided.\nSpeaker A: They can be basically the shape of the cases.\nSpeaker A: They can be flat.\nSpeaker A: They can be curved with one sided curved and one sided flat.\nSpeaker A: And they can be curved with on both the sides.\nSpeaker A: These are the three options.\nSpeaker D: I mean, this would be like the overall shape of the remote control.\nSpeaker D: Yeah.\nSpeaker A: Would it be flat on both the sides?\nSpeaker A: Would it be curved from one side?\nSpeaker A: There are different kinds of supplements available.\nSpeaker A: Like it can be in plastic, rubber, wood or titanium.\nSpeaker A: Okay.\nSpeaker B: Would you say woah?\nSpeaker B: Wood?\nSpeaker B: Wood.\nSpeaker E: Oh.\nSpeaker E: It can be remote.\nSpeaker A: Yeah, you understand why when we get to work with it.\nSpeaker A: So we can use even, titanium is also used in the company to make some space design equipment.\nSpeaker A: So it's kind of, it'll be probably nicer to use because it relates to the overall image of the company.\nSpeaker A: But it cannot be used on a double curved surface.\nSpeaker A: If you choose this, we cannot use titanium.\nSpeaker A: For these two, we can use titanium, wood, rubber or plastic.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: The interface options now.\nSpeaker A: So we can have push buttons like most remote do.\nSpeaker A: And our company is an expert in making push buttons.\nSpeaker A: We can have scroll wheels like the ones on mouse pointer.\nSpeaker E: And the...\nSpeaker A: So do you actually have a phone?\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Something like that.\nSpeaker A: So, and they have, they can even have an integrated push button inside the scrolling.\nSpeaker A: Okay.\nSpeaker A: The scroll plus push.\nSpeaker A: So this is something that has been recently developed by the company in the last decade.\nSpeaker A: So not too recent.\nSpeaker A: And LCDs.\nSpeaker A: We can have LCDs.\nSpeaker A: These two are recent.\nSpeaker A: And this is quite old.\nSpeaker A: The various electronic options are...\nSpeaker A: So this concerns first of all the chips I showed you that...\nSpeaker A: So there is a chip behind this one, right?\nSpeaker A: The PCB is inexpensive.\nSpeaker A: So we can put in whatever we want.\nSpeaker A: But the various integrated circuit options are...\nSpeaker A: We have either a simple one or a regular or advanced.\nSpeaker A: And the price goes up as we go down, obviously.\nSpeaker A: Okay.\nSpeaker A: So the good thing about why we would want to use advanced...\nSpeaker A: Why we might want to use advanced is that LCDs can only come with the advanced chip.\nSpeaker A: We need regular or advanced for scroll wheels.\nSpeaker A: And the chip basically includes the infrared sender.\nSpeaker A: Besides this, under electronics, also the company has started making a sample sender, which did not explain what it was.\nSpeaker A: But I'm guessing that...\nSpeaker A: So they have a sample sender and a sample speaker.\nSpeaker A: So I'm guessing that the sample speaker is probably something like, you know, as soon as you press a button, it gives you feedback.\nSpeaker A: One, five or whatever.\nSpeaker A: Yeah, on.\nSpeaker A: And I don't know whether a sample sender has to do something with voice recognition or not.\nSpeaker A: But anyway, so these are the different options that we have.\nSpeaker A: Okay.\nSpeaker A: So that's basically...\nSpeaker A: Now I think that we can integrate the user interface and the marketing things in that...\nSpeaker A: To keep taking out things from this and underlining things that are important.\nSpeaker C: Excellent.\nSpeaker C: Do you want to stay somewhere near the board so that if we need to...\nSpeaker A: Yeah, sure.\nSpeaker C: You can sit down, but just...\nSpeaker C: We might need you to leap up.\nSpeaker D: What do you have now?\nSpeaker D: I have some PowerPoint here.\nNone: Okay.\nSpeaker D: Oh, yeah.\nSpeaker C: Can these pens give you a cancer overhand?\nSpeaker E: I suppose we'll come in and click in and say...\nSpeaker A: Yeah, it should do it.\nNone: Right.\nSpeaker C: Interface concept.\nSpeaker D: Okay.\nSpeaker D: To be honest, actually, I mentioned some of the things which could fit under this talk.\nSpeaker D: This time I mentioned the more ID in the previous talk.\nSpeaker D: So, yeah, this time I might not have them on the slides, but I can just mention them again.\nSpeaker D: Okay.\nSpeaker D: So I thought I would also include the definition of using interface.\nSpeaker D: So it's the aspects of a computer system or a program which can be seen by the user.\nSpeaker D: And which the mechanism that the user uses to control is operation and input data.\nSpeaker D: So this would include things like shape and size and buttons and voice recognition as well and color and so on.\nSpeaker D: The method I employed this time was again having a look to related products and mainly on the internet.\nSpeaker D: Then analyzed them from the point of view of user friendliness and also whether their appearance was pleasant.\nSpeaker D: And then this can help us to decide which reaches we want to incorporate in our product.\nSpeaker D: So some findings.\nSpeaker D: So in the case of many user interfaces, they are just so full of buttons that it's actually hard to find the ones you really want to use.\nSpeaker D: And it's just confusing.\nSpeaker D: It takes time to learn.\nSpeaker D: Okay.\nSpeaker D: And I thought I would just quickly show some of them that I found.\nSpeaker D: Okay.\nSpeaker D: Some of them are here.\nSpeaker D: The picture is not very clear.\nSpeaker D: But as you can see there are actually, oh, oh, sorry for that.\nSpeaker D: Let's go back.\nSpeaker D: Oh, please.\nSpeaker D: Okay.\nSpeaker D: So yeah, they are quite big and have many, many buttons.\nSpeaker D: Actually, all these I personally prefer this one because it's the smallest and with least with the smallest number of buttons.\nSpeaker D: And I would say even the appearance of some of them is kind of not so nice.\nSpeaker D: Okay.\nSpeaker D: So let's carry on with this.\nSpeaker D: So other findings.\nSpeaker D: Some new things used.\nSpeaker D: Some of them were mentioned already by our technical designer.\nSpeaker D: Our own company has developed a new user interface.\nSpeaker D: Wait.\nSpeaker D: No, this is not the one.\nSpeaker D: Okay.\nSpeaker D: There is a, we can include voice recognition and it allows.\nSpeaker D: It's possible to record 80 different voice samples on it.\nSpeaker D: So this one was already mentioned, the LC display.\nSpeaker D: Yeah.\nSpeaker D: Another new development is a scroll button, which was also also already mentioned.\nSpeaker D: And our own manufacturing division has designed a new programmable speech.\nSpeaker D: Sorry, a speaker unit, I guess it should be.\nSpeaker D: And this means that once it comes together with a voice recognition, but it's once the, once the, it recognizes the voice of the speaker, there can be a pre-programmed answer.\nSpeaker D: For example, you can pick up the remote control and say something to it like hello and it says hello and your name or whatever.\nSpeaker D: So, I mean, this is also one of the new developments which we might consider if we wanted to.\nSpeaker A: Sorry, can you go back for a second?\nSpeaker A: Are you sure what this means, the spinning wheel with the LC display?\nSpeaker C: It's like the, like you said, no, the scroll button.\nSpeaker D: No, no, the scroll button is a different thing.\nSpeaker D: I have a picture if you, just a moment, I'll show you.\nSpeaker D: I wasn't completely sure myself, but I think it's just like, it's a wheel, it's like not separate buttons.\nSpeaker D: Look, this one here, but I'm not really sure whether you can really turn it around.\nSpeaker D: It's like a press this or this.\nSpeaker E: It's like, you know how you have your mouse and you go around and it's kind of like that and you spin around and it's okay.\nSpeaker A: So instead of going down, you just spin.\nSpeaker E: You just go around and it's a bit weird at first, but it's actually very fast.\nSpeaker E: Like the wheels that click on the side, you get a much slower set.\nSpeaker E: It's quite good if you're like searching quite a lot of stuff.\nSpeaker E: So, you know, if you're looking, you're scrolling through the A to Z and you're looking for something at T, then it's a lot faster than the wheel, but you've got a lot less control over it.\nSpeaker A: I think that you include that here as well as CVs plus spinning.\nSpeaker D: Okay, and the personal preference is pretty much the same as last time.\nSpeaker D: It has to be a small, simple, okay, we decided to include voice recognition.\nSpeaker D: So, the standard major buttons like gone off, the channels and then volume, and then there will be menu on the screen.\nSpeaker D: And I also thought if we want to put small and nice, and actually I quite like the idea of a scrolling button, with voice like, I don't know, like on the movements or something.\nSpeaker D: I think there is no reason why we couldn't do something like this for the remote control.\nSpeaker D: So, yeah, that's it. That's it. Excellent.\nSpeaker C: Okay, straight to trains and then we can discuss it all at once.\nSpeaker B: I can't wait. I have a little bit of a nice, and a little bit of a quick difference.\nSpeaker B: Yeah, if you're too good by stealing it.\nSpeaker B: No, no, no, no.\nNone: It's not the same thing.\nNone: No, no, no, no.\nSpeaker B: No, no, no.\nNone: No, no, no, no, no, no, no, no.\nNone: Oh, no, no, no.\nSpeaker B: Okay, cool.\nSpeaker E: What I always do is search the inserts, come up with market trends, and you know what users are going to be wanting in the near future.\nSpeaker E: Okay.\nSpeaker E: So, the first aspect is about only twice as important as the second aspect, which is twice as important as the third aspect.\nSpeaker E: So, I mean, the easy to use thing is fairly low down on there, which I think given the target group is what you would expect.\nSpeaker E: You know, people want something new, something technologically innovative and different, so the whole idea with the LCDs and the spinning and the colours and the voice recognition is like quite a thing to go for.\nSpeaker E: And yeah, it would still look funny.\nSpeaker A: So, maybe as you're discussing things, is it okay if we just keep highlighting things there?\nSpeaker A: Yeah, yeah.\nSpeaker A: So, so, so, so, so, so probably voice recognition is kind of important, right?\nSpeaker A: Maybe the LCD and spinning.\nSpeaker A: Yeah, okay, I have a point about LCD, I don't know if it is the right point to take it up.\nSpeaker A: LCDs are basically for feedback, right, to the user who is pressing buttons, and the feedback can come through the television itself, so do we need a hand sitting on the remote?\nSpeaker E: It depends how fast your television runs really, don't you think?\nSpeaker E: I only have one of those, tell you where spot is, and you put the number in the remote, and then you wait, and then it goes to the TV, and then you wait, and then it comes, so it actually takes quite a long time.\nSpeaker A: Right.\nSpeaker E: So, if you get the number wrong, then it's a bit of a pain, so I think, you know, a screen on the remote would probably get down the time of that, but like, the roads do tend to get thrown about a bit.\nSpeaker E: Right.\nSpeaker C: It is also quite nice, though, to have something here so you don't interrupt the picture on the screen, so if you're watching something.\nSpeaker C: Yeah, that's true.\nSpeaker E: I think it would be like, I mean, if you can make it integrate with the TV, then it can come up with your information about what's on, and you could just see that on the remote\nSpeaker C: rather than having to interrupt your viewing pleasure.\nSpeaker E: But I think maybe a way to do it would be a similar way to how you have your mobile phone, you know, like you have the slide you want, and you have the flip you want, and then the screen is protected, so it doesn't actually get scratched, so you can have like, what looks like a normal remote control, you know, or like a minimalist remote control, so you got your buttons one to nine, you're on and off, and your volume on that, and then if you want to mess about it, you flip it open, and, yeah.\nSpeaker A: Okay.\nSpeaker A: So we seem to have a consensus that LCDs are definitely the way to go because of style,\nSpeaker E: and yeah, so that kind of decides your whole chip thing.\nSpeaker A: Right. You agree?\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Right.\nSpeaker A: So, okay.\nSpeaker A: LCDs.\nSpeaker A: Definitely.\nSpeaker A: Go on.\nNone: Okay.\nSpeaker E: Okay.\nSpeaker E: Apparently, these are vegetables we'll be providing inspiration, so I discovered your part.\nSpeaker E: So these will be an important feature for clothes, shoes, and furniture.\nSpeaker E: So, I mean, I'm taking this to mean, you know, carviness, you know, because you don't tend to get flat vegetables, you know, possibly even uneven, like, a bit of asymmetry and stuff, but that would be a good way to get in the whole RSI issue in the videos.\nSpeaker E: I mean, if you think most people use the remote control with their right hand, right hand, so you want to, you can't have it so that it's too small for you to use your right hand.\nSpeaker E: Yeah.\nSpeaker E: I'm not quite sure about the relevance of material will be spongy.\nSpeaker C: Something a bit squishy.\nSpeaker C: Yeah.\nSpeaker E: Yeah.\nSpeaker C: We have.\nSpeaker A: It'd be like a robbery.\nSpeaker A: We have a robbery.\nSpeaker A: But, I think it's a good case with using.\nSpeaker E: Well, I suppose you wouldn't get a remote control.\nSpeaker E: Yeah.\nSpeaker C: Yeah.\nSpeaker C: And it'd help if you drop it, to protect it as well.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, if we use latex cases, they won't allow us to use solar cells as an emergency.\nSpeaker A: We could use titanium, food, or plastic.\nSpeaker C: Or if we want to use the latex, then we have to go with one of the other.\nSpeaker C: Yeah.\nSpeaker E: We'll have a nice show.\nSpeaker E: We could get the kinetic energy fairly easily there.\nSpeaker C: From down to down.\nSpeaker C: You can have it as like a little bald.\nSpeaker C: Yeah.\nSpeaker E: Yeah.\nSpeaker E: So, yeah.\nSpeaker A: So, probably double curved surfaces the way to go.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, we're curved at one end and flat on the top, because I'm not sure if it is flat on both the sides, then how much easy would it be to reach for buttons, etc.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: It depends on the whole.\nSpeaker E: Yeah.\nSpeaker E: It's like how you put your hand on a single side curve.\nSpeaker A: Single side curve, double side curve does not say too much does it?\nSpeaker E: No, I don't think it makes a lot of difference.\nSpeaker E: I have one of those slidey bones and the back is essentially straight, but it's curvy.\nSpeaker E: Besides, you have four sides to a thing.\nSpeaker E: I think it's curved.\nSpeaker E: One side means one side is straight.\nSpeaker E: Right.\nSpeaker E: And curved two sides means the whole thing is just a big curvy thing.\nSpeaker A: Did it say anywhere in your research material about the sliding stuff?\nSpeaker A: Because according to the information that I have, I think the only options that we have with the case is that we used to be.\nSpeaker A: Yeah.\nSpeaker A: We either have a black surface to the case or a curved surface case.\nSpeaker A: It does not say anything about whether technically this stuff is available at all.\nSpeaker E: Yeah.\nSpeaker E: It's more about the protection of the LCD.\nSpeaker E: We have one of the keys where it came on.\nSpeaker E: Yeah.\nSpeaker E: My research didn't tell me anything, which is why we have all the pictures.\nSpeaker E: That's what I had to do with my time.\nSpeaker E: Okay.\nSpeaker E: Cool.\nSpeaker E: What have we got?\nSpeaker E: Combined style with a level of functionality, beauty and practicality.\nSpeaker E: And a piece of functionality.\nSpeaker E: Okay.\nSpeaker C: Yes.\nSpeaker C: So, looking at what we've got, we want to now set it up to display with a spinning wheel.\nSpeaker A: Let's try to rub off things.\nSpeaker A: Rub off some of those.\nSpeaker A: So, hand dynamics are definitely out, right?\nSpeaker A: Yeah.\nSpeaker C: Yeah.\nSpeaker C: That's not streamlined and sexy having a wind up.\nSpeaker A: Um, kinetic energy does seem to have some kind of a appeal, but...\nSpeaker A: It's rather practicality really, isn't it?\nSpeaker A: Yeah.\nSpeaker A: As against a watch, which constantly keeps moving, this thing will have to be tapped every time, which might be very frustrating for the user.\nSpeaker C: Depends how much movement it really needs.\nSpeaker B: I don't have to mention that.\nSpeaker C: Presumably, if they're suggesting it, then we could use it.\nSpeaker A: Okay.\nSpeaker A: Let's keep it off.\nSpeaker A: I'll keep it off.\nSpeaker A: Yeah.\nSpeaker A: Um, the flat, completely flat case is definitely out, right?\nSpeaker A: Yeah.\nSpeaker B: It has to be removed from one side.\nSpeaker A: Not vegetable.\nSpeaker A: Um, okay.\nSpeaker A: We still have all the options.\nSpeaker A: What do you think would be very...\nSpeaker D: What is...\nSpeaker D: I can't...\nSpeaker D: How do you...\nSpeaker D: I mean, you can't keep it really small.\nSpeaker D: You can't make it like thin and...\nSpeaker D: Right.\nSpeaker D: I would think because you need to put all the technology in.\nSpeaker D: So, I mean, if the case, you had the case.\nSpeaker D: Yeah.\nSpeaker A: If it is really...\nSpeaker A: If it is really...\nSpeaker A: If it is really thin, it's like you to break.\nSpeaker A: It's like it's much more...\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Given that we're looking at more spongy material preferences, would think maybe rubber or plastic as well.\nSpeaker E: Yeah.\nSpeaker E: That's true.\nSpeaker E: Yeah.\nSpeaker E: It's not a practical...\nSpeaker E: I mean, it's right for a table, but for a micro control, you know?\nSpeaker E: Yeah.\nSpeaker E: And splinters and stuff.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker E: Okay.\nSpeaker E: What is that?\nSpeaker D: It doesn't make any sense, I think.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: In the case of remote control, remote really.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Now for the really interesting stuff, the interface.\nSpeaker A: Yeah.\nSpeaker A: Right.\nSpeaker A: So, the push buttons is our expertise in the industry.\nSpeaker A: But it seems to be out of trend, you know, no big thing.\nSpeaker E: Yeah, but I have some push buttons.\nSpeaker D: I think for the channel numbers, you still need them, wouldn't you?\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker A: Right.\nSpeaker A: So, for channel numbers...\nSpeaker A: If you have LCD displays that opens up a whole world, you know, if you have an LCD display, then you can select almost everything on the LCD display.\nSpeaker D: Yeah, but I think the LCD display is kind of...\nSpeaker D: Yeah, it's faster with...\nSpeaker D: And when we discuss that, we might like the flipping open thing.\nSpeaker D: I mean, you can use it as a normal remote control.\nSpeaker D: Yeah.\nSpeaker D: But if you do want to use LCD, then you flip it open, but it's more time consuming.\nSpeaker E: Yeah.\nSpeaker E: I think this is going back to the graph at the beginning that I made, where, you know, the buttons, the people use all this honey on buttons for them.\nSpeaker E: Yeah.\nSpeaker A: And everything else.\nSpeaker A: Yeah.\nSpeaker A: Many don't.\nSpeaker A: So, in the buttons we have, for the channels also, we have options.\nSpeaker A: Do we enumerate everything from zero to nine?\nSpeaker A: Or do we have just a channel plus channel minus?\nSpeaker D: No, no, I mean, we definitely need the numbers because it's...\nSpeaker D: Otherwise, people don't want to flip through all the channels.\nSpeaker C: Do we need them as buttons or do we need them as LCD?\nSpeaker A: On the LCD we can, you know...\nSpeaker D: Yeah, I would say buttons because it's...\nSpeaker D: Yeah.\nSpeaker E: It's...\nSpeaker E: I think the thing is, if someone just wants to turn on the TV and put on a channel, then it should be easier to use than any other remote.\nSpeaker E: And then if someone wants to, you know, change the contrast on their TV and...\nSpeaker E: Right.\nSpeaker E: They should be able to do that and it should be accessible, but, you know, I mean, most of the time, I mean, there's a limit to how much the biggest technology can spend fiddling with the TV.\nSpeaker E: Right.\nSpeaker E: I think it's the issue here.\nSpeaker A: Okay.\nSpeaker A: So, buttons definitely, but...\nSpeaker A: Shall we try to draw...\nSpeaker C: I think that's what you guys are going to do next.\nSpeaker C: Okay.\nSpeaker C: So, I'm going to put down the key.\nSpeaker A: Okay.\nSpeaker A: So, components.\nSpeaker A: Right.\nSpeaker A: So, what about the scrolling?\nSpeaker D: Yeah, but I'm not completely clear.\nSpeaker D: Yeah.\nSpeaker D: About the spinning wheel.\nSpeaker D: So, I think it doesn't make sense to have both like a scrolling and spinning thing.\nSpeaker D: You cannot include everything in the spinning, if you just spin it.\nSpeaker D: Yeah.\nSpeaker E: I would say the spinning goes at a higher speed to the scrolling wheel.\nSpeaker E: So, you have to decide whether you want to be going so fast or not.\nSpeaker E: But, I mean, the thing with this hole, you're planning on making it out of rubber, and the basis that it's spondy.\nSpeaker E: And I'm not sure how well a scrolling wheel would work.\nSpeaker C: But if you've got a...\nSpeaker C: I mean, you can...\nSpeaker C: If you've got a flip thing, effectively, it's something that's curved on one side and plait on the other side, but you fold a little in half.\nSpeaker E: Yeah, but your spinning wheel tends to go to one side.\nSpeaker C: That would be on one side.\nSpeaker A: I'm not sure it'll be a good idea to construct the whole thing out of rubber.\nSpeaker A: Yeah, I think so too.\nSpeaker A: I mean, the case would be rubber and the buttons.\nSpeaker A: Or at the corner edges, just the edges covered by rubber or something like that, everything else in plastic.\nSpeaker A: Or even titanium if you want to use it.\nSpeaker E: Or maybe like interchangeable cases.\nSpeaker E: Because I know, like, we're going back to eye pods again in the whole spinning wheel, but I have like a...\nSpeaker E: Now, obviously my eye pod is not made of rubber, and then I have a little rubber case that goes over the top of it, and I can change the color.\nSpeaker A: Right.\nSpeaker E: The U.S.A.T. to march my outfit.\nNone: Right.\nSpeaker A: Okay.\nSpeaker A: So that gives us a more trendy look as well.\nSpeaker E: Yeah, I think the spinning wheel is definitely very narrow.\nSpeaker A: Right.\nSpeaker A: Yeah, and we're going more for the trend stand for the usability anyway.\nSpeaker A: That's right.\nSpeaker A: Yeah, after.\nSpeaker A: So, rub that out.\nSpeaker A: And colors can be provided with the case rather than...\nSpeaker A: But we still need to think about the color of our moda such.\nSpeaker D: Yeah, I think it was a requirement that we use our...\nSpeaker D: The colors of our company.\nSpeaker D: So would it be like yellow, gray and black or something?\nSpeaker D: I guess.\nSpeaker E: That doesn't fit in with the whole vegetable thing.\nSpeaker D: Bananas?\nSpeaker D: Bananas?\nSpeaker D: Yellow.\nSpeaker E: Yeah, definitely.\nSpeaker E: I mean, do you think we could incorporate the colors of the company into the buttons and then make the colors the main remote.\nSpeaker E: The color like vegetable colors, you know?\nSpeaker E: So you could have like...\nSpeaker E: I mean, I suppose vegetable colors would be all in green and some reds and maybe purple and that.\nSpeaker E: And then you'd paint the buttons and company colors to match the color.\nSpeaker E: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: If you go over to the integrated circuits, since we are having LCDs, there's no way that we will be able to.\nSpeaker A: What we do need to consider however is that the price is going up with every such thing that we are considering.\nSpeaker A: But since LCDs seems to be a definite yes.\nSpeaker A: So it seems to be one area where we would want to spend.\nSpeaker A: So I'll rub off the other two.\nSpeaker E: So we're discounting solar energy because rubber is going to be used in there somewhere.\nSpeaker E: Or...\nSpeaker C: That was the...\nSpeaker A: Oh, so the constraint was...\nSpeaker A: We can't have solar panels.\nSpeaker A: Solar panels with rubber.\nSpeaker B: So...\nSpeaker E: Yeah, okay.\nSpeaker E: So you really lose that, I think.\nSpeaker C: Should we go for...\nSpeaker C: If we're going for rubber, we think of it as our case.\nSpeaker C: And then...\nSpeaker D: And the buttons as well.\nSpeaker C: Yeah.\nSpeaker A: I think...\nSpeaker A: We'll have...\nSpeaker A: Using the simple battery will be a safer option as compared to the kinetic energy one.\nSpeaker A: I mean, although it does seem interesting.\nSpeaker A: Yeah.\nSpeaker A: But it does not hold any advantages as that.\nSpeaker A: Yeah, it's just to get it.\nSpeaker E: Yeah.\nSpeaker A: Okay.\nSpeaker A: Okay, so we understand this better now that the speaker is for the feedback, right?\nSpeaker A: It says the things that you type in or something like that.\nSpeaker A: Yeah.\nSpeaker C: So...\nSpeaker C: I think if we can include them not too much extra cost, then I'd put them in...\nSpeaker A: We don't have too much information about it.\nSpeaker D: Yeah, but I think it should be quite cheap because it's from a company company.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So this is in as well then.\nSpeaker A: Yeah.\nSpeaker A: So I'm supposed to be good.\nSpeaker C: Okay.\nSpeaker C: And the case is curved on one side but in flat.\nSpeaker C: Flat on the top.\nSpeaker C: Flat. So it's flipped.\nSpeaker C: Into each other.\nSpeaker C: Okay.\nSpeaker C: Can I pull the...\nSpeaker C: Yeah, sure.\nSpeaker C:...the computer.\nSpeaker B: Just so I can...\nSpeaker E: Sorry, Joey.\nSpeaker C: Nothing gets right.\nSpeaker C: I'll just...\nNone: Okay.\nSpeaker E: What is ICSC?\nSpeaker A: ICs, integrated circuits.\nSpeaker E: Okay.\nSpeaker E: So it's advanced integrated circuits?\nSpeaker E: Yeah.\nSpeaker E: Welcome to the...\nSpeaker A: We're definitely going in for voice recognition as well as on CDs.\nSpeaker A: Yeah.\nSpeaker B: Okay.\nSpeaker C: So we've basically worked out that we're going with a simple battery.\nSpeaker C: The advanced chip.\nSpeaker C: And curved on one side case which is folded in on itself.\nSpeaker C: Made out of rubber.\nSpeaker C: And the buttons are also rubber.\nSpeaker C: We're having push buttons on the outside and then on the inside an LCD with spinning wheel.\nSpeaker C: And we're incorporating voice recognition.\nSpeaker C: That's however all contact.\nSpeaker C: And it's going to look sort of digital and be in bright vegetable colours.\nSpeaker D: So would we have the spinning wheel inside with the LCD or would it be on the altar?\nSpeaker D: Imagine being inside.\nSpeaker E: Okay.\nSpeaker E: So actually that could like really cut down your thing.\nSpeaker E: So you've got your outside which is like minimalist and then you open it up and you've got a screen and a spinning wheel which you can incorporate buttons into.\nSpeaker E: So you've still not got like a lot of stuff in there.\nSpeaker E: You've maybe got like if you're modelling on an iPod you've got five buttons and a wheel and fours of it and doing the wheel.\nSpeaker E: And the other one's a little bit inside.\nSpeaker C: Yeah.\nSpeaker C: Sure.\nSpeaker C: Okay.\nSpeaker C: So now we've got 30 minutes before our next meeting.\nSpeaker C: In the meantime the industrial designer over here is going to work on the look and feel design which I presume will work out what that means.\nSpeaker C: The user interface designer will work on the user interface design and the marketing expert is going to work on productive evaluation.\nSpeaker C: And as well as that the two designers are going to work together on our prototype following those instructions that we've just come up with using modelling clay and you will get extra instructions from your personal coach.\nSpeaker C: That all okay.\nSpeaker C: And anyone who hasn't put their presentation in the project documents folder it would be good just so in case we have to refer to it.\nSpeaker E: I'm going to go and sit on my own.\nSpeaker C: Nobody wants to talk to you.\nSpeaker C: I know.\nSpeaker E: I'm plugging in.\nSpeaker E: I've got a bit tangled up in here.\nSpeaker D: Shall I move away first or shall I stay here?\nSpeaker D: I don't know.\nSpeaker D: Maybe.\nSpeaker D: I would care.\n"}, {"meeting_id": "Bmr015", "summary": "The group talked about the status of the first test set of digits data, naming conventions for files, speaker identification tags, and encoding files with details about the recording. The group also discussed a proposal for a grant from the NSF's ITR (Information Technology Research) program, transcriptions, and efforts by speaker mn005 to detect speaker overlap using harmonicity-related features. Particular focus was paid to questions about transcription procedures, i.e. how to deal with overlooked backchannels, and audible breaths.", "dialogue": "Speaker G: We're not crashing anymore.\nSpeaker G: It really bothers me.\nSpeaker G: I crashed.\nSpeaker G: You crashed this morning?\nSpeaker G: I did not crash this morning.\nSpeaker F: Oh, well, maybe it's just how many times you crashed in a day.\nSpeaker F: First time in the day.\nSpeaker J: Maybe it's one to do enough meeting to a crash.\nSpeaker J: No matter of experience.\nSpeaker E: Yeah.\nSpeaker F: That's great.\nSpeaker F: Do we have an agenda?\nSpeaker F: I have an address.\nSpeaker F: Can't come.\nSpeaker G: I have a agenda and it's all me because no one sent me anything else.\nSpeaker J: Did they send the messages to you?\nSpeaker G: I have no idea, but I just got a few minutes ago, right when you were in my office it arrived.\nSpeaker G: Does anybody have any agenda items other than me?\nSpeaker G: I have one more also which is to talk about the digits.\nSpeaker F: Great.\nSpeaker F: I'm just going to talk briefly about the NSFITR.\nSpeaker F: Oh, great.\nSpeaker F: You won't see much, but you said one time about digits.\nSpeaker G: I have a short thing about digits and then I want to talk a little bit about naming conventions, although it's unclear whether this is the right place to talk about it.\nSpeaker G: Maybe just talk about it very briefly and take the details to the people for whom it's relevant.\nSpeaker C: I could always say something about transcription.\nSpeaker F: Well, if we, yeah, we shouldn't add things and just add things in.\nSpeaker F: I actually pre-visit you, so if we're short meeting, would we find?\nSpeaker G: So the only thing I want to say about digits is we are pretty much done with the first test set.\nSpeaker G: There are probably forms here and there that are marked as having been read that weren't really read.\nSpeaker G: So I won't really know until I go through all the transcriber forms and extract out pieces that are an error.\nSpeaker G: So two things.\nSpeaker G: The first is what should we do about digits that were misread?\nSpeaker G: My opinion is we should just throw them out completely and have them read again by someone else.\nSpeaker G: The grouping is completely random, so it's perfectly fine to put a group together again of errors and have them read just to finish out the test set.\nSpeaker G: The other thing you could do is change the transcript to match what they really said.\nSpeaker G: So there's the two options.\nSpeaker F: But there's often things where people do false starts.\nSpeaker F: I know I've done it where I say, say, yeah.\nSpeaker G: What the transcribers did with that is if they did a correction and they eventually did read the right string, you extract the right string.\nSpeaker G: Wait, were they completely wrong?\nSpeaker G: Yeah, didn't correct.\nSpeaker G: And didn't notice, which happens in a few places.\nSpeaker G: So, so.\nSpeaker C: So, so.\nNone: Correct.\nSpeaker G: And so the two options are change the transcript to match what they really said.\nSpeaker G: But then the transcript isn't the error test set anymore.\nSpeaker G: I don't think that really matters because the conditions are so different.\nSpeaker G: And that would be a little easier.\nSpeaker J: How many are, how often is that happen?\nSpeaker G: Five or six times.\nSpeaker J: Oh, so it's not very much.\nSpeaker J: No, it's not much at all.\nSpeaker J: There's a question just change the transcript.\nSpeaker J: Yeah.\nSpeaker F: Okay.\nSpeaker F: Yeah, it's five or six times out of thousands.\nSpeaker F: Four thousand.\nSpeaker F: Four thousand.\nSpeaker F: Yeah, I would do the easy way.\nSpeaker F: Okay.\nSpeaker F: Yeah.\nSpeaker F: It's kind of nice.\nSpeaker F: I mean, who knows what studies people will be doing on speaker dependent things.\nSpeaker F: And so I think having having it all the speakers who we had is least interesting.\nSpeaker J: So, how many digits have been transferred?\nSpeaker G: And so, I think that's a thousand lines in each line is between one and about ten digits.\nSpeaker G: I didn't compute the average.\nSpeaker G: I think the average was around four or five.\nSpeaker F: So that's a couple hours of speech probably.\nSpeaker F: Yeah.\nSpeaker F: Which is a reasonable, reasonable test set.\nSpeaker G: And Jane, I do have a set of forms which I think you have copies of somewhere.\nSpeaker G: Oh, you do.\nSpeaker G: Oh, okay, good.\nSpeaker G: Good.\nSpeaker G: I thought I had had all of them back from you.\nSpeaker G: And then the other thing is that the forms in front of us here that we're going to read later, were suggested by Liz because she wanted to elicit some different prosotics from digits.\nSpeaker G: And so, I just wanted people to take a quick look at the instructions and the way it worked and see if it makes sense.\nSpeaker G: And if anyone has any comments on it.\nSpeaker F: I see.\nSpeaker F: And the decision here was to continue with the words rather than the numerics.\nSpeaker G: Yes, although we could switch it back.\nSpeaker G: The problem was 0 and 0.\nSpeaker G: Although we could switch it back and tell them always to say 0 or always to say 0.\nSpeaker F: Or neither, but it's just two things ways that you can say it.\nSpeaker F: Sure.\nSpeaker F: That's the only thought I have because if you start talking about these, you know, she's trying to get it natural groupings.\nSpeaker F: But there's nothing natural about reading numbers this way.\nSpeaker F: I mean, if you saw telephone number, you would never see it this way.\nSpeaker G: The problem also is she did want to stick with digits.\nSpeaker G: I mean, I'm speaking for her.\nSpeaker G: She's not here.\nSpeaker G: But the other problem we were thinking about is if you just put the numerals, they might say 43 instead of 43.\nSpeaker C: Well, the space though between them.\nSpeaker C: When you space them out, they don't look like 43 anymore.\nSpeaker G: Well, she and I were talking about it.\nSpeaker G: And she felt that it's very, very natural to do that sort of.\nSpeaker G: She's right.\nSpeaker F: It's a different problem.\nSpeaker F: I mean, it's an interesting problem.\nSpeaker F: I mean, we've done stuff with numbers before.\nSpeaker F: And yeah, sometimes people, if you say 3981, sometimes people will say 3981 or 3891.\nSpeaker F: I don't think they'd say that.\nSpeaker F: Not very frequently.\nSpeaker F: But certainly could.\nSpeaker F: Yeah.\nSpeaker F: 3891 is probably how they do that.\nSpeaker G: So, I mean, this is something that Liz and I spoke about.\nSpeaker G: Nice.\nSpeaker G: This was something that Liz asked for specifically.\nSpeaker G: I think we need to defer to her.\nSpeaker F: Okay. Well, we're probably going to be collecting meetings for a while.\nSpeaker F: If we decide we still want to do some digits later, we might be able to do some different versions.\nSpeaker F: But this is the next suggestion.\nSpeaker F: Okay.\nSpeaker F: Okay. So, I guess, let me get my short thing out about the NSF.\nSpeaker F: I said this.\nSpeaker F: Actually, this is maybe a little side thing.\nSpeaker F: I sent to what I thought we had in some previous mail as the right joint thing to send to.\nSpeaker F: It was.\nSpeaker F: MTG, our CDR hyphen joint.\nSpeaker F: Yep.\nSpeaker F: But then I got some sort of funny mail saying that the moderator was going to.\nSpeaker G: That's because they set the one up at UW.\nSpeaker G: That's not on our side.\nSpeaker G: That's on the UW website.\nSpeaker G: Oh.\nSpeaker G: And so UW set it up as a moderated list.\nSpeaker G: Oh.\nSpeaker G: And I have no idea whether it actually ever goes to anyone.\nSpeaker G: So you might want to just mail to Mari.\nSpeaker F: No, I got a little excited notes from Mari and Jeff and so on.\nSpeaker F: Okay. Good.\nSpeaker G: Yeah.\nSpeaker G: So the moderator actually did repost it.\nSpeaker G: Yeah.\nSpeaker G: Because I had sent one earlier.\nSpeaker G: Actually, the same thing happened to me.\nSpeaker G: I had sent one earlier.\nSpeaker G: The message says you'll be informed and then I was never informed.\nSpeaker G: But I got replies from people indicating that they had gotten it.\nSpeaker G: Right.\nSpeaker G: It's just to prevent spam.\nSpeaker G: I see.\nSpeaker F: Yeah.\nSpeaker F: So, okay.\nSpeaker F: Anyway, I guess everybody here, you are on that list, right?\nSpeaker F: So you got to know.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: So this was a proposal that we put in before on more, more higher level issues in meetings from, I guess, higher level from my point of view.\nSpeaker F: Yeah.\nSpeaker F: And meeting mappings.\nSpeaker F: And so it was a proposal for the ITR program, the Information Technology Research Program, as part of the National Science Foundation.\nSpeaker F: It's the second year of there doing these grants.\nSpeaker F: There are a lot of them, some of them anyway, but they're larger grants than usual, small NSF grants.\nSpeaker F: So they're very competitive and they have a first phase where you put in pre-proposals and we got through that.\nSpeaker F: And so the next phase will be, you will actually be doing a larger proposal.\nSpeaker F: And I hope to be doing very little of it, which was also true for the pre-proposals.\nSpeaker F: So, there are a bunch of people working on it.\nSpeaker G: When's the full proposal, Tim?\nSpeaker F: I think April 9th or something.\nSpeaker F: So that's about a month.\nSpeaker G: And they said, end of business day, you could check on the reviewer forms.\nSpeaker F: Tomorrow.\nSpeaker F: Tomorrow, March 2nd.\nSpeaker G: Might be in a day off all week.\nSpeaker G: I guess that's a good thing, because I mean, I got my papers on.\nSpeaker F: So it's amazing you showed up at this meeting.\nSpeaker G: It is, it is actually quite amazing.\nSpeaker J: Let me just see the reviewers comment.\nSpeaker J: Yeah.\nSpeaker F: Yeah.\nSpeaker F: My favorite is when one reviewer says, this should be far more detailed.\nSpeaker F: And the next reviewer says, there's way too much detail.\nSpeaker F: Yes.\nSpeaker G: This is way too general.\nSpeaker G: The other reviewer says this is way too specific.\nSpeaker G: Yeah.\nSpeaker G: It's way too hard.\nSpeaker F: Way too easy.\nSpeaker F: We'll see.\nSpeaker G: Maybe there'll be something useful.\nSpeaker G: Well, it sounded like the first gate was pretty easy.\nSpeaker G: Is that right?\nSpeaker G: That they didn't reject a lot of the pre-proposals.\nSpeaker F: Do you know anything about the numbers?\nSpeaker J: No.\nSpeaker J: I don't think that's true.\nSpeaker J: He said the next phase will be very competitive because we didn't want to weed out much in the first phase.\nSpeaker J: Well, I have to see what the numbers are.\nSpeaker J: Yeah.\nSpeaker J: But they have to weed out enough so that they have enough reviewers.\nSpeaker F: Right.\nSpeaker F: So maybe they weed out as much as they want.\nSpeaker F: But it's usually pretty.\nSpeaker F: Yeah, it's certainly not.\nSpeaker F: I'm sure that it's not down to one and two or something.\nSpeaker F: Right.\nSpeaker F: It was left.\nSpeaker F: I'm sure it's.\nSpeaker G: How many awards are there?\nSpeaker F: Well, there's different numbers of awards for different size.\nSpeaker F: They have three size grants.\nSpeaker F: Let's find this one.\nSpeaker F: Let's see.\nSpeaker F: The small ones are less than 500,000 total over three years.\nSpeaker F: And that they have a fair number of them.\nSpeaker F: And the large ones are.\nSpeaker F: Oh, I forget.\nSpeaker F: I think more than.\nSpeaker F: More than a million and a half, more than two million or something like that.\nSpeaker F: And we're in the middle.\nSpeaker F: Middle category.\nSpeaker F: I think we're.\nSpeaker F: I forget it was.\nSpeaker F: But.\nSpeaker F: I don't remember.\nSpeaker F: But it's probably.\nSpeaker F: I don't remember.\nSpeaker F: But it's probably along the line.\nSpeaker F: I could be wrong in this.\nSpeaker F: It would probably along lines of 15, that they'll find the 20.\nSpeaker F: I mean, when they, do you know how many they funded when they've been in, in, in Chucks?\nSpeaker F: They got.\nSpeaker F: I see.\nSpeaker F: Yeah.\nSpeaker G: I thought it was smaller and that was like four or five.\nSpeaker G: Well, they fun.\nSpeaker F: They, yeah.\nSpeaker F: I mean, we'll find out one more.\nSpeaker F: Yeah.\nSpeaker F: I mean, last time I think they just had two category small and big.\nSpeaker F: And this time they came up with a middle one.\nSpeaker F: So it'll, they'll be more of them that they fund than of the big.\nSpeaker J: If we end up getting.\nSpeaker J: What will it be in the context in terms of where will the money go to what we'll be doing with it?\nNone: Exactly.\nSpeaker G: We say in the proposal.\nSpeaker J: I mean, I was far enough.\nSpeaker F: You know, none of it will go for the yachts that we've been talking about.\nSpeaker F: Well, you know, I mean, it's just for the research.\nSpeaker J: It's extending the research, right?\nSpeaker G: Because the other higher level stuff than we've been talking about for a record.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: The other things that we have been working on with the communicator, especially with the newer things, with the more acoustically oriented things, are lower level.\nSpeaker F: And this is dealing with mapping on the level of the conversation, mapping the conversation to different kind of planes.\nSpeaker F: So.\nSpeaker F: But so it's all stuff that none of us are doing right now.\nSpeaker F: None of us are funded for.\nSpeaker F: So it's, so it's, it would be new.\nSpeaker J: So assuming everybody's busy now.\nSpeaker J: I mean, it's going to be higher more students.\nSpeaker F: Well, there's evenings in this room.\nSpeaker F: Yeah, there would be, there would be cars.\nSpeaker F: And there would be expansion.\nSpeaker F: But also, there's always, for everybody, there's always things that are dropping off, grants that are renting or other things.\nSpeaker F: So there's a continual need to bring in new things.\nSpeaker F: But there definitely would be new students and so forth, both of them.\nSpeaker G: Are there any students in your class who are expressing interest?\nSpeaker F: Not clear yet.\nSpeaker F: Other than the one who's already here.\nSpeaker F: I mean, we got, yeah, two in the, two in the, two in the class already here.\nSpeaker F: And then, and then there's a third who's doing a project here.\nSpeaker F: But he won't be in the country that long.\nSpeaker F: Maybe another one.\nSpeaker F: Yeah, actually, there's one other guy who's looking at that, that, that, Jeremy, I think.\nSpeaker F: Anyway, yeah, that's, that's how I was going to say is that, that's, you know, that's nice and we're sort of proceeding to next step.\nSpeaker F: And it'll mean some more work, you know, in March and getting a proposal out.\nSpeaker F: And then it's, you know, we'll see what happens.\nSpeaker F: The last one was that he had their whistle naming.\nSpeaker G: Yeah, it just, we've been cutting up sound files and, for both digits and for doing recognition.\nSpeaker G: And Liz had some suggestions on naming and it just brought up the whole issue that hasn't really been resolved about naming.\nSpeaker G: So one thing she would like to have is for all the names to be the same length.\nSpeaker G: So that sorting is easier.\nSpeaker G: Same number of characters so that when you're sorting file names, you can easily extract out bits and pieces that you want.\nSpeaker G: That's easy enough to do.\nSpeaker G: And I don't think we have so many meetings that that's a big deal just to change the names.\nSpeaker G: So that means instead of calling it MR1, MR2, you call it MRM001, MRM002, things like that, just so that they're all the same length.\nSpeaker C: But you know, when you do things like that, you can always, as long as you have, you can always search from the beginning of the end of the string.\nSpeaker G: The problem is that there are a lot of fields.\nSpeaker G: Right, so we have, we're going to have the speaker ID, the session, information on the microphones, information on the channels and all that.\nSpeaker G: And so if each one of those is a fixed length, the sorting comes a lot easier.\nSpeaker B: She wanted to keep the same length across different meetings also.\nSpeaker B: So like the NSA meeting links, file names are going to be the same length as the media recording meeting names.\nSpeaker G: And as I said, we just don't have that many that that's a big deal.\nSpeaker G: And so at some point we have to sort of take a few days off, like the transcribers have a few days off, make sure no one's touching the data and reorganize the file structures.\nSpeaker G: And when we do that, we can also rationalize some of the naming.\nSpeaker C: I would think though that the transcribes themselves wouldn't need to have such lengthy names.\nSpeaker C: So I mean, you're dealing with a different domain there, I mean, with starting n times and all that, channels and stuff.\nSpeaker G: So the only thing we would change with that is just the directory names, I would change them to match.\nSpeaker G: So instead of being MR1, it would be MRM001, but I don't think it's a big deal.\nSpeaker G: So for the meetings, we were thinking about three letters and three numbers for meeting IDs, for speakers, M or F, and then three numbers.\nSpeaker G: And that also brings up the point that we have to start assembling a speaker database so that we get those links back and forth and keep it consistent.\nSpeaker G: And then the microphone issues, we want some way of specifying more than looking in the key file, what channel and what mic, what channel, what mic, and what broadcast, or I don't know how to say it.\nSpeaker G: So with this one, it's this particular headset, with this particular transmitter, as a wireless, and you know, that one is a different headset and different channel.\nSpeaker G: And so we just need some naming conventions on that. And that's going to be come especially important once we start changing the microphone setup.\nSpeaker G: We have some new microphones that I'd like to start trying out once I test them, and then we'll need to specify that somewhere.\nSpeaker G: So I was just going to do a fixed list of microphones and types.\nSpeaker F: As I said, yeah. I'm sure it's such a short agenda list, I guess I will ask how are the transcription skills.\nSpeaker C: But the news is that I switched to start my news sense. I switched to doing the channel by channel transcriptions to provide a tighter time bins for partly for use in Teal's work and also its environments and other people in the project.\nSpeaker C: And I discovered in the process a couple of interesting things, which one of them is that it seems that there are time lags involved in doing this using an interface that has so much more complexity to it.\nSpeaker C: And I wanted to maybe ask Chuck to help me with some of the questions and efficiency. I was thinking maybe the best way to do this in the long run, maybe to give them single channel parts and then piece them together later.\nSpeaker C: And I have a script I can piece them together. So it's like I know that I can take them apart and put them together and end up with the representation, which is where the real power of that interface is.\nSpeaker C: And it may be it's faster to transcribe the channel at a time with only one sound violin, one set of utterances to check through.\nSpeaker F: I'm a little confused. I thought that that one of the reasons we thought we were so much faster than than the other transcription thing was that that we were using the mix.\nSpeaker C: Oh yes. Okay. But with the mix, when you have an overlap, you only have a choice of one start and end time for that entire overlap, which means that you're not tightly tuning the individual parts of that overlap by your foot speaker.\nSpeaker C: So someone may have only said two words in that entire big chunk of overlap. And for purposes of things like, well, some things like training the speech non speech segmentation thing.\nSpeaker C: It's necessary to have it more tightly tuned than that. And you know, it would be wonderful if it's possible then to use that algorithm to more tight be tying all the channels after that.\nSpeaker C: But I don't know exactly where that's going at this point, but I was experimenting with doing this by hand. And I really do think that it's wise that we've had them start the way we have with working off the mix signal, having the interface that doesn't require them to do the type, the type in every single channel.\nSpeaker C: The entire interaction. I did discover a couple other things by doing this though. And one of them is that once in a while a back channel will be overlooked by the transgarber as you might expect because when a back channel could well happen in a very densely populated overlap.\nSpeaker C: And if we're going to study types of our laps, which is what I want to do analysis of that, then that really tests require listening to every single channel all the way through the entire length for all the different speakers. Now, for only four speakers, that's not going to be too much time. But if it's nine speakers, then that is more time.\nSpeaker C: So it's like, you know, kind of wondering. And I think again, it's like this, it's really valuable that he was working on the speech non-speech segmentation because maybe we can close it out without having to actually go to the time that would take to listen to every single channel from start to finish the every single need.\nSpeaker H: Yeah, but those spectacles will always be a problem. I think, especially if they're really short and they're not very loud. And so it can, it will always happen that also the automatic detection system will miss some of them.\nSpeaker C: Also, then maybe the answer is to listen, especially densely in places of overlap, just so that there is not a overlooked because of that and count on accuracy during the sparse faces.\nSpeaker C: Because there are large spaces at a, that's a good point. There are large spaces where there's no overlap at all. Someone's giving a presentation.\nSpeaker C: Yeah. Whatever. That's a good, that's a good point. And let's see, there was only a thing I was going to say.\nSpeaker C: I think it's really interesting data to work with. I have to say it's very enjoyable. Really, not a problem spending time with these data.\nSpeaker C: I'm not just because I'm in there.\nSpeaker F: Well, I think it's a short meeting. You're still on the midst of what you're doing from the script last time, I assume.\nSpeaker D: I have a result, but I'm continuing working with the mix side now, after the last experience.\nSpeaker D: And I tried to adjust to improve in our own city, the test, that I implement.\nSpeaker D: But I have a problem because I get very much harmonics now.\nSpeaker D: I'm only possible harmonics. And now I'm trying to find some kind of help using the energy to distinguish between possible harmonics and other frequency picks.\nSpeaker D: And I have to talk with you, with the group, about the instantaneous frequency, because I have an algorithm.\nSpeaker D: And I get similar results, like the paper that I am following. But the rules that people use in the paper to distinguish the harmonics doesn't work well.\nSpeaker D: And I'm not sure that the way to obtain the instantaneous frequency is right. It's not right.\nSpeaker D: I haven't enough feeling to distinguish what happened.\nSpeaker F: I'd like to talk with you about it. If I don't have enough time, and you want to discuss with someone else, besides us, that you might want to talk to, might be Stefan.\nSpeaker J: I'm not going to do experience for this. The experience is not enough.\nSpeaker D: I don't process the fundamental. I calculate the phase derivate using the F50. The algorithm said that if you change the frequency X using the instantaneous frequency, you can find how in several frequencies, the rest of the frequency picks, the frequency picks, the frequency harmonic, and if you compare the instantaneous frequency of the continuous filters, the use to get the instantaneous frequency is probably to you can find that instantaneous frequency for continuous output continuous filters are brilliant.\nSpeaker F: I'd have to look at that and think about it. I haven't worked with that either. The simple-minded way I suggested was what Doug was just saying is that you can make a sieve.\nSpeaker F: Let's hypothesize that it's this frequency. Maybe you could use some other cute methods to shortcut it by making some guesses.\nSpeaker F: I mean, you could make some guesses from the autocorrelation or something, but then, given those guesses, try only looking at the energy at multiples of that frequency and seeing how much it takes to one that's maximum.\nSpeaker D: Using the energy of the multiples of frequencies. You have to do some kind of low pass filter before you do that. I don't use.\nSpeaker D: I know many people use low pass filter to get the pitch.\nSpeaker J: I'm going to try the vocal track, the response of the vocal track. Just looking at the energy and those, that's the harmonic.\nSpeaker F: The thing is that this is for a...\nSpeaker F: I don't need to get rid of it. I mean, that'd be nice, but I don't know if it's essential. I mean, because I think the main thing is that you're trying... what are you doing this for?\nSpeaker F: You're trying to distinguish between the case where there are more than one speaker. And the case where there's only one speaker.\nSpeaker F: You're not distinguishing voice to non-voice. If you don't care about that, see if you also want to determine whether it's on voice, then I think you want to look at high frequencies also because the fact that there's more energy in high frequencies is going to be sort of obvious cue that it's on voice.\nSpeaker F: But other than that, I guess, as far as the one person versus two persons, it would be primarily low frequency phenomenon.\nSpeaker F: And if you look at the low frequencies, yes, the higher frequencies are going to be a spectral slope. The higher frequencies would be lower energy.\nSpeaker D: I would be there for the next week, all my results about the Dharmonicity and we try to comment on this case here because I have enough feeling to understand what happened with so many peaks.\nSpeaker D: I see Dharmonics in many times, but there are a lot of peaks that have no harmonics.\nSpeaker D: I have to discover what is the best way to...\nSpeaker F: I don't think you're not going to be able to look at every frame. I really thought that the best way to do it, and I'm speaking with no experience on this to a good point, but my impression was the best way to do it was however you use instantaneous frequency, however you clope with your candidates, you want to see how much of the energy is in that.\nSpeaker F: It's supposed to all of the total energy. And if it's voiced, I guess so. So I think maybe you do need a voiced and voiced determination too, but if it's voiced and the fraction of the energy that's in the harmonic sequence that you're looking at is relatively low, then it should be more likely to be an overlap.\nSpeaker D: This is the idea I had to compare the ratio of the energy of the harmonics with the total energy in the spectrum, try to get the ratio to distinguish between overlap and speech.\nSpeaker F: But you're looking at... let's take a second of this. You're looking at the phase derivative in what domain... I mean this is in bands or...\nSpeaker D: No, no, no. Just overall...\nSpeaker D: The band is from 0 to 4 kilohertz. You just take the instantaneous frequency.\nSpeaker D: I use two methods. One, basically, on FFT to FFT to obtain the... or to study the harmonics from the spectrum directly, and to study the energy and the multiples of frequency. And another... another algorithm I have is the instantaneous frequency.\nSpeaker D: I use FFT to calculate the phase derivative in the time. I mean, I have two algorithms. But in my opinion, the instantaneous frequency, the behavior was very interesting, because I saw how the spectrum concentrates around the harmonic.\nSpeaker D: When I apply the rule of the instantaneous frequency of the continuous filter, the rule that people propose in the paper doesn't work, and I don't know why.\nSpeaker F: The instantaneous frequency wouldn't give you something more like the central frequency of the way most of the energy is. I mean, I think, why would it correspond to pitch?\nSpeaker D: I'm not sure. I try to... First, I calculate using the FFT, I get the spectrum, I represent all the frequency.\nSpeaker D: When I obtain the instantaneous frequency, and I change the disease using the instantaneous frequency here.\nSpeaker D: I use a scaling along that axis according to the instantaneous frequency. I use this frequency. The range is different, and the resolution is different.\nSpeaker D: More or less, seeing like this. And the paper said that these frequencies are probably harmonics.\nSpeaker D: But they used a rule based on the... Because to calculate the instantaneous frequency, they used a handing window.\nSpeaker D: They said that if these big harmonics, the instantaneous frequency of the continuous filters are very near.\nSpeaker D: I don't know what is the distance. I try to put different distance, to put different length of the window, different frequency.\nSpeaker F: I guess I'm not following enough. I'm not going to have time to do any of these.\nSpeaker C: I get it in the return of the transition. There's one third thing I wanted to raise is an issue, which is how to handle breaths.\nSpeaker C: The reason I asked the question is, aside from the fact that there are times when we took code, the fact that I have the indication from down ellis in the email that I sent to you.\nSpeaker C: I think that the question is, whether it would be possible to eliminate them from the audio signal, which would be the ideal situation.\nSpeaker C: I don't think it would be ideal. We're dealing with real speech. We're trying to have these real as possible and breaths are part of real speech.\nSpeaker C: We're hearing you breathing as if you're hearing our ear. It's like, I mean, breath is natural, but not...\nSpeaker G: I think the BDA application would have to cope with breath. The BDA might not have to, but more people than just BDA users are interested in this corpus.\nSpeaker G: We could remove it, but I think we don't want to remove it from the corpus in terms of delivering it because people will want it in there.\nSpeaker F: If it gets in the way of what somebody is doing with it, then you might want to have some method which will allow you to block it, but it's real data.\nSpeaker F: If there's a little bit of noise out there and somebody is talking about something they're doing, that's part of what we accept is part of a real meeting.\nSpeaker F: We have the fan and the projector up there. This is actual stuff that we want to work with.\nSpeaker C: This is very interesting because it shows very clearly the contrast between speech recognition research and discourse research.\nSpeaker C: Discourse in linguistic research is what's communicative. Once in a while, breath is communicative, but very rarely.\nSpeaker C: I had a discussion with Chuck about the data structure and the idea is that the transcripts will get stored as a master.\nSpeaker C: There will be a master transcript which has in it everything that's needed for both of these uses.\nSpeaker C: The one that's used for speech recognition will be processed by a script. Don's been writing scripts and two processes for speech recognition side.\nSpeaker C: Discourse side will have this side over here. The discourse side will have a script which will strip away the things which are non-communicative.\nSpeaker C: Let's think about the practicalities of how we get to that master copy with reference to breaths. What I would wonder is would it be possible to encode those automatically?\nSpeaker C: Could we get a breath detector?\nSpeaker G: Just to save the transcribers.\nSpeaker C: You just have no idea. If you're getting a breath several times every minute and just simply the keystrokes it takes to negotiate to put the boundaries in to type it in, it's just a huge amount of time.\nSpeaker C: You want to be sure it's used and you want to be sure it's done as efficiently as possible and it's done automatically. That would be ideal.\nSpeaker F: What if you put it in but put the boundaries?\nSpeaker C: You just know it's between these other things. The time boundaries could mark off words from non-words. That would be extremely time effective if that's sufficient.\nSpeaker F: If it's too hard for us to annotate the breaths per se, we are going to be building up models for these things and these things are somewhat self-aligning.\nSpeaker F: If we say there is some kind of a thing which we call a breath or a breath in or breath out, the models will learn that sort of thing.\nSpeaker F: But you do want them to point them at some region where the breaths really are.\nSpeaker C: That would maybe include a pause as well and that wouldn't be a problem.\nSpeaker F: There's this dynamic tension between marking absolutely everything and marking just a little bit and touting on the statistical methods.\nSpeaker F: Basically the more we can mark the better. But if there seems to be a lot of effort for a small amount of reward in some area, this might be one like this.\nSpeaker F: Although I'd be interested to get input from those Andreas on this to see if they've got lots of experience with breaths in...\nSpeaker F: I have lots of experience breathing.\nSpeaker F: Well, yes, they do but we can handle that without it here. But you're going to say something about that.\nSpeaker J: I think one possible way to handle it is that as the transcribers are going through and if they get a hunk of speech, they're going to transcribe it.\nSpeaker J: They're going to transcribe it because there's words in there or whatnot.\nSpeaker J: If there's a breath in there, they could transcribe that.\nSpeaker C: That's what they've been doing.\nSpeaker C: So within overlap segments.\nSpeaker J: But if there's a big hunk of speech, let's say on Morgan's mic where he's not talking at all, don't worry about that.\nSpeaker J: So what we're saying is there's no guarantee that...\nSpeaker J: So for the chunks that are transcribed, everything's transcribed.\nSpeaker J: But outside of those boundaries, there could have been stuff that wasn't transcribed.\nSpeaker J: So you just... somebody can't rely on that data and say that's perfectly clean data.\nSpeaker J: Do you see what I'm saying?\nSpeaker J: I haven't said, don't tell the transcribe anything that's outside of...\nSpeaker F: That sounds like a reasonable compromise.\nSpeaker I: And that's quite a corresponds to the way I try to train the speech and speech detector.\nSpeaker I: I really try not to detect those breaths, which are not with a speech-strung, but with just an silence region.\nSpeaker I: So they hopefully won't be mocked in those channel-specific files.\nSpeaker F: I wanted to comment a little more just for clarification about this business, about the different purposes.\nSpeaker F: In a way, this is a really key point.\nSpeaker F: That for speech recognition research, it's not just a minor part.\nSpeaker F: In fact, I would say the core thing that we're trying to do is to recognize the actual meaningful component.\nSpeaker F: The meaningful components in the midst of other things that are not meaningful.\nSpeaker F: So it's critical, it's not just incidental, it's critical for us to get these other components that are not meaningful.\nSpeaker F: Because that's what we're trying to pull the other out. That's our problem.\nSpeaker F: If we had only linguistically relevant things, if we only had changes in the spectrum that were associated with words with different spectral components, and we didn't have noise, we didn't have convolutional errors, we didn't have extraneous behaviors and so forth, moving your head and all these sorts of things, then actually speech recognition isn't that bad right now.\nSpeaker F: I mean, you know, it's the technology has come along pretty well.\nSpeaker F: The reason we so complain about it is because when you have more realistic conditions, then things fall apart.\nSpeaker C: Okay, I guess what I was wondering is what at what level does the breathing aspect enter into the problem?\nSpeaker C: Because if it were likely that a PDA would be able to be built, which would get rid of the breathing, so it wouldn't even have to be processed at this computational, it would have to be computationally processed, rid of it.\nSpeaker C: But if there were likely on the frontier, a good breath extractor, then...\nSpeaker C: So that's a research question, you know?\nSpeaker F: I've seen that's what I wouldn't know. And we don't either.\nSpeaker F: So the thing is, right now it's just data that we're collecting, and so we don't want to presuppose that people will be able to get rid of particular degradation, because that's actually the research that we're trying to feed.\nSpeaker F: So, you know, maybe in five years it'll work really well, and it'll only mess up 10% of the time, but then we would still want to account for that 10%.\nSpeaker C: I guess there's another aspect which is that as we've improved our micro-technique, we have a lot less breadth in the more recent recordings. So it's in a way it's an artifact that there's so much on the earlier ones.\nSpeaker J: I see.\nSpeaker J: One of the... just to add to this, one of the ways that we will be able to get rid of breath is by having models, I mean that's what a lot of people do now.\nSpeaker J: In order to build a model, you need to have some amount of it, right?\nSpeaker J: I don't think we need to worry a lot about breaths that are happening outside of a conversation.\nSpeaker J: You don't have to go and search for them to mark them at all, but they're there while they're transcribing some comfort that put them in possible.\nNone: Okay.\nSpeaker C: It's also the fact that they do for a lot in one channel to the other because of the way the microphone suggests.\nSpeaker F: Should we do a diddys?\nSpeaker F: Yep.\nSpeaker G: Okay, this is transcript. L173, 4368, 136593, 7308, 591761, 56010, 6395, 038510, 868, 64651, 878, 873, 482434, 6432, 0259, 42086, 289, 671, 2916, 4399, 76303, 277, 556, 390.\nSpeaker J: Transcript L-257, 6200, 0129, 0236, 7034, 6368, 463, 236, 418, 0649, 58, 2316, 775, 907, 724, 4, 014, 264, 845, 1, 877, 45, 872, 2, 495, 41015, 5105, 022, 0206, 6556, 1, 858,\nSpeaker D: 1, 888, 1, 877, 1, 847, 234, 603, 939, 25, 666, 640, 3, 8, 8, 8, 8, 975, 3-4, 6-0-3, 9-39-25, 6-6, 6-4-0, 3-5-13, 9-8, 6-5-3, 3-7-1-3-0, 5-6-0-1, 4-9-8-7, 7-29, 6-1-1, 4-6-3, 4-3-5, 1-6-9-1-0, 0-2, 9-0, 0-0, 3-1, 7-3.\nSpeaker I: Transcript L559, 4-2, 9-1, 8-8, 4-2, 2-9, 5-2, 3-5, 2-8, 3-6, 3-8, 6-8, 4-9, 1-4, 7-1, 5-6, 4-8-3, 7-2, 4-1-8-0, 0-1-1-3, 5-2-1-6, 3-4, 7-5-07, 3-5-20, 1-19-1, 0-9, 7-0-5-1, 2-5-1-3, 5-2-7, 3-4-6-3, 5-3-0-7, 4-0-1-2, 1-8-0, 5-9-5, 9-3-9.\nSpeaker F: Transcript L493, 9-3-4, 6-8, 5-2-6-8, 0-4-3, 8-6-4, 2-3-9-5, 4-9-2, 1-1-0, 8-7-1-0, 9-5-3, 0-3-0, 6-8-4, 8-5-7-1, 5-0-9-2, 9-3-8-9, 8-3-2-1, 3-1-2-9, 2-7-3-6, 9-9-8-9, 9-5-7-4, 7-3-0-5-4-3-5-3-3-1, 6-6-1-3-9-5-5-4-2, 6-5-1-9-5-8-9-3-8-5.\nSpeaker A: Transcript L-31, 3-271-9-6-6-5, 6-08-3-1-6-8-8-5-1-1, 7-09-4-9-2-1-6-9-7, 4-7-01-4-9-9-0-2-5, 3-7-9-2-4-2-7-9-0, 8-6-1-6-4-0-7-3-8, 5-8-6-4-4-3-8-7-7, 2-8-7-6-8-5-9-9-2, 4-8-6-8-5, 0-8-1-6-3-9-1-6-8-9.\nSpeaker C: Transcript L-6-2-1, 0-2-3-1-9-5-8-5-4, 1-5-0-1-1-8-8-3, 9-6-6-3-2-5-7-8-7-9, 2-0-6-4-3-4-6-6-0, 3-6-2-3-6-8-3-5-2-4, 1-7-2-6-1-9-7-2-6-5, 8-9-8-7-3-2-2-1-3-8, 2-3-3-7-7-1-9-5, 4-8-8-7-6-1-3-5-3-3-5-7, 9-5-8-5-1-5-4-2-2.\n"}, {"meeting_id": "IS1006c", "summary": "The meeting was mainly about conceptual design. In the beginning, the group focused on industrial components: energy, material and interface. The industrial designer preferred traditional batteries, titanium and simple push buttons so that they would have enough money for speech recognition. After that, the group talked about LCD display and locating methods. The second topic was user interface concept, which was basically a good-looking design with few buttons. The group decided to change the colour of the assembly at an extra charge to meet special needs. To make a smart controller, they mentioned speech recognition and user identification. When it came to trend watching, the marketing put forward three noteworthy aspects - being fancy, technologically innovative and useful. The group agreed to follow the trend but the specific design was not decided.", "dialogue": "Speaker C: So now we are in the conceptual design meeting.\nSpeaker C: It's been close to last.\nSpeaker C: It's the penalty meeting.\nSpeaker E: How was lunch?\nSpeaker E: Great!\nSpeaker E: Thanks!\nSpeaker F: Don't be sarcastic!\nSpeaker D: Spark!\nSpeaker C: So, I will again go to the secretary part.\nSpeaker C: We will have a presentation first.\nSpeaker C: We will have a design, then mark, and then Sami.\nSpeaker C: We have to take a decision on remote control concepts.\nSpeaker C: We have 40 minutes.\nSpeaker C: So, what we want to take on this meeting are on the first component concept.\nSpeaker C: So, what kind of energy we use, what kind of chip on print, and what kind of case, and also on user interface concept, what kind of interface we use, and if there is some supplements.\nSpeaker C: At the end, Sami will give a trend watching what is being doing.\nSpeaker C: So, let's go.\nSpeaker C: First, what is it?\nSpeaker C: Ramana.\nSpeaker A: Two.\nSpeaker A: Yeah, but it's different.\nSpeaker A: Component?\nSpeaker A: So, we have to mainly design, mainly need to know which components we will use for energy and the material and the interface.\nSpeaker A: For energy, there are maybe two or three possibilities.\nSpeaker A: First, one, we can use simple battery, or we can use traditional solar cells, and then material, we can have plastic, rubber, which is good for this RSI.\nSpeaker A: And then a titanium, which can be, which have very good look.\nSpeaker A: And then interface, we have to use push buttons or liquid crystal LCD display, and we can use some jacks, moving jacks kind of thing.\nSpeaker A: So, as we discussed before, we would like to have some feature recognition chip in our remote control.\nSpeaker A: So, this can be simple kind of programmable chip, which can use microphone as well as some sound sensors.\nSpeaker A: And we also want to locate our remote control.\nSpeaker A: So, still we are looking for possible technical specifications and how easy we can do and within our range, like we have to do in our 12 euros or around that.\nSpeaker A: So, we are looking for simple devices or simple technology to do the location of remote control in a room or in house.\nSpeaker A: So, if we discussed and, excuse me, so we would like to propose better instead of solar cells and it would be problematic to have enough energy with solar cells and so we would like to just use simple battery.\nSpeaker A: And also we want to go for titanium design instead of rubber or hand.\nSpeaker A: But the problem is with this design, we found that we can't use double curved shapes.\nSpeaker A: What is a double curved shape?\nSpeaker A: Like you can have two curves.\nSpeaker A: Aha, wow.\nSpeaker A: It's, I think in manufacturing, I guess it's problematic.\nSpeaker A: So, we want to go for simple push buttons because it needs simple chip and it's really less expensive compared to LCD, which need advanced chip technology and it's more expensive.\nSpeaker A: Since we want to put some other features such as speech recognition, we want to reduce the process.\nSpeaker C: I want to know why, just to, sorry, for the point before, why not the rubber?\nSpeaker C: It is something that it seems to be...\nSpeaker A: And also in if you put a bird, it's a bit difficult to do all the molding of buttons and these things.\nSpeaker C: And you, titanium, it's more...\nSpeaker E: Yeah.\nSpeaker E: What can you use something like, you know, the whole body's titanium, but there are some rubber, you know, some rubber parts like...\nSpeaker A: Yeah.\nSpeaker A: To make it feel better.\nSpeaker A: Like in cell phones recently, you can move with the rubber in four directions.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: But full assembly, we will use mainly for titanium.\nSpeaker A: Okay.\nSpeaker A: Rubber is expensive.\nSpeaker A: Okay.\nSpeaker A: And also it's a bit difficult to do all the shit.\nSpeaker A: Okay.\nSpeaker A: And this push button, we would like to use push button since LCD is in.\nSpeaker A: So we want to move...\nSpeaker A: I mean, we want to put speech recognition, so we want to reduce price on this technology and so that we can have enough space or enough money.\nSpeaker C: So we have a simple button and speech recognition for the more complicated.\nSpeaker A: Yeah.\nSpeaker A: We have simple buttons and speech recognition technology.\nSpeaker A: Okay.\nSpeaker E: And still we have...\nSpeaker E: Can we still include the LCD display?\nSpeaker E: LCD.\nSpeaker E: LCD, yeah.\nSpeaker E: It seems not.\nSpeaker E: It's either.\nSpeaker E: So it's like...\nSpeaker E: It's not going to be, you know, touchable, but still like a source of information or a source for many use.\nSpeaker E: Yeah.\nSpeaker A: So maybe, maybe we can see depending on how we will come up with our full design then if you have enough money or like...\nSpeaker A: Okay.\nSpeaker A: So that's not enough.\nSpeaker A: Because speech recognition technology will take at least five euros or batchy or something.\nSpeaker A: So we want to reduce the cost on display or this...\nSpeaker F: The display would only be display and not...\nSpeaker E: Yeah.\nSpeaker E: That's what it's going to be.\nSpeaker E: It's not going to be a touch spot.\nSpeaker E: Just display for giving you information.\nSpeaker A: Yeah. That can be we can consider because like it won't take much money.\nSpeaker A: Okay.\nNone: Okay.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker A: Do you have any further questions?\nSpeaker F: I guess no.\nSpeaker F: So the batteries are going to be very light.\nSpeaker F: Yeah.\nSpeaker A: We have to go for...\nSpeaker A: And now I think we have many options in the market.\nSpeaker A: So we can go for small nickel or alkaline batteries for really light batteries and with good price.\nSpeaker F: So this device that can be used for speech recognition could also be used for this finding it basically instead of clapping wide.\nSpeaker A: Yeah.\nSpeaker A: Just the...\nSpeaker A: The one thing we want to know is like because remote control is used for like in household.\nSpeaker A: So it will be...\nSpeaker A: It may be at least five, six people want to use it.\nSpeaker A: So how to define our speech recognition whether we want to do speaker independent or speaker dependent.\nSpeaker A: If you're going for more speaker independent then it would be like again cumbersome.\nSpeaker A: It would be really more technology.\nSpeaker A: Okay.\nSpeaker A: For the location.\nSpeaker A: Yeah.\nSpeaker A: If everybody in the house they want to look at them where to go for some speaker independent technology.\nSpeaker A: Okay.\nSpeaker C: Okay.\nSpeaker C: So it's now good to do.\nSpeaker C: You do have a question.\nSpeaker C: No, it's a no more question.\nSpeaker F: Thank you.\nSpeaker F: But it's less of a constraint on what we can do.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: It's always like that.\nSpeaker F: We have dreams and in the end we find out that it's not feasible.\nSpeaker B: Yeah.\nSpeaker B: Anyway.\nSpeaker A: We have some limitations on this.\nSpeaker E: LST is already quite nice.\nSpeaker E: LST is something else.\nSpeaker E: It's quite nice.\nSpeaker F: I'm an artist.\nSpeaker D: Sorry.\nSpeaker D: So good.\nSpeaker C: That's not the impression.\nSpeaker E: No, let's talk about the interface.\nSpeaker E: This is about number three.\nSpeaker E: The...\nSpeaker E: Which one?\nSpeaker E: Hmm.\nSpeaker E: Hmm.\nSpeaker E: Hmm.\nSpeaker E: Oh, look at this.\nSpeaker E: No.\nSpeaker E: Yeah.\nSpeaker E: And five.\nSpeaker E: So the concept of the interface.\nSpeaker E: Generally I developed quite a broad concept.\nSpeaker E: Not only for the interface but for possible instruction or use of the interface.\nSpeaker E: Or use of manual and all the complex things that come together with your TV and remote controls.\nSpeaker E: So let's start with this.\nSpeaker E: We got our perfect remote control with a lot of buttons.\nSpeaker E: And we got explanation for every button and you can use your time and it will take some days to learn all these buttons.\nSpeaker E: And the LCD is going to be somewhere here.\nSpeaker E: And go back button.\nSpeaker E: I don't know really where it is.\nSpeaker E: Maybe one of these buttons.\nSpeaker E: And power on and off.\nSpeaker E: I don't remember.\nSpeaker E: So it should be...\nSpeaker E: Maybe this button is power on and off or no info.\nSpeaker E: I can see nothing.\nSpeaker E: So that's our concept.\nSpeaker E: It's called the Millennium Remote Control.\nSpeaker F: Let's change millenials.\nSpeaker A: Yeah.\nSpeaker A: So maybe you can use for those in the audience.\nSpeaker F: Oh, that doesn't make sense.\nSpeaker F: This is very...\nSpeaker E: Really?\nSpeaker E: I'll touch your knackers.\nSpeaker D: Oh, I don't know.\nSpeaker E: Okay, just press the button.\nSpeaker E: Oh, no.\nSpeaker E: Yeah.\nSpeaker E: We will not use this.\nSpeaker E: We will not use this.\nSpeaker E: Instead of this ugly device, that's our concept.\nSpeaker E: And it's got just few buttons quite well looking and all this stuff we already...\nSpeaker E: We already discussed.\nSpeaker E: And what will people say?\nSpeaker E: They will say it's perfect.\nSpeaker E: Or what will say?\nSpeaker E: They will say it's blended.\nSpeaker E: Everyone will say I'll buy it.\nSpeaker E: And everyone's going to be satisfied.\nSpeaker F: Do you think it can come in several colors?\nSpeaker E: Or did I would make a backlight of the LCD screen with different colors?\nSpeaker E: But not the case.\nSpeaker E: Not the case.\nSpeaker F: The case would only be in that aluminium titanium stuff.\nSpeaker C: Because apparently from your survey people like color.\nSpeaker F: Yeah, they like something which is...\nSpeaker E: Okay, so let's remember there's a Nokia phone which changeable panels.\nSpeaker F: Yeah, okay.\nSpeaker F: So that would be the option.\nSpeaker F: I don't know.\nSpeaker F: I don't have a Nokia phone.\nSpeaker F: But it's...\nSpeaker A: I don't use that, but again...\nSpeaker E: That's why you don't have it.\nSpeaker D: Yeah.\nSpeaker D: That's why.\nSpeaker A: Of course it's nasty.\nSpeaker A: But it would be expensive.\nSpeaker A: Now if you use color LCD...\nSpeaker A: No.\nSpeaker A: Instead of that, maybe we can change the color of the assembly.\nSpeaker A: Maybe we can just...\nSpeaker A: If you just want more colors they can pay more money to get this...\nSpeaker A: Oh.\nSpeaker A: The shapes.\nSpeaker A: I mean they can have different assembly vibe here.\nSpeaker A: So users have different...\nSpeaker A: They have their own color interest and so on.\nSpeaker A: So if they want they can just pay another...\nSpeaker C: Okay, so you propose something with option...\nSpeaker C: Yeah.\nSpeaker C: That increase the price if we...\nSpeaker A: Yeah, yeah.\nSpeaker A: If they want like...\nSpeaker A: Of course, kind of...\nSpeaker A: Upgradable...\nSpeaker F: Yeah.\nSpeaker F: Remote control.\nSpeaker F: Okay.\nSpeaker A: Just they will get a few more things and a few more colors.\nSpeaker E: But there is one more solution.\nSpeaker E: Because there are some paints that can change color according to where they are.\nSpeaker E: Like they can reflect different colors.\nSpeaker C: And depending on what is around.\nSpeaker C: Like what color is around.\nSpeaker E: And depending on the temperature.\nSpeaker E: Yeah.\nSpeaker E: Kind of a good effect.\nSpeaker E: Yeah, but that may be too much expensive.\nSpeaker C: But it can be in maybe in an upgradeable version.\nSpeaker F: Because I think there are two kinds of people.\nSpeaker F: Those for which the remote control is to be something useful.\nSpeaker F: I am going to talk about this later.\nSpeaker F: And those for which it's something that is specific to them.\nSpeaker F: So it's like a signature.\nSpeaker F: My remote control is pink.\nSpeaker F: Nobody else than me has a pink remote control.\nSpeaker F: And that makes me special.\nSpeaker F: Okay.\nSpeaker C: And you seem that we don't have to make them pay more.\nSpeaker F: I think they would be ready to pay more for that.\nSpeaker F: Okay.\nSpeaker F: So those who wanted to have it be.\nSpeaker C: So it's not a base service.\nSpeaker E: No.\nSpeaker E: Not an option.\nSpeaker E: Might be optional.\nSpeaker A: But those people will be really few.\nSpeaker F: The young people want to be different from their friends.\nSpeaker F: Although similar, but have something just slightly better.\nSpeaker F: Pink.\nSpeaker C: So maybe that if that is a selling point.\nSpeaker E: Maybe it has to be the base.\nSpeaker E: Yeah.\nSpeaker E: But you know if you want to be different.\nSpeaker E: You just take your remote control with you all the time.\nSpeaker E: And it's based on different.\nSpeaker D: You know?\nSpeaker E: And you want to pass the remote controls here.\nSpeaker C: You don't have to say hello.\nSpeaker C: I don't know.\nSpeaker F: You know, for instance, take the iPod.\nSpeaker F: It's a kind of remote control.\nSpeaker F: It's white.\nSpeaker F: And it's so white that you see it from anywhere.\nSpeaker F: It has this distinctive look and feel and look video.\nSpeaker F: Which people seem to like.\nSpeaker F: Just because it's a color that we don't usually see in a remote control.\nSpeaker F: White.\nSpeaker E: Could we integrate something into our remote control?\nSpeaker E: Something like light.\nSpeaker E: That I can use it in darkness.\nSpeaker E: Oh, the dark.\nSpeaker A: Yeah.\nSpeaker A: Maybe like the infrared.\nSpeaker A: Like we can put some radium chips or something.\nSpeaker A: So that like at least some.\nSpeaker A: Enredium?\nSpeaker A: Yeah.\nSpeaker D: I'll work with you, Enium.\nSpeaker A: This is radium.\nSpeaker A: Sorry.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker C: Okay.\nSpeaker C: So.\nSpeaker E: Well, let's go on maybe with the presentation.\nSpeaker E: Yeah.\nSpeaker E: Sure.\nSpeaker E: And the remote control is going to be smart.\nSpeaker E: But how smart should it be to not to complicate things too much.\nSpeaker E: And I don't know.\nSpeaker E: That's a question to you.\nSpeaker F: Well, so I heard that it seems that speech recognition is something that can be done.\nSpeaker F: So that's the smartness of the thing currently.\nSpeaker F: We don't have guns with speech recognition or beer cans with speech recognition.\nSpeaker F: But we may have remote controls with speech recognition.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker E: That's a nice one.\nSpeaker E: Just take a bite.\nSpeaker F: Don't touch the room.\nSpeaker F: Thank you.\nSpeaker F: But yeah.\nSpeaker F: Are you saying here that the remote control should be aware of who's using it?\nSpeaker F: So for instance, the young guy would not be able to use it because his father doesn't want.\nSpeaker F: Yeah.\nSpeaker F: I just want to say it should be a fingerprint recognition.\nSpeaker E: The worst recognition is quite enough.\nSpeaker E: I say don't use it and the control just locks.\nSpeaker E: Yeah.\nSpeaker E: Because I ordered.\nSpeaker A: Just that could be feasible like this.\nSpeaker A: Okay.\nSpeaker A: So since we want to do some speaker dependent speech recognition also.\nSpeaker F: So it could be smart in that way.\nSpeaker B: Okay.\nSpeaker F: But for instance, I'm thinking about the other.\nSpeaker F: Particularity that the remote control could have since it knows who is using it.\nSpeaker F: It might also record the kind of channels you are more often using.\nSpeaker F: Levels of volumes that you're more often.\nSpeaker F: You know, some things like that and provide you ways of using them.\nSpeaker F: I don't know.\nSpeaker F: I don't know.\nSpeaker F: That might be expensive.\nSpeaker F: But that might also be a good sales speech again.\nSpeaker F: The remote that knows you.\nSpeaker B: Okay.\nSpeaker E: Okay.\nNone: Thanks.\nSpeaker C: So.\nSpeaker C: My turn.\nSpeaker C: Yeah.\nSpeaker C: Marketing expert.\nSpeaker F: Okay.\nSpeaker F: Sorry.\nSpeaker C: But spent four I think.\nSpeaker F: Trend.\nSpeaker F: Yeah.\nSpeaker F: No.\nSpeaker F: Yes.\nSpeaker F: Okay.\nSpeaker F: So I'm going to talk about trends.\nSpeaker F: And then I hope this can help us to understand how we should design our remote control.\nSpeaker F: So next slide please.\nSpeaker F: So first maybe just a small recap on how do we watch trends.\nSpeaker F: So it's not so simple.\nSpeaker F: You might think that it's easy but it's not so simple.\nSpeaker F: Anyway, these days the best source of information is the web as you know.\nSpeaker F: So have to go often on the web and look at what the others are doing.\nSpeaker F: And ask real people who are using real remote controls every day.\nSpeaker F: Or any other tool that is similar to a remote control, which basically is a small device that people may have with them.\nSpeaker F: Always like a phone.\nSpeaker F: We can use the phone as a good example of where to be inspired.\nSpeaker F: Of course those who are watching the trends are also have also to be inspired because in fact they are not only watching the trends.\nSpeaker F: They are inventing it.\nSpeaker F: They are creating the trend.\nSpeaker F: I hope I'm going to try to help you on that.\nSpeaker F: This is more risky because you are not following the trend you try to invent it.\nSpeaker F: Which means either you succeed and you make a lot of money or you don't and you are out of business.\nSpeaker F: So anyway, next slide please.\nSpeaker F: To be quick.\nSpeaker F: There are a lot of words here but basically they are in the market of remote controls.\nSpeaker F: There are three aspects that we should very pay much attention to.\nSpeaker F: The first one which seems to be the most important one is that it has to be fancy.\nSpeaker F: It has to have a fancy look and feel.\nSpeaker F: Interestingly this is the very most important thing.\nSpeaker F: It has to be fancy.\nSpeaker F: Strangely enough it's more important to be fancy than to be with and now that's the second thing it has to be.\nSpeaker F: It has to be technologically innovative.\nSpeaker F: It has to be new with some of new technology inside.\nSpeaker F: This is also more important than the last thing which we may think would have been the most important.\nSpeaker F: It should be easy to use and it should be easy to use as a remote control.\nSpeaker F: As you see it first has to be very nice.\nSpeaker F: People are proud of.\nSpeaker F: They can be identified with and then something that contains very novel stuff that they can talk about with their friends.\nSpeaker F: Mine has this and not yours.\nSpeaker F: And finally of course it has to be useful as a remote control.\nSpeaker F: But it seems that it's not so important that it's used to this remote control.\nSpeaker F: Next slide please.\nSpeaker F: Now in a more general broad way of seeing the thing.\nSpeaker F: If we look back and not look at only remote controls I think it's important to see that the trends are quite the same in many areas.\nSpeaker F: So currently the trends that we see in big cities like Paris and Milan.\nSpeaker F: Well it seems that this year things should have a fruit and vegetable way of Luca feel.\nSpeaker F: And I think of course it applies to everything that's the thing with trends.\nSpeaker F: It can travel from cloth to furniture.\nSpeaker F: Same idea fruit and vegetable.\nSpeaker F: And if we compare to last year now it has to be spongy.\nSpeaker F: So I think when we were talking about rubber I think the rubber aspect might be important because it's what is pretty more visible in terms of sponginess.\nSpeaker C: So maybe Titanyam is not a good idea.\nSpeaker A: Seems not.\nSpeaker A: We need to think about...\nSpeaker F: I'm sorry Mark.\nSpeaker F: I think more of something in the colors of fruits and vegetables and spongy.\nSpeaker F: Even in the shape it has to be more round and more colorful.\nSpeaker F: It looks natural somehow.\nSpeaker F: And not those futuristic remote controls with angles and Titanyam like.\nSpeaker F: So that's what people seem to...\nSpeaker C: All fashion.\nSpeaker F: Quite far from what you thought but that's fashion man.\nSpeaker F: Okay.\nSpeaker F: Okay that's all I have to say.\nSpeaker A: So these update is mainly addressed by people or...\nSpeaker A: I'm sorry.\nSpeaker A: You got some update, fashion updates.\nSpeaker F: Yeah we have people listening to the trends everywhere in the world of course.\nSpeaker F: Our company is quite big.\nSpeaker F: It's not fun.\nSpeaker F: I think it's more general trend.\nSpeaker A: It's more general trend.\nSpeaker A: It's not particular to remote controls.\nSpeaker F: No it's not.\nSpeaker F: This is very general.\nSpeaker F: But it seems that trends travel across things.\nSpeaker A: Yeah but some materials there to be something like solid.\nSpeaker A: We can't build this conduit.\nSpeaker F: I think we have to have the look of fruit and vegetables.\nSpeaker A: Yeah at least I...\nSpeaker F: But we still have to put our chips inside.\nSpeaker F: Yeah of course.\nSpeaker F: This is your problem.\nSpeaker F: This is not mine.\nSpeaker A: Yeah looking fruit.\nSpeaker A: These teams can be easily incorporated.\nSpeaker A: I think in the colors or the kind of material.\nSpeaker F: If it's something like rubber made or...\nSpeaker F: Yeah I think it's also going to be good.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker C: One more question.\nNone: Okay.\nSpeaker D: Thanks.\nSpeaker D: Okay.\nSpeaker C: So we will have the next meeting in 30 minutes again.\nSpeaker C: And now you have to concentrate on this individual action which are for...\nSpeaker C: Rama...\nSpeaker C: You can feel it.\nSpeaker C: Design...\nSpeaker C: Mark the user interface design and Sami the product evaluation.\nSpeaker C: You will work together on a prototype using modeling clay.\nSpeaker C: Sounds interesting.\nSpeaker C: And I...\nSpeaker C: As always your personal coach will send you specific instruction.\nNone: Okay.\nSpeaker F: Okay.\nSpeaker F: Thanks.\nSpeaker A: So can we highlight the specific features of our...\nSpeaker A: Yeah you're right.\nSpeaker A: Yeah.\nSpeaker A: So we want the speech recognizer and we want...\nSpeaker A: Some kind of buttons and we want some teams like fruits or vegetables.\nSpeaker A: So you say...\nSpeaker A: We also follow the general trend.\nSpeaker A: Spongy.\nSpeaker C: Yeah.\nSpeaker C: Do we agree on that?\nSpeaker C: Yeah.\nSpeaker C: You have to.\nSpeaker E: So do you think...\nSpeaker E: So we have to...\nSpeaker F: No we don't have to.\nSpeaker F: No.\nSpeaker F: Seems it's the trend.\nSpeaker F: Again as I said we can also try to make it.\nSpeaker F: Yeah.\nSpeaker E: So we get the trend.\nSpeaker E: So there's no way confident enough in creating trends.\nSpeaker F: Well that's you can try to convince us.\nSpeaker E: Wow.\nSpeaker E: We can make it smell like fruit.\nSpeaker E: Okay.\nSpeaker C: That's a good idea.\nSpeaker C: So titanium smell like fruits.\nSpeaker C: Yeah.\nSpeaker A: So what about location and these people are really interesting on those features or they really...\nSpeaker A: Like there are more on these fancy features.\nSpeaker F: I think it's again in this...\nSpeaker F: What I said first it has to be fancy.\nSpeaker F: So I think if nobody else provides currently remote control with that kind of feature.\nSpeaker F: And if we can provide it I think it's a good sell for us.\nSpeaker F: Because we have it and others don't.\nSpeaker F: It's fancy.\nSpeaker F: Whether it's useful or not doesn't seem to be very important.\nSpeaker F: Yeah.\nSpeaker C: I agree with this.\nSpeaker C: Now we have to decide on what kind of fanciness.\nSpeaker C: Do we take titanium smelling like fruit?\nSpeaker C: Or do we make spongy fruit like fruit?\nSpeaker A: We will try to explore these two options.\nSpeaker A: Maybe we can explore the two options.\nSpeaker A: Yeah.\nSpeaker E: Could we make a titanium shape?\nSpeaker A: Yeah.\nSpeaker A: Yeah at least we can put fruit shapes.\nSpeaker C: But you say that you cannot do double shaped...\nSpeaker A: Yeah it beats.\nSpeaker A: Yeah that's a way to look for alternatives and...\nSpeaker A: Okay.\nSpeaker A: We have to see whether rubber is expensive and...\nSpeaker E: In fact I just agreed to make titanium panels on a whole rubber body.\nSpeaker E: Well okay.\nSpeaker C: Okay so you explore now that you're going to work together.\nSpeaker C: Yeah.\nSpeaker C: These two are spongy and...\nSpeaker C: Yeah.\nSpeaker A: Maybe we can have two different disassembly.\nSpeaker A: Also like one spongy and one kind of titanium.\nSpeaker A: But since they're chips we have only the plastic or that...\nSpeaker A: Oh for the size.\nSpeaker A: Yeah fiber chips or...\nSpeaker C: Yeah I don't think... I think we have to choose...\nSpeaker C: If we choose titanium or if we choose spongy but it cannot be both.\nSpeaker E: I really don't like this model in clay because you know it makes some balance for...\nSpeaker E: I don't know.\nSpeaker E: For creation.\nSpeaker E: Yeah.\nSpeaker E: Oh look.\nSpeaker E: Yeah.\nSpeaker C: Yeah you can't do the exact tenures.\nSpeaker F: You can paint it after work.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: We have a very large department of...\nSpeaker F: Yeah don't worry.\nSpeaker C: You will do it.\nSpeaker C: You will do it.\nSpeaker C: You will do it.\nSpeaker E: Alright.\nNone: Alright.\nSpeaker C: Okay.\nSpeaker C: So explore a shape.\nSpeaker A: So still we want to keep LCD or...\nSpeaker C: I think it's what we say that we have a new CD with information.\nSpeaker F: The thing is that if we want to have as he says, if we want to have a small number of buttons...\nSpeaker F: Yeah.\nSpeaker F: Then we need to have a kind of output that says currently what their actions are.\nSpeaker A: Yeah.\nSpeaker A: User friendly.\nSpeaker F: Yeah.\nSpeaker F: That's the converse to having the zillions of buttons where each button does only one thing.\nNone: Yeah.\nSpeaker A: And we just stick with two batteries so it's not any...\nSpeaker A: Yeah.\nSpeaker A: So we stick less time in that.\nSpeaker F: We don't reinvent the wheel.\nSpeaker F: Anyway it's very...\nSpeaker F: In general I think people change TV more often than batteries of their remote control.\nSpeaker F: So because the trend goes faster than the life of them.\nSpeaker F: So it's too low.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Now we...\nSpeaker A: Yeah.\nSpeaker A: We're done.\nSpeaker C: We're done.\nSpeaker C: Okay.\nSpeaker C: So see you in 30 minutes.\nSpeaker A: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "ES2010c", "summary": "This was the conceptual design meeting for the remote. The first presentation was given by Marketing on trend watching. Marketing pointed out that customers wanted the remote to be technologically innovative and yet simple to use, and that the current fashion trend was fruits and vegetable patterns. Clearly labelled large buttons with the company logo and colour scheme were also preferred. User Interface followed with the presentation on interface concepts, proposing to have a scroll-down design and a distinctly recognizable demographic for specific buttons. Industrial Designer's presentation on component design provided information for the material needed for the remote, and certain restrictions coming with different options. After User Interface's additional thoughts on voice recognition, the team discussed designs of the energy source, the material, and the user interface concept, deciding on latex covers, the gun-metal grey colour, and blue and yellow touch buttons.", "dialogue": "Speaker C: On graves I'm not singing it yet.\nNone: Okay.\nNone: Hi, team.\nSpeaker C: Hope you had a good lunch.\nSpeaker C: Okay, we're back for the conceptual design meeting.\nSpeaker C: Let's get started.\nSpeaker C: Okay, here's the agenda for today's meeting.\nSpeaker C: We're going to open it, and I'm going to keep the minutes as project manager.\nSpeaker C: We're going to have three presentations, one from each of you again.\nSpeaker C: Then we are going to come to decision on their mode control concepts, and then we're going to close it up.\nSpeaker C: We have 40 minutes again.\nSpeaker C: Okay, and just to reiterate, after this meeting, the team will reach a decision on the concepts of their mode control.\nSpeaker C: Okay, let's go ahead and start off with your presentations.\nSpeaker C: Who would like to go first?\nSpeaker B: Just trying to move mine right now.\nSpeaker C: Courtney, would you mind starting us off?\nNone: Yeah.\nSpeaker C: Are you trying to do the training?\nSpeaker D: Yeah.\nSpeaker D: Okay, so, trend watching.\nSpeaker D: Since we do put the fashion electronics, it is kind of important how our product looks.\nSpeaker D: So, I guess we can go ahead and go to the next.\nSpeaker D: So, what they want right now, customers want fancy versus functional.\nSpeaker D: Basically, about 58% of what they like of the product that they want, describe me like the, in order of how much they want, 58% of the decision of what it should look like, fancy versus functional.\nSpeaker D: And then it has to also be technologically innovative and yet easy to use.\nSpeaker D: So, the customer basically is confused.\nSpeaker D: They don't know exactly what they want.\nSpeaker D: They want us to tell them.\nSpeaker B: They want everything.\nSpeaker D: Yes, exactly.\nSpeaker D: So, we can go next.\nSpeaker D: Okay, so, in Milan and Paris recently, the trends have been showing that clothing, shoes and furniture are basically just covered with fruits and vegetable patterns.\nSpeaker D: So, I don't know if we want to go in that.\nSpeaker D: And also, the spongy feels in contrast to last year.\nSpeaker D: I don't know, really.\nSpeaker D: I mean, I guess the spongy could relate to the buttons if we want to, rather than like a hard clicky button that you find on like some mobiles and stuff.\nSpeaker D: You'd want like a softer touch.\nSpeaker D: I mean, do you guys know what I mean?\nSpeaker D: Right, yeah.\nSpeaker D: But as for the fruits and vegetable patterns, I don't know if we really want to go with that, because it is just a trend.\nSpeaker D: And our product, we want to stay around for much longer than just a few months.\nSpeaker D: Right, because people don't buy a new remember every month.\nSpeaker D: Yeah, I mean, that could just be a spring thing right now.\nSpeaker B: I can address some of that issue, I think, with my presentation.\nSpeaker D: Okay, awesome.\nSpeaker D: So, design preferences, we need easy to read, like large buttons, clearly labeled.\nSpeaker D: So, I mean, because we talked about that being a problem.\nSpeaker D: And then also buttons illuminating upon touch.\nSpeaker D: You said that in your design with the bulb.\nSpeaker D: And that could also tie in with the color scheme.\nSpeaker D: We need the real reaction logo, the color scheme, obviously.\nSpeaker D: That's one of our key goals, we want to promote our product.\nSpeaker D: And I was thinking about different types of designs, and I came up with something actually right here.\nSpeaker D: So, what we could do is something like an old fashioned telephone, like this, where we put the buttons around, like we put a big on-off button or something else in the middle.\nSpeaker D: I mean, it could be the arrows or whatever for channel up and down, and then put the numbers around in like an old fashioned dial shape.\nSpeaker D: Because then it'll appeal to older generation and, like you said, retro.\nSpeaker D: So, it's cool.\nSpeaker D: So, it's classically retro.\nSpeaker D: So, I mean, that's just an idea.\nSpeaker C: I like it.\nSpeaker C: Ready for the next slide?\nSpeaker D: Yep.\nSpeaker D: And that's it.\nSpeaker C: Great.\nSpeaker C: Great presentation.\nSpeaker C: Ready?\nSpeaker C: Hang on.\nSpeaker A: Let's see if it's there.\nSpeaker A: Which one is it?\nSpeaker A: I don't know.\nSpeaker A: Interface concept?\nNone: No?\nSpeaker C: Interface concept.\nSpeaker A: Either refresh it or, oh wait, maybe I didn't put it in there.\nSpeaker A: Hang on.\nSpeaker B: I don't know if mine will always read copy of something, brother.\nSpeaker B: Sorry?\nSpeaker B: I copied mine before I sent it over.\nSpeaker A: Oh, okay.\nSpeaker A: Sorry.\nSpeaker A: Hang on.\nSpeaker A: Don't know where to go.\nNone: There we go.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nNone: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Looking at the interface concept is going to be mostly examples of possibilities of where we can go with this.\nSpeaker A: If you want to start the next slide, sure.\nSpeaker A: I can't really see it, but there's two possible ways on the left.\nSpeaker A: If you see it on the sides of the remote, you have the sort of scroll down.\nSpeaker A: So you have that option right there.\nSpeaker A: And then also there's the idea of the base.\nSpeaker A: That's sort of like an idea there.\nSpeaker A: And then on the right, you have what's really big trend right now.\nSpeaker A: It's the iPod.\nSpeaker A: It's becoming really fast thing.\nSpeaker A: And so you have this sort of very, very simplistic menu section with the round buttons.\nSpeaker A: And it's sort of like you have the both kind of trendy and hip, but also very sleek and very simple, but technologically advanced.\nSpeaker A: So if you wanted to do that, and we could find a way of sort of like using that idea into remote control and sort of look into it, but anyway, next.\nSpeaker A: There's the idea of like being able to do it by feel as well as by sight.\nSpeaker A: You know, you're in the dark, you don't want to be looking at your remote control.\nSpeaker A: And the picture particularly is pointing out if you look at the top volume button, it's a V.\nSpeaker A: And so you're kind of feeling a V like volume up.\nSpeaker A: What it really is is a V and what you think it is is down because it's the down arrow.\nSpeaker A: And so it's like sort of quizzes and you'd probably turn that the other way up.\nSpeaker A: And you could either do it by raised type, which could be, you know, if he sort of old fashioned in a way.\nSpeaker A: Either that or just have it by shape, for example, you have a specific triangular shape that you know you're looking at the up and down arrow.\nSpeaker A: And then the round ones you sort of feel by, you know, the second one down, that sort of thing.\nSpeaker A: So it's sort of looking into how we wanted to do it by feel.\nSpeaker A: Okay.\nSpeaker A: So this is sort of an example of going for a certain demographic, particularly the gear towards children.\nSpeaker A: That's cute.\nSpeaker A: It's very cute.\nSpeaker A: We could probably change it to yellow, bright yellow for the company logo.\nSpeaker A: You have the shape since it's very simplistic and friendly looking.\nSpeaker A: And then the other thing that you'll be able to do is just to be able to program certain channels that only these children watch.\nSpeaker A: You know, the CBB or something like that.\nSpeaker A: Keep them away from other channels.\nSpeaker A: So that's like another error.\nSpeaker A: I mean, these are three examples sort of looking at it.\nSpeaker A: You have the wider section for the main controls there.\nSpeaker A: You can see how many buttons there are.\nSpeaker A: And then on the left you have an example of the round buttons and a simpler design.\nSpeaker A: On the bottom we probably wouldn't need that because it's more for like a DVD function which we're not going to be using.\nSpeaker A: So again, it's sort of like just giving you ideas.\nSpeaker A: And then down at the bottom you have the logos and that's where you can put the R-R.\nSpeaker A: Real reaction.\nSpeaker A: And then finally these are like the sort of same examples but also some more just possibilities that we can go with.\nSpeaker A: None of them are particularly keen on by the way.\nSpeaker A: But it's sort of like just taking aspects out of that and saying well out of this one we like the round section or we like the button size on this.\nSpeaker C: Or I like the black finish or the silver finish or something.\nSpeaker D: I have four of those from out.\nSpeaker C: Okay, ready?\nSpeaker C: Let's go.\nSpeaker C: Okay, great job.\nSpeaker B: Okay, my turn.\nSpeaker B: Okay, what's the title?\nSpeaker B: It'll be copy of component design.\nSpeaker B: Yeah, it looks like it.\nSpeaker B: Okay, so basic remote operation runs as follows.\nSpeaker B: Press button makes connection with the power source and the rest of the circuit.\nSpeaker B: Chip sends is the connection chip produces a Morse code infrared signal specific to that button.\nSpeaker B: So you press the button and it produces a signal that's encoded specifically for that button.\nSpeaker B: Transistors amplify that signal and it goes to the TV center which interprets the signal and responds accordingly changes channel, etc.\nSpeaker B: So that being said, next slide please.\nSpeaker B: Findings which were the required materials for the basic internal construction.\nSpeaker B: So all the really simplistic functions that we just discussed.\nSpeaker B: We need rubber for buttons, aluminum for battery contacts, integrated circuit which consists of a diode transistor resonator, resistors and a capacitor.\nSpeaker B: All those basic things that make a circuit function.\nSpeaker B: Fiberglass and thin copper wire to create the actual circuit board itself.\nSpeaker B: And LED which is a light emitting diode.\nSpeaker B: Contact disks for the buttons, plastic for the casing and a power source, whatever power source we've actually determined we want.\nSpeaker B: Next slide please.\nSpeaker B: Thank you.\nSpeaker B: Personal preferences.\nSpeaker B: To save money for the components, the remote should be mass produced and basic materials should be bought on mass.\nSpeaker B: If we find another company who can produce the required chips casing LED any additional materials we decide we require at a less expensive rate than we ourselves are producing, we should go for it.\nSpeaker B: Next slide please.\nSpeaker B: Just talking to the manufacturing division.\nSpeaker B: They suggested power options, solar cells, hand dynamo and kinetic power so you shake it and increases the power.\nSpeaker B: I'm not sure how the hand dynamo works, they have yet to get back to me on that.\nSpeaker B: Next slide please.\nSpeaker B: Suggested casing options.\nSpeaker B: Okay.\nSpeaker B: We can offer options for casing such as straight curve, double curved, you know, very specific to the customer.\nSpeaker B: Options for materials, plastic, rubber, titanium, wood.\nSpeaker B: I don't think anyone's going to go for a wood one because winter is.\nSpeaker B: Oh yeah.\nSpeaker B: No, it's winter's in.\nSpeaker B: Certain restrictions do apply here though. Latex you can't do solar power with a latex one.\nSpeaker B: So if they want soft squishy rubber, they can't have the solar power adoption.\nSpeaker B: Double curved.\nSpeaker B: You can't do titanium.\nSpeaker B: That would be two curvatures.\nSpeaker B: So it would actually, if the shape of your hand you curve here and you curve here.\nSpeaker B: So you could have two curves that match the shape of your hand to make it more comfortable to hold.\nSpeaker B: Now if you wanted that, you can't do titanium.\nSpeaker B: And so adjusted functions for the buttons, scrolling function could be very beneficial to us instead of actual buttons themselves.\nSpeaker B: I think I have one more slide.\nSpeaker B: No, I don't.\nSpeaker B: The manufacturing division also has said that they have several types of chips and they've just developed a sample sensor or sample speaker chip, which we could utilize.\nSpeaker B: Push button requires a simple chip and scroll requires more complicated chip.\nSpeaker B: So it depends on what we decide we want to do.\nSpeaker B: In addition to that, if we're offering all those different options to the customer for producing their remote, we're going to have to have multiples of each type like a double curved in rubber.\nSpeaker B: Each option should have a certain select number produced with all those options.\nSpeaker B: So we'll have to mix it up, make sure we produce enough of everyone.\nSpeaker B: But that could also drive up the price of the actual remote itself if they know that we only produced 5,000 double curved wooden remotes.\nSpeaker C: That's all I got.\nSpeaker C: Alright, well thank you for those informative presentations.\nSpeaker C: Let's go back to now we have to make some decisions.\nSpeaker A: Let me just add one more thing that I couldn't say for sure.\nSpeaker A: And that's just that new technology that I've developed on the voice rescue mission.\nSpeaker A: Oh, this is the thing we were talking about earlier.\nSpeaker A: Except it's sort of odd and not exactly sure why they're explaining it in the way they are.\nSpeaker A: There's a sample sensor and there's a sample speaker unit.\nSpeaker A: So you would say like good morning coffee maker and it would respond good morning Joe.\nSpeaker A: But I'm not sure exactly how it's going to work because do you program the we program the responses and the questions.\nSpeaker A: Does that mean that the user then has to ask the specific question and can't change it in order for it to be recognized or can it be altered in a certain way or just the actual user program it channel means this.\nSpeaker D: Yeah, like using the menu to be like enter your name into the screen like on the menu option so that way the so it's got like a limited memory and so you'd program it.\nSpeaker A: So it's sort of if you know that's kind of what you say.\nSpeaker D: I feel like voice recognition would be.\nSpeaker D: I don't know.\nSpeaker D: It would be too hard to really probably.\nSpeaker C: I mean we could do it.\nSpeaker C: But if it's within our price to get that kind of chip that would you know.\nSpeaker B: Well we are making the chip.\nSpeaker B: So I mean.\nSpeaker D: I guess we have to look at what our production cost is for the chip itself and it is a growing trend the higher technological like the I'm just like the more advanced it is.\nSpeaker B: Yeah better.\nSpeaker B: It'll sell.\nSpeaker B: I thought offering some of those options for different materials that it could be made of different you know.\nSpeaker B: I think we'd have to decide on the power options.\nSpeaker B: Maybe.\nSpeaker B: Yeah we could reduce.\nSpeaker C: Yeah.\nSpeaker C: I'll have the okay that kind of brings us to this let's let's think we can decide what kind of energy source we want to have first and foremost.\nSpeaker C: Do we want to go for batteries or a stand like the one that we saw illustrated earlier.\nSpeaker C: Oh the base yeah the base the charging base.\nSpeaker A: I don't know what it looks like before.\nSpeaker A: Is this something really really small.\nSpeaker A: Then it's sort of harder to imagine the base where that was quite a substantial size sort of standing up.\nSpeaker C: Yeah and we don't have multiple things that has to control just has to control the TV.\nSpeaker D: It's not going to be a huge universal about.\nSpeaker A: We need to decide well so we can figure out how big it's going to be like what exactly what buttons we want and exactly.\nSpeaker A: We want to figure this yes I won't because it still fits in your hands you still want it something that's comfortable and substantial but not necessarily full of buttons.\nSpeaker D: This one is really comfortable like I like the sides.\nSpeaker D: Whatever it'll lose it easier but if we have the the locator that we don't have to worry about that that's true.\nSpeaker D: So we can make it small if we have like a locating device.\nSpeaker B: We do a voice activated locator though we're going to be looking at a more substantial chip.\nSpeaker A: So that's the other thing it's like you know what are we going to have certain chips are going to require a bigger size period.\nSpeaker D: Two double A's for the size.\nSpeaker A: But like you know if we get more complicated then it's going to be bigger to just accommodate the chip size.\nSpeaker B: Honestly I think the customer would be kind of irritated by the fact that it has a base if we did do a nice small compact.\nSpeaker C: It's either going to be bigger with a base or smaller with just a battery like this guy.\nSpeaker C: Alright so what direction do you want to go and you want to vote?\nSpeaker D: I think if we had a locating device with a small one I think that seems way more advanced.\nSpeaker C: I'm kind of leaning in the direction of the base kind of bigger and the base.\nSpeaker D: That just seems clunky and yeah because I mean even looking at cell phones right now those trends the smaller the hotter it is.\nSpeaker B: Okay the only problem with that is you forget to take it out of your pocket and it goes in the wash.\nSpeaker B: Oh kidding.\nSpeaker B: You know what happens.\nSpeaker B: Yeah.\nSpeaker B: Three watches go that way too.\nSpeaker A: Oh watch.\nSpeaker D: Never watch the phone.\nSpeaker D: Oh no that would wow.\nSpeaker D: That would hurt.\nSpeaker C: Okay so what kind of material do we want to do?\nSpeaker B: Well we have lots of options.\nSpeaker B: I don't think wood is a viable option.\nSpeaker C: No.\nSpeaker C: What did you decide?\nSpeaker A: Go ahead.\nSpeaker A: I think that's high tanning if we're being restricted then.\nSpeaker A: I hope to lean away from that.\nSpeaker D: Yeah because if it's going to cause us more to produce a chip titanium would be more expensive.\nSpeaker D: Right.\nSpeaker D: However.\nSpeaker B: Oh would you recommend?\nSpeaker B: Oh we only want to sell it for 25 euro right?\nSpeaker B: Yeah.\nSpeaker B: Because I was thinking if we wanted to get the high end market that you could produce a few in titanium and make them a rarity.\nSpeaker D: So we could do that because all our research shows that people are definitely willing to spend more.\nSpeaker A: How are you restricted by this?\nSpeaker E: Well the original.\nSpeaker C: I think we should just focus on one design and one concept right now.\nSpeaker C: I'm not sure that we'll have the time and money to produce a whole array of remotes.\nSpeaker C: If this was a successful remote we might then produce a higher end version of it.\nSpeaker C: I think.\nSpeaker C: Good plan.\nSpeaker C: Good plan.\nSpeaker C: Okay so we want to go for plastic or what would you recommend for materials?\nSpeaker B: Honestly I'd recommend like since we're going with batteries instead of solar power.\nSpeaker B: I'd recommend maybe a soft latex because we could produce you know how cell phones have those overlays that you can change the color.\nSpeaker B: We could do one that fits in with the trends of the year.\nSpeaker B: So because this year is all fruit.\nSpeaker B: Got on the news.\nSpeaker B: Who knows.\nSpeaker B: We could do a cherry cover for this year and then if next year is stripes or solids.\nSpeaker B: You know.\nSpeaker A: Okay.\nSpeaker A: So you're talking about like when it leaves our sales room then it's all going to be cherry colored or is it going to be the kind of thing where people come back.\nSpeaker A: Or we could like take it back.\nSpeaker C: Cases maybe.\nSpeaker C: And by the extra case.\nSpeaker C: If they wanted.\nSpeaker C: I think it's good to sell a basic thing and then sell.\nSpeaker B: So we could do like a hard base plastic and then we could give.\nSpeaker B: Two latex covers.\nSpeaker B: Yes.\nSpeaker B: Okay.\nSpeaker B: To start.\nNone: Okay.\nSpeaker D: Yeah because the soft latex definitely is squishy.\nSpeaker D: That's in.\nSpeaker D: Well I mean squishier than like.\nSpeaker D: Right.\nSpeaker D: Yeah.\nSpeaker D: Right.\nSpeaker C: It's just a hard plastic.\nSpeaker C: Okay.\nSpeaker C: And what kind of chip would we need for this guy.\nSpeaker B: How complicated are we going to go with the voice activated.\nSpeaker C: I don't think we should do voice.\nSpeaker C: I think we should just do the recognition for when it's lost.\nSpeaker C: You know.\nSpeaker C: Could we.\nSpeaker D: Yeah.\nSpeaker D: Because that what type of yeah for voice activation would it be like a certain term.\nSpeaker D: What would we say like.\nSpeaker D: Because people could just be talking and we don't want it going off all the time.\nSpeaker D: Right.\nSpeaker B: Well we could give it a specific code.\nSpeaker B: You know remote missing.\nSpeaker D: Ooh.\nSpeaker D: See I'm strangely trying to because I know that's.\nSpeaker D: It's definitely going to be big because it's.\nSpeaker D: I don't know.\nSpeaker A: It's just so high tech.\nSpeaker A: My little sister got for Christmas.\nSpeaker A: She got one of those key finders.\nSpeaker A: That's like a key ring.\nSpeaker A: And you have to whistle in a certain frequency for it to work.\nSpeaker A: Yes.\nSpeaker A: See that would not be a laugh.\nSpeaker A: And it would start going off and her person you couldn't turn it off.\nSpeaker A: Oh.\nSpeaker A: It's a little bit.\nSpeaker A: Okay.\nSpeaker B: So I think having a key phrase is much better.\nSpeaker C: Okay.\nSpeaker C: All right.\nSpeaker C: But it's not going to be voice activated in the fact that you would say channel up and it would work.\nSpeaker C: Right.\nSpeaker C: Yeah.\nSpeaker D: No, no, we don't know.\nSpeaker D: Okay.\nSpeaker D: But do you can your can the department make that would be like a mid class.\nSpeaker B: Oh, really then so we don't actually have to go for.\nSpeaker B: Well, if they've just about the sample sensor sample speaker.\nSpeaker B: It's a brand new chip.\nSpeaker B: Why not introduce it in this way?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: And what size batteries?\nSpeaker C: Double A triple A.\nSpeaker D: I think triple A.\nSpeaker D: It'll be lighter too.\nSpeaker A: More than a package.\nSpeaker A: More than a package.\nSpeaker A: More than a package.\nSpeaker B: I think well, we could do two.\nSpeaker B: We could do one small lithium.\nSpeaker B: Because you know, the lithium batteries are doing quite well in most other electronic products.\nSpeaker B: Right.\nSpeaker B: So they're more widely available now.\nSpeaker B: And they also have a longer battery life than most batteries.\nSpeaker C: They're more expensive than a two.\nSpeaker B: But if you only have to replace it every five years.\nSpeaker A: That's the good thoughts.\nSpeaker B: Well, how about initial you get one battery when you buy it.\nSpeaker B: Because I'm pretty sure we can get them.\nSpeaker D: We could get a cheap bottle.\nSpeaker C: Come back to it next meeting.\nSpeaker C: Yeah.\nNone: All right.\nSpeaker C: Okay.\nSpeaker C: So we've covered that first category.\nSpeaker C: User interface concept.\nSpeaker C: Okay.\nSpeaker C: I kind of like your idea about the retro phone dial.\nSpeaker C: And that the central button could have maybe our logo on it.\nSpeaker C: It might be the four way scroll too.\nSpeaker C: Yeah.\nSpeaker C: It could be whatever.\nSpeaker D: As long as there's something big in the middle because like the old phones.\nSpeaker D: Yeah.\nSpeaker D: There's like that just like piece of metal or like a picture or something.\nSpeaker A: I think that is if it got too big though.\nSpeaker A: So if you have the circle and the button in the middle then is it going to get wider than your hands are?\nSpeaker A: Because and then would the button be too small if it was enough to fit on it?\nSpeaker A: Complete.\nSpeaker B: In the sample ones that you showed us there was one that had the scroll buttons on the side.\nSpeaker B: Yeah.\nSpeaker B: Which I think if we make it curved like a and like a hand shape.\nSpeaker B: Like if we put the the scroll button on the side.\nSpeaker B: Oh, I see.\nSpeaker B: That could be particularly useful.\nSpeaker D: I think so.\nSpeaker D: So scroll buttons on the side.\nSpeaker D: Yeah.\nSpeaker D: But it's on the top.\nNone: Okay.\nSpeaker D: But we definitely if we have scroll thing on the side we definitely have to have them labeled.\nSpeaker D: Yeah.\nSpeaker D: Like on the side of it.\nSpeaker D: It's just up and down.\nSpeaker D: If it's just up and down.\nSpeaker A: Is that for volume or channel?\nSpeaker B: Which I don't know.\nSpeaker B: Well, you could do.\nSpeaker B: We have both sides.\nSpeaker C: Can we?\nSpeaker B: Yeah.\nSpeaker B: We should probably make it that you have to depress it to activate it then.\nSpeaker B: Yeah.\nSpeaker B: Because otherwise you're not just holding it and going like this.\nSpeaker A: You know.\nSpeaker A: Instead of a scroll you just have a button up on the side.\nSpeaker A: Which are on the side.\nSpeaker A: Okay.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker C: Any other ideas?\nSpeaker C: What color?\nSpeaker C: Oh, yeah.\nSpeaker C: Latex covers.\nSpeaker B: We have to make sure that the logo always sticks out when we put the latex covers on.\nSpeaker B: So we'll have to like have a little square or something.\nSpeaker B: So that the logo is available.\nSpeaker A: So having a yellow strip at the bottom with the R.R. like that.\nSpeaker A: Yeah.\nSpeaker C: So the bottom of it.\nSpeaker C: I think maybe we should do it on a button itself though because people are able to change the covers.\nSpeaker C: I don't know.\nSpeaker C: Maybe the on-off button something.\nSpeaker C: Some, the menu button.\nSpeaker C: I don't know.\nSpeaker C: But you know if we're going to put our company logo on there and somebody could just get another one.\nSpeaker C: Are they all going to have our company logo on them?\nSpeaker C: Every cover?\nSpeaker B: Yeah.\nSpeaker B: I don't think we should do that because that would just be icky.\nSpeaker B: Yeah.\nSpeaker B: So I think maybe putting it on a button is probably a good idea.\nSpeaker C: If we want it to be visible.\nSpeaker C: Are all of those, those ones that you showed where they were silver metallic looking?\nSpeaker C: Yeah.\nSpeaker C: But those are plastic, right?\nSpeaker C: They're not titanium.\nSpeaker C: I kind of like that look.\nSpeaker C: But or if it was really fun?\nSpeaker C: Yeah.\nSpeaker C: Or if we're going for the retro look, I think like a really shiny black would be cool.\nSpeaker C: Yeah.\nSpeaker C: Or a good gunmetal gray.\nSpeaker C: Because that would be great.\nSpeaker D: That would be great.\nSpeaker D: Cibund, silver and black.\nSpeaker C: Here you go.\nSpeaker C: Good metal gray.\nNone: I'm just really wary of putting anything on a button.\nNone: Why?\nNone: It'll wear off.\nNone: Yeah, buttons.\nSpeaker A: Well, then what's the button do?\nSpeaker A: And then you're like, oh, I'm just going to put it on a button.\nSpeaker A: Why?\nSpeaker A: It'll wear off.\nSpeaker A: Yeah, buttons.\nSpeaker A: Well, then what's the button do?\nSpeaker A: And how do you know what that button does?\nSpeaker A: Yes.\nSpeaker A: Just looking at your examples, you just don't ever see the logo on a button.\nSpeaker A: It's always on the actual casing.\nSpeaker E: Right.\nSpeaker B: There's nothing saying that we have to put the logo on the front.\nSpeaker B: We want to be seen.\nSpeaker B: It's visible.\nSpeaker D: It's visible.\nSpeaker C: It's visible, though, because if it was only on the back, really, the only time you're going to see it is when you drop it or when you're changing the battery.\nSpeaker A: The other option is, I don't know if you can see it, but it's like, if I can find it again.\nSpeaker A: Yeah.\nSpeaker A: It's like the second to last slide.\nSpeaker C: Okay.\nSpeaker C: And yours was called interface.\nSpeaker C: Yeah.\nSpeaker C: This one?\nSpeaker C: Yeah.\nSpeaker C: For some reason, I can't get it to just go to that slide directly.\nSpeaker A: Okay.\nSpeaker A: It's the very right one.\nSpeaker A: You see at the bottom, it's kind of difficult to see, but you have like a sort of division between the bottom, like where the logo is.\nSpeaker A: And if we have the replaceable section, it's like the top.\nSpeaker A: It doesn't necessarily replace the entire top.\nSpeaker A: So you have that one piece that stays in the rest, sort of clips in.\nSpeaker A: Okay.\nSpeaker A: Yes.\nSpeaker A: You envisioning it.\nSpeaker A: And so that stays to stay when you have the logo.\nSpeaker A: And then you have the slipback client clips in, and that's a bit of changes.\nSpeaker B: We're using a latex overlay.\nSpeaker B: So that actually would go over top of everything and have holes for the buttons.\nSpeaker B: So I was thinking maybe instead of doing that, what we could do is leave a space for where the logo should be.\nSpeaker C: Yeah.\nSpeaker C: Like a little cutout kind of.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: You know, for like a cell phone, it's like the screen.\nSpeaker A: Right.\nSpeaker A: It's always just left open.\nSpeaker A: And so what are we going to do it like right yellow?\nSpeaker A: Yeah.\nSpeaker D: Anybody see anything that they liked in the buttons that are blue based?\nSpeaker D: Well, a lot of those buttons are blue based.\nSpeaker D: Well, kind of.\nSpeaker D: And then if we do have them illuminate upon contact, they can illuminate yellow.\nSpeaker D: Yellow.\nSpeaker D: I like that idea.\nSpeaker D: Like the one all the way on the left, you can see it on your computer better.\nSpeaker D: Where the button is actually blue, but the number itself is clear or white or whatever.\nSpeaker D: So if you pressed it, it would illuminate yellow.\nSpeaker D: So we'd have blue and yellow for the touch.\nSpeaker D: Oh, that one.\nSpeaker C: I like the yellow illumination idea.\nSpeaker C: Very good.\nSpeaker C: Okay.\nSpeaker C: Any other ideas or thoughts?\nSpeaker C: Yeah.\nSpeaker C: I asked him to be fairly in agreement about what we want to do with his project.\nSpeaker C: So let me catch up.\nNone: Okay.\nSpeaker A: Do we finalize what buttons we're including?\nSpeaker A: It's just everything that we've said before.\nSpeaker C: I think there will be time for that later.\nSpeaker C: I'm guessing.\nSpeaker C: Okay.\nSpeaker C: Next meeting is going to start in 30 minutes.\nSpeaker C: And here's what we're each of us going to do.\nSpeaker C: The ID is going to do the look and feel design, the UID, the user interface design.\nSpeaker C: I think you're going to get a lot of, I mean, the final say on what buttons get put.\nSpeaker C: We'll all talk about it, but I think that's pretty much what you're going to do, right?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: And you're going to do some product evaluation.\nSpeaker C: Okay. And right now, the ID and UID, you two are going to work together on a prototype using modeling clay.\nSpeaker B: Great.\nSpeaker B: Later.\nSpeaker C: Yep.\nSpeaker C: Okay.\nSpeaker C: And you should all be getting an email pretty soon.\nSpeaker C: All right.\nSpeaker C: Thank you for a very productive meeting.\nSpeaker C: Bye.\n"}, {"meeting_id": "Bed017", "summary": "The first phase of the data collection has finished. There is a new wizard for phase two, during which subjects will be given more complex scenarios. An alternative representation of the Bayes-net, it depicts context features as classes, and dependencies as relations between them. Following this, a visiting researcher presented an overview of a parallel project at the International University. It attempts to build a smart tutoring system for a computer science course. The assumption is that document searches can give more personalised results, if they take into account contextual parameters (user, situation). There were also further suggestions for meetings with ICSI researchers.", "dialogue": "None: You Why I know She already told me it before he told you no, she told me a long time ago She's been like two weeks ago\nSpeaker B: You know how to talk about this thing with me\nSpeaker E: What is it? Let me explain some video. Oh, I'm just moving to the nearest You're You have to see something genuinely funny before you get an example\nSpeaker F: To the next page here at least we sell the slide and have to laugh at your own jokes\nSpeaker D: It's a different laugh Oh\nSpeaker F: Oh\nSpeaker B: Well, you've been doing anything. Okay, you've just got a laptop. I'm just trying to show it off\nSpeaker F: That was actually open study, but hey\nSpeaker C: Okay, so once again So we haven't had a meeting for a while and Probably won't have one next week. I think I remember people are going So remember why don't you bring us up the date on where we are with you\nSpeaker B: In a smaller group we had talked and decided about continuation of the data collection so face time with us is almost officially over and She brought us some 30 subjects and collected data and Ten dialogues have been transcribed and can be looked at if you're interested in that talk to me And we found another Coxie student who's interested in playing wizard for us We're gonna make it a little bit more complicated for the subjects This round she's actually suggested Look at the psychology department students because they have to partake into experiments in order to fulfill some requirements so they have to be subject it before they can actually graduate and We want to design it so that they really have to think about having some time today's For example, plan certain things and figure out which can be done at what time and Sort of package the whole thing and a few more complicated Structure that's for the data collection as for smart com The last smart com meeting I mentioned that we have some problems with the synthesis which as of this morning should be resolved and so Should be means they aren't yet, but but I think I have the info now that I need Plus John and I am meeting tomorrow, so maybe When tomorrow is over we're done and I will never have to look at it again Maybe I'll take some more time We realistic but at least we're seeing the end of the tunnel there There was that the Don't think we need to discuss the formalism that will be done officially once we're done Something happened you know on either side with the PRM that we're gonna look at today and We have a visitor from Rochstor from the International University Andreas. I think you've met everyone except Nancy Andy you actually go by Andy\nSpeaker A: There's another Andreas around so it's on confusion That would be right. Yeah\nSpeaker B: Okay, so my scientific director of the email is also the dean of the International University One of his many occupations that just contributes to the fact that he is very occupied and the He might tell us a little bit about what he's actually doing and why it is somewhat related and Using maybe some of the same technologies that we are using and And was that enough of an update?\nSpeaker B: But all of you will proceed okay, maybe you have your online\nSpeaker F: Yeah, sure, so it's just been looking at Okay Been looking at the PRM stuff So this is sort of like the latest thing I have on it and I Sort of constructed a couple of classes like a user class a site class and You know a time around and then a query class and I Try to simplify it down a little bit so that I can actually look at it more as the same paper that I gave to Jerry last time So basically I took out a lot of stuff a lot of the decision notes and then try the the red lines on the graph of the Relations between the different Classes like a user has like a query and then also has you know Reference lots to its preferences the special needs and you know money and The user interests and so This is more or less similar to the flat base net that I have you know what the input nodes and all that and So I try to construct the dependency models and A lot of these stuff I got from the flat base net on what they depend on and turns out, you know The CPT is a really big if I do that. So I try to see how I can do Put in the computational notes in between and what that would look like in a PRM and so I ended up making several classes Actually, you know a class of with different attributes that Are they intermediate notes and one of them is like time affordability money affordability site availability and The travel compelability and And so some of these classes are Some of these attributes only depend on stuff from say the user or just from I don't know like the site like These here it's only like user but if you look at travel compelability For each of these factors you need to look at a pair of you know what the Preference of the user is Versus, you know what type of an event it is or you know What form of transportation the user has and whether you know the on-site parking matters to the user in that case and That makes a scenario a little different in the PRM because Then you have one user objects and potentially you can have many different sites in in mine and And so for each of the site you'll come up with this rating of travel compelability and They all depend on same users but different sites and that makes a I will have been trying to see Whether the PRM would make it more efficient if we do inferencing like that and so I Guess you end up having If you're a number of notes and a flat-based net because otherwise you would Now you would definitely have be able to reuse like All the user stuff and not not having to recompute a lot of those stuff because it's all from the user side So if you change sites you You can you know save some work on that but You know in the case where it depends on both the user and the site then I'm still having a hard time trying to see how Using the PRM will help So anyhow using those intermediate notes then this this would be the class that Represented intermediate notes and it basically is just another class in the model with you know references to the user and the site and the time and Then after you group them together this no the dependencies with of the queries would be Reduced to this and so you know it's easier to specify the CPT and all So I think that's about as far as I've gone on the PRM stuff The output Okay, so it only makes two decisions in this model and one is Basically how desirable a site is meaning How good it matches the needs of a user and the other is the mode of the visit whether it's the UVA decision So instead of Doing a lot of you know computation about you know which one site it wants to the user wants to visit I'll come more try to come up with like sort of a list of sites and For each site you know how how well it fits and basically a rating of how it fits and what to do with it So anything else I missed No, not yet\nNone: That's Will I think Rebuilding The notion of instantiating elements from This very nice It looked to me like if you were able to With a value for each of the With some other information about how much time the person is and then plan\nSpeaker B: Well, personally green looks much cleaner Certain beauty in it so the beauty is truth and we're good shape but As mentioned before we probably should look at the details so if you have a write-up then I love to read it and because Can you go all the way back to the very top?\nSpeaker B: Yeah This when these are instantiated they take on the same values that we had before or are they they changed?\nSpeaker F: I think I'd basically leave them to similar things Some of the things might be different maybe like Or that the hours are inside and eventually I meant that to me whether they're open at this hour or not And status would be more or less like with their under construction\nSpeaker B: The other question I would have is that presumably from the way the Stanford people talk about it You can put the probabilities also on the relations\nSpeaker F: Which is the structural uncertainty?\nSpeaker C: Yeah, that's That I think is actually in the previous the ubin stuff I don't remember whether they carried that over to this or not\nSpeaker B: It's sort of in the definition or in the definition of a PRM is that Classes and relations and you know have CPTs over the classes and the relations All right, uncertainty\nSpeaker F: I remember the learning when you know you don't know the structure for sure but I don't remember reading how you specify\nSpeaker B: That would exactly my question\nSpeaker C: So the plan is when definitely gets back we'll get in touch and supposedly we'll actually get seriously connected to their work If somebody was you know it's a group meeting once a week probably someone will go down whatever Whatever, figure all this out\nSpeaker B: And I think the long term perspective is pretty clear We get rock n' rollin' on this again once we get a package if when and how then this becomes Full ground that profile focused again designated And And then we'll come up with something that's way more complicated for you Because this was laughing the easy right?\nSpeaker F: Actually I had to take all the complicated stuff because I made a really complicated into beginning\nSpeaker D: So it was like this is too much\nSpeaker C: Yeah, so you could from this go on and start to people traveling together And you wanted to plan something that somehow was some free-hub\nSpeaker E: Definitely shop for artificial intelligence\nSpeaker B: Well that's not even something humans say\nSpeaker C: Well that would be like you could sell it You don't have to fight it but you speak your purpose And then you can blame the computer\nSpeaker B: But what potential results be to split up and never talk to each other again?\nSpeaker E: That should be one of them\nSpeaker C: That would be nice So there are some elaborations of this that you could try to put in to this structure But I don't think it's worth it now because we're going to see what else is going to do\nNone: Let's go to the other, there were a couple others Thanks for able to look at Can we enter them?\nSpeaker B: Then we can move on and see what Andreas has got to have his sleep\nSpeaker A: So thanks for having me here first of all So maybe just a little background on my visits I'm not really involved in any project that's relevant to you at the moment The reason is really for me to have an opportunity to talk to some of the researchers in the field And so I'll just give you a real quick introduction to what I'm working on I just hope that you have some comments or maybe you're interested in finding out more And so I'll be happy to talk to you And I'd also like to find out some more and maybe I'll just walk around the office And then ask some questions in the next couple of days So I'll be here for tomorrow and then the reiner of next week Okay, so what I started looking at to begin with is just content management systems in general So what's sort of the state of the art there is to you have a bunch of documents or learning units or learning objects And you store metadata associated to them So there's some international standards like the IEEE, it's an IEEE LOM standard And these fields are pretty straightforward, you have author information, you have size information, format information, and so on But there are two fields that are more interesting One is you store keywords associated with the document and one is you have sort of what is the document about So it's some sort of taxonomic ordering of the units Now if you sort of put on your semantic glasses, you say, well, that's not all that easy Because there's an implicit assumption behind that is that all the users of this system share the same interpretation of the keyword and same interpretation of whichever taxonomy is used So I think that's a key point of these systems and this sort of always brush over this real quickly Without really elaborating much of that As a matter of fact, the only thing that apparently really works out so far are library ordering codes Which are very coarse grains, so you have some like science biology But that's really all that we have at the moment, so I think there's a huge need for improvement there Now what this standard like this would give us is we could sort of with the search engine just query different repositories all over the world But we can't really, so what I try to do is to have, so the scenarios following you, you're working on some sort of project and you encounter a certain problem Now what we have at our university quite a bit is that students try to program a certain assignment For example, they always run into the same problems and they always come running to us and say, why is that it's not working?\nSpeaker A: We always give out the same answers without well, it would be nice to have a system that could sort of take care of this So what I want to build is basically a smart FAQ system Now what you need to do here is you need to provide some context information which is more elaborate than I'm looking for this in this keyword So, and I think I don't need to tell you this, I'm sure you have the same way when somebody else has a sentence and a certain context And the same sentence and another context makes a huge difference So I want to be able to model information like in the concepts of developing distributed systems at a computer science school What kind of software is the person using which homework assignment is he or she working on at the moment? Maybe what's the background of the students?\nSpeaker A: Which error message was encountered? So this sort of information I think should be transmitted when a certain document is retrieved Now, basically giving this, so we somehow need to have a formalized way of writing this down basically, and that's where the shared interpretation of certain terms and keywords comes in again And using this and some knowledge about the domain, I think you can do some simple inferences like you know that when somebody's working on serverlets for example, it's using Java, like a server, so I used a written in Java, so some inferences like that now using this you can infer more information and you could then match this to the metadata of the documents you're searching against So what I want to do is basically have some sort of given these inputs and then I can compute how many documents match and use this as a metric in the search Now, what I plan to do is I want to sort of do a try to improve the quality of the search result and I want to do this by having a steepest descent approach So if I knew which operating system the person was working on would disimprove my search result and having a symbolic formalized model of this, I could simply compute that and find out which questions are worth asking and that would have been propagated back to the user and sort of tried to optimize the search in this way Now the big problem that I'm facing right now is it's fairly easy to hack up a system quickly that works in the small domain but the problem is obviously the scalability So what I was mentioning earlier today is that Microsoft for example with their printer setup program has a Bayesian network which does exactly this but there you face a problem that these are very hard to extend And so what I'm what I try to do is basically try to model this in a way that you could really combine knowledge from very different sources and sort of looking into some of the ideas that the semantic web community came up with trying to have an approach how to integrate certain representation of certain concepts and also some computational rules, what you can do with those what I'm also looking into is a probabilistic approach because document retrieval is a very fuzzy procedure so it's probably not that easy to simply have a symbolic computational model that probably isn't expressive enough so that's another thing which I think you're also looking into right now and then sort of as an add-on to this whole idea that would be now depending on what the search engine or the content repository depending on which rules and which ontologies it uses or basically its view of the world and you can get very different results so it might make a lot of sense to actually query a lot of different search engines and there you could have an idea where you actually have sort of a peer-to-peer approach where we're all sort of carrying around our individual bookshelves and if you have a question about a homework it's probably make sense to ask somebody who's in your class with you sort of the guru in a certain area rather than going to some Yahoo-like search engine so this is some of the just an unnatural some of the ideas and I think a lot of the even though it's a very different domain rather than a lot of the issues are fairly similar okay\nSpeaker E: so some of the I don't know how much you know about the larger project I know yeah I know some of the issues are the same as like you know context-based factors influence how you interpret in this case and knowing one you know what kinds of things to ask you can talk about that but we haven't worried too much about the scores but maybe you guys had that in the previous\nSpeaker B: well in one small difference in a way is that he doesn't have to come up with an answer but he wants to point to the\nSpeaker A: yeah so I'm not I'm not building an expert I want to build a smart librarian basically I can point you to the right reference I don't want to compute the answer so it's a little bit easier for me\nSpeaker B: well you have to still understand what the content says about itself and then match it to what you think the information on\nSpeaker E: yes you also don't have to figure out what the content is you're just taking the key word is the topic\nSpeaker A: I assume that there will be learning systems that that tag their content and basically what I envision is that you rather it's just applying a bunch of keywords you could basically for an FAQ for example you could state sort of like a logic condition when this document applies so this document explains how to set up your male account on line-up stuff like this so something very specific that you can then but I think that the key point with these learning systems is that a learning system is only as good as the amount of content it carries you can have the best learning system with the best search interface if there's no content inside of it it's not very useful so I think ultimately because developing these rules and these inference inferences I think is very costly so I think you must be able to reuse some existing domain information or ontologies that other people wrote and tried to integrate them and then also search the entire web basically run in just a small content management system so I think that's crucial for the success of\nSpeaker E: or so you're not I guess I'm trying to figure out how how I maps to some things that we've talked about in this group and actually associated groups some of us do pretty detailed linguistic analyses and I'm guessing that you won't be doing that?\nSpeaker E: No.\nSpeaker E: Okay so you take the query and on the other hand frame that could well be useful.\nSpeaker C: Is that the frame that's great?\nSpeaker A: Yeah not too much but I have a rough way over here.\nSpeaker C: Standard story is that keywords evoke frames and the frames may well give you additional keywords or if you know that a bunch of keywords indicate a frame that you can find documents that actually have the whole frame rather than just so there's a lot of stuff and people are looking at that most of the work here is just trying to get the frames right there's language and stuff there's a lot of it there's a lot of it there are some application efforts trying to exploit it and this looks like it seems to be the place where\nSpeaker A: yeah yeah I'm sure I could learn a lot about yeah just how to how to come up with these structures because it's very easy to whip up something quickly but it maybe then makes sense to me but not to anybody else and if we want to share and integrate things they must all they must be well designed really.\nSpeaker B: Remember the prescient story?\nSpeaker B: Right.\nSpeaker B: Absolutely no linguistic background person that the IU sent over here and Andreas and I tried to come up with it or we had come up actually with him working on an interface for frame it as it was back then that would do some of the work for this machine which never got them because prescient found a happy object\nSpeaker C: and did he do what he did was much more sensible frame? Absolutely.\nSpeaker B: But so I've just said we had that idea.\nSpeaker B: The idea was there.\nSpeaker B: To exploit frame it there as well and yeah until you guys never...\nSpeaker B: And it's free news doing information extraction also right?\nSpeaker B: Right.\nSpeaker B: And with that frame it basically.\nSpeaker C: You guys never sent anybody else from IU?\nSpeaker A: Except prescient?\nSpeaker A: Yeah.\nSpeaker C: It's mostly an exchange program.\nSpeaker C: No it's fine I don't care but it's just a little surprise.\nSpeaker C: I don't know if you guys totally forgot we had a program.\nSpeaker A: I think it's really the lack of students at IU at the moment.\nSpeaker A: I think it's more the lack of students really and we have all these sponsors that are always sort of eager to get some teams and if I were a student I'd love to come here rather than for some company.\nSpeaker B: You are being recorded right now so you're working.\nSpeaker A: Oh right.\nSpeaker A: Well I didn't say anybody to anything to offend well except for sponsors maybe.\nSpeaker C: Right so that's one of the things that might be more useful.\nSpeaker C: Unfortunately Sreeni who is heavily involved in the label and all this sort of stuff is himself out of town.\nSpeaker A: But I'll go to the semantic workshop in two weeks.\nSpeaker C: Yeah for some reason he's not.\nSpeaker C: Yeah.\nSpeaker C: Who knows.\nSpeaker C: Anyway you'll see it certainly.\nSpeaker D: The other person I thought I was Dan Gilday because he did some work on topic spotting.\nSpeaker E: That would be a very good idea.\nSpeaker E: Which is I mean you, depending on how well you want to integrate with that and you know like taking the data and you said the learning systems at.\nSpeaker E: You got we there someone in XC who actually has been working on has worked on that cast of and he's worked with Frey Matt so you can have them.\nSpeaker E: Over those things I said.\nSpeaker E: And he just finished writing a draft of his thesis.\nSpeaker A: So who was that again?\nSpeaker D: Dan Gilday G I L D E A. He's in one of the.\nSpeaker B: I can take your do his office.\nSpeaker B: Okay just around the corner.\nSpeaker E: I hope you solve the problem.\nSpeaker E: Hope you can do one for us too.\nNone: All right we do anything else?\nSpeaker C: One of these times we're going to hear about Construel.\nSpeaker B: Yeah I'm sure I have.\nSpeaker B: I think it was November 2003 or something.\nSpeaker B: I had something in my calendar.\nSpeaker C: Wait a second that's a long way away.\nSpeaker B: Well maybe I could I could bribe my way out of this so I did some double checking and it seems like spring break in Trito.\nNone: That's not too much of a topic.\nNone: No.\nSpeaker B: He's trying to offer some right.\nSpeaker B: Oh they refused the battery again.\nSpeaker B: So about citrus still nothing.\nSpeaker C: We're involved in a literally $300 million program with the state of California.\nSpeaker C: And the state of California is now a month and a half behind its legally required date to approve a budget.\nSpeaker C: So the budget has not been approved.\nSpeaker C: And two days ago there's two branches of legislation.\nSpeaker C: One branch approved it.\nSpeaker C: Yesterday there was a...\nSpeaker C: I thought that the other branch was just approved but now there's actually a little backsliding people who approved it got to slide from there.\nSpeaker C: Anyway.\nSpeaker C: Well I have to tell you a wonderful story about this.\nSpeaker C: So I turned out I wound up having lunch today with a guy named Tom Colliel K-I-L-L K-A-L-I-L.\nSpeaker C: And you know works at Berkeley.\nSpeaker C: In fact he's hired to run citrus even though we don't have the money they've been hiring people right and left.\nSpeaker C: And he was, I think, the chief staffer at Clinton on technology matters.\nSpeaker C: He was in the White House and I would say.\nSpeaker C: And he liked that.\nSpeaker C: And is now doing all the politics for citrus and also has a lot of interest in actually doing things for society.\nSpeaker C: So digital divide and stuff like that.\nSpeaker C: So that's interesting to me but maybe not to you.\nSpeaker C: But the really interesting thing was he said something about, you know I'm interested in things that have high social model players.\nSpeaker C: Something that is a great social value.\nSpeaker C: He said, for example, this was his only example.\nSpeaker C: If you had an adult literacy program that was as good as an individual tutor and as compelling as a video game, then that would have a huge social impact.\nSpeaker C: It's not great. It's a good problem to work on.\nSpeaker C: Anyway, so it was nice that he's got this view of, hey that's what you should try to do and be language of people going to do it.\nSpeaker C: Definitely.\nSpeaker D: So anyway, that's the end of the story.\nSpeaker D: For adults and not for the...\nSpeaker C: I didn't push some on the child thing but, you know, again, if you get to...\nSpeaker C: And this was literacy, potentially somewhat different problem.\nSpeaker C: Maybe easier, I don't know.\nSpeaker C: So this is reading rather than teaching another project that we started on in 2004 was to try to build an automatic tutoring program for kids whose first language was in English.\nSpeaker C: Just like half the school population in California.\nSpeaker C: Something like that, isn't it?\nSpeaker C: So enormous problem in California. And the idea was, of course, it was smart about language understanding and speech understanding.\nSpeaker C: But we build programs that would be tutors for the kids.\nSpeaker C: We think we could.\nSpeaker C: Anyway, so this is a slightly different problem.\nSpeaker C: None of us have the spare time to look at it right now, but it's interesting and I may talk to you more about, is that somebody doing this?\nSpeaker B: So I did manage to get put my head out of the sling by sidetracking it through citrus, but temporarily putting it out of the sling.\nSpeaker B: But I'll volunteer to put it right back in by stating that I am amongst some other things in the process of writing up stuff that we have been discussing at our daily meetings.\nSpeaker B: And also revising, thanks for all the comments, the original construal proposal.\nSpeaker B: And if I put one in one together, I may end up with a number that's greater than one and that I can potentially present once you get back.\nSpeaker B: It's better than two.\nSpeaker B: Sometimes the sum is not less than...\nSpeaker C: Anyway, yeah, so, okay, so that would be great.\nSpeaker C: I think it's time again, right?\nSpeaker B: Absolutely, but hopefully all sidetracking, other things will have disappeared soon.\nNone: So anyway, people want to see your paper.\nNone: Thank you.\nNone: Thank you.\n"}, {"meeting_id": "Bed003", "summary": "The group discussed the first version of the Bayes-net used to work out a user's intentions when asking for directions from a navigation device. Three intentions were identified: Vista (to view), Enter (to visit) and Tango (to approach). The structure of the belief-net comprises, firstly, a feature layer, which includes linguistic, discourse and world knowledge information that can be gleaned from the data. It is possible for these variables to form thematic clusters( eg \"entrance\", \"type of object\", \"verb\"), each one with a separate middle layer.  At this stage, all the actual probabilities are ad-hoc and hand-coded. However, there has been progress in the design and organisation of experiments, that will eventually provide data more useful and appropriate for this task.", "dialogue": "None: you Sureer!\nSpeaker D: OK.\nSpeaker D: True\u4e86\nNone: Good? That's cool Go teeth Kans\nSpeaker E: Did people do these things? They know he paints the nests\nSpeaker D: But they're hound duringbring Okay.\nSpeaker D: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: I hit F7 to do that.\nSpeaker C: I'm not in there.\nSpeaker C: Robert.\nSpeaker C: Oh, the removal.\nSpeaker C: Oh, the removal.\nSpeaker B: Okay.\nSpeaker C: Because I'm already up there.\nSpeaker C: I'm in control here.\nSpeaker C: You are in control.\nSpeaker C: I'm in control.\nSpeaker C: Already?\nNone: Wow.\nSpeaker D: We're also high tech here.\nSpeaker D: Get another PowerPoint presentation.\nSpeaker D: Oh, it makes it easier to do PowerPoint.\nSpeaker D: It certainly does.\nSpeaker D: So we were...\nSpeaker C: I don't know what I mean.\nSpeaker C: Okay.\nSpeaker C: So...\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker C: Okay.\nSpeaker C: So...\nSpeaker C: Let's see.\nSpeaker C: Which one of these buttons will do this for me?\nSpeaker C: Uh-huh.\nSpeaker A: Okay.\nSpeaker C: Should we go back to the first one?\nSpeaker C: Don't want to go back to the first one?\nSpeaker C: Okay.\nSpeaker C: I'm sorry.\nSpeaker C: Well, let me...\nSpeaker C: Just do it.\nNone: Okay.\nSpeaker A: Introduce.\nSpeaker A: Yeah.\nSpeaker A: Well, the search for the middle layer.\nSpeaker A: Basically, uh...\nSpeaker A: I can read.\nSpeaker A: That's what...\nSpeaker A: It's just a first to the fact that...\nSpeaker A: One of the main things we had to do was to decide what the intermediate sort of nodes were.\nSpeaker A: Uh-huh.\nSpeaker B: But if you really want to find out what the body of the body have to click on the little black bulb.\nSpeaker B: I've never...\nSpeaker C: I don't know what the light bulb was for.\nSpeaker C: I didn't install that into my PowerPoint presentation.\nSpeaker B: It obviously assistant that tells you that the font type is too small.\nSpeaker C: Uh...\nSpeaker C: Try it?\nSpeaker C: I prefer not to.\nSpeaker C: Okay.\nNone: Continue.\nSpeaker D: It's a needless good idea, is that the idea?\nSpeaker B: Why are you doing this in this mode and then the presentation mode?\nSpeaker C: Because I'm going to switch to the Java base.\nSpeaker C: Oh, okay.\nSpeaker C: Programming.\nSpeaker C: If I do that...\nSpeaker C: I was wondering.\nSpeaker C: Is that okay?\nNone: Yeah, it's okay.\nSpeaker D: Proceed.\nSpeaker D: Sumize the window.\nSpeaker A: Proceed.\nSpeaker A: Wait, what do you want to do?\nSpeaker A: You maximize the window so that stuff on the side isn't...\nSpeaker A: No, it's for kids.\nSpeaker C: Well, I can do that, but then I have to end the presentation in the middle so I can go back to open up Java base.\nSpeaker C: Okay, fine.\nSpeaker C: Here.\nSpeaker C: Let's see if I can...\nSpeaker C: Alright.\nSpeaker C: Very nice.\nSpeaker C: Is that better?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: I'll also get rid of this click to add nodes.\nSpeaker C: Okay, perfect.\nSpeaker C: So then the features we decided we were talked about, right?\nSpeaker C: The prosody to discourse, verb choice.\nSpeaker C: Now we had a list of things like to go and to visit and whatnot.\nSpeaker C: The landmarkiness of a...\nSpeaker C: I need you to like...\nSpeaker C: Nice coinage.\nSpeaker C: Thank you.\nSpeaker C: Of a building.\nSpeaker C: Well, we actually have a separate feature, but I decided to put on the same line for space.\nSpeaker C: Nice walls, which we can look up because I mean if you're going to get real close to a building in the tango mode, right?\nSpeaker C: It's going to be a reason for it.\nSpeaker C: And it's either because you're in route to something else or you want to look at the walls.\nSpeaker C: The context, which in this case we've limited to businessperson, tourist or unknown.\nSpeaker C: The time of day, and open to suggestions is an actually feature.\nSpeaker C: We are open to suggestions.\nSpeaker D: Can I just ask the nice walls part of it is that in this particular domain, you said it could be on two different lines.\nSpeaker D: But are you saying that in this particular domain, it happens to the landmarkiness correlated with...\nSpeaker C: Oh, yes, that's correct.\nSpeaker C:...there are separate things.\nSpeaker C: They're being nice.\nSpeaker C: Yeah.\nSpeaker C: I either could put nice walls on its own line or open to suggestion.\nSpeaker A: Like you could have a post office with nice murals or something.\nSpeaker A: Right.\nSpeaker A: Okay.\nSpeaker D: So nice walls is a stand-in for like architecturally significant.\nSpeaker D: I see the thing is if you're outside.\nSpeaker C: Okay.\nSpeaker C: Yeah, but if it's architecturally significant, you might be able to see it from...\nSpeaker C: You might be able to visit it, right?\nSpeaker C: And be able to...\nSpeaker B: Appreciate it.\nSpeaker C: Yeah, versus...\nSpeaker C: Like I was at this place in Europe or the head of little carvings.\nSpeaker C: And like dead people in the walls or something.\nSpeaker C: I don't remember.\nSpeaker C: It was a long time ago.\nSpeaker C: There's a lot of those.\nSpeaker C: But if you look at it real close, you can see the intricacy of the wall.\nSpeaker D: Okay.\nSpeaker D: So that counts as nice wall.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: Something you want to inspect at close range is interesting.\nSpeaker D: Exactly.\nSpeaker B: Okay.\nSpeaker B: Robert.\nSpeaker B: Well, there is a term that's often used at saliency or the salience of an object.\nSpeaker B: And I was just wondering whether that's the same as what you described as landmarkiness.\nSpeaker B: But it's really not.\nSpeaker B: I mean, an object can be very salient, but not a landmark at all.\nSpeaker D: There's landmark for tourist degrees and landmark for navigational reasons or something.\nSpeaker A: Yeah, we went tourist degrees.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker D: But you can imagine maybe wanting the other, both kinds of things there for different goals.\nSpeaker C: Yeah.\nSpeaker C: Right.\nSpeaker C: But, yeah.\nSpeaker C: Tourist line marks also happen to be...\nSpeaker C: Wouldn't they also be...\nSpeaker C: They're not exclusive groups, are they?\nSpeaker C: Like, non-tourist line marks and...\nSpeaker C: They're not mutually exclusive?\nSpeaker C: Yeah.\nSpeaker D: Right.\nSpeaker D: Definitely.\nSpeaker C: Okay.\nSpeaker C: So our initial idea was not very satisfying because our initial idea was basically all the features pointing to the output node.\nSpeaker D: So a big flat structure.\nSpeaker C: Right.\nSpeaker C: And so we...\nSpeaker C: Reasons being, you know, it'd be a pain to set up all the probabilities for that.\nSpeaker C: If we moved on to the next step and did learning of some sort according to Busker, we'd be handicapped.\nSpeaker C: Well, he was very well.\nSpeaker A: And if you have any features in the end, it's exponential.\nSpeaker C: And it wouldn't look pretty.\nSpeaker A: Yeah.\nSpeaker C: They'd all be like pointing to the one node.\nSpeaker C: So our next idea was that in middle layer, right?\nSpeaker C: So the thinking behind that was we have the features that we've drawn from the communication of some...\nSpeaker C: Like, the person at the screen is trying to communicate some abstract idea.\nSpeaker C: Like, I'm...\nSpeaker C: The abstract idea being, I am a tourist.\nSpeaker C: I want to go to this place.\nSpeaker C: Right.\nSpeaker C: So we can attract features along lines of where they want to go and what they've said previously and whatnot.\nSpeaker C: And then we have the means that they should use, right?\nSpeaker C: But the middle thing we were thinking of along lines of maybe trying to figure out like the concept of whether they're a tourist or whether they're running an errand or something like that along those lines.\nSpeaker C: Or, yes, we could...\nSpeaker C: Things we could extract from the data of the hidden variables.\nSpeaker C: Yes, good.\nSpeaker C: So then the hidden variables we came up with or whether someone was on a tour running an errand, they were there in a hurry because we were thinking if they're in a hurry, they'd be less likely to...\nSpeaker C: Like, want to do this to right?\nSpeaker C: If you want to view things, you wouldn't be in a hurry.\nSpeaker C: Or they might be more likely to be using the place that they want to go to as a...\nSpeaker C: Like, a navigational point to go to another place.\nSpeaker C: Whether the destination was their final destination, whether the destination was closed was real.\nSpeaker C: And then let's look at the belief net.\nSpeaker C: Okay, so that means I should switch to the other program.\nSpeaker C: Right now it's still kind of in a toy version of it because we didn't know the probabilities of...\nSpeaker C: Well, I'll talk about it when I get that picture up.\nSpeaker B: No one knows it.\nSpeaker C: Okay, so this is what we...\nSpeaker C: Let's see, that's what I maximize this.\nSpeaker C: There we go.\nSpeaker C: So the mode basically has three different outputs, whether the probability of Vista Tango or Enter.\nSpeaker C: The context we simplified, basically it's just the businessman, the tourist unknown.\nSpeaker C: Verbe used is actually personally amusing.\nSpeaker C: And then because it's just whether the verb is a Tango verb, an Interverb or a Vista verb.\nSpeaker D: Yeah, that one is a verb.\nSpeaker D: And are those mutually exclusive?\nSpeaker A: Not at all, that is a lot of work.\nSpeaker A: That would have made the probability significantly more complicated to enter.\nSpeaker A: So we decided that for the purposes of this...\nSpeaker C: Yeah, that's pretty much it.\nSpeaker C: Right.\nSpeaker C: Why don't you mention things about this busker that I am...\nSpeaker A: That are not coming to my mind right now.\nSpeaker A: Yeah, so the four notes down there are the sort of things that are not directly extracted.\nSpeaker A: Actually the five things, the closed is also not directly extracted, I guess.\nSpeaker A: From the...\nSpeaker A: From the...\nSpeaker C: No, it's sort of it is because it's...\nSpeaker C: Okay, closed, we have the time of day and the close, it just had...\nSpeaker C: Right.\nSpeaker A:...for time it closed.\nSpeaker A: Right, but the other one is the final destination, the whether they're doing business, whether they're in a hurry and whether they're tours.\nSpeaker A: That kind of thing is all sort of...\nSpeaker A: You know, probabilistically...\nSpeaker A: Inferred from the other ones?\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: And the mode, you know, depends on all those things.\nSpeaker C: Yeah, the actual parts are somewhere.\nSpeaker C: Yeah, so we have an intervention.\nSpeaker A: Like we don't have notes for discourse in parts, although like in some sense they are parts of this belief net.\nSpeaker A: But the idea is that we just extract those features from them so we don't actually have a node for the entire part.\nSpeaker A: Right.\nSpeaker A: Because we'd never do inference on it anyway.\nSpeaker D: So some of the top row of things, what's disc admission?\nSpeaker A: Whether they discuss the admission fee.\nSpeaker A: So we looked at the data, a lot of the data people were saying things like, can I get to this place? What is the admission fee?\nSpeaker A: So that's like a huge...\nSpeaker A: They're trying to enter the place rather than to...\nSpeaker A: Right.\nSpeaker A: Time goes.\nSpeaker C: Okay, I see.\nSpeaker C: There would be other things besides just the admission fee.\nSpeaker C: But, you know, we didn't know that.\nSpeaker C: That was like our example.\nSpeaker C: That was the initial one I was telling.\nSpeaker D: Okay, so there are certain cues that are very strong...\nSpeaker D: From the discourse.\nSpeaker D: Either like school or topic based content cues for one of those.\nSpeaker D: And then that second row, or whatever that row of time of day through that.\nSpeaker D: So all of those, some of them come from the utterance and some of them are sort of either world knowledge or situational things.\nSpeaker D: Right.\nSpeaker C: So the new distinction between those and everything else we want to say about Skye.\nSpeaker C: Yeah, I mean...\nSpeaker A: One thing...\nSpeaker B: There are a couple more things.\nSpeaker B: I mean, I would actually suggest we go through this one more time.\nSpeaker B: So we all agree on what the meaning of these things is at the moment and maybe what changes.\nSpeaker C: Yeah, okay.\nSpeaker C: So one thing I'm sure about is how we have the admission fee thing set up.\nSpeaker C: So one thing that we were thinking was by doing the layers like this, we kept things from directly affecting the mode beyond the concept.\nSpeaker C: But you could see perhaps the admission fee going directly to the mode, pointing it in to right versus pointing just at tourist.\nSpeaker C: Okay.\nSpeaker C: But we just decided to keep all the things we extracted to point at the middle and then down.\nSpeaker B: Why is the landmark...\nSpeaker B: Okay, the landmark is facing to the tourists.\nSpeaker B: That's because we're talking about landmarks, the tourist landmarks.\nSpeaker B: Right, yeah.\nSpeaker B: As possible...\nSpeaker B: Navigation...\nSpeaker B: Navigation landmarks.\nSpeaker B: So...\nSpeaker C: Yeah, that would be whatever building they referred to.\nSpeaker A: Right.\nSpeaker A: So the variable admission fee is a binary thing, time of day is like morning, afternoon, night, is that the deal?\nSpeaker C: That's how we have it currently set up, but it could be based upon hour or...\nSpeaker C: Or we could just re-hidden it, discretize it.\nSpeaker A: Normally, context would include a huge amount of information, but we are just using the particular part of the context, which consists of the switch that they flick to indicate whether they're tourist or not.\nSpeaker A: Okay, so it's given in there.\nSpeaker A: Right, so it's not really all of context.\nSpeaker A: The velocity is not all of the velocity, but simply for our purposes, whether or not the appear tends to relax.\nSpeaker B: That's very nice.\nSpeaker B: Okay.\nSpeaker B: So the context is the switch between tourist or non-tourist or unknown?\nSpeaker C: Or unknown.\nSpeaker D: Yeah, unknown.\nSpeaker D: So final desk...\nSpeaker D: So it seems like that would really help you for doing business versus tourist.\nSpeaker D: Which one?\nSpeaker D: Okay, so the context being...\nSpeaker D: I don't know if that question is sort of in general, are you...\nSpeaker D: Does it allow business people to be doing non-business things at the moment?\nSpeaker A: Yeah, it does.\nSpeaker D: Okay, so then you just have some probabilities over which of those it is.\nSpeaker A: Right, so then landmark is...\nSpeaker A: Oh, sorry, verb used is like...\nSpeaker A: Right now we only have three values, but in general there would be a probability distribution overall.\nSpeaker A: Verbs.\nSpeaker A: Rather, let me rephrase that.\nSpeaker A: It can take values in the set of all verbs that they could possibly use.\nSpeaker A: Nice walls, this binary, closed this binary, final destination.\nSpeaker A: Again, yeah, all those are binary, I guess.\nSpeaker A: And mode is one of three things.\nSpeaker B: So the middle layer is also binary now?\nSpeaker A: Yeah, anything with the question mark after it in that picture is a binary node.\nSpeaker B: Yeah, but also things without question marks are also binary.\nSpeaker B: Which things?\nSpeaker A: Nice walls.\nSpeaker A: Oh, nice walls is something that we extract from our world knowledge.\nSpeaker A: Yeah, sorry, it is binary.\nSpeaker A: It is binary, but it doesn't have a question mark, because it's very important.\nSpeaker B: Yeah, similarly closed, I guess.\nSpeaker B: So we can either be in a hurry or not, but we cannot be in a medium hurry at the moment?\nSpeaker A: Well, to do that we would add another value for that.\nSpeaker A: And that would require updating the probability distribution for mode as well, because it would now have to take that possibility into account.\nSpeaker D: So, of course this will happen when we think more about the kinds of verbs that are used in each case, but you can imagine that it's verb plus various other things that are also not in the bottom layer that would help you.\nSpeaker D: It's a conjunction of the verb used and some other stuff that would...\nSpeaker D: Right, other conjunction.\nSpeaker B:...declination in the exact way.\nSpeaker B: Well, the sort of landmark is sort of the object, right?\nSpeaker B: The argument.\nSpeaker D: Usually.\nSpeaker D: I don't know if that's always the case, I guess I haven't looked at the data as much as you guys have.\nSpeaker B: It's always worth putting in something.\nSpeaker B: Maybe at this stage we do want to sort of get modifiers in there, because they may also tell us whether the person is in a hurry or not.\nSpeaker C: Yeah, I want to get to the church quickly.\nSpeaker B: That would be a cue.\nSpeaker B: What's the fastest way?\nSpeaker B: Yeah, correct.\nSpeaker C: Alright, excellent.\nSpeaker C: Do we have anything else to say about this?\nSpeaker C: We can do a little demo.\nSpeaker C: Oh, yeah, we could.\nSpeaker C: But the demo doesn't work very well.\nSpeaker B: Okay.\nSpeaker B: Then it wouldn't be a demo.\nSpeaker A: I mean, we can do a demo in the sense that we can observe the fact that this will in fact do inference, so we can set some of the nodes and then find the probability of other nodes.\nSpeaker A: What should I observe?\nSpeaker A: Just set a few of them.\nSpeaker A: You don't have to do the whole thing.\nSpeaker A: Maybe the fact that they use a certain verb.\nSpeaker A: Actually, forget the verb.\nSpeaker A: I don't know.\nSpeaker A: See, they discussed the admission fee.\nSpeaker A: And the place has nice walls.\nSpeaker C: I love nice walls.\nSpeaker C: I'm a big fan.\nSpeaker C: I started to grow on it.\nSpeaker A: And at the time of day's night.\nSpeaker A: Yeah, no wait.\nSpeaker A: That doesn't sound inconsistent.\nSpeaker A: They don't discuss the admission fee.\nSpeaker A: Make that false.\nSpeaker A: Alright.\nSpeaker A: It's night.\nSpeaker C: Was it okay?\nSpeaker C: I forgot to.\nSpeaker C: Nothing to bite me, a job of basis.\nSpeaker C: You have to click that and do this.\nSpeaker D: Yeah, that seems kind of redundant.\nSpeaker C: Okay.\nSpeaker C: That all you want?\nSpeaker C: Yes.\nSpeaker C: Okay.\nSpeaker C: So let's see.\nSpeaker C: I want to query, right?\nSpeaker C: The mode.\nSpeaker C: Okay, and then on here.\nSpeaker C: So let's see.\nSpeaker A: So that is probably the entering, Vistai, or a tag word.\nSpeaker D: And so slightly biased toward tangoing?\nSpeaker C: Yeah.\nSpeaker C: If it's night time, they have not discussed admission fee and the net walls are nice.\nSpeaker C: So yeah, I guess that sort of makes sense.\nSpeaker C: The reason I say that demo doesn't work very well is that we observed everything in favor of taking a tour.\nSpeaker C: And it came up as tango, right?\nSpeaker C: Over and over.\nSpeaker C: We couldn't figure out how to turn it off of tango.\nSpeaker A: Huh.\nSpeaker A: It loves the tango.\nSpeaker A: But that's obviously just to do with our probabilities.\nSpeaker A: Yeah, yeah.\nSpeaker A: We totally hand to the problem.\nSpeaker A: Okay, well, the question is this in this news.\nSpeaker A: Let's say 40% for this.\nSpeaker A: So obviously that's going to happen.\nSpeaker A: Right.\nSpeaker D: Yeah, I put it.\nSpeaker D: Maybe the biased toward tangoing was yours then.\nSpeaker D: Yeah.\nSpeaker A: So we have to like fit the probabilities.\nSpeaker C: Spent my youth practicing the tango delamuerte.\nSpeaker B: However, you know, the purpose was not really at this stage to come up with the penalties.\nSpeaker B: But to get thinking about that hidden middle layer.\nSpeaker C: We would actually, I guess we're going to see, look at the data more.\nSpeaker C: We'll get more hidden nodes.\nSpeaker C: Yeah.\nSpeaker C: But I'd like to see more.\nSpeaker C: Not because it would expedite the probabilities.\nSpeaker C: Because it wouldn't.\nSpeaker C: It would actually slow that down tremendously.\nSpeaker A: Well, yeah, I guess.\nSpeaker A: But not that much though.\nSpeaker C: When do you literally?\nSpeaker C: I think we should have exponentially more middle nodes than features we've extracted.\nSpeaker C: I'm just interested.\nSpeaker D: So are doing business versus tourists?\nSpeaker D: They prefer your current task.\nSpeaker A: Like with current what you want to do at this moment.\nSpeaker A: That's an interesting point.\nSpeaker A: Whether you're, whether it's not, I think it's more like are you a tourist?\nSpeaker A: Are you in Ham, like, Hydealberg for?\nSpeaker D: I thought that's directly given by the context.\nSpeaker D: That's the difference.\nSpeaker A: What if the context which is not said, but still they say things like, I want to go see the castle and.\nSpeaker C: I kind of said, not of doing business is more of running an errand type thing.\nSpeaker A: Yeah, business on the other hand is definitely what you're doing.\nSpeaker B: So if you're running a tourist and.\nSpeaker D: Oh, I see.\nSpeaker D: You may have a task.\nSpeaker D: You have to go get money.\nSpeaker D: So you're doing business at that stage.\nSpeaker D: Right.\nSpeaker D: How do I get the money?\nNone: I see.\nSpeaker A: And that'll affect whether you want to enter.\nSpeaker D: Okay.\nSpeaker D: So the tourist node should be very consistent with the context node.\nSpeaker D: Right.\nSpeaker D: If you say that's more there.\nSpeaker D: Yeah.\nSpeaker A: In general, in this context node is a bit of a background.\nSpeaker A: I don't know.\nSpeaker A: Like, do we want to have like, are you assuming that or not?\nSpeaker D: Like, is that to be?\nSpeaker D: I mean, if that's accurate, then that would be.\nSpeaker A: The context node is said one way or another that like strong the.\nSpeaker A: So something or whether or not there to us.\nSpeaker B: So what's interesting is when it's not when set to unknown.\nSpeaker B: We're okay.\nSpeaker A: Set the set the context to unknown.\nSpeaker A: Okay.\nSpeaker A: Right now we haven't observed it.\nSpeaker A: So I guess it's sort of averaging over all those three possibilities.\nSpeaker B: But yes, you can set it to unknown.\nSpeaker B: And if we know to do everything else.\nSpeaker B: As is.\nSpeaker B: The result should be the same, right?\nSpeaker A: No.\nSpeaker A: Well, no, because we the way we set the probabilities.\nSpeaker A: My nod if.\nSpeaker A: Yeah, it's it's an issue, right?\nSpeaker A: Like.\nSpeaker B: Pretty much the same.\nSpeaker A: Yeah, it is.\nSpeaker A: So the issue is that in beliefness is not common to do what we did.\nSpeaker A: Like having, you know, a bunch of values and then unknown as an actual value.\nSpeaker A: Yeah.\nSpeaker A: What's common is you just like don't observe the variable right and then just marginalizes.\nSpeaker A: But we didn't do this because we felt that they're I guess we were thinking in terms of a switch.\nSpeaker A: Yeah, yeah.\nSpeaker A: But I don't know what the right things to do for that.\nSpeaker A: I don't know if I told you I'm happy with the way it is.\nSpeaker B: Why don't we, how long would it take to add another note on the observatory and play around with it?\nSpeaker C: Another note on what?\nSpeaker C: Well, it depends on how many things it's linked to.\nSpeaker B: Let's just say make it really simple if we create something that for example would be.\nSpeaker B: So somethings can be landmarks in your sense, but they can never be entered.\nSpeaker B: So for example, statue.\nSpeaker B: Right.\nSpeaker B: So maybe we want to have landmark meaning now, interrubble landmark versus something that's simply just a vista point for example.\nSpeaker B: Yeah, statue or.\nSpeaker A: So basically it's just a variable that's entrubble or not.\nSpeaker A: So like an intruple question mark.\nSpeaker C: Also, you know, didn't we have a size this one?\nSpeaker C: What size of the landmark?\nSpeaker C: I guess it's not when we were doing this, but I guess it's a point.\nSpeaker C: It's a reason I had that.\nSpeaker C: Okay, that was a thought that I had at one point, but then went away.\nSpeaker A: So you want to have a note for like whether or not it can be entered?\nSpeaker B: Well, for example, if we include that.\nSpeaker B: Yeah.\nSpeaker B: Accessibility or something.\nSpeaker B: Can it be entered?\nSpeaker B: Then of course this is sort of binary as well.\nSpeaker B: And then is also the question whether it may be entered in the sense that, you know, if it's from the house of Tom Cruise, you know, it's entrubble, but you may not enter it.\nSpeaker B: You're not allowed to.\nSpeaker B: Yeah.\nSpeaker B: And as you are whatever is divorce lawyer or.\nSpeaker B: And these are very observable sort of from the from the ontology sort of things.\nSpeaker C: Does it actually help to distinguish between those two cases though?\nSpeaker C: Whether it's practically speaking intribule or actually physically intribule or not?\nSpeaker D: It seems like a word for determining whether they want to go into or not.\nSpeaker B: Well, I can see why.\nSpeaker B: If you're running an errand, you may be more likely to be able to enter places that are usually not.\nSpeaker B: And you know, usually not allowed to out.\nSpeaker B: Let's get this clear.\nSpeaker B: So it's a matrix between.\nSpeaker B: If it's not enterable, whether it's whether it's a public building.\nSpeaker C: And whether it's actually has a door.\nSpeaker B: Yeah, exactly.\nSpeaker B: Okay.\nSpeaker C: So Tom Cruise's house is not a public building, but it has a door.\nSpeaker C: But the thing is, okay, explain to me why it's necessary to distinguish between whether something has a door and is not public.\nSpeaker C: Or something, it seems like it's equivalent to say that it doesn't have a door and it or not public and not a door equivalent of things.\nSpeaker C: It seems like in practice.\nSpeaker B: Right?\nSpeaker B: Yeah.\nSpeaker B: So we would have, what does that mean then that we have to, we have an object type statue that really is an object type.\nSpeaker B: So there's going to be a bunch of statues.\nSpeaker B: And then we have, for example, an object type, let's say hotel.\nSpeaker B: How about hotels?\nSpeaker B: Okay.\nSpeaker B: So the most famous building in Heidelberg is actually a hotel.\nSpeaker B: So a hotel Sumrita, which is the only Renaissance building in Heidelberg that was left after the big destruction and for the 30 years wall.\nSpeaker B: Does it have nice walls?\nSpeaker B: It has wonderful walls.\nSpeaker B: Excellent.\nSpeaker B: And lots of detail, carvings and cravings and so on.\nSpeaker B: But still an unlikely candidate for the Tango boat, I must say.\nSpeaker B: But so if you are a, well, it's very tricky. So I guess your question is so far I have no real argument why to differentiate between statues and houses of celebrities.\nSpeaker B: Okay, let's do a, can we add to so I can see how it's done a has door property or what would it connect to?\nSpeaker B: Like what would affect it might affect?\nSpeaker B: Oh, actually, it wouldn't affect any of our notes, right?\nSpeaker B: What I was thinking was if you had a, like, it affects the doing business.\nSpeaker C: You could affect, theoretically, you could affect doing business with has door.\nSpeaker A: Yeah.\nSpeaker B: Okay.\nSpeaker B: Right. That's a, yeah.\nSpeaker A: I don't know if Java base is nice about that.\nSpeaker A: It might be that you add a new thing pointing to a variable.\nSpeaker A: It just like it just overrides everything.\nSpeaker A: But you can check.\nSpeaker C: Oh, we have it saved.\nSpeaker C: So we can open it up again.\nSpeaker C: It's true.\nSpeaker C: The safety net.\nSpeaker D: I think you could just add it.\nSpeaker D: I mean, I have to be working.\nSpeaker A: Well, that's fine.\nSpeaker A: But we have to see the function now.\nSpeaker A: Has it become all 0.5's or not?\nSpeaker A: Let's see.\nSpeaker C: So this is has door.\nSpeaker C: It's true false.\nSpeaker C: That's acceptable.\nSpeaker C: I want to edit the function going to that, right?\nSpeaker C: No, this is fine.\nSpeaker C: This is fine.\nSpeaker C: Right.\nNone: That is fine.\nSpeaker A: I'm going to edit this one.\nSpeaker A: Yep.\nSpeaker A: What would be nice if it is if it just like kept the old function for right away?\nSpeaker C: But nope.\nSpeaker C: Oh, I should do it.\nSpeaker C: Oh, wait.\nSpeaker C: It might be that we...\nSpeaker C: Yes.\nSpeaker C: That's not good.\nSpeaker A: That's kind of annoying.\nSpeaker A: Okay.\nNone: So just just just miss everything.\nSpeaker B: Close it and load up the old state.\nSpeaker B: So.\nSpeaker B: That's it.\nSpeaker B: Doesn't screw that up.\nSpeaker B: Okay.\nSpeaker B: Maybe you can read it.\nSpeaker B: So I've used Java Beaslot.\nSpeaker B: I haven't used it a lot.\nSpeaker B: And I haven't used it a lot.\nSpeaker D: I haven't used it a lot.\nNone: You know, many months.\nNone: So.\nNone: Okay.\nNone: We can ask someone.\nNone: It might be worth asking around.\nSpeaker D: Like, I'm not going to be able to do it.\nSpeaker A: We can ask someone.\nSpeaker A: It might be worth asking around.\nSpeaker A: Yeah.\nSpeaker A: We looked at sort of a...\nSpeaker D: To me.\nSpeaker D: To me, it's a one to ask I would say.\nSpeaker D: Yeah.\nSpeaker A: He might know.\nSpeaker A: Because, yeah.\nSpeaker A: I mean, in a way this is a lot of good features in Java.\nSpeaker A: It has a gooey and it's...\nSpeaker A: It's those are the main two things.\nSpeaker A: It does learning.\nSpeaker A: It doesn't actually.\nSpeaker C: What?\nSpeaker C: I didn't think it did learning.\nSpeaker C: Okay.\nSpeaker A: Maybe I did a little bit of learning.\nSpeaker A: Okay.\nSpeaker A: Maybe you're right.\nSpeaker A: Okay.\nSpeaker A: Right.\nSpeaker A: But it's free.\nSpeaker A: Which is quite positive here.\nSpeaker A: But, yeah.\nSpeaker A: Maybe another thing that...\nSpeaker A: But I mean, is interface is not degrees.\nSpeaker C: But it actually had an interface.\nSpeaker C: A lot of them were like, you know.\nSpeaker C: Yeah.\nSpeaker C: Command line.\nSpeaker B: What is the code?\nSpeaker B: Can we see that?\nSpeaker B: How do you write the code?\nSpeaker B: Or do you actually never have to write any code there?\nSpeaker A: Yeah.\nSpeaker A: There is actually a text file that you can edit.\nSpeaker A: But it's...\nSpeaker C: You don't have to do that.\nSpeaker C: It's like an XML format for base nets.\nSpeaker C: Is it XML?\nSpeaker C: There is one.\nSpeaker C: I don't know if this uses...\nSpeaker C: Oh, I see no.\nSpeaker A: This doesn't use it.\nSpeaker A: But...\nSpeaker A: You can look at the text file.\nSpeaker A: Yeah.\nSpeaker A: But do you have it here?\nSpeaker A: Yes, I do, actually.\nSpeaker C: Oh, yes, of course.\nSpeaker C: Oh, man.\nSpeaker C: Is there an ampersand in DOS?\nSpeaker C: Nope.\nSpeaker C: Just start up a new DOS.\nSpeaker C: That's all right.\nSpeaker C: I can probably double click.\nSpeaker C: Click on it.\nSpeaker C: Or...\nSpeaker A: Yeah, right.\nSpeaker C: Let's see.\nSpeaker C: Yep.\nSpeaker A: Let's see.\nSpeaker A: Come on.\nSpeaker A: You'll ask you what you wanted once...\nSpeaker A: What you want to open it with and see...\nSpeaker A: What bad, I guess?\nSpeaker C: These days, it should open this theoretically.\nSpeaker C: Go with the right mouse.\nSpeaker B: Open with.\nSpeaker C: That's all.\nSpeaker C: There we go.\nSpeaker C: I was just...\nSpeaker C: Oh.\nSpeaker C: I was dead.\nSpeaker C: To the world.\nSpeaker B: Okay.\nSpeaker B: To the old note pet, that's my favorite pet.\nSpeaker B: I like word pet?\nSpeaker C: I like word pet because it has the returns, the carriage returns on some of them.\nSpeaker C: And how do I get autofills, I guess, or whatever you call it?\nSpeaker C: Anyway, there it is.\nSpeaker B: So this sort of list-be?\nSpeaker E: Oh, yeah.\nSpeaker C: It just basically looks like it is as much as tracks.\nSpeaker A: Yeah, that's how actual probability tables are specified.\nSpeaker A: It's like lists of numbers.\nSpeaker A: So theoretically, you would edit that.\nSpeaker A: They're not very friendly.\nSpeaker A: Yeah, the ordering is very clear.\nSpeaker A: So you'd have to figure out the table.\nSpeaker A: Yeah.\nSpeaker C: Well, if we could write a program, we could generate this.\nSpeaker C: I think so.\nSpeaker A: Yeah, you could.\nSpeaker A: Yeah, we can maybe write an interface for entering probability distributions easily.\nSpeaker A: Like a little script.\nSpeaker A: That might be worth it.\nSpeaker D: Yeah, I actually seem to recall extremely complaining about something to do with entering probability.\nSpeaker D: So this is probably-\nSpeaker A: The other thing is it isn't Java.\nSpeaker C: So we could manipulate the source itself. Or-\nSpeaker B: Do you have the source file?\nSpeaker A: I don't know if you actually have it. I saw it directly called source.\nSpeaker A: I didn't- Yeah.\nSpeaker C: Go up one.\nSpeaker C: Up one.\nSpeaker C: Yeah.\nSpeaker C: Oh, yes.\nSpeaker C: Good source.\nSpeaker C: That's quite nice.\nSpeaker A: I don't know if you actually manipulate the source though.\nSpeaker A: That might be a bit complicated.\nSpeaker A: I think it might be simply just- Have a script that, you know, it's like friendly and logical.\nSpeaker E: The two types of things as well.\nSpeaker E: Yeah.\nSpeaker B: Right.\nSpeaker B: But if there is an XML file that a format that it can also read, I mean, it just reads this right when it starts.\nSpeaker C: Yeah, I know there is an- I was looking on the web page and he's updated it for an XML version of, I guess, Bayesnet.\nSpeaker C: Social Bayesnet is back for it in the next model.\nSpeaker A: He's- Like this guy has?\nSpeaker A: So, but he doesn't use it.\nSpeaker A: So in what sense is he updated it?\nSpeaker C: Well, you can either- You can read both.\nSpeaker C: Oh, I see.\nSpeaker A: To my understanding.\nSpeaker C: No, that would be awesome.\nSpeaker C: Because, well, at least I could have misread the web page I haven't been doing that, but-\nSpeaker B: Good, wonderful.\nSpeaker C: So you got more slides? Do I have more slides?\nSpeaker C: Um, yes, one more.\nSpeaker C: Future work.\nSpeaker C: I think every presentation should have a future work slide.\nSpeaker C: But, uh, it's basically already talked about all this stuff, so-\nSpeaker A: The addition of things I guess learning the probabilities also. That's maybe, I don't know.\nSpeaker C: That's future-future work.\nSpeaker C: That's, yeah.\nSpeaker C: Right.\nNone: Future.\nSpeaker C: And of course, if you have a presentation that doesn't- That doesn't work at all, then you have what I learned.\nSpeaker C: As a slide.\nSpeaker D: Can't you have both?\nSpeaker C: You could.\nSpeaker C: My first approach failed.\nSpeaker C: What I learned.\nSpeaker C: Okay, so I think that- I had a presentation finished.\nSpeaker C: Good.\nSpeaker C: I don't know what I'd like about these meetings is one personal nod, and then the next personal nod just goes all the way around the room.\nSpeaker C: So, yeah.\nSpeaker C: Yeah, earlier I went and busker went.\nSpeaker C: And he did.\nSpeaker B: Like yawning.\nSpeaker B: This and this one wasn't straight away, yo.\nSpeaker B: Okay, so this means-\nSpeaker C: Should I pull up the-\nSpeaker D: Yeah, could you put the net up again? There we go.\nSpeaker C: And then, actually, is that gonna wireless mic on?\nSpeaker D: So, a more general thing than discussed admission fee could be- I'm just wondering whether the context, the background context of the discourse might be, I don't know if there's a way to divide it, or maybe, you know, generalize it in some way.\nSpeaker D: There might be other cues that say, in the last few utterances, there has been something that is strongly associated with, say, one of the particular modes.\nSpeaker D: I don't know if that might be- And into that node would be various things that could just specifically come up.\nSpeaker B: I think a sort of general strategy here, you know, this is excellent because that's your thinking along these terms is that maybe we could observe a couple of discourse phenomena such as the admission fee and something else and something else that happened in the discourse before.\nSpeaker B: And let's make those four.\nSpeaker B: And maybe there are two- So maybe this could be sort of a separate region of the net, which has two- has its own middle layer.\nSpeaker B: Maybe this, you know, has some kind of funky thing that if this and this may influence these hidden nodes of the discourse, which is maybe something that is a more general version of the actual phenomenon that you can observe.\nSpeaker B: So things that point towards-\nSpeaker C: So instead of a single node for, like if they said the word admission fee- Exactly, yeah.\nSpeaker C: Admission fee or maybe, you know, how much to enter or, you know, something other cues- Open hours exactly.\nSpeaker C: They would all funnel into one node that would constitute entrance requirements or something like that.\nSpeaker C: Yeah, visit.\nSpeaker B: Sure.\nSpeaker D: I mean, it sort of gets into plan recognition kinds of things in the discourse.\nSpeaker D: I mean, that's like the big thing, you know.\nSpeaker B: And then maybe there are some discourse acts if they happen before.\nSpeaker B: It's more for a cue that the person actually wants to get somewhere else and that you're in a route, sort of proceeding past these things.\nSpeaker B: So this would be just something that where you want to pass it.\nSpeaker B: Is it it?\nSpeaker B: However, these are of course then the nodes, the observed nodes for your middle layer.\nSpeaker B: So this again points to final destination doing business, tourist, hurry and so forth.\nSpeaker B: Okay.\nSpeaker B: And so then we can say, okay, we have a whole-\nSpeaker D: That's a whole set of discourse related cues to your middle layer.\nSpeaker B: Yeah, exactly. And this is just then just one.\nSpeaker B: So because at the end of the more we add, you know, the more spider webbish it's going to become in the middle and the more of hand editing, it's going to get very ugly.\nSpeaker B: But with this way we can say, okay, these are the discourse phenomena.\nSpeaker B: They may have their own hidden layer that points to some of the real hidden layer or the general hidden layer.\nSpeaker B: And the same we will be able to do for syntactic information, the verbs used, the object types used, modifiers.\nSpeaker B: And maybe there's a hidden layer for that.\nSpeaker B: And so for that we have context.\nSpeaker A: Yeah, so essentially a lot of those nodes can be expanded into little based nets of their own.\nSpeaker A: Yep.\nSpeaker E: Precisely.\nSpeaker C: One thing that's kind of been bugging me when I'm more I look at this is that, I guess the fact that there's a complete separation between the observed features and the output.\nSpeaker A: Yeah.\nSpeaker C: I mean it makes it cleaner but then, I mean, for instance if the discourse does, well for instance the discourse admission fee node seems like it should point directly to the, or increase the probability of enter directly versus going there via tourist.\nSpeaker A: Yeah, or we could add more sort of middle nodes.\nSpeaker A: Like we could add a node, like do they want to enter it which is affected by admission fee and whether it's closed and whether it has a door.\nSpeaker A: So it's like, there are those the two options either like make an arrow directly or.\nSpeaker A: Yeah, that makes sense.\nSpeaker B: And if you do, if you connect it too hard you may get such phenomena back.\nSpeaker B: So how much does it cost to enter?\nSpeaker B: The answer is $250 and then the person says, yeah, I want to see it.\nSpeaker B: Meaning it's way out of my budget.\nSpeaker C: There are places in Germany where it costs $250 to enter.\nSpeaker B: Nothing comes to mind without thinking too hard.\nSpeaker B: Maybe, yeah, of course, opera premieres.\nSpeaker B: Really?\nSpeaker B: So, you know, or any good old people.\nSpeaker C: You want to see the magic fluid or something?\nSpeaker D: Or maybe a famous restaurant or I don't know, various things.\nSpeaker D: The spy goes to the high level.\nSpeaker B: I think that the, I mean, nothing beats the admission charge prices in Japan.\nSpeaker B: So there are $200 is moderate for getting into a discotech.\nSpeaker B: Then again, everything else is free then once you're in the food and drink and so on.\nSpeaker B: But you know, we can, somebody can have discussed the admission fee and the answer is if we still, based on that result, is never going to enter that building.\nSpeaker B: Because it's just too expensive.\nSpeaker C: All right, I think I see.\nSpeaker C: So the discourse refers to admission fee, but it just turns out that they changed our mind in the middle of the discourse.\nSpeaker D: You have to have some notion of not just, I mean, there's change across several terms of discourse.\nSpeaker D: So I don't know how, if any of this was discussed, but how, if at all, this is going to interact with whatever general, other, other discourse processing that might happen.\nSpeaker C: What sort of discourse processing is, how much is built in this Markham?\nSpeaker B: It works like this.\nSpeaker B: The first thing we get is that already the intention is sort of they try to figure out the intention, right?\nSpeaker B: Simply by parsing it.\nSpeaker B: And this won't differentiate between our modes, but at least it'll tell us, okay, here we have something that somebody that wants to go someplace.\nSpeaker B: Now it's up for us to figure out what kind of going there is happening.\nSpeaker B: And if the discourse takes a couple of turns before everything or the information is needed, what happens is, you know, the parser parses it, and then it's handed on to the discourse history, which is one of the most elaborate, elaborate modules.\nSpeaker B: It's actually the whole memory of the entire system that knows who said what was presented.\nSpeaker B: It helps in an effort resolution, and it fills in all the structures that are omitted.\nSpeaker B: So, because you say, okay, how can I get to the castle?\nSpeaker B: How much is it?\nSpeaker B: And yeah, let's do it, and so forth, so even without it, and that for somebody has to make sure that the information we had earlier on is still here, because not every module keeps a memory of everything that happened.\nSpeaker B: So whenever the person is not actually rejecting what happened before, so as in, no, I really don't want to see that movie.\nSpeaker B: I'd rather stay home and watch TV.\nSpeaker B: What movie was selected and what's in it and what tone is going to be sort of added into the representations, every, each dialogue step by the discourse model.\nSpeaker B: And it does some help in the effort resolution, and it also helps in coordinating the gesture screen issue.\nSpeaker B: So, the person pointing to something on the screen, the discourse model actually stores what was presented at what location on the screen.\nSpeaker B: So, it's rather huge thing.\nSpeaker B: We can sort of have a very clear interface, we can query it, whether admission fees were discussed in the last turn, and the turn before that, or, you know, how deep we want to search.\nSpeaker B: We're just a question, how do we want to search, you know.\nSpeaker B: But we should try to keep in mind that, you know, we're doing this sort of for research, so we should find a limit that's reasonable and not go all the way back to Adam and Eve.\nSpeaker B: You know, the person ever discussed admissions fees in his terror life.\nSpeaker B: And the dialogues are pretty, pretty concise in any way.\nSpeaker D: So, one thing that might be helpful, which is implicit in the use of admission fee discussion as a queue for entry, is thinking about the plans that various people might have.\nSpeaker D: Like all the different sort of general schemas that they might be following.\nSpeaker D: Okay, this person is finding out information about this thing in order to go in as a tourist or finding out how to get to this place in order to do business.\nSpeaker D: Because then anything that's a queue for one of the steps would be slight evidence for that overall plan.\nSpeaker D: I don't know, in none, in sort of more traditional AI kinds of plan recognition things, you sort of have, you know, some idea at each turn of doing something.\nSpeaker D: Okay, what plans is consistent with, and then you get some more information, and then you see, here's a sequence that this sort of roughly fits into.\nSpeaker D: It might be useful here too, I don't know how, you know, you'd have to figure out what knowledge representation would work on that.\nSpeaker B: This plan scheme, I mean, there are some of my extremely elaborate, you know, what do you need to buy a ticket?\nSpeaker B: You know, and it's 50 steps, just for buying a ticket, a ticket counter.\nSpeaker B: You know, and maybe that's helpful to look at those.\nSpeaker B: It's amazing what human beings can do.\nSpeaker B: When we talked, we had the example, you know, of you being a person on a ticket counter working at a railway station, and somebody runs up to you with a suitcase in a sense as New York.\nSpeaker B: You say, track seven.\nSpeaker B: And it's because you know that that person actually is following execute a whole plan of going through 150 steps without any information, other than New York.\nSpeaker B: And for everything, for the context.\nSpeaker B: Even though there is probably no train from New York, right?\nSpeaker D: Not direct.\nSpeaker C: Did I probably have to transfer in Chicago?\nSpeaker B: It's possible.\nSpeaker B: No, you probably have to transfer also somewhere else.\nSpeaker B: Is that something?\nSpeaker B: I think Scotie Carvell, it's not possible.\nSpeaker C: One time I saw a report on trains.\nSpeaker C: I think there is a, I don't know, I thought there was a line that went from somewhere, maybe it was Sacramento to Chicago, but there was like a California to Chicago line of some sort.\nSpeaker C: I could be wrong now.\nSpeaker C: It was a while ago.\nSpeaker D: The Transcontinental Railroad, isn't that Ring of Bell?\nSpeaker C: I don't know if it's still there.\nSpeaker C: I might have blown it up.\nSpeaker B: Well, it never went all the way, right?\nSpeaker B: I always had to change trains in Omaha. One track ended there and the other one started five meters away from that.\nSpeaker D: You seem to know where it is in the media.\nSpeaker B: Has anybody ever been on an app track?\nSpeaker D: I have, but not Transcontinental.\nSpeaker C: I'm frightened by Amtrak myself.\nSpeaker C: What?\nSpeaker C: They seem to have a lot of accidents on the Amtrak.\nSpeaker B: The reputation is very bad.\nSpeaker B: It's not like German chains.\nSpeaker B: German chains are really great.\nSpeaker B: I don't know whether it's which ones are safer, statistically.\nSpeaker E: But they're faster.\nNone: Yeah.\nSpeaker A: Much faster.\nSpeaker A: There's much more than, yeah, they're way better.\nSpeaker B: I use Amtrak quite a bit on the East Coast.\nSpeaker B: I was surprised it was actually okay, from Boston, New York, New York, Rhode Island, whatever.\nSpeaker B: That kind of thing.\nSpeaker C: But that's a different issue.\nSpeaker C: That's going to be an interesting transcript.\nSpeaker A: I want to see what it does with the landmarkiness.\nSpeaker D: That's all I say at a few more times.\nSpeaker C: It helps to figure it out.\nSpeaker D: That structure that Robert Jenner on the board is more q-type based.\nSpeaker D: Here's like, we're going to segment off a bit of stuff that comes from discourse.\nSpeaker D: And then some of the things we're talking about here are more, you know, we mentioned maybe if they talk about, I don't know, entering or something, you know, like, they might be more task-based.\nSpeaker D: So I don't know.\nSpeaker D: There's obviously some more than one way of organizing the variables into something.\nSpeaker B: I think that what you guys did is really nicely sketching out different tasks.\nSpeaker B: And maybe some of their conditions.\nSpeaker B: One task is more likely you're in a hurry when you do that kind of doing business and less in a hurry when you're a tourist.\nSpeaker B: Tourists may have never had final destinations, you know, because they are eternally traveling around.\nSpeaker B: So maybe what might happen is that we do get this sort of task-based middle layer and then we get the sub-middle layers that are more key to this.\nSpeaker B: That's heated to this.\nSpeaker B: Might be a nice dichotomy of the world.\nSpeaker B: So I suggest to proceed with this in a sense that maybe throughout this week, the three of us will talk some more about maybe segmenting of different regions.\nSpeaker B: And we make up some toy observable notes.\nSpeaker B: What's the technical term for which?\nSpeaker B: The artificial notes, the evidence of features.\nSpeaker B: Identify four regions, maybe make up some features for each region and a middle layer for those.\nSpeaker B: And then these should then connect somehow to the more planned based deep space.\nSpeaker A: This is just refined some of the problems in all that completely ad hoc.\nSpeaker B: There will be ad hoc for some time.\nSpeaker A: I mean, close to the end we were like, you know, you're like, really ad hoc.\nNone: It was an even distribution.\nSpeaker A: Right, because it was like four things coming in, right, and some of them have like three possibilities and all that.\nSpeaker C: They want to enter but it's closed, it's nighttime, you know, they're adorists and all this weird stuff happens.\nSpeaker C: The possibility of the possibility of the competition is that they're like, confused.\nSpeaker A: I'm here just to rob the museum.\nSpeaker D: In which case you're supposed to alert the authority.\nSpeaker A: I don't think the other thing to do is also to ask around people about other business packages.\nSpeaker A: It's really going to be the meeting tomorrow, you know.\nSpeaker A: Maybe.\nSpeaker E: Oh, sorry, sorry, Wednesday.\nSpeaker D: Maybe we can talk.\nSpeaker D: Who's the most?\nSpeaker C: I've never said I don't have any other.\nSpeaker A: No, but you mentioned the last meeting that someone was going to be talking.\nSpeaker A: I forget who.\nSpeaker B: Oh, isn't Ben?\nSpeaker D: Ben, I think it's Ben actually.\nSpeaker D: Giving this job talk, I think.\nNone: Sorry, just screen the screen.\nSpeaker B: Okay.\nSpeaker B: So that will be one thing we could do.\nSpeaker B: We can start looking at the smart com tables.\nSpeaker B: I actually wanted to show that to you guys now.\nSpeaker B: Do you want to trade?\nSpeaker B: No, I actually made a mistake because it fell asleep.\nSpeaker B: Linux falls asleep on my machine.\nSpeaker B: It doesn't wake up ever.\nSpeaker B: I have to reboot and reboot without a network.\nSpeaker B: I will not be able to start smart com.\nSpeaker A: So we'll do that maybe.\nSpeaker A: But, okay.\nSpeaker A: But once you start smart com, you can be on.\nSpeaker A: You don't have to be on an network anymore.\nSpeaker A: Is that the deal?\nSpeaker C: Yeah.\nSpeaker C: Interesting.\nSpeaker B: What does smart com need a network?\nSpeaker B: It looks up some stuff that is written by the operating system.\nSpeaker B: Only if you get a DHCP request.\nSpeaker B: So my computer does not know its IP address.\nSpeaker B: Unless.\nSpeaker B: Who's up with that?\nSpeaker B: If I don't have an IP address, they can't look up.\nSpeaker B: They don't know who local host is.\nSpeaker B: Always.\nSpeaker B: But it's a simple solution.\nSpeaker B: We can just go downstairs and look at this.\nSpeaker B: But maybe not today.\nSpeaker B: The other thing.\nSpeaker B: I have to report data collection.\nSpeaker B: We interviewed Fay.\nSpeaker B: She's willing to do it.\nSpeaker B: Meaning be the wizard for the data collection.\nSpeaker B: Also maybe transcribe a little bit if she has to.\nSpeaker B: But also recruiting subjects, organizing them and so forth.\nSpeaker B: That looks good.\nSpeaker B: Jerry however suggested that we should have a trial run with her.\nSpeaker B: See whether she can actually do all the spontaneous eloquent and creativeness that we expect of the wizard.\nSpeaker B: And I talked to Liz about this and it looks as if Friday afternoon will be the time when we have a first trial run.\nSpeaker A: Who will be the subject of this trial run?\nSpeaker A: Will there be one of you going to be the subject like are you?\nSpeaker B: Liz also volunteered to be the first subject.\nSpeaker B: Which I think might be better than other guys.\nSpeaker B: If we do need her for the technical stuff then of course one of you has to sort of trump it.\nSpeaker C: I like how we have successfully narrated down.\nSpeaker C: Is one of you going to be the subject?\nSpeaker C: I don't want to jump in.\nSpeaker A: Reference.\nSpeaker A: I haven't done it yet.\nSpeaker A: Well then she great has to be someone who is familiar enough with the data to cause problems with the wizard so you can see if they're good.\nSpeaker D: Oh plant.\nSpeaker A: Someone who can plant difficult.\nSpeaker A: Yeah.\nSpeaker D: I mean that's what we want to check.\nSpeaker D: It's a sort of testing of the wizard rather than something.\nSpeaker D: Is that what it is?\nSpeaker D: Subject.\nSpeaker B: Yes we would like to test the wizard but if we take a subject that is completely unfamiliar with the task.\nSpeaker B: I think that's the setup.\nSpeaker B: We get a more realistic setup.\nSpeaker C: And that's probably a good enough test of having an actively antagonistic.\nSpeaker A: That might be a little unfair.\nSpeaker D: I'm sure if we, you think there's a chance we might need lives for whatever the technical side of things.\nSpeaker D: I'm sure we can get other people around who don't know anything.\nSpeaker D: If we want another subject, I can drag Ben into it or something.\nSpeaker D: Although he might cause problems.\nNone: Okay.\nSpeaker D: Is the experimental setup for the data collection?\nSpeaker D: That on T-shirt.\nSpeaker D: Totally ready.\nSpeaker B: I think determined.\nSpeaker B: Experimental setup.\nSpeaker B: On the technical issue, yes, except we still need a recording device for the wizard.\nSpeaker B: Just a tape recorder that's running in a room.\nSpeaker B: But in terms of specifying the scenario.\nSpeaker B: We've gotten a little further but we wanted to wait until we know who is the wizard and have the wizard partake in the ultimate sort of definition.\nSpeaker B: So if on Friday it turns out that she really likes it and we really like her.\nSpeaker B: Then nothing should stop us from sitting down next week and getting all the details completely figured out.\nSpeaker B: Okay.\nSpeaker D: The ideal task will have whatever, I don't know how much the structure of the evolving base net will affect.\nSpeaker D: We want to be able to collect as much of the variables that are needed for that in the course of the task.\nSpeaker D: We're not all of them but you know.\nSpeaker B: So this tango enter Vista is sort of itself an ad hoc scenario.\nSpeaker B: The basic idea behind the data collection was the following.\nSpeaker B: The data we get from Munich is very command line simple linguistics stuff.\nSpeaker B: Hardly anything complicated, no matter for us whatsoever, not a rich language.\nSpeaker B: So we want to just to collect data to get that that that elicits more that elicits richer language.\nSpeaker B: And we actually did not want to constrain it too much.\nSpeaker B: Just see what people say.\nSpeaker B: We discovered the phenomenon, the phenomena that we want to solve with whatever engine we come up with.\nSpeaker B: So this is a parallel track.\nSpeaker D: So in other words, the state collection is more general.\nSpeaker D: It could be used for not just this kind of phenomenon.\nSpeaker B: It could tell us also maybe something about the difference between people who think they speak to a computer.\nSpeaker B: People who think they speak to human beings and the sort of differences there.\nSpeaker B: So it may get us some more information on the human machine pragmatics that no one knows anything about as of yesterday.\nSpeaker B: And nothing is changed since then.\nSpeaker B: And secondly, now that of course we have sort of started to lick blood with us, especially since Donald can't stop tangoing, we may actually include those intentions.\nSpeaker B: So now I think we should maybe have at least one navigational task with sort of explicit, not explicit, that the person wants to enter.\nSpeaker B: And maybe some task where it's more or less explicit that the person wants to take a picture or see it or something.\nSpeaker B: So that we can label it. That's how we get a copy of it.\nSpeaker B: Whereas if we just get data and we never know what they actually wanted, get no cues.\nSpeaker C: Already.\nSpeaker C: That was the official end of the meeting now.\nSpeaker A: So what's economics the fallacy?\nSpeaker C: I just randomly label things.\nSpeaker C: So that has nothing to do with economics or anything.\nNone: Okay, switching on.\nSpeaker B: I don't know what it is.\nSpeaker C: The secret plans.\nSpeaker C: So what happens is whatever I'm thinking about at the time happens to be the title of whatever I start writing.\nSpeaker C: And then I start writing.\nSpeaker C: But a lot of it is bad.\nSpeaker C: If you want to see what I write, it's on jigskacy.com.\nSpeaker C: I have a web page that I manage with some other people.\nSpeaker C: And I will start out writing a piece, realize it's bad and stop, but never deleted off my desktop.\nSpeaker C: That's a humor website.\nSpeaker C: The finger quotes are in there because it appears that it's only funny to me and two other people in the United States.\nSpeaker C: So how do you spell it?\nSpeaker C: J-I-G-G-S-C-A-S-E-Y.\nSpeaker C: Oh man, we should have put that in the transcript.\nSpeaker E: What was the website again?\nSpeaker B: I don't know.\nSpeaker C: How do you stop?\nSpeaker B: I'll send it again.\nSpeaker C: Oh, I see.\nSpeaker B: Clever.\nSpeaker B: You see this in a single image?\nSpeaker B: Yes, the blue thing here.\nSpeaker D: We'll turn it around.\nSpeaker B: This will come off soon and it's about the same day.\nSpeaker B: Is iti\u00e3o?\nSpeaker D: The red thing is that you have used it, er...\nSpeaker C: I was on the list but I got it indirectly.\nSpeaker D: So, um, definitely.\nSpeaker D: I mean, sometimes when this has happened before, like, you know, it was there in the 85, I think it's really something that I think is really, really, I think, yeah, I think it's really something that I think is really, really good.\nSpeaker D: So, I mean, it's definitely, I think.\nSpeaker B: It's also very cultural, different, you know.\nSpeaker B: I mean, you get the individual difference with some people,\nNone: the issues of, really? Any kind of sense?\nSpeaker B: Um, sensible cultural philosophy, very, really.\nSpeaker D: I think it's culturally appropriate and it's really not in a bit, but, yeah.\nSpeaker D: Do you have a bit?\nSpeaker C: I would volunteer but I don't know what goes into, uh, I've never done a condolence card before.\nSpeaker D: Yeah, I don't know if there's something that, um, should be, certainly, or made available for everyone.\nSpeaker D: Yeah, of the other issues.\nSpeaker D: In the past, I would be, you know, there's a big no card, but that's just, you know, it's not funny.\nSpeaker B: I mean, they can go for two at the same one.\nSpeaker B: Oh, you know, you can go for a channeling?\nSpeaker B: I'm trying to, in terms of exclusive, only to people that are in the Zoom, that the presence of them, whatever, N-T-L, or, you know, whoever we are,\nSpeaker D: I read this by myself. Yeah.\nSpeaker D: Um, yeah.\nSpeaker D: Maybe, um, maybe I'll show you the data, I'll just go ahead and try.\nSpeaker D: Thank you for having me.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker D: Um, I guess I should, I said, I know, there's a thing, it's like, because for N-T-L-S, I mean, if I could say, I'd vote for a channel, but, you know, it's like, oh, yeah.\nSpeaker A: I don't know about mediums, I think,\nSpeaker D: it's, I think. I never can't think you want to have a big one,\nSpeaker E: I don't know. So, um,\nSpeaker C: it's like, if everyone could just sign their name\nSpeaker D: with something, it would be nice. Yeah.\nSpeaker D: That's how it would be, I think, um, I can get one.\nSpeaker D: I don't want one.\nSpeaker B: Okay, one for one thing.\nSpeaker B: That's what he was referring to,\nSpeaker D: by the way,\nSpeaker B: I think he was referring to it,\nSpeaker D: as well, as he was referring to it. So, of course, he was referring to it.\nSpeaker D: Yeah, and his, um, his vote, in fact, I definitely thought I was referring to it, but over the last few years, he's been getting interested in that.\nSpeaker B: And, every time he's been, I'm clear whether, how long it can last, or any position he's been.\nSpeaker B: So, he's been doing better.\nSpeaker B: This is the, this is the, uh, in Spencer, if you notice the matter is going to be, you're in it, and in the second four weeks, then try to keep your Saturday, or something Saturday.\nSpeaker B: I wonder if you're just interested about the date.\nSpeaker B: Yeah, the body, the body, the body, so, um, sometimes, I wonder whether we do something like that, just when I'm 15.\nSpeaker C: That would be acceptable, yes.\nSpeaker A: And we could, just, um, yeah, but, uh,\nSpeaker E: Betty Brothers.\nSpeaker B: Yeah, we, um, we do a lot of, um, I mean, I, if you mention that my, um, X-Men, the heart-cannon, you know, that's not my X-Men Brothers, I won't say this.\nSpeaker D: All right.\nSpeaker D: Did you say A-Clover?\nSpeaker B: Oh, yeah.\nSpeaker B: What can you do when I'm talking about what you do?\nSpeaker B: This is, okay, you kind of think that X-Men,\nNone: you solve this. Now, you do what?\nSpeaker A: You plan it up and you solve it?\nSpeaker A: Oh, okay.\nSpeaker A: Okay.\nSpeaker A: You don't have an X-Men?\nSpeaker B: So, it's a really young X-Men to go, you can solve it.\nSpeaker B: Okay.\nSpeaker B: You know, it's primers, and then, and then, please, I'll be right back.\nSpeaker B: Okay.\nSpeaker B: You know?\nSpeaker B: It's the first thing.\nSpeaker B: The first thing,\nSpeaker A: you know, the whole thing.\nSpeaker A: It's like, you're cubes.\nSpeaker A: You do most of the slides.\nSpeaker C: No, I assume you just slide this in.\nSpeaker B: Like press, okay.\nNone: Like press, press, okay.\nSpeaker C: Like, press, press in terms of, width and length.\nSpeaker C: Yeah.\nSpeaker B: Take your average.\nSpeaker B: So, it's the next one.\nSpeaker B: Okay, I see.\nSpeaker B: Okay.\nSpeaker B: I didn't have.\nSpeaker B: What does it look like?\nSpeaker B: Okay.\nSpeaker C: Okay.\nSpeaker B: And then, you, you add, you add, you press, so some, what's the name of that?\nSpeaker D: Different, different.\nSpeaker D: Extmarine.\nNone: Straight up.\nSpeaker B: Okay.\nSpeaker B: And, where can you pass me, the answer, that would be, some researches that are caring for you.\nSpeaker B: Yeah.\nSpeaker B: And,\nNone: Pendant. Pendant, Pendant.\nSpeaker D: Pendant.\nSpeaker B: Pendant.\nSpeaker B: And some people use that.\nSpeaker D: And what do you think is appropriate?\nSpeaker B: Actually, it's a function of the end result.\nSpeaker B: If it's really dry, spoons are better.\nSpeaker B: If it's not too dry, it's best to put the water in your hands, make them boil and then do the bit.\nSpeaker B: Then you're stored in the pan with all the water.\nSpeaker D: And then you're stored in the stove.\nSpeaker D: Leave.\nSpeaker D: I hope it gives us a big heart.\nSpeaker E: Next weekend.\nSpeaker E: I don't know yet.\nSpeaker B: It's a moment.\nSpeaker D: It is being.\nSpeaker B: You know, but there are some types of mushrooms.\nSpeaker B: I agree.\nSpeaker D: But come on.\nSpeaker D: Mushroom is not a vegetable.\nSpeaker B: Okay, that's actually the mushroom that's not a vegetable.\nSpeaker A: Exactly.\nSpeaker A: You can put it in the mushrooms.\nSpeaker D: So you can put it in the mushrooms, too?\nSpeaker D: Yeah.\nSpeaker B: That's like the big...\nSpeaker D: They're like this big and grow as it's like out there.\nSpeaker B: Yeah.\nSpeaker B: So, some of the other mushrooms.\nSpeaker B: Okay.\nSpeaker D: I've heard that some of the other mushrooms.\nSpeaker B: Why would you say good?\nSpeaker B: You find mushrooms, for example, morsels.\nSpeaker B: You can't even pay a living, pay a box of five bucks for the ranch.\nSpeaker D: There's a really expensive...\nSpeaker D: Yeah.\nSpeaker B: So, if you have to pay for a hundred dollars for maybe a half...\nSpeaker D: That's really expensive.\nSpeaker B: That's really, really expensive.\nSpeaker B: They only grow.\nSpeaker B: You can't cultivate more mushrooms.\nSpeaker B: You can only choose what they can for mushrooms.\nSpeaker B: Like, actually, cultivates?\nSpeaker B: Fine.\nSpeaker B: Exactly.\nSpeaker B: These are really the bad ones.\nSpeaker B: They never sell people that.\nSpeaker B: But the real kind just grows out in the woods and cannot...\nSpeaker D: Real mushrooms don't...\nSpeaker D: They'll want to be a completely different thing.\nSpeaker C: You have to have them with pigs, right?\nSpeaker B: It's a couple of months of trouble.\nSpeaker B: Oh, just troubles.\nSpeaker C: I'm not the biggest mushroom fan.\nSpeaker B: I'm a big...\nSpeaker C: I replace mushroom with green beans.\nSpeaker D: My...\nSpeaker D: Green beans are fine, but...\nSpeaker D: No, green beans are great.\nSpeaker C: Green beans are the best thing in the world.\nSpeaker A: Never heard of this.\nSpeaker A: Anyway, it's a question of...\nSpeaker B: Actually, I think I'm a...\nSpeaker B: I mean, if you think of a green pepper gun...\nSpeaker B: I make some good pepper gun.\nSpeaker B: But the...\nSpeaker B: Is a mushroom and a green thing?\nSpeaker B: Maybe it's available.\nSpeaker C: Is mushroom and a green?\nSpeaker D: Yes.\nSpeaker D: Is mushroom and a green?\nSpeaker D: Yes.\nSpeaker D: Is there anything in the woods?\nSpeaker D: Bart, when I said, sorry, you don't like your time in here.\nSpeaker D: You're like, no, but...\nSpeaker D: Let me have a fun guy.\nSpeaker D: But it's normal because it's a floral thingy.\nSpeaker D: There's no way to make it worse.\nSpeaker D: We've tried...\nSpeaker D: We've been in an hour and a half.\nSpeaker D: Okay, mushrooms, like...\nSpeaker D: Like, Chinese twins.\nSpeaker A: They've been watching all these...\nSpeaker E: Because then you're going to fun guys.\nSpeaker E: Yeah, exactly.\nSpeaker D: Guys, fun guys?\nSpeaker E: Yeah.\nSpeaker D: Anyway, mushroom being Gus walking on a fun guy.\nSpeaker D: Anyway, I'm just a little...\nSpeaker A: Why does he have a few mushrooms?\nSpeaker A: Like, I need to be a coward or something.\nSpeaker D: But it doesn't...\nSpeaker D: Fun guy.\nSpeaker A: Fun guy.\nSpeaker A: His name is Gus anyway.\nSpeaker C: So, what about if the mushrooms on the date...\nSpeaker C: Or no, the mushroom is being described to someone else.\nSpeaker C: And you say, he is of the...\nSpeaker C: Violum fun guy is out.\nSpeaker E: Yeah, I've tried that one.\nSpeaker D: It's that.\nSpeaker D: So what do you say?\nSpeaker D: Violum fun guy is out.\nSpeaker D: Anyway.\nNone: So what about this?\nNone: No.\nNone: Okay.\nSpeaker A: The only similar joke I've heard of is...\nSpeaker A: A rope walks into the bar and the bartender says...\nSpeaker A: Sorry, you're kind of here.\nSpeaker A: So what's the rope?\nSpeaker A: A rope.\nSpeaker A: So then the rope goes out to stand...\nSpeaker A: The middle of the street and gets run over.\nSpeaker A: And then he comes back all torn and stuff.\nSpeaker A: And the bartender is like, hey, aren't you the rope who's here a minute ago?\nSpeaker A: He's like, no, I'm afraid not.\nSpeaker D: He has a time for a rope for a two.\nNone: Oh, yeah.\nNone: Anyway, it doesn't matter.\nSpeaker D: Yeah, I'm afraid not.\nSpeaker D: No, I'm afraid not.\nSpeaker D: He's got a big classic.\nSpeaker D: He classes.\nSpeaker D: That's more like better than a washer.\nSpeaker B: Yeah, it's more like a washer.\nSpeaker B: That's what I'm talking about.\nSpeaker A: That's it.\nSpeaker A: I like that.\nNone: I like that.\nSpeaker A: I like the street.\nSpeaker D: So...\nSpeaker D: So...\nSpeaker D: Is it recording?\nSpeaker B: I turned my knob.\nSpeaker B: You have a department of the barcode?\nSpeaker E: I have a department of the barcode.\nSpeaker E: I've never entered my knob, actually.\nSpeaker D: It's an appropriate.\nSpeaker D: Like, we've gotten rid of the barcode.\nSpeaker D: It looks like...\nSpeaker E: Get one.\nNone: That's it.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\n"}, {"meeting_id": "Bed015", "summary": "The revised semantic specification and construction formalism are more stable than the previous versions. Semantic constraints also come into play. The semantic specification, on the other hand, is split into three levels: \"scenario\" is a list of schemas and bindings between them, which describes the current event in terms of Source-Path-Goal, Container, etc.; \"referent\" is about the entities in the discourse and includes grammatical information and pointers to the ontology; \"discourse segment\" comprises utterance-specific things. Apart from the presentation, JavaBayes can now run through the modified web page of the project.", "dialogue": "Speaker G: what things to talk about.\nSpeaker G: Really?\nSpeaker G: Oh, it was horrible.\nSpeaker G: It was incentive.\nSpeaker G: You recorded it?\nSpeaker A: Hello?\nSpeaker A: Hello?\nSpeaker A: Yeah.\nSpeaker A: Which one?\nSpeaker D: All right, good.\nSpeaker D: OK.\nSpeaker D: OK.\nNone: Did you do something?\nNone: OK.\nSpeaker G: And I guess I'm doing something.\nSpeaker G: So basically, the result of much thinking since the last time we met, but not as much writing, is a sheet that I have a lot of thoughts and justification to comment on.\nSpeaker G: But I'll just pass out as is right now.\nSpeaker G: So here's what's around.\nSpeaker G: And there's two things.\nSpeaker G: So one on one side is a revised, updated semantics\nNone: specification. The other side is revised.\nSpeaker F: This is just one sheet, right?\nSpeaker A: No, just one sheet.\nSpeaker A: OK.\nSpeaker A: So this thing also.\nSpeaker G: And it's very similar to the reverse.\nSpeaker G: If you change it, it's always more.\nSpeaker G: But before I don't think everyone here is seen all of this.\nSpeaker G: So sure, here, begin.\nSpeaker G: As usual, the disclaimers are there.\nSpeaker G: All these things are, it's only slightly worse people than it was before.\nSpeaker G: And after a little bit more discussion, especially that people that I have formerly missed at all in the next release, probably changing that so far.\nSpeaker G: Maybe I will, let's start with number two, actually, on the notation.\nSpeaker G: Because that's thinking possible, and there are more familiar to people.\nSpeaker G: OK.\nSpeaker G: So the top block is just sort of a abstract technique, sort of like those things are the kinds of things that we can have.\nSpeaker G: And certain things that have changed, to have changed back to this, that there's been a little bit of going back and forth.\nSpeaker G: But basically, obviously, all the constructions have some kind of name.\nSpeaker G: I forgot to include that you could have a type included in this line, so it's like, well, there's an example.\nSpeaker G: The text for example, the end, it has a causal construction.\nSpeaker G: So just to show it doesn't have to be a beautiful way to get people to be able to send text as well.\nSpeaker G: These three have various ways of doing certain things.\nSpeaker G: I'll just try to go through them.\nSpeaker G: So they could all have a type at the beginning.\nSpeaker G: And then they say the keyword construction.\nSpeaker C: Oh, I see.\nSpeaker C: So the current syntax is, if there's a type, it's before construct.\nSpeaker C: OK, that's fine.\nSpeaker G: OK.\nSpeaker G: And then it has a block that is constituents.\nSpeaker G: And as usual, I guess all the constructions here, all the examples here have only one type of constituent that is a constructional constituent.\nSpeaker G: I think that's actually going to turn out to be certainly the most common kind.\nSpeaker G: But in general, instead of the word construct, there are clear who might have meaning or form as well.\nSpeaker G: So there's some element that isn't yet constructional, and this is the maps one and meaning.\nSpeaker G: OK.\nSpeaker G: The main change with the constructs, which each of which has a keyword construct, and then some name, and then some type of classification, is that it's often, sometimes the case in the first case here that you know what kind of construction is.\nSpeaker G: So for example, whatever I have here, it's going to be a form of the word throw, or it's going to be a word, you know, happy, or something like that.\nSpeaker G: Some people may be a specific word, or maybe you'll have the type, you'll say, I need a spatial relation for easier.\nSpeaker G: I need a directional decimal right here.\nSpeaker G: You could have an actual type here.\nSpeaker G: Or you could just say, in the same case, that you only know the meaning of type.\nSpeaker G: So a very common example is this one that, you know, in directed motion, the first person to do some nature of being agent of some kind, often a human, right?\nSpeaker G: So if I run down the street, then I, I run as a chief, it's type, I mean, the category is what's there.\nSpeaker G: But the new kind is this one is sort of a pair.\nSpeaker G: And sort of skipping the moths and whatever.\nSpeaker G: The idea is that sometimes there are general constructors that you know that you're going to be equivalent of a noun phrase, or a noun phrase, or something like that.\nSpeaker G: And usually it has a whole consideration that I don't want with it.\nSpeaker G: And then you might know something much more specific, depending on what construction you're going to go about, what meaning, what specific you want.\nSpeaker G: So the example, again, at the bottom of a directed motion, you might need a nominal expression to take the place of, you know, the big,\nNone: the big, the tall, dark man walked into the room. But because of the nature of this particular construction, you know, not just that it's nominal or some kind, but in particular it's some kind of anemone, nominal, like usual, like just as well as a light.\nSpeaker G: So for now, I know we're just having complicated expression.\nSpeaker G: So I don't know if this is textable, but something that gives you a way to go with the structural and the new types.\nSpeaker G: OK, then I don't think the, at least, yeah, none of these examples have anything different for formal constraints, but you've been referred to any of the available elements in scope, which here are the constructs, to say something about the relation.\nSpeaker G: And I think, if you know if you compare the top block and the textual block, we dropped like the little F subscript.\nSpeaker G: F subscript 3, 4, 2, the form piece of the construct.\nSpeaker G: Good.\nSpeaker G: And I think that the general will be unended US.\nSpeaker G: Like if you're giving a formal constraint, then you're referring to the normal pull of that.\nSpeaker G: So by saying, if I just said name one, then that means name one, formal, full time level, which makes sense.\nSpeaker G: There are certain times when we're having a discussion to that, which case you could just communicate here, I think, meaningful for some reason, right?\nSpeaker G: Or actually, as we're often that, only to handle this, one special case of George and Jerry walking to the room in that order.\nSpeaker G: So we have a few funny things where something in the meaning might refer to something in the form.\nSpeaker G: But we're not going to really worry about that from right now.\nSpeaker G: And there are ways we can be more specific if we have to use later on.\nSpeaker G: OK.\nSpeaker G: And so in terms of the relations, as usual in every four, and I should have put in a couple of things in something that isn't an interval relation, but in four, you might also have a value binding.\nSpeaker G: You know, you can say that name one dot number of goals, plural or something like that.\nSpeaker G: There are certain things that are actually value similar to the bindings below.\nSpeaker G: But usually, they're going to be value and the alert, right?\nSpeaker G: OK.\nSpeaker G: And then again, semantic constraints here are just bindings.\nSpeaker G: There was talk of changing the name of that.\nSpeaker G: John, you're not going to be like, fight about that.\nSpeaker G: You like about changing it to semantic effects, which I thought in a little bit to order bias and semantic bindings, which I thought might be too restrictive in case we don't have only binding.\nSpeaker G: And so it was an issue whether constraints, there were some linguists who reacted against constraints saying, oh, it is not used for matching, then it shouldn't be called a constraint.\nSpeaker G: But I think we want to be uncommitted about whether it's used for matching or not, because I think we thought it was some situations where it was useful to use whatever bindings are for actually modified constraints.\nSpeaker C: Well, you definitely want to decouple the formalism from the parsing strategy.\nSpeaker C: So whether or not it's used for matching or only for verification, for sure.\nSpeaker C: I don't know what term we want to use, but we don't want to.\nSpeaker G: There was one time when the context explained why constraints wasn't misleading or it's in.\nSpeaker G: And I think the reason that he gave was similar to the reason why John thought it was like, the record was just an interesting coincidence.\nSpeaker G: But until I was like, OK, well, both of you don't like it.\nSpeaker G: I'm going to change it.\nSpeaker G: But I'm starting to like it again.\nSpeaker G: So I'm going to be like, what?\nSpeaker E: Well, you know what?\nSpeaker E: If then-trace.\nSpeaker E: You're what the then-trace is called?\nSpeaker G: What?\nSpeaker G: Concument?\nSpeaker G: Yeah.\nSpeaker G: Yeah, but it's not if then.\nSpeaker C: Anyway, so the other strategy you guys can consider is when you don't know what word you could put no word, just meaning.\nSpeaker C: OK, and let-\nSpeaker G: Yeah, that's true.\nSpeaker D: So that's why you put semantic constraints up top and meaning bindings done here.\nSpeaker G: Oh, oops, nope.\nSpeaker G: That was a mistake of cutting paste.\nSpeaker G: OK.\nSpeaker G: I'm sorry.\nSpeaker G: That one's an unintentional.\nSpeaker D: So there should be semitic.\nSpeaker G: And unintentional, because it's not Jordan.\nSpeaker G: You're actually-\nSpeaker D: So there should be semitic constraints down at the bottom?\nSpeaker G: Yeah.\nSpeaker G: OK.\nSpeaker G: Well, unless I go with meaning, whatever.\nSpeaker G: Or whatever.\nSpeaker G: I'm going to be good at semantic, but I think the best is of other people's biases.\nSpeaker C: Right.\nSpeaker C: Minor pros.\nSpeaker C: Miner pros.\nSpeaker G: OK.\nSpeaker G: So I think the middle block doesn't really give you any more information next to the top block.\nSpeaker G: And the bottom block similarly, only just illicit- so all it does is illustrate that you can drop the subscripts and that you can drop the- that you can give dual types.\nSpeaker G: One thing I should mention about designates, I think, actually assistant cross-leads as well.\nSpeaker G: So strike out the M subscript on the middle block.\nSpeaker G: So basically now, this little change actually goes along with a big linguistic change, which is that designates isn't only something for this and mandates to worry about now.\nSpeaker G: So we want designates to actually- one of the constituents, which acts like a head and some respect, but is sort of really important for a composition later on.\nSpeaker G: So for instance, if some other construction says, argue of type, is this part of type whatever, the designate tells you which sort of part is the minimum part.\nSpeaker G: So if you have the big red ball, you want to know if this object's where noun, but ball is going to be the designated element of that kind of phrase.\nSpeaker G: There is a slight complication here, which is that when we talk about form, it's useful sometimes to talk about- to talk about there also being a designated object.\nSpeaker G: And we think that there will be the same one.\nSpeaker G: So the ball is the head of the phrase, the big red ball, and the entity, to know if it's by the word ball, is sort of the semantic head in some ways.\nSpeaker C: And there's some cases where the grammar depends on some form property of the head.\nSpeaker C: And this enables you to get that.\nSpeaker C: If I understand your right side, yeah, yeah.\nSpeaker G: And you might be able to say things like if the head has to go last and the head final language, you can refer to the head as a formal head as opposed to the rest of the form having to be at the end of that.\nSpeaker G: Right.\nSpeaker G: So that's a useful thing to get some general structure.\nSpeaker C: OK, so that all looks good.\nSpeaker C: Let me- oh, I don't know.\nSpeaker G: Are you finished?\nSpeaker G: There is a list of things that isn't included, but you can ask questions that might.\nSpeaker C: OK.\nSpeaker C: So if I understand this, aside from construde and all that sort of stuff, the differences are mainly that we've gone to the possibility of having form meaning pairs for a type, or we actually gone back to if we go back far enough.\nSpeaker G: It's a construction meeting.\nSpeaker G: So it's not clear that what now is a construction type\nNone: and the type. So I don't know what a form is.\nSpeaker C: Oh, I see.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: I'm sorry.\nSpeaker C: You're right.\nSpeaker C: Construction type.\nSpeaker C: That's fine.\nSpeaker C: But it-\nSpeaker G: Well, and a previous version of the division certainly allowed you to single out the meaning bit by- so you could say constructs of type whatever does it mean something.\nSpeaker G: Yeah.\nSpeaker G: But that was mostly for reference purposes, just to refer to the mean pool.\nSpeaker G: I don't think that it was often used to give an extra meaning to type constraints on the meeting,\nNone: which is really what we want most of the time I think. Mm-hmm.\nSpeaker G: I don't know if we'll ever have a case where we actually- if there is a form category constraint, you could imagine having a triple there.\nSpeaker C: No, no, no, no.\nSpeaker C: I don't think so.\nSpeaker C: I think that you'll do fine.\nSpeaker C: In fact, these are as long as- as Mark is- these are form constraints.\nSpeaker C: So nominal expression is- the fact that this animate is- man, the fact that it's a nominal expression, I would say, on most people's notion of higher form types, this is one.\nSpeaker C: I think that's just fine.\nSpeaker C: Yeah.\nSpeaker F: Which is fine.\nSpeaker F: Yeah.\nSpeaker F: Now, I mentioned this.\nSpeaker F: I don't know if I ever explained this, but the point of- I mentioned in the last meeting, the point of having something called nominal expression is because it seems like having the verb subcategorized, for, you know, like, say, taking as its object, just some expression which designates an object, or designates a thing, or whatever.\nSpeaker F: That leads to some syntactic problems, basically.\nSpeaker F: So you want to, you know, sort of have this problem like, OK, well, I'll put the word, let's say the word dog.\nSpeaker F: You know, that has to come right after the verb, because we know verb meets its object.\nSpeaker F: And then we have a construction that says, oh, you can have the proceeding a noun.\nSpeaker F: And so you have this sort of problem that the verb has to meet that does it not come in.\nSpeaker F: And you can get, you know, the kicked dog or something like that meaning kicked the dog.\nSpeaker F: So you kind of have to let this phrase idea in there.\nSpeaker C: But I have no problem with what I think it's fine.\nSpeaker G: Right.\nSpeaker G: You may not be like everyone else in Berkeley.\nSpeaker F: Yeah.\nSpeaker F: But that's OK.\nSpeaker F: I mean, we sort of thought we were getting away with, I mean, this is not reverting to the X-bar theory of phrase structure, but I just know that this is, like we didn't originally have in mind that verbs would subcategorize for a particular sort of form.\nSpeaker G: Well, there's no arrangement to this.\nSpeaker G: The question was, did we want directed motion, which is an argument structure construction?\nSpeaker G: Did we want it to worry about anything more than the fact that it has semantic, you know, it's sort of a frame-based construction?\nSpeaker G: So one option that Keith had mentioned also was like, well, if you have more abstract construction, such as subject predicate, basically things like grammatical relations, those could intersect with these in such a way that subject predicate or subject predicate verb object, would require that those things fill subject object are non-expressions.\nSpeaker G: And that would be a little bit cleaner in some way.\nSpeaker G: But for now, I mean.\nSpeaker C: But it's just moving it.\nSpeaker C: Moving it differently.\nSpeaker C: It constraints around.\nSpeaker G: OK, so that's.\nSpeaker G: Basically, the point is there has to be that constraint somewhere.\nSpeaker G: Right.\nSpeaker C: Yeah.\nSpeaker C: And so that was the.\nSpeaker G: Reverb is not having now.\nSpeaker C: OK.\nSpeaker C: OK.\nSpeaker C: And sort of going with that is that the designatim also now is a pair instead of just the meaning.\nSpeaker C: And aside from some terminology, that's basically yet.\nSpeaker C: I'm asking.\nSpeaker G: Yeah.\nSpeaker G: Yeah.\nSpeaker G: The unaddressed questions in this definitely would, for instance, be semantic constraints we talked about.\nSpeaker G: Here are just bindings.\nSpeaker G: But we might want to introduce mental spaces.\nSpeaker G: There's all these things.\nSpeaker C: The mental spacing is clearly not.\nNone: Right.\nSpeaker G: So there's going to be some extra definitely other notation only for that, which we skip for now.\nSpeaker C: By the way, I do want to get on that as soon as rubber gets back.\nSpeaker C: So the mental space thing.\nSpeaker C: Obviously, constroul is a big component of that.\nSpeaker C: So there's probably not worth trying to do anything till it gets back.\nSpeaker C: But sort of as soon as it gets back, I think we ought to.\nSpeaker F: So what's the time frame I forgot again when you're going away for a lot?\nSpeaker E: As a mental bridge, I'm skipping forth of July.\nSpeaker E: OK.\nSpeaker E: Right afterwards.\nSpeaker E: OK.\nSpeaker G: What?\nSpeaker G: You're missing the premiering America.\nSpeaker G: It's a point of spending a year here.\nSpeaker D: Well, he went to college here.\nSpeaker D: Oh, yeah, I forgot.\nSpeaker C: It furthermore, it's well worth missing.\nSpeaker C: Yes.\nSpeaker C: Thank you.\nSpeaker G: I like spending forth of July in other countries.\nSpeaker G: Winner, right?\nSpeaker G: OK.\nSpeaker G: So that's great.\nSpeaker G: Construct, OK, so there was one question that came out.\nSpeaker G: I hate this thing.\nSpeaker G: This is right.\nSpeaker G: So something like past, which we think is a very simple.\nSpeaker G: We've often just stuck in it as a feature.\nSpeaker G: This event takes place before it's each time.\nSpeaker G: OK, it's what this means.\nSpeaker G: It's often thought of as it is also considered a mental space by lots of people around.\nSpeaker G: So there's issue of well, sometimes there are really exotic, explicit space builders that say in France blah, blah, blah.\nSpeaker G: And you have to build up.\nSpeaker G: You would imagine that would require you be very specific about machinery.\nSpeaker G: Whereas past is a very conventionalized one.\nSpeaker G: We sort of know what it means.\nSpeaker G: But we don't necessarily want to unload all the notation every time we see that it's past tense.\nSpeaker G: So we can think of our, just like X-Siem of walk, we first do this kind of thing.\nSpeaker C: I think it's exactly right.\nSpeaker G: Past refers to a certain configuration of this thing with respect to it.\nSpeaker G: So we're kind of like having an arcade.\nSpeaker G: Yeah, in both ways.\nSpeaker G: Right?\nSpeaker C: So I think that we'll have to see how it works out when we do the details.\nSpeaker C: But my intuition would be that that's right.\nSpeaker C: Yeah.\nSpeaker E: OK.\nSpeaker E: You want to do the same for space?\nSpeaker E: Sorry?\nSpeaker E: Space?\nSpeaker G: Space?\nSpeaker E: Here?\nSpeaker E: No?\nSpeaker G: Oh, oh, oh, oh, instead of just time?\nSpeaker G: Yeah, yeah.\nSpeaker G: Yeah, same thing.\nSpeaker G: So there are very conventionalized like dakedic ones, right?\nSpeaker G: And then I think for other spaces that you introduce, you could just attach whatever you could build up and appropriately structure according to them.\nSpeaker E: Now, this basically would involve everything you can imagine to fit under your seat, thought to something.\nSpeaker E: Yeah.\nSpeaker E: Where it's actually dependent on us now, what was passed, what is in the future, versus what is here, what is there, what is.\nSpeaker G: Yeah.\nSpeaker G: So time and space, we'll get that on the other side a little, like very minimally.\nSpeaker G: There's a slot for setting time and setting place.\nSpeaker G: And you know, you can imagine for both of those, there are absolute things you could say about the time and place.\nSpeaker G: And then there are many more interestingly, linguistically, anyway, there are relative things that you relate the event in time and space to where you are now.\nSpeaker G: If there's something a lot more complicated, like hypothetical, whatever, then you have to do your job.\nSpeaker G: Or somebody's job, and then get up like this.\nSpeaker F: Yeah, I mean, I'm curious about how much of the mental, I mean, I'm not sure that the formalism, sort of the grammatical side of things, is going to have that much going on in terms of the mental space stuff.\nSpeaker F: Basically, all of the so-called space builders that are in the sentence are going to sort of, I think of it as sort of giving you the coordinates of, you know, assuming that at any point in discourse there's the possibility that we could be sort of talking about a bunch of different world scenarios, whatever.\nSpeaker F: And the speaker is supposed to be keeping track of those.\nSpeaker F: The construction that you actually get is just going to sort of give you a cue as to which one of those that you've already got going.\nSpeaker F: You're supposed to add structure to.\nSpeaker F: So in France, Watergate wouldn't have heard Nixon or something like that.\nSpeaker F: Well, you say, all right, I'm supposed to add some structure to my model of this hypothetical past France universe or something like that.\nSpeaker F: The information in the sentence tells you that much, but it doesn't tell you exactly what the point of doing so is.\nSpeaker F: So for example, depending on the linguist context, it could be like the question is, for example, what does Watergate refer to there?\nSpeaker F: Does it refer to, if you just hear that sentence cold, the assumption is that when you say Watergate, you're referring to a Watergate-like scandal, as we might imagine it happening in France, but in a different context, oh, you know, if Nixon had apologized right away, I wouldn't, you know, Watergate wouldn't have heard him so badly in the US and in France, it wouldn't have heard him at all.\nSpeaker F: Now that Watergate, we're now talking about the real one and the wood, sort of, it's a sort of different dimension of hypotheticality.\nSpeaker F: I mean, we're not saying what's hypothetical about this world.\nSpeaker F: In the first case, hypothetically, we're imagining that Watergate happened in France.\nSpeaker F: In the second case, we're imagining hypothetically that Nixon had apologized right away or something, right?\nSpeaker F: So a lot of this isn't happening at the grammatical level.\nSpeaker F: And so I don't know where that sits then, sort of the idea of sorting out what the person meant.\nSpeaker G: The grammatical things, such as the acceleraries that introduce these conditions, whatever, give you sort of the most basic, those we, I think we can figure out what the possibilities are, right?\nSpeaker G: They're sort of relatively limited number.\nSpeaker G: And then how they interact with some extra thing like in France, or if such and such, that's like there are certain ways that they can, you know, one is a more specific version of the general pattern that the grammatical grammar gives you.\nSpeaker G: Yeah.\nSpeaker G: But, you know, whatever.\nSpeaker C: Yeah, in the short run, all we need is enough mechanism on the form side to get things going.\nSpeaker F: But the whole point of what Focundian Turner have to say about mental spaces and blending and all that stuff is that you don't really get that much out of the sentence.\nSpeaker F: You know, there's not that much information contained in the sentence that just says here, add the structure to the space.\nSpeaker F: And exactly what that means for the overall ongoing interpretation is quite open.\nSpeaker F: An individual sentence could mean a hundred different things depending on quote what the space configuration is at the time of utterance.\nSpeaker F: And so somebody's going to have to be doing a whole lot of work, but not me, I think.\nSpeaker C: I think that's not, I don't think it's completely right.\nSpeaker C: I mean, in fact, the sentence examples you gave, it did constrain the meaning, the form did constrain the meaning.\nSpeaker C: Yeah.\nSpeaker C: And so it isn't sure, but like what was the point of saying\nSpeaker F: that sentence about Nixon and France? That is not, there's nothing about that in the sentence.\nSpeaker F: That's like it usually don't know the point of it.\nSpeaker G: Yeah.\nSpeaker G: The sentence is all, but we know what it's trying to say.\nSpeaker G: We know what what predication is sending out.\nSpeaker C: But bottom line I agree with you that we're not expecting much out of the purely linguistic cues, right?\nSpeaker C: The purely formed cues.\nSpeaker C: And I mean, you're the linguist, but it seems to me that these, you know, we've talked about maybe half a dozen linguistics these these in the last few minutes or something.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Oh yeah.\nSpeaker C: And that's my feeling that these are really hard problems that decide exactly what's going on.\nSpeaker C: Yeah.\nSpeaker G: OK.\nSpeaker G: OK.\nSpeaker G: One other thing I just want to point out is there's a lot of confusion about the terms like profile, designate, focus, etc.\nSpeaker G: Right, right, right.\nSpeaker G: For now, I'm going to say like profiles often use like two uses that come to mind immediately.\nSpeaker G: One is in the traditional like semantic highlight of one element with respect to everything else.\nSpeaker G: So hypotenuse, you profile this guy against the background of the right triangle.\nSpeaker G: OK.\nSpeaker G: And the second use is in frame met.\nSpeaker G: It's slightly different.\nSpeaker G: And I was asking tons about this.\nSpeaker G: They use it to really mean this in a frame.\nSpeaker G: This is the profile element.\nSpeaker G: These are the ones that are required.\nSpeaker G: So they have to be there or expressed in some way, which, which nothing one and two are mutually exclusive, but they're different meanings.\nSpeaker G: So the closest thing, so I was thinking about how it relates to this notation for us.\nSpeaker G: OK, so how is it?\nSpeaker C: So does it really what they mean in frame met?\nSpeaker C: Yeah, I didn't know that.\nSpeaker G: Yeah.\nSpeaker G: I mean, I was a little bit surprised about it.\nSpeaker G: I knew that I thought that that would be something like, there's another term that I've heard for that.\nSpeaker C: Right.\nSpeaker G: Well, at least Hans does use it that way.\nSpeaker G: Well, I'll check.\nSpeaker G: Anyway, so I think the designate that we have in terms of meaning is really the highlight this thing with respect to everything else.\nSpeaker G: OK, so this is what it means.\nSpeaker G: But the second one seems to be useful, but we might not need a notation for it.\nSpeaker G: We don't have a notation for it, but we might want one.\nSpeaker G: So for example, we've talked about, if you're talking about the electrical item walk, it's an action.\nSpeaker G: Well, it also has this idea.\nSpeaker G: Carries along with it the idea of an actor.\nSpeaker G: Everybody's going to do the walking.\nSpeaker G: Or if you talk about an adjective red, it carries along the idea of the thing that has a property of having color red.\nSpeaker G: So we used to use the notation with for this.\nSpeaker G: And I think that's closest to the second one.\nSpeaker G: So I don't yet know I have no commitment as to whether we need it.\nSpeaker G: It might be, it's kind of thing that parser might want to think about whether we require.\nSpeaker G: These things are like, it's semantically.\nSpeaker C: No, no, no.\nSpeaker C: Critically, they're not required syntactically.\nSpeaker C: Often they're presuppose than all that sort of stuff.\nSpeaker G: Definitely.\nSpeaker G: So in, was a good example.\nSpeaker G: If you walk in, like, well, in what?\nSpeaker G: You know, like, what is that?\nSpeaker G: So it's only semantically is it?\nSpeaker G: It is still required, say, by simulation time, though, to have something.\nSpeaker G: So I meant the idea of that the semantic value is filled in by some, similar.\nSpeaker G: I don't know if that's something we need to say ever as part of the requirement, or the construction, or not, what will again, do for it.\nSpeaker C: I mean, or have it construed, is that the point?\nSpeaker G: Point it, Robert.\nSpeaker G: Have you ever had a views of this point?\nSpeaker E: It's his thesis, right?\nSpeaker C: Anyway, right.\nSpeaker C: Yeah, this is going to be a good example of this a bit of the mess.\nSpeaker C: And we still have emphasis as well, or stress, or whatever.\nSpeaker G: OK, well, we'll get, we have thoughts about those, as well.\nSpeaker G: I would just say, some of this is just like my, you know, by feet, I'm going to say, this is how we use these terms.\nSpeaker G: I don't, you know, there's lots of these stories in the world that people use it.\nSpeaker G: I think that the other terms that are related are like focus and stress.\nSpeaker G: So I think that the way we would like to think about, I think, is focus is something that comes up in, I mean, lots of basic information structure.\nSpeaker G: OK, it's like, it's not.\nSpeaker G: It might be that there's a syntactic device that you use to indicate focus, or that there are things like, you know, I think you've been telling me things toward the end of the sentence, post-verbal, tend to be the focused, focused on the new information.\nSpeaker G: You know, if I, you know, I walked into the room, you tend to think that whatever into the room is sort of like the more focused kind of thing.\nSpeaker G: And when you, you have stress on something that might be, you know, a cue that the stressed element, or for instance, the negated element, is kind of related to information structure.\nSpeaker G: So that's like the new, the sort of like, import whatever of this thing.\nSpeaker G: So I think that's kind of nice to keep focused on being an information structure term.\nSpeaker G: Stress, I think, and then there are different kinds of focus that you can bring to it.\nSpeaker G: So like stress, the stress is kind of a pun on, and you might have like whatever, like accent kind of stress.\nSpeaker G: And that's just a, we want to stinger stress as a form device, you know, like, oh, high volume, or whatever.\nSpeaker G: And to stinger that from its effect, which is, oh, the kind of focus we have is we're emphasizing this value often as opposed to other values, right?\nSpeaker G: So focus carries along a scope.\nSpeaker G: Like if you're going to focus on this thing, and you want to know, it sort of evokes all the other possibilities that it wasn't.\nSpeaker G: So my classic, my now classic example of saying, oh, he did go to the meeting.\nSpeaker G: That was my way of saying, as opposed to, you know, oh, he didn't, or there was a meeting, I think that was the example it was caught on by the link with the immediate, and so the, like, if you said he, you know, there's all these different things that, if you put stress on different part of it, then you're focusing, whatever, on depending.\nSpeaker G: He walked to the meeting as opposed to he ran, or he did walk to the meeting as opposed to he didn't walk.\nSpeaker G: You know, so we need to have a notation for that, which I think that's still in progress, so it's sort of still working it out.\nSpeaker G: But it did, one implication it does have for the other side, which we'll get to in a minute, is that I couldn't think of a good way to say, here are the possible things that you could focus on, because it seems like any entity in any sentence, or any meaning component of anything, you know, all the possible meaning you could have, any of them could be the subject of focus.\nSpeaker G: But I think one thing you can't skimitize is the kind of focus, right?\nSpeaker G: So for instance, you could say it's the tense on this as opposed to the action, okay?\nSpeaker G: Or it's an identity thing, or contrast with other things, or stress as value as opposed to other things.\nSpeaker G: So it is sort of like profile background thing, but I can't think of like the limited set of possible meanings that you would, that you would, highlight, as opposed to other ones.\nSpeaker G: So it has some sort of complications for the later on.\nSpeaker G: I mean, the best thing I come up with is that information structure has a list of focused elements.\nSpeaker G: For instance, one other type that I've mentioned is like query elements, and that's probably relevant for the, like where is, you know, the castle kind of thing, because you might wanna say that location, or certain WH words, bring, you know, sort of automatically, focus, you know, I don't know the identity of this thing kind of way on certain elements.\nSpeaker G: So, okay, anyway, so that's only, there are many more things that are, there's sort of like a little bit on stable other notation, but it's mostly, I think this is, you know, the current form, other things we didn't totally deal with.\nSpeaker G: Well, we have a lot of other stuff that you can have in working on in terms of like, how you deal with an adjective, you know, a nominal expression.\nSpeaker G: I mean, we should put an example of this, and we could do that later.\nSpeaker G: But I think then, inherently, like the general principle still work though, that we can have constructions that have sort of constituent structure in that there is like, you know, for instance, one, you know, they have constituents, right?\nSpeaker G: So you can like nest things when you need to, but they can also overlap in a sort of flatter way.\nSpeaker G: So if you don't have like a lot of grammar extracts, like this might, you know, be a little, okay, but, you know, we have the properties of dependency grammars and some properties of constituents constituent based grammar.\nSpeaker G: So that's, I think that's the main thing we wanted to aim for.\nSpeaker G: And so far, it's worked out, okay.\nSpeaker G: Good.\nSpeaker G: Okay.\nSpeaker E: I can say two things.\nSpeaker E: Yes.\nSpeaker E: Maybe you want to forget stress.\nSpeaker E: As a word?\nSpeaker E: No, as, as, just don't think about it.\nSpeaker E: Sorry.\nSpeaker E: Canonically speaking, you can, if you look at a curve or a sentence, you can find out where it's a certain stress is, and say, hey, that's my focus exponent.\nSpeaker E: Right.\nSpeaker E: It doesn't tell you anything what the focus is.\nSpeaker E: If it's just that thing, or the constituent that it pulls it.\nSpeaker E: Or the whole phrase.\nSpeaker G: You mean to forget about stress the form?\nSpeaker E: Yeah, because as a form cue, not even trained experts can always, well, they can tell you where the focus exponent is sometimes.\nSpeaker E: And that's also mostly true for red speech.\nSpeaker E: And real speech, people may put stress.\nSpeaker E: It's so the context dependent on what was there before, phrase-based breaks, restarts.\nSpeaker E: It's just, it's absurd.\nSpeaker E: It's complicated.\nSpeaker F: Yeah, I mean, I'm sort of inclined to say, let's worry about specifying the information, the structure, focus of the sentence.\nSpeaker F: And then- Ways that you can get it.\nSpeaker F: The phonology component can handle actually assigning an intonation contour to that.\nSpeaker F: I mean, later on, we'll worry about it.\nSpeaker E: I don't really contour to what the focus exactly is.\nSpeaker E: I'll figure out how the, yeah.\nSpeaker E: But if you don't know what your focus is, then your focus, you lost anyways.\nSpeaker E: And the only way of figuring out what that is is by sort of generating all the possible alternatives to each focused element, decide which one in that context makes sense of which one doesn't.\nSpeaker E: And then you left with a couple of three.\nSpeaker E: So, you know, again, that's something that humans can do.\nSpeaker E: But, paracetoscope, anything.\nSpeaker E: So-\nSpeaker G: OK. Well, yeah, I wouldn't have assumed that it's an easy problem in absence of all the, you need all the other information,\nSpeaker E: I guess. But it's pretty easy to put it in the formalism, though.\nSpeaker E: I mean, because you can just say whatever stuff is the container being focused or the entire whatever, both, and so forth.\nSpeaker G: Yeah, exactly.\nSpeaker G: So the sort of effect of it is something we want to keep.\nSpeaker C: But I think, I'm not sure I understand, but here's what I think is going on.\nSpeaker C: That if we do the constructions right, when a particular construction matches, it, the fact that it matches does in fact specify the focus.\nSpeaker G: I'm not sure about that.\nSpeaker C: OK.\nSpeaker G: It certainly constrains the elements.\nSpeaker G: It's the very least.\nSpeaker G: That's certainly true.\nSpeaker G: And depending on the construction, it may or may not.\nSpeaker C: Oh, for sure.\nSpeaker C: Yes, there are, yeah, it's not every, but there are constructions where you explicitly take into account those considerations that you need to take into account in order to decide what is being focused.\nSpeaker E: So we talked about this morning.\nSpeaker E: John is on the bus.\nSpeaker E: Not Nancy.\nSpeaker E: Right.\nSpeaker E: Focus is on John.\nSpeaker E: John is on the bus and not on the train.\nSpeaker E: All right.\nSpeaker E: John is on the bus versus John is on the train.\nSpeaker B: John is on the bus.\nSpeaker E: John is on the bus.\nSpeaker E: Right.\nSpeaker E: All of those.\nSpeaker E: Right.\nSpeaker E: All of these.\nSpeaker E: And will we have, is it all the same constructions?\nSpeaker E: Just with a different one.\nSpeaker G: Yeah, I would say that argument structure, in terms of like the main, sort of, I don't know, the fact that you can get it without any stress and you have some, whatever is predicated anyway, should be the same set of constructions.\nSpeaker G: That's why I was talking about overlapping constructions.\nSpeaker G: So then you have a separate thing that picks out, you know, stress on something relative to everything else.\nSpeaker G: So the question is actually, I'm sorry, go ahead.\nSpeaker G: And that would have to, it might be ambiguous as whether it picks out that element or the phrase or so on that, but it still is limited possibility.\nSpeaker G: So that should, you know, interact with, it should overlap with whatever other construction is.\nSpeaker C: The question is, do we have a way on the other page, when we get to this semantic side of saying what the stressed element was or stressed phrase or something?\nSpeaker G: So that's why I was saying how, since I couldn't think of an easy, like, limited way of doing it, all I can say is that information structure has a focus slot and I think that should be a little inferred.\nSpeaker C: So that's down at the bottom here when we get up.\nSpeaker G: Yeah, and I don't have a, I don't have a great way or great examples, but I think that's something like that is probably going to be more, more what we have to do.\nSpeaker G: Okay.\nSpeaker G: Okay, that was one comment.\nSpeaker E: Yeah, well, once you know what the focus is, the everything else is background.\nSpeaker E: How about topic command, it's the other side of, how about what?\nSpeaker E: topic command?\nSpeaker G: Yeah, so that was the other thing.\nSpeaker G: So I didn't realize before, it's like, oh, I have a piphony that, you know, topic and focus are a contrast set.\nSpeaker G: So topic is, topic focus seems to me like background profile, okay, or landmark trajectory or something like that.\nSpeaker G: There's definitely that kind of thing going on.\nSpeaker G: I don't know whether, I don't have as many great examples of like topic indicating constructions on like focus, right?\nSpeaker G: Topic, it seems to kind of, you know, something that might be an ongoing kind of thing.\nSpeaker F: Japanese has this though, you know, yeah, that's what is, just to mark which thing is the topic, it doesn't always have to be the subject.\nSpeaker G: So again, information structure has a topics lot.\nSpeaker G: I stuck it in thinking that we might use it.\nSpeaker G: I guess I did.\nSpeaker G: Yep, it's there.\nSpeaker G: And one thing that I didn't do consistently is, when we get there, is like indicate what kind of thing fits into every role.\nSpeaker G: I think I have an idea of what should be, but there, you know, so far we've been getting away with like, either a type constraint or, you know, whatever, it'll be a frame, you know, it'll be, it'll be another predication or it'll be, I don't know, some value from something, some variable and scope or something like that.\nSpeaker G: Or a slot chain, please, not a variable.\nSpeaker G: Okay, so that's, should we flip over to the other side, officially then?\nSpeaker G: Okay, side one.\nSpeaker G: Pointing forward to it, yeah.\nSpeaker G: Now, flip back step.\nSpeaker G: Okay, so this doesn't include something which might, may have some effect on it, which is the, discourse situation context record.\nSpeaker G: Right, so I did, I meant just like draw line and like, you know, you also have some tracking of what was going on.\nSpeaker G: And sort of, this is a big spell comment before I, you know, look into the details of this.\nSpeaker G: But for instance, you can imagine instead of having, I changed the name of, I used to be entities, so you can see it's scenario, reference and discourse segment.\nSpeaker G: And scenario is essentially what kind of, what's the basic predication, what event happened?\nSpeaker G: And actually it's just a list of various slots from which you would draw, draw in order to paint your picture, bunch of frames by, and findings, right?\nSpeaker G: And obviously there are other ones that are not included here, general, cultural frames in general, like, other action, you know, specific X-key maverings, okay, whatever.\nSpeaker G: The middle thing used to be entities, because you could imagine it should be like, really a list where here was various information.\nSpeaker G: And this is intended to be grammatically specifiable information about a reference, you know, about some entity that you were going to talk about.\nSpeaker G: So Harry walked into the room, Harry and room, you know, the room, but they'd be represented in this list somehow.\nSpeaker G: And it could also have, for instance, has this category slot, should be either category in more instance, basically, could be a point or terminology, so that everything you know about this could be drawn.\nSpeaker G: But the important things for grammatical purposes are things like number, gender.\nSpeaker G: The ones I included here are slightly arbitrary, but you could imagine that you need to figure out whether, if it's a group, whether some event is happening linear time, linear space, is like, you know, are they doing something serially, or is it like, I'm not sure, this partly came from a Tommi's schema, I'm not sure we'll need all of these actually, but, and then the status I used was like, again, in some languages, you know, like, for instance, in the child language, you might distinguish between different status.\nSpeaker G: So the big, and finally, discourse segment is about, sort of, bejack the information structure, like utterance specific kinds of things.\nSpeaker G: So the common I was going to make about changing entity, the entity's block to reference is that, you can imagine your discourse, so like situation context, you have a set of entities that you're sort of referring to, and you might, that might be your general, I don't know, database of all the things in this discourse that you could refer to, and I changed a reference because I would say, for a particular utterance, you have particular referring expressions in it, and those are the ones that you get information about that you stick in here.\nSpeaker G: For instance, I know it's going to be plural, I know it's going to be feminine, or something like that.\nSpeaker G: And these could actually just point to, you know, the ID in my other list of active entities, right?\nSpeaker G: So there's all this stuff about discourse status, we've talked about, I almost listed discourse status as a slot where you could say it's active, you know, there's this hierarchy, there's a schemingization of, you know, things can be active, or they can be accessible, inaccessible, there's someone that, you know, keep email to us once, do some of us not a while.\nSpeaker G: And the thing is that that, I noticed that that list was sort of discourse dependence, like in this particular set instance, it has been referred to recently, or it hasn't been, or this is something that's left in my world knowledge, but not active.\nSpeaker C: Yeah, there seems to be context properties.\nSpeaker G: Yeah, and for instance, I used to have a location thing there, but actually that's a property of the situation, and again, at certain points, things are located near or far away.\nSpeaker C: Well, this is recursive, because until we do the mental space story, we're not quite sure, which is fine.\nSpeaker C: Yeah, so some of these are.\nSpeaker G: We just don't know yet.\nSpeaker G: So for now, I thought, well, maybe I'll just have in this list the things that are relevant to this particular utterance, right, everything else here is utterant specific.\nSpeaker G: And I left this slot, predications open, because you can have things like the guy I know from school, or you're referring a fashion might be constrained by certain unbounded amounts of predications that you might make.\nSpeaker G: And it's unclear whether, I mean, you could just have in your scenarios, here are some extra few things that are true, right?\nSpeaker G: And then you could just sort of not have this slot here.\nSpeaker G: But it's used for identification purposes.\nSpeaker G: It's a little bit different just saying all these things are true from my other.\nSpeaker F: Right, this guy I know from school came for dinner.\nSpeaker F: Does not mean there is a guy I know him from school and he came over for dinner.\nSpeaker F: It's not the same.\nSpeaker G: Yeah, it's a little bit different, right?\nSpeaker G: So maybe that's like the restrictive non-restrictive.\nSpeaker G: You know, it gets into that kind of thing.\nSpeaker G: But maybe I'm mixing, you know, this is kind of like the final result after parsing the sentence.\nSpeaker G: So you might imagine that the information you passed to, you know, identifying a particular reference would be, oh, it's a guy and it's someone I know from school.\nSpeaker G: So maybe that would be sort of intermediate structure that you would pass into the disc, to whatever a continual engine or whatever discos context to find, you know, either create this reference, in which case it would be created here.\nSpeaker G: And you know, so you can imagine that this might not.\nSpeaker G: So I uncommitted to a couple of these things.\nSpeaker E: I just make it pretty much at least in my mind, this is not too much.\nSpeaker E: So how is gender looter?\nSpeaker C: In reality?\nSpeaker G: It could be in semantically.\nSpeaker G: Semantically, yeah.\nSpeaker G: So a table, you know, I think.\nSpeaker G: It doesn't have a gender or so.\nSpeaker G: It could be that, I mean, maybe not all of these.\nSpeaker G: I mean, I would say that I tried to keep lots of you that were potentially relevant to most things.\nSpeaker E: Just to make sure that everybody that we read the question.\nSpeaker E: OK, that is semantic as opposed to.\nSpeaker E: It has nothing to do with warm.\nSpeaker G: Yeah, that's right.\nSpeaker E: So again, that purification makes sense to have it open before something like an extreme.\nSpeaker E: Yeah, not.\nSpeaker G: OK, so let's see.\nSpeaker G: So maybe having made that big scale, sort of like large scale comment, should I just go through each of these slots, each of these blocks a little bit.\nSpeaker G: Mostly the top one is sort of image schematic.\nSpeaker G: And just a note, which is that when we actually have some of them seem more inherently static, like a container or sort of support dish.\nSpeaker G: And others are a little bit seemingly inherently dynamic, like source path goal, is often thought of that way, or force or something like that.\nSpeaker G: But in actual fact, I think that they're intended to be sort of neutral with respect to that.\nSpeaker G: And different ex-chemists use them in a way that's either static or dynamic.\nSpeaker G: So path, you could just be talking about the path between this and this and container.\nSpeaker G: And then you go in and out all these things.\nSpeaker G: So I think this came up when Ben and I were working with the Spaniards on the other day.\nSpeaker G: Spaniards, this is what we built them, to decide how you want to split up image schematic contributions versus ex-chematic contributions and how you link them up.\nSpeaker G: And I think again, it's going to be something in the ex-chemist that tells you is a static or a system.\nSpeaker G: So we definitely need that sort of a spectral type gives you some of that.\nSpeaker G: Is it a state or is it a change of state or is it an action of some kind?\nSpeaker E: Is there any meaning to where you have sort of parameters behind it and when you don't?\nSpeaker G: Oh, you mean in the slot?\nSpeaker G: No, it's like I was thinking of type constraints.\nSpeaker G: But ex-chemist obviously has been ex-chemist.\nSpeaker G: Agents, I mean, the performer of the ex-chemist, that depends on the ex-chemist.\nSpeaker G: In general, probably, the...\nSpeaker F: So the difference is basically whether you thought it was obvious what the possible filler is for.\nSpeaker G: Basically, as a spectral type probably isn't obvious, but I should have.\nSpeaker G: So I just neglected the sticks and things.\nSpeaker G: Perspective actor, underwear, observer.\nSpeaker G: I think we've often used...\nSpeaker G: Wee!\nSpeaker G: That's that one, right?\nSpeaker G: Exactly.\nSpeaker G: And so one nice thing that we had talked about is this example.\nSpeaker G: If you have a passive construction, then one thing it does is definitely, it is one way for you to specifically take the perspective of the undergoing kind of object.\nSpeaker G: And so then we talked about whether, well, does that specify a topic as well?\nSpeaker G: Well, maybe there are other things.\nSpeaker G: Now that it's subject, it's more like a topic.\nSpeaker G: And now that...\nSpeaker G: Anyway, so I'm going to trail off on that one so it's not that important right now.\nSpeaker C: No, for the moment, we just need the ability to write it down.\nSpeaker C: If somebody figured out what the rules were.\nSpeaker G: Yeah, exactly.\nSpeaker G: Some of these other ones, let's see.\nSpeaker G: So one thing I'm uncertain about is how polarity interacts.\nSpeaker G: So polarity is using...\nSpeaker G: For like action did not take place, for instance.\nSpeaker G: So by default, it'll be like true, I guess.\nSpeaker G: If you're specifying events, it did happen.\nSpeaker G: You could imagine that you skip out this.\nSpeaker G: Leave off this polarity, don't have it here.\nSpeaker G: And then have it part of the speech act in some way, in some negation.\nSpeaker G: But the reason why I left it is, because you might have a change of state, let's say, where some state holds and then some state doesn't hold.\nSpeaker G: And you're just talking, you know, if you're trying to have the nuts bolt of simulation, you need to know that whatever.\nSpeaker G: No, I think this level, which is where you have it.\nSpeaker G: OK, so it's fine where it is.\nSpeaker G: So how do you get it?\nSpeaker C: It may blow off and involve the discourse.\nSpeaker C: But by the time you're simulating, you should know that.\nSpeaker F: So I'm still just really not clear on what I'm looking at, the scenario box.\nSpeaker F: Like, what does that look like for an example?\nSpeaker F: Like, not all of these things are going to be here.\nSpeaker F: This is just basically says...\nSpeaker F: It's a grab bag.\nSpeaker F: Part of what I'm going to hand you is a whole bunch of schemas, image and ex schemas.\nSpeaker F: Here are some examples of the sorts of things you might have in there.\nSpeaker G: So that's exactly what it is.\nSpeaker G: OK.\nSpeaker G: And for a particular instance, which I make an example of that, is that you might have an instance of container and path, let's say, as part of your into definition.\nSpeaker G: So you would eventually have instances filled in with various values for all the different slots.\nSpeaker G: And they're bound up.\nSpeaker G: And yeah, they're bindings.\nSpeaker F: OK.\nSpeaker F: Do you have to say about the binding in your...\nSpeaker F: Is there a slot in here for the tells you how the bindings are done?\nSpeaker C: No, no, no.\nSpeaker C: Let's say I think we're not... another way of quite right yet.\nSpeaker C: So what this is, let's suppose for the moment it's complete.\nSpeaker C: OK.\nSpeaker C: OK.\nSpeaker C: Then this says that when an analysis is finished, the whole analysis is finished, you'll have, as a result, some spec for that utterance and context, which is made up entirely of these things and bindings among them and bindings to ontology items.\nSpeaker C: So that this is the toolkit out of which you can make a semantic specification.\nSpeaker C: So that's A. But D, which is more relevant to your life, is this is also the toolkit that is used in the semantic side of constructions.\nSpeaker C: So this is...\nSpeaker C: And anything you have in the party line, anything you have a semantic side of construction comes from pieces this, ignoring a lot...\nSpeaker C: I mean, in general, you ignore lots of it.\nSpeaker C: But it's got to be pieces of this along with constraints among them so that the goal of the...\nSpeaker C: a sort of that goal has to be the landmark of the interior, the container, or whatever.\nSpeaker C: So those constraints appear in constructions.\nSpeaker C: But pretty much this is the full range of semantic structures available to you.\nSpeaker C: OK.\nSpeaker C: For a cause.\nSpeaker G: Thank you.\nSpeaker G: Anyway, some kind of causal structure.\nSpeaker C: OK, good.\nSpeaker C: Let's mark that.\nSpeaker C: So we need to...\nSpeaker G: I mean, so it gets a little bit funny.\nSpeaker G: These are all so far with these structures, especially from path and on down.\nSpeaker G: These are sort of relatively familiar image schematic kind of slots.\nSpeaker G: Now, with cause, the fillers will actually be the cell strings.\nSpeaker G: So you'll say event one causes meant B.\nSpeaker C: And this again may get...\nSpeaker C: And of course, worlds.\nSpeaker G: Yeah.\nSpeaker G: So these are all implicitly within one role, even though saying that place takes place, whatever, if I said time is passed, that would say set this world somewhere before the world that corresponds to a current speech time.\nSpeaker G: So that's sort of OK.\nSpeaker G: Within the event, it's still one role.\nSpeaker G: Yeah, so cause and other frames that could come in.\nSpeaker G: I mean, unfortunately, you could bring in, say, for instance, a desire or something that I kind of want.\nSpeaker G: And actually, there is right now under discourse segment that's attitude, volition, could fill that.\nSpeaker G: There are a couple things where I'm like, oh, not sure if I want to have there.\nSpeaker G: Or basically, there was a whole list of possible speaker attitudes that, say, told me listed.\nSpeaker G: I'm like, well, I don't know.\nSpeaker G: It was like hope, wish, desire, blah, blah, blah.\nSpeaker G: And so I'm like, well, I feel like if I wanted to have an extra meaning, I don't know if those are grammatically marked in the first place.\nSpeaker G: They're more lexically marked, right?\nSpeaker G: At least in English.\nSpeaker G: So if I wanted to, I would stick in an extra frame, in my meaning saying, so there would be a hierarchical frame that, you know, like Naomi wants such a certain situation.\nSpeaker G: And that situation itself is a state of affairs.\nSpeaker C: Right.\nSpeaker C: So want itself can be.\nSpeaker C: It can be just another frame.\nSpeaker C: Well, basically, it's an action in our situation.\nSpeaker C: In our terminology, one could be an action in what you want is a world.\nSpeaker C: So that's, I mean, it's certainly one way to do it.\nSpeaker C: Yeah, there are other things, causal stuff we absolutely need, the mental space we need, the context we need.\nSpeaker C: So anyway, Keith, so is this comfortable to you that once we have this defined, it is your toolkit for building the semantic part of constructions.\nSpeaker C: And then when we combine constructions semantically, the goal is going to be to fill out more and more of the bindings needed in order to come up with the final one.\nSpeaker C: Yeah.\nSpeaker C: And that's the, I mean, according to the party line, that's the whole story.\nSpeaker F: Yeah.\nSpeaker F: Right.\nSpeaker F: That makes sense.\nSpeaker F: So I mean, there's this stuff in the, up in the scenario, which just tells you how various what's gave us you're using in our Heather bound together.\nSpeaker F: And I guess that some of the discourse segment stuff, is that where you would, I mean, that's, OK, that's where the information structure is, which sort of is a kind of profiling on different parts of this.\nSpeaker F: I mean, what's interesting is that the information structure is stuff.\nSpeaker F: There's almost, I mean, we keep coming back to how focus is like this, this trajectory landmark thing.\nSpeaker F: So if I say, in France, it's like this.\nSpeaker F: Great.\nSpeaker F: We've learned something about France, but the fact is that utterances of that sort are generally used to help you draw conclusion also about some implicit contrast.\nSpeaker F: Like in France, it's like this.\nSpeaker F: And therefore, you're supposed to say, boy, like sure, you know, in France, kids are allowed to drink at age three.\nSpeaker F: And that's not just the fact about France.\nSpeaker F: You also can do something about how boring it is here in the US.\nSpeaker F: Right.\nSpeaker F: And so.\nSpeaker G: So I would prefer not to worry about that for right now.\nSpeaker G: OK.\nSpeaker G: And to think that there are, that comes in.\nSpeaker G: This first level constructions, in a sense, topic, focus constructions that would say, oh, when you focus something, then just in the same way, just actually, in the same way, it's the lower level.\nSpeaker G: If you stressed, you know, John went to the bar, whatever.\nSpeaker G: You're focusing that and it's possible in France is in contrast to other things.\nSpeaker G: So similarly, for a whole sentence, in France, such and such happens.\nSpeaker G: So the whole thing is sort of like, again, implicitly, as opposed to other things that are possible.\nSpeaker E: So just look at the same formal maths route.\nSpeaker E: It's nice to just pick any paper on alternatively.\nSpeaker E: OK.\nSpeaker E: So that's the best way of talking about focus.\nSpeaker E: I think it's the way.\nSpeaker F: OK.\nSpeaker F: What was the name?\nSpeaker E: M-A-T-S.\nSpeaker E: OK.\nSpeaker E: Rooft.\nSpeaker E: T2O.\nSpeaker E: OK.\nSpeaker E: T-H.\nSpeaker E: I never know what you're going to know, so it's not.\nSpeaker E: Because it's sweet.\nSpeaker E: E-Dutch.\nSpeaker E: Dutch?\nSpeaker E: Yeah.\nSpeaker E: And very confused by the way.\nNone: So.\nSpeaker F: And maths route.\nSpeaker E: It's had enough.\nSpeaker E: You also just left the IMS in Stuttgart.\nSpeaker E: So it's not there anymore.\nSpeaker E: But I don't know where you said now.\nSpeaker E: But alternative semantics.\nSpeaker E: So you type that into a browser search list.\nSpeaker E: OK.\nSpeaker E: And you get that.\nSpeaker F: OK.\nSpeaker F: Thanks.\nSpeaker E: And what I'm kind of confused about is this.\nSpeaker E: What's the speaker and the hearers?\nSpeaker E: This is what I'm doing here.\nSpeaker G: So for a particular segment, it's really just a reference to some other entity, again, in the situation.\nSpeaker G: So for a particular segment, the speaker might be you or might be me.\nSpeaker G: Here is a little bit harder.\nSpeaker G: It could be like multiple people.\nSpeaker G: I guess that that's not very clear for me.\nSpeaker G: I mean, that's not clear.\nSpeaker E: Don't we ultimately want to handle that analogously in the way we handle time and space?\nSpeaker E: Because you, me, he, they, these guys, all these expressions are much, say, we're contextually dependent as here and now and there.\nSpeaker C: You know, this is assuming you've already solved that.\nSpeaker C: So it's Fred and Mary.\nSpeaker C: So the speaker would be Fred and the.\nSpeaker G: Right.\nSpeaker G: So the constructions, of course, will refer using pronouns or whatever.\nSpeaker G: In which case, they have to check to see who the speaker in here would in order to resolve those.\nSpeaker G: But when you actually say that he walked into whatever, he will refer to a particular.\nSpeaker G: You will already figure out who he or you or I, maybe a better example, who I refers to.\nSpeaker G: And then you just be able to refer to Harry, in wherever that person, whatever role that person was playing in the event.\nSpeaker G: That's up at the reference.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: And don't there in this speaker here apart?\nSpeaker G: So that's, I think that's just, for instance, speaker is known from the situation.\nSpeaker G: Right.\nSpeaker G: When you hear something, you're told that the speaker is.\nSpeaker G: I mean, you know who the speaker is.\nSpeaker G: In fact, that's kind of constraining how, in some ways, you know this before you get to the, you feel in all the rest of it.\nSpeaker G: I think.\nSpeaker G: All right.\nSpeaker G: How else would you?\nSpeaker G: Very.\nSpeaker E: Yeah.\nSpeaker E: Well, I hear.\nSpeaker E: In the English, it's about to say, I.\nSpeaker E: Yeah.\nSpeaker E: I'm on the 25% level.\nSpeaker E: I mean, wouldn't the eye then set up the reference?\nSpeaker E: It happens to be the speaker of the type of number.\nSpeaker E: Right.\nSpeaker E: They, whoever they are, you.\nSpeaker E: Much like the you.\nSpeaker G: So, OK.\nSpeaker G: So I would say, under reference would be something that corresponds to eye.\nSpeaker G: And maybe each reference should probably have a list of whatever the way it was referred to.\nSpeaker G: So that's eye, but it should be say, it refers to what?\nSpeaker G: If it were Harry, it would refer to like some ontology thing.\nSpeaker G: If it were, if it's eye, it would refer to the current speaker, OK, which is given to be like, you know, whatever it is.\nSpeaker E: Not always.\nSpeaker E: And then he said, I.\nSpeaker G: Within the current world.\nSpeaker G: Yeah.\nSpeaker C: Yeah.\nSpeaker C: So, again, this is going to get us into the mental space stuff.\nSpeaker C: And because Fred said that Mary said in whatever.\nSpeaker C: And so we're going to have to chain those as well.\nSpeaker G: So this entire thing is inside a world.\nSpeaker G: Not just like the top part.\nSpeaker C: I think it's a bit of a problem.\nSpeaker C: Except it's trickier than that because the reference, for example, is where it gets really tricky.\nNone: Yeah.\nSpeaker C: Is there some things, and this is where blends and all terapathy, you know, some things which really are meant to be identified in some things which aren't.\nSpeaker C: And again, all we need for the moment is some way to say that.\nSpeaker G: So I thought I'd have like for each referendum having the list of other things, which is identified.\nSpeaker G: You know, which you could do that.\nSpeaker G: You, for instance.\nSpeaker G: So I guess it sort of depends on if it is referring to the fits identifiable already or if it's a new thing.\nSpeaker G: If it's a new thing, you have to create a structure or whatever.\nSpeaker G: If it's old thing, it could be referring to usually something in a situation or something in ontology.\nSpeaker G: So there's a, you know, whatever.\nSpeaker C: I just had an idea that would be very nice if it works.\nSpeaker C: Or what?\nSpeaker C: It works.\nSpeaker C: I even told you what it is.\nSpeaker C: It just was mine.\nSpeaker C: Build up.\nSpeaker C: An idea that would be nice.\nSpeaker C: We're crossing our fingers.\nSpeaker D: If it works.\nSpeaker D: The billion mental space.\nSpeaker D: Good.\nSpeaker D: Right.\nSpeaker D: It was a space builder.\nSpeaker C: We might be able to handle context in the same way that we handle mental spaces.\nSpeaker C: Because you have somewhat the same things going on of things being accessible or not.\nSpeaker C: And so I think if we did it right, we might be able to get at least a lot of the same structure.\nSpeaker C: So pulling something out of a discourse context is, I think, similar to other kinds of mental space.\nSpeaker C: Well, I've never seen anybody write that up.\nSpeaker C: And maybe they did.\nSpeaker C: I don't know.\nSpeaker C: Maybe all of the literature.\nSpeaker F: So there's things like, there's all kinds of stuff like in, I think I mentioned last time, in check if you have a verb of saying, then you say something like, or of thinking you can say something like, oh, I thought you are a Republican or something like that.\nSpeaker F: Whereas in English, you would say, I thought you were.\nSpeaker F: Sort of the past tense being copied onto the lower verb doesn't happen there.\nSpeaker F: So you have to say something about tense is determined relative to current blah, blah, blah.\nSpeaker F: The same thing happens with pronouns.\nSpeaker F: There's languages where if you have a verb of saying, then OK, so a situation like Bob said he was going to the movies where that lower subject is the same as the person who is saying or thinking.\nSpeaker F: You're actually required to have I there.\nSpeaker F: And it's sort of an extended approach.\nSpeaker F: Yeah.\nSpeaker F: But it's not perceived as a quotative construction.\nSpeaker F: I mean, it's been analyzed by the formalist as being a logophoric pronoun, which means a pronoun which refers back to the person who is speaking, or that sort of thing.\nSpeaker F: But that happens to sound like the word for I, but is actually semantically unrelated to it.\nSpeaker F: Oh, good.\nSpeaker F: I love the formalist.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker F: Kidding.\nSpeaker F: There's a whole book which basically operates on this assumption, Mary Del Rinpoel, this book, 93 book on pronouns stuff.\nSpeaker F: Well, yeah.\nSpeaker F: And then same thing for ASL, where you're signing.\nSpeaker F: And someone says something.\nSpeaker F: And then so he say, and then you sort of do a role shift.\nSpeaker F: And then you sign, I, this, that, and the other.\nSpeaker F: I did this.\nSpeaker F: That's also been analyzed as logophoric and having nothing to do with I.\nSpeaker F: And the role shift thing has completely left out and so on.\nSpeaker F: I mean, the point is that pronoun reference is sort of ties in with all this mental space stuff and so on and so forth.\nSpeaker C: So yeah.\nSpeaker C: So that does sound like consistent with what we're saying.\nSpeaker C: Yeah.\nSpeaker G: OK.\nSpeaker G: So it's kind of like the unspecified mental spaces just are current context.\nSpeaker G: And then when you embed them, sometimes you have to pop up to the, you know, depending on the construction or whatever.\nSpeaker G: Your scope might re-extend out to the base one.\nSpeaker G: It would be nice to actually use the same mechanism.\nSpeaker G: Since there are so many cases where you actually need, it'll be one or the other.\nSpeaker G: It's like, oh, actually it's the same.\nSpeaker F: OK.\nSpeaker F: So this is where it's like what's happening.\nSpeaker F: Yeah.\nSpeaker F: What's happening there is that you're moving the base space, something like that, right?\nSpeaker F: So that's how we're going to talk about it.\nSpeaker F: And it happens under different circumstances and different languages.\nSpeaker F: So things like pronoun reference and tense, which we're thinking of as being these discoursey things, actually, are relative to a base space, which can change.\nSpeaker F: And we need all the same machinery.\nSpeaker F: OK.\nNone: OK.\nSpeaker C: Well, this is very good, actually, because it's an extent that it works.\nSpeaker C: Ties together several of these things.\nSpeaker E: And I'm sure going to read the transcript of this one.\nSpeaker E: But the, let's comment that we don't have a camera.\nSpeaker E: You know, all the partings.\nSpeaker E: Oh, yeah.\nSpeaker D: Yeah.\nSpeaker D: Every time Nancy giggles it means that it's your job.\nSpeaker E: Listen, I mean, I was sort of dubious why he even introduces this sort of reality as your basic space in Androidism.\nSpeaker E: Doesn't start with something because it's so obvious.\nSpeaker E: It should be so obvious.\nSpeaker E: It's to me that whenever I say something, I could grab this that with, I think.\nSpeaker E: Yeah.\nSpeaker E: So there should be no pedagogical difference between you.\nSpeaker E: Yeah.\nSpeaker E: And all the others.\nSpeaker C: But there's a grisly thing going on there that when you say, I think you're actually hedging.\nSpeaker C: Yeah.\nSpeaker G: I mean, I don't totally think.\nSpeaker G: Yeah.\nSpeaker G: Yeah, I usually think.\nSpeaker F: Yeah, it's an evidential.\nSpeaker F: It's sort of semi-grammaticalized.\nSpeaker F: People have talked about it this way.\nSpeaker F: And you can do sort of special things.\nSpeaker F: You can put just the phrase, I think, as a parenthetical in the middle of sense and so on.\nSpeaker F: I actually wanted the child on the researchers who\nSpeaker G: worked with Homosolo study a bunch of these constructions. And it was like, it's not using any kind of interesting embedded ways just to mark uncertainty or something like that.\nSpeaker E: So.\nSpeaker E: About linguistic hedges, I mean, those tend to be quite the anyways because we don't have that in here either doing.\nSpeaker C: Hedges?\nSpeaker G: Yeah.\nSpeaker G: There used to be a slot for speaker.\nSpeaker G: There's something like activity.\nSpeaker G: I couldn't remember what it meant.\nSpeaker G: So it took me.\nSpeaker F: Yeah.\nSpeaker F: Well, we were just talking about this sort of evidentiality and stuff like that, right?\nSpeaker F: I mean, that's what I think is sort of telling you.\nSpeaker F: So we probably should have the scent reality or should give this or, you know, confidence or something.\nSpeaker F: Yeah.\nSpeaker F: And the fact that I'm, you know, the fact maybe if I think it versus key things that might, you know, depend on how much you trust the two of us or whatever.\nSpeaker E: You know, you're right.\nSpeaker E: We're in the English language, but about what the study help people use it.\nSpeaker E: What's the word about?\nSpeaker E: About.\nSpeaker E: It's about.\nSpeaker C: Oh, that use of about that.\nSpeaker C: I think I think if you want to spend the plus and six or seven hours, you can get George started on that.\nSpeaker F: You wrote a paper about 35 years ago on that one.\nSpeaker D: I read that paper.\nSpeaker D: The head just paper.\nSpeaker D: Yeah.\nSpeaker D: I read some of that paper, actually.\nSpeaker F: Would you believe that paper led directly to the development of anti-lock breaks?\nSpeaker F: No.\nSpeaker F: Ask me about it later.\nSpeaker F: I'll tell you how.\nSpeaker F: When we're not on time.\nSpeaker F: I'd love to know.\nSpeaker D: Oh, man.\nSpeaker G: So I think someone had raised like sarcasm.\nSpeaker C: It was all that stuff.\nSpeaker C: I don't think.\nSpeaker G: I think we just won't deal with sarcasm.\nSpeaker F: I don't really know why.\nSpeaker F: We don't have to care too much about the speaker attitude, right?\nSpeaker F: Like there's not so many different.\nSpeaker G: I know.\nSpeaker G: Well, there are international markers, I think,\nSpeaker F: that was part. I don't know.\nSpeaker F: I just mean.\nSpeaker G: So much about the grammatical.\nSpeaker F: There's lots of different attitudes that the speaker could have, and then we can clearly identify and tone and so forth.\nSpeaker F: But what are the distinctions among those that we actually care about for our current purposes?\nSpeaker C: Right.\nSpeaker C: So this was the first question of what are our current purposes.\nSpeaker F: Oh, shoot.\nSpeaker F: Here it is 315 already.\nSpeaker C: So I don't know the answer, but it does seem that this is coming along.\nSpeaker C: I think it's as far as I can tell, there's one major thing we have to do, which is the mental whole, spindle space thing.\nSpeaker C: And then there's some other minor things.\nSpeaker C: And we're going to have to sort of bound the complexity.\nSpeaker C: I mean, if you get everything that anybody ever thought about, you know, we're going to have to.\nSpeaker C: So we had started with the idea that the actual constraint was related to this tourist domain.\nSpeaker C: And the kinds of interactions that might occur in the tourist domain, assuming that people were being helpful and weren't trying to do.\nSpeaker C: You know, there's all sorts of cognitive irony.\nSpeaker C: And so like what you mean, isn't probably as much use in dealing with a tourist guide.\nNone: Lockery.\nSpeaker C: So no end of things that we don't deal with.\nSpeaker C: And go ahead.\nSpeaker E: This is not easy, though, because in terms of the sum figure, which is mean you put one more set of brackets around it, just to sort of negate whatever.\nSpeaker E: Yeah.\nSpeaker E: I'm going to have to see if I can go around your.\nSpeaker G: Right.\nSpeaker G: Maybe.\nSpeaker G: Yeah.\nSpeaker G: In model theory, semantic is always like speaker, believe it's not P. You know, like speakers is P.\nSpeaker F: We have a theoretical model of sarcasm now.\nSpeaker F: Right.\nSpeaker F: Now.\nSpeaker C: Anyway.\nSpeaker C: So I guess let me make a proposal on how to proceed on that, which is that it was Keith's sort of job over the summer to come up with this set of constructions.\nSpeaker C: And the most interesting Keith is that you over the next couple of weeks.\nSpeaker C: Don't try to do them in detail or formerly, but just try to describe which ones you think we ought to have.\nSpeaker C: And then when Robert gets back, we'll look at this set of them.\nSpeaker C: Just sort of define your space.\nSpeaker C: Yeah.\nSpeaker C: OK.\nSpeaker C: And so these are the set of things that I think we ought to deal with.\nSpeaker C: And then we'll go back over it.\nSpeaker C: OK.\nSpeaker C: We'll give feedback.\nSpeaker C: And then we'll have a, at least, initial speck of what we're actually trying to do.\nSpeaker C: And that'll also be useful for anybody who's trying to write a parser.\nSpeaker C: Kisses in your mouth.\nSpeaker C: Right.\nSpeaker C: Who might want it, et cetera.\nSpeaker C: So you know, we get this four-tooled fixed.\nSpeaker C: And then we have an idea of the initial range.\nSpeaker C: And then of course, Nancy, you're going to have to do your set of, but you don't have to do that anyway.\nSpeaker C: So we're going to get, they're basically dealing with two domains, the first domain in the child language learning.\nSpeaker C: And we'll see what we need for those two.\nSpeaker C: And then my proposal would be to not totally cut off more general discussion, but to focus really detailed work on the subset of things that we really want to get done.\nSpeaker C: And then as a kind of separate thread, think about the more general things and all that.\nSpeaker E: Well, I also think the detailed discussion would bring us to problems that are all for general information.\nSpeaker C: Without doubt.\nSpeaker C: Yeah.\nSpeaker C: But what I want to do is to constrain the things that we really feel responsible for.\nSpeaker C: So that we said, these are the things we're really going to try to do by the end of the summer.\nSpeaker C: And other things we'll put on a list of research problems or something, because you can easily get the point where nothing gets done because every time you start to do something, you say, oh, yeah, but what about this case?\nSpeaker C: This is called being a linguist.\nNone: Yeah.\nSpeaker G: And that's what I want to do.\nSpeaker G: For me.\nSpeaker D: As I quote in Dr. Schumer, where some guy goes, every time I fire a linguist, the performance other recognize her goes up.\nSpeaker C: Exactly.\nSpeaker C: But anyway, is that make sense as a general way to proceed?\nSpeaker F: Yeah, we'll start with that just figuring out what needs to be done.\nSpeaker F: Then actually, the next step is to start trying to do it.\nSpeaker F: Exactly right, right?\nSpeaker F: OK.\nSpeaker E: We have a little bit of news with this minor stuff.\nSpeaker D: Who can I add out of power?\nSpeaker D: Can I ask a quick question about this?\nSpeaker D: Is this, was it intentional to leave off things like inherits and?\nSpeaker D: Oops.\nSpeaker F: No.\nSpeaker G: No, really?\nSpeaker G: Just on the construction, right?\nSpeaker D: Yeah.\nSpeaker D: Like, construction's going to hurt from other things in my right.\nSpeaker G: Yeah.\nSpeaker G: I didn't want to think too much about that for now.\nSpeaker G: So maybe it was subconsciously intentional.\nSpeaker F: Yeah.\nSpeaker F: Yeah, there should be.\nSpeaker F: I wanted to find out someday if there was going to be some way of dealing with, if this is the right term, multiple inheritance, where one corporation is inheriting from both parents or different ones, or three or four different ones.\nSpeaker F: Yeah.\nSpeaker F: The problem is that then you have to reach of how they're getting bound together.\nSpeaker G: Yeah.\nSpeaker G: And there are certainly cases like that, even, just semantics chemos we have, some examples.\nSpeaker G: Right.\nSpeaker G: And we've been talking a little bit about that.\nSpeaker C: So what I would like to do is separate that problem out.\nSpeaker C: So my argument is there's nothing you can do with that that you can't do by just having more construction.\nSpeaker C: It's like we're in, doesn't have the deep linguistic insights and stuff.\nSpeaker C: Ah, whatever.\nSpeaker C: Yeah, no, no, no, no, by all means.\nSpeaker F: Right.\nSpeaker C: And so what I'd like to do is in the short run, focus on getting it right.\nSpeaker C: And when we think we have it right, then saying, ah-ha, can we make it more elegant?\nSpeaker C: Yeah.\nSpeaker C: Can we connect it?\nSpeaker C: What are the generalizations and stuff?\nSpeaker C: But rather than trying to guess a inheritance structure and all that sort of stuff before we know what we're doing.\nSpeaker C: So I would say in the short run, we're not going to, first of all, not doing them yet at all.\nSpeaker C: And it could be that halfway through we say, ah-ha, we now see how we want to clean it up.\nSpeaker C: And inheritance is only one, I mean, that's one way to organize it, but there are others.\nSpeaker C: And it may or may not be the best way.\nSpeaker C: I'm sorry, you had news.\nSpeaker E: Oh, just small stuff.\nSpeaker E: Thanks to Eva.\nSpeaker E: On our website, we can know if you want to run Java based, you can see, download these classes.\nSpeaker E: And then it will enable you.\nSpeaker E: She modified the query so it has no button menu item or saving it into the embedded Java based.\nSpeaker E: So that's wonderful.\nSpeaker E: Great.\nSpeaker E: And you tested it out.\nSpeaker E: Do you want to say something about that?\nSpeaker A: I was just checking like, when we want to get the posterior probability of variables, you know how you ask whether we can just observe all the variables in the same way so you can't have to make separate queries every time.\nSpeaker B: So yeah.\nSpeaker B: What's the purpose?\nSpeaker A: For the time being, it's very important.\nSpeaker A: You just have a long list of all the variables.\nSpeaker G: All the things you want to query.\nSpeaker G: Yeah, you just have to pass it to them.\nSpeaker G: Yeah.\nSpeaker E: Other, maybe in a long term, that's good news because of course, that's the thinking.\nSpeaker E: I don't know how we can do that at all.\nSpeaker E: That's a pretty good discussion for it.\nSpeaker E: And they're running late, so I had an idea yesterday, but I don't know what the machine can start.\nSpeaker C: What do you mean, sure, tell us what it is.\nSpeaker E: The construal that has been pointed to, but this may be made precise by any means, may work as follows.\nSpeaker E: I thought that we would, the problem I think would be a little bit nice, and I've no clue whether it's work or nothing so it's just a pattern.\nSpeaker E: A couple of matters, this is a famous here.\nSpeaker E: Imagine you write a base net.\nNone: Base?\nSpeaker E: Base netting.\nSpeaker E: Completely from scratch.\nSpeaker E: Every time you do a construal.\nSpeaker E: So you have nothing.\nSpeaker E: This is quite a piece of paper.\nSpeaker E: You consult your ontology, which will tell you a bunch of stuff and parts of the product.\nSpeaker E: Grab out the things that you need.\nSpeaker E: Right.\nSpeaker E: Then you simply write a piece into your real white piece of paper.\nSpeaker E: And you will get a lot of notes and stuff out of there.\nSpeaker E: You won't really get any CPTs.\nSpeaker E: Therefore, we need everything that triggers to what the situation is.\nSpeaker E: I need the context in this stuff.\nSpeaker E: So you get whatever comes from discourse.\nSpeaker E: I've also filtered.\nSpeaker E: So only the ontology relevance stuff from the discourse does the situation in the user model.\nSpeaker E: And that fills in your CPTs with which you can then query the net that you just wrote and find out how the X is construed as in utterance, U.\nSpeaker E: And the embedded Java base works exactly like that.\nSpeaker E: But once you have the precise form it in which to write it, so we write it down, you query.\nSpeaker E: You get the result, you get thrown away.\nSpeaker E: And the nice thing about this idea is that you don't ever have to sit down and think about it or write about it.\nSpeaker E: You may have some general rules as to how things will be equals to this.\nSpeaker E: What so that will allow you to craft the initial notes.\nSpeaker E: But in that respect, it's completely scalable, because it doesn't have any prior configuration.\nSpeaker E: It's just you need an ontology of the domain, and you need the context to pen modules.\nSpeaker E: And if this can be made to work at all,\nSpeaker C: let me kind of thank you. It sounds to me like you want PRM's.\nSpeaker C: PRM's?\nSpeaker E: PRM?\nSpeaker E: I think it's a great forward thing.\nSpeaker C: No, no, you can't.\nSpeaker C: See, the critical thing about the PRM is it gives these relations in general form.\nSpeaker C: So once you have instantiated the PRM with the instances, and then you can unfold it.\nSpeaker E: No, I was using a generic.\nSpeaker E: So probabilistic whatever additional models, whatever you write at it.\nSpeaker C: What matters a lot?\nSpeaker C: Because what you want are these generalized rules about the way things relate that you then instantiate in each case.\nSpeaker C: Yeah.\nSpeaker E: Maybe the way it works.\nSpeaker C: Yeah, that's the only way it could work.\nSpeaker C: We have our local expert on PRM's.\nSpeaker C: My guess is that they're not currently good enough to do that.\nSpeaker C: But we will have to see.\nSpeaker C: Yeah, that would be a good thing to try.\nSpeaker C: It's related to the Hobbes abduction story.\nSpeaker C: In that, you throw everything into a pot and you try to come up with the.\nSpeaker C: That's the question.\nSpeaker C: There's no theorem.\nSpeaker C: No, there isn't a theorem proofer.\nSpeaker C: But the PRMs are like rules of inference.\nSpeaker C: And you're coupling a bunch of them together.\nSpeaker C: And then instead of proving you're trying to compute the most likely.\nSpeaker C: Trick.\nSpeaker C: Yeah, it's a good thing to put in your thesis proposal.\nSpeaker C: So you're going to write something for us before you go?\nSpeaker C: You have something.\nSpeaker E: In the process of.\nSpeaker C: So what are we going to meet again?\nSpeaker A: When are you leaving?\nSpeaker E: There's it, right?\nSpeaker E: Persons my last day here.\nSpeaker E: So I would suggest this one as a puzzle.\nSpeaker C: Do you mean where we the whole thing?\nSpeaker C: I didn't mean you just two of us.\nSpeaker C: Obviously we can do this.\nSpeaker C: But the question is, do you want to, for example, send a little group a draft your thesis proposal and get another session on feedback on that?\nSpeaker C: Or.\nSpeaker E: We can go through Thursday again.\nSpeaker F: Find anything?\nSpeaker F: Should we do the 1vm time for Thursday since we were on that before?\nSpeaker F: Sure.\nSpeaker E: OK.\nSpeaker E: All right.\nSpeaker E: First, then one.\nSpeaker E: I can also maybe then run through the talk I have to give as email, which highlights all of our work.\nSpeaker C: OK.\nSpeaker E: And we can make some last few changes on that.\nSpeaker C: OK.\nSpeaker D: You can just give them the abstract that we wrote for the paper.\nSpeaker C: That'll tell them exactly what's going on.\nSpeaker C: Yeah.\nSpeaker G: All right.\nSpeaker G: Can we do 130?\nSpeaker G: No.\nSpeaker G: Oh, you already told me no.\nSpeaker G: One.\nSpeaker A: It's fine.\nSpeaker A: We can do one.\nSpeaker A: It's fine.\nSpeaker F: To me, this is equal.\nSpeaker F: I don't care.\nSpeaker F: For all?\nSpeaker G: Yeah, it's fine.\nNone: For?\nNone: Yeah.\nSpeaker E: No, no.\nSpeaker E: I don't care.\nSpeaker E: It's equal to all of us.\nSpeaker E: You can decide one or four.\nSpeaker D: The person is ununancy.\nSpeaker E: This actually is actually what she likes for, because it forces the meeting record of people to put discussion to vote.\nSpeaker A: OK.\nSpeaker A: Four.\nSpeaker F: Well, if you insist, then.\nSpeaker F: Thank you.\nNone: Sorry.\nNone: Sorry.\nNone: Sorry.\nNone: You can't call that.\nNone: Pr\u0101nir Hell, msra.\nNone: You\n"}, {"meeting_id": "Bmr003", "summary": "The team was starting a transcription project for which they would have to collect a lot of speech data and then transcribe it. The meeting covered the pipeline they would use to create their data set. The team started by discussing the technical issues with their own recording equipment and then moved onto a general overview of the entire pipeline. Then, the team narrowed down on how they could carry out transcriptions. Various ideas were thrown around, but the team did not seem convinced that a software could do the task. They would need human annotators due to the granularity they were aiming for. Then, the team talked abut transcription conventions; what should be annotated and how? The team also talked about the interface they could use for transcriptions. Finally, the team ended with a discussion on the time it takes to annotate and the electronics involved.", "dialogue": "Speaker F: I see.\nSpeaker G: Okay, this is one channel.\nSpeaker G: Can you say your name and talk into your mic?\nSpeaker G: One at a time.\nSpeaker D: Mr. Micah.\nSpeaker D: This is Eric on channel three, I believe.\nSpeaker E: Okay.\nSpeaker G: Is it Jane on channel five?\nSpeaker G: I don't think it's on there, Jane.\nSpeaker F: That's the on switch.\nSpeaker F: Tasting one, two, three, tasting.\nSpeaker B: Okay, this is Jane on channel five.\nSpeaker G: I still don't see you, Jane.\nSpeaker F: Can you see me on channel four?\nSpeaker D: Really?\nSpeaker D: No, no.\nSpeaker B: I see.\nSpeaker B: No.\nSpeaker B: Is it three, really?\nSpeaker B: No.\nSpeaker B: No.\nSpeaker B: Oh, don't you, you can't see channel five yet?\nSpeaker G: Well, the mic isn't close enough to your mouth, so.\nSpeaker B: I should be okay.\nSpeaker B: Is that better?\nSpeaker B: I'd like to talk to you later.\nSpeaker B: I'd like to talk to you later.\nSpeaker G: I'd like to talk to you later.\nSpeaker F: Hello.\nSpeaker F: Hello.\nSpeaker F: Thank you.\nSpeaker F: We are your labeling machine to improve the quality of the labeling.\nNone: We'll put you in the video.\nNone: I'm going to just need four lines.\nNone: I'm just one, two, three, two, three, two, one.\nSpeaker C: One, two, three.\nSpeaker C: How many other, one to five?\nSpeaker C: One to five, yeah.\nSpeaker C: Yeah.\nSpeaker B: Would you like to join the meeting?\nSpeaker G: Well, we don't want to re-number them because we've already have like forms filled out with the numbers on them.\nSpeaker G: Okay, that's a good idea.\nSpeaker G: Okay, Dan.\nSpeaker G: Are you on?\nSpeaker C: I'm on two and I should be on.\nSpeaker C: Good.\nSpeaker C: We're going to join the meeting, Dave.\nSpeaker F: We have a spare room.\nSpeaker F: And I'm getting lots of responses on different ones, so I assume.\nSpeaker G: The various, sort of, be the answer.\nSpeaker G: This is a meeting.\nSpeaker B: We're mainly being taped, but we're going to talk about transcription for the future meeting meetings.\nSpeaker B: This is not something you need to attend.\nSpeaker G: So I should really.\nSpeaker G: I would have had a lot of diamonds.\nSpeaker C: You're always having one of those things.\nSpeaker G: Besides, I don't want anyone who has a weird accent, right, Dan?\nSpeaker F: So I don't understand.\nSpeaker F: Next, Madeline, you don't get very good performance.\nSpeaker D: It's not next planet.\nSpeaker D: It's supposed to be head mounted.\nSpeaker D: Should be head mounted.\nSpeaker D: Well, then put it on your head.\nSpeaker D: What are you doing?\nSpeaker D: I'm never saying that.\nSpeaker G: I could hear you really well on that and then transcription.\nSpeaker B: Well, I would prefer that people were on their head, but they were complaining about it.\nSpeaker G: Because it doesn't go over the years.\nSpeaker G: It's badly designed.\nSpeaker G: It's very badly designed.\nSpeaker G: It's very badly designed.\nSpeaker G: It's very badly designed.\nSpeaker C: It's not the best cover up you are.\nSpeaker G: There's nowhere to put the pads.\nSpeaker G: That's comfortable.\nSpeaker B: He's got it on his temple, so it cuts off the circumference.\nSpeaker B: Yeah, that's right.\nSpeaker G: It feels so good that way.\nSpeaker D: It feels so good when I stop.\nSpeaker G: So I again would like to do some digits.\nSpeaker G: Should I run?\nSpeaker A: Do it with noise.\nSpeaker A: So let me.\nSpeaker G: Obviously, digits.\nSpeaker G: Well, I'm just that sort of.\nSpeaker G: Did you think I saw the guy?\nSpeaker G: I don't know man.\nSpeaker G: Okay, so this is Adam.\nSpeaker G: I don't know what I had before.\nSpeaker G: I doubt it.\nSpeaker G: So this is Adam Janan on mic number one transcript, 371-390, wireless head mic.\nSpeaker G: 6817-890-01295-1383-277-34515.\nSpeaker G: 60107-824-7649-935-06340-071-1230-51065.\nSpeaker G: I think we're session four, by the way.\nSpeaker G: Or it might be five.\nSpeaker G: Oh, that's good.\nSpeaker G: I didn't bring my previous thing.\nSpeaker B: Just to be sure, the numbers on the back, this is the channel.\nSpeaker B: That's the microphone.\nSpeaker G: That's the microphone.\nSpeaker G: I believe the channel blank.\nSpeaker F: Okay, good.\nSpeaker F: But number has to be, so we have to look up the roof.\nNone: Right.\nSpeaker C: Okay.\nSpeaker C: Alright, let me do one.\nSpeaker C: This is Dan L.S. on mic to transcript 471-490-0500-546-2371-32227-473-2011.\nSpeaker C: 596-740-979-8058-028-1301-041788042-237073-4014-5-736-8517-2961-0.\nSpeaker B: Well, this is Jane on mic number five.\nSpeaker B: I just start.\nSpeaker B: Do I need to say anything more?\nSpeaker B: Transcript number five, four, seven, zero.\nSpeaker B: 907-042-312-978-70288-3456.\nSpeaker B: 8134-91802-0601-2030306060603060307589-997.\nSpeaker D: Okay, this is Eric on microphone number three.\nSpeaker D: Reading transcript 491-510-011-2029-413-524-6795-797-894-909-00.\nSpeaker D: 1-390-270-6571-34625-50217-608-0064-83692-9443-0789-0.\nSpeaker F: This is Beck on mic four.\nSpeaker F: Transcript 511-530-channel blank.\nSpeaker F: 1-2-015-0.\nSpeaker F: 3-537-686-9.\nSpeaker F: 626-0.\nSpeaker F: 776-4214-89-0.\nSpeaker F: 2149-3202-4308-3-59356-670-9278-0.\nSpeaker F: 590-796-2.\nSpeaker F: 1-5000-277.\nSpeaker F: Thanks.\nSpeaker G: Should I turn off the view meter, can you think that makes any difference?\nSpeaker G: Let me do that.\nSpeaker G: Are you going to do something other than head quit?\nSpeaker C: No, but I'm going to look at the logs as well.\nSpeaker C: Oh.\nSpeaker G: The view meter, which tells you what the levels on the various mics are, and there was one hypothesis that perhaps the active recording of the view meter was one of the things that contributed to the errors.\nSpeaker F: Oh, I see.\nSpeaker F: You did dig a reasonable hypothesis, right?\nSpeaker F: That was me.\nSpeaker F: Oh, sorry.\nSpeaker F: That was Mallard.\nSpeaker G: Well, the only reason that could be is if the driver has a bug, right?\nSpeaker G: Because the machine just isn't very capable of that.\nSpeaker G: No chance of that, just because it's made it.\nSpeaker C: Yeah, that was a bug that was a glitch last time you ran.\nSpeaker F: Are you recording more of the cable mics or are they way?\nSpeaker C: No.\nSpeaker G: Yeah, we usually do that.\nSpeaker C: No, we don't.\nSpeaker C: Why not?\nSpeaker C: We all stand in eyes.\nSpeaker C: I think I spoke to somebody more than about that.\nSpeaker C: We just did it.\nSpeaker C: We had to look down before we did that.\nSpeaker C: I know what they're there.\nSpeaker C: 4, 3, 3, 2, 1.\nSpeaker C: But I think we should put them in stand-up positions.\nSpeaker C: I think we should make little marks on the table top.\nSpeaker G: Which means we need to move this thing and sort of decide how we're actually going to do things.\nSpeaker C: I guess that's the point.\nSpeaker C: It'll be a lot easier if we have them in place.\nSpeaker B: I do wish there were big booms coming down from the ceiling.\nSpeaker B: You do?\nSpeaker D: Would it make you feel more important?\nSpeaker D: Yeah, yeah, yeah.\nSpeaker F: Wait a little projector gets installed.\nSpeaker F: That'll work.\nSpeaker C: That'll be good.\nSpeaker C: It's going to hang down, make noise.\nSpeaker C: When's it going to be installed?\nSpeaker F: Wow, it depends.\nSpeaker F: Is this being recorded yet?\nSpeaker F: I think Lila actually is almost getting pretty close to even getting ready to put out the purchase order.\nSpeaker F: I handed it off to her about a month ago.\nSpeaker G: Okay, so topic of this meeting is I want to talk a little bit about transcription.\nSpeaker G: I've looked a little bit into commercial transcription services.\nSpeaker G: Jane has been working on doing transcription.\nSpeaker G: We want to decide what we're going to do with that.\nSpeaker G: Then get an update on the electronics.\nSpeaker G: Then maybe also talk a little bit about some infrastructure and tools.\nSpeaker G: Eventually, we're probably going to want to distribute this thing.\nSpeaker G: We should decide how we're going to handle some of these factors.\nSpeaker C: Distribute what?\nSpeaker C: The data.\nSpeaker G: Right.\nSpeaker G: We're collecting a corpus and I think it's going to be generally useful.\nSpeaker G: It seems like it's not a corpus which has been done before.\nSpeaker G: I think people will be interested in having it.\nSpeaker F: You think audio DVDs are something like that?\nSpeaker G: Audio DVDs?\nSpeaker G: Well, or something.\nSpeaker G: Audio DVDs, CDs, tapes.\nSpeaker G: How do we distribute the transcripts?\nSpeaker G: How do we distribute the audio files?\nSpeaker G: How do we just do all that infrastructure?\nSpeaker D: Well, I think, for that particular issue, there are known sources where people go to to find these kind of things like the LDC, for instance.\nSpeaker G: So should we do it in the same form as the LDC?\nSpeaker C: What does that mean to what we've done already?\nSpeaker C: It's not so much the logistics of distribution, the secondary, to preferring the data in a suitable form of distribution.\nSpeaker G: As it is, it's sort of a ad hoc combination of stuff, dance, set up, and so if I set up, between May 1, make a little more formal.\nSpeaker C: The other thing is that University of Washington may want to start recording things as well in which case we'll have to decide what we actually got so that we can give them a copy.\nSpeaker G: The field trip.\nSpeaker G: I was actually thinking I wouldn't mind spending the summer up there.\nSpeaker G: I really, because of my friends.\nSpeaker G: It's been some time.\nSpeaker G: Well, and then also I have a bunch of stuff for doing this digits.\nSpeaker G: So I have a bunch of scripts with x-waves and some purl scripts and other things that make it really easy to extract out and align where the digits are.\nSpeaker G: And if UW is going to do the same thing, I think it's worthwhile for them to do these digits tasks as well.\nSpeaker G: And what I've done is pretty ad hoc.\nSpeaker G: So we might want to change it over to something a little more standard.\nSpeaker G: You know, STM files.\nSpeaker F: There's interest up there.\nSpeaker G: What's that?\nSpeaker G: There's interest up there.\nSpeaker G: Well, they certainly want to collect more data.\nSpeaker G: And so they're applying, I think, IBM is that right?\nSpeaker G: Something like that.\nSpeaker G: For some more money to do more data.\nSpeaker G: So we were planning to do like 30 or 40 hours worth of meetings.\nSpeaker G: They want to do an additional 100 or so hours.\nSpeaker G: So they want very large data set.\nSpeaker G: But of course we're not going to do that if we don't get money.\nSpeaker C: I see.\nSpeaker G: And I would like that just to get a disjoint speaker set in a disjoint room.\nSpeaker G: I mean, one of the things Morgan and I were talking about is we're going to get to know this room really well.\nSpeaker G: The acoustics of this room.\nSpeaker G: Including the fan.\nSpeaker C: Did you notice the fan difference?\nSpeaker C: Oh, no, you've touched the fan control now a lot.\nSpeaker C: It's going to be.\nSpeaker C: Here's a difference.\nSpeaker C: Oh, that's enormous.\nSpeaker F: Oh, it's enormous.\nSpeaker F: You want to leave it off or not?\nSpeaker G: All the others have been on.\nSpeaker G: Yeah, that's true.\nSpeaker F: Oh, yeah.\nSpeaker F: You can't think of something.\nSpeaker F: This fan's wired backwards, by the way.\nSpeaker F: I think this is high speed here.\nSpeaker B: Yes, that is possible.\nSpeaker C: Well, not clear.\nSpeaker C: It was like low is mid.\nSpeaker C: Mid scale.\nSpeaker C: I'll say it was not actually what I've acted with.\nSpeaker C: I was wondering also, you ready?\nSpeaker F: With a light.\nSpeaker F: Made a noise.\nSpeaker F: This, that, yeah.\nSpeaker G: Yeah, I pitch home.\nSpeaker G: So do our meetings in the dark with no air conditioning?\nSpeaker F: Yeah, just get a variety.\nSpeaker B: Candles would be nice.\nSpeaker B: They don't make noise.\nSpeaker B: Very good.\nSpeaker D: You know, it really really means to do short meetings.\nSpeaker D: Short meetings.\nNone: Short meetings.\nSpeaker D: Yes, I remember.\nSpeaker D: Alright, do you have the air conditioning?\nSpeaker D: Yes, I do.\nSpeaker D: You have the air conditioning?\nSpeaker F: I've finished the meeting.\nSpeaker F: Yes, I do.\nSpeaker F: You have the air conditioning.\nSpeaker F: You have the air conditioning.\nSpeaker F: Actually, the air conditioning still working, it's just an auxiliary fan.\nSpeaker D: I see.\nSpeaker D: So in addition to this issue about the UW stuff, there was an ounce today via the LDC corpus from, I believe, Santa Barbara.\nSpeaker D: Yeah.\nSpeaker B: I saw it.\nSpeaker D: I've been watching for that corpus of General Spoken English.\nSpeaker D: English and I don't know exactly how they recorded it but apparently there's a lot of different\nSpeaker B: styles of speech and whatnot. They have people come in to a certain degree and they and they have to add\nSpeaker D: Recorders. I see. So it is sort of far-field stuff. Right? I\nSpeaker B: assumed so actually I hadn't thought about that unless they added close field later on but I've listened to some of those data and I I was actually on the advisory board for when they set the project up. Oh, okay. I'm glad to see that it got released.\nSpeaker G: So it's a very nice thing. I wish we had someone here working on adaptation because it would be nice to be able to take that stuff and Adapted to a meeting setting Mechanical adaptation. No software to adapt the speech recognition. Well, what I was thinking is it may be useful in\nSpeaker D: Transcribing if it's far-field stuff right in doing some of our first automatic speech recognition models It may be useful to have that kind of data because that's very different than any kind of data that we have so far.\nSpeaker B: That's true and their recording conditions are really clean. I mean, I've heard\nSpeaker F: That's not good. Well, but what I mean is that far-field means great distance. I mean just these not head-mounted Yeah, and so that's why they're getting away with just two channels or something or they're using multiple deaths. Oh\nSpeaker B: Good question and I can't answer it. No, we can look at the web page to the answer\nSpeaker D: So I was thinking that we should contact them. So it's sort of a beside the point point, but so we can get that just with\nSpeaker G: Media costs. Is that right?\nSpeaker D: In fact, we get it for free because they're distributing 3D LDC. So that would be yeah, that would be something looking to. I can actually arrange for it to arrive in short order if we're\nSpeaker G: Wow, it's still is doing unless we're gonna have someone to work on it. So maybe we need to\nSpeaker B: Think about it a little bit. The thing too is that they're just their transcription format is really nice and simple and in the discourse domain But they also mention that they have time aligned. I mean, I\nSpeaker D: Maybe we should maybe we should get a copy of it just to see what we can do. Yeah, we can compare. Hey, why don't you go ahead and do that then?\nSpeaker G: I\nSpeaker B: See SAE corpus is spoken American English Yeah, I was really pleased to see that I knew that they They had had some funding problems in completing it. Well, this is clever. Apparently this was like phase one\nSpeaker D: And there's no more that they're gonna do apparently or something like that Unless of course they have funding issues and then they may not do phase two but from all the web documentation and it looked like oh, this is phase one\nSpeaker B: So whatever that means super great. Yeah, that's I mean, they're really well respected in the linguistics side too on the discourse area\nSpeaker D: So this is very good. It would also maybe help be helpful for Liz if she wanted to start working on some discourse Issues, you know looking at some of this data and then You know, right so once you get here\nSpeaker G: Actually, that's another thing I was thinking about is that maybe Jane should talk to Liz To see if there any transcription issues related to discourse that she needs to get marked\nSpeaker D: Okay, maybe we should have a big meeting meeting\nSpeaker G: That would be a meeting meeting meeting meeting me. Yeah, well, this is the meeting about the meeting meeting meeting\nSpeaker D: But maybe we should find some day that it was a little is an hydrate seem to be around more often So I should find a day when they're gonna be here and And Morgan's gonna be here and we can meet at least this subgroup. I mean not necessarily have the UW people now\nSpeaker G: Well, I was even thinking that maybe we need to at least painly you dub we need to see you You know say this is what we're thinking about for our transcription if nothing else So what will show we move on and talk a little bit about transcription there. Okay, so since that's what we're talking about What we're using right now as a tool From this French group called transcriber that seems to work very well So has a nice useful tickle decay user interface and\nSpeaker F: This is the process of converting audio to text right and just requires humans just like so yeah\nSpeaker G: Yes, you piece right right so we're at this point only looking for word level So all so what you have to do is just identify a segment of speech in time and then write down what was said within it and identify the speaker And so the things we that we know that I know I want are the text the start and end and the speaker But other people are interested in for example stress marking and so Jane is doing primary stress Stress marks as well, and then things like repairs and false starts and Field pauses and all that other sort of stuff. We have to decide how much of that we want to do I did include a\nSpeaker B: Certain first pass my my view on it was when you have a repair then It seems I mean we saw there was this presentation in one of the speech group meetings about how and I think Liz has done some stuff too on that that It that you get it bracketed in terms of like well if it's parenthetical, which I know that was his work done then You'll have different Prasadic aspects and then also if it's if it's a repair where they're like what I just did Then it's nice to have sort of a sense of the continuity of the utterance with the start to the finish and It's a little bit deceptive if you include the repair the pre-repair part Sometimes or it's the middle anyway, so what I was doing is bracketing them to indicate that they were repairs which isn't Very time consuming\nSpeaker F: Is there already some sort of plan in place for how this is going to be staffed or done or is it real well talking about here?\nSpeaker G: That's part of the thing we're talking about so what we wanted to do is have Jane do basically one meetings worth\nSpeaker F: You know 40 minutes to an hour and yourself. Yeah, this is this is like five times real time or 10 times real time about\nSpeaker G: Is and so one of the things was to get an estimate of how long it would take and then also what tools would use and so the next Decision which has to be made actually pretty soon is how are we gonna do it?\nSpeaker F: So you may change to the first one so then she can decide oh, we don't need all this stuff just the words are fine\nSpeaker B: But we have a good you were continuing with the transcription right so so one one option is to get\nSpeaker G: Wingwistics grad students and undergrads to do it and apparently that's happened in the past and I think that's probably the right way to do it It will require a post pass. I mean people will have to look at it more than once to make sure that it's been done correctly But I just can't imagine that we're gonna get anything that much better from a commercial one and the commercial ones I'm sure will be much more expensive can we could join it do it all?\nSpeaker G: Yeah, right We just get joy and Jane to do everything that's right But you know, that's what we're talking about is getting some slaves who need money and I can't get that characterization\nSpeaker B: I\nSpeaker G: Mention I if so again, I have to say are we recording and then say Morgan has has consistently Resisted telling me how much money we have well the answer is zero there's a reason why it's resistant Well if it's zero then we can't do any transcription because we I mean I can't imagine us doing it ourselves well, we already we already have a plan in place for the first meeting\nSpeaker B: There's also the other possibility which is if you can provide not money, but instructional experience or some other perks You could get people to\nSpeaker F: Well, but seriously, I mean work is obviously in a bind over this and the thing to do is just do the field dream Theory which is we go ahead is though there will be money at the time we need to money and that's that's the best we can do We should not do anything until we get money is ridiculous. We're not gonna do anything done if we do that\nSpeaker G: So at any rate Jane was looking into the possibility of getting students at that right talking to people\nSpeaker B: Right, any progress in that front yet. I should have sent email on I haven't yet, right?\nSpeaker F: So do it so until you actually have a little experience with what this this French thing does we don't even have And I do have she's already done. Oh, yeah, I'm sorry. Yeah, so that's where you came over the 10x number\nSpeaker B: Or is that the one people usually use 10x and I haven't really calculated how fast I'm like I haven't done as you see I've been at the same time doing kind of a bootstrapping and deciding on the transcription conventions that are no one and Stuff like you know how much There's some interest in you and batteries problems like you know what span of time is it useful to segment the thing into in order to Transcribe it that most quickly because you know you get like you get to span the five words. That's easy But then you have to take the time to market and then there's a issue of it's easier to Hear it the first time if you've wanted it a boundary instead of somewhere in the middle because in the words I sectored or whatever and And so I mean I've been sort of playing with different ways of Because I think and you know I mean if you could get optimal instructions you could cut back on the number of hours or\nSpeaker F: Yeah, this tool you're using is strictly it doesn't do any speech recognition doesn't know it doesn't but what a super tool It's a great environment, but is there any way to do our speech recognizer up to it and actually run it through\nSpeaker G: We've we've thought about doing that but the recognition quality is going to be horrendous\nSpeaker F: Well a couple things first of all the time marking you'd get you could get by a tool And so if the issue really I'm thinking about the closed caption you see running by a live newscast You know most of those are done by a person. I know I know that no I understand it and a lot of them you see typos and things like Yeah, but it occurs to me that it may be a lot easier to correct things and it is to do things from scratch No matter how wonderful the tool is but if there was a way to merge the two\nSpeaker D: Well, I mean we talked about it's easier to type out something instead of going through and figuring out which is the right I mean it depends on the error rate\nSpeaker F: But but again the timing is for should be for free the timing should be we don't care about the timing of the words Oh, I thought you just said that was we don't care about the time of the words just\nSpeaker C: We don't we don't know actually we haven't decided which which time we care about this kind of one of things you're saying is like you have the option to put in more or less timing data and In the absence of more specific instructions\nSpeaker G: You're trying to figure out what the message is so what which is done so far is sort of more or less breath not breathable Sort phrases continuous phrases and so That's nice because you you Separate when you do and extract you get a little silence on either end so that seems to work really well That's ideal\nSpeaker B: I know the alternative which I was sort of experimenting with before I ran out of time Recently was that you know if it were like an arbitrary segment of time Pre-marked because it does take time to put those markings in It's really the interface is wonderful because you know the time it takes is you listen to it and then you press the return key But then you know it's like you press the tab key to stop the flow and and The return key to put in a marking of the boundary But you know obviously there's a lag between when you hear it when you can press the return key so it's slightly delayed So then you listen to it a second time and move it over to here So that takes time now if it could all be pre-marked at some\nSpeaker F: Are those delays adjustable Those delays adjustable see a lot of people who actually build stuff with human computer interfaces Understand that delay and yeah, so when you by the time you click it It'll be right on because it'll go back at time put\nSpeaker G: We could program that pretty easily couldn't we Dan? Yeah, Mr. G\nSpeaker B: That would make a difference but\nSpeaker G: It's not bad, but it does we tried to do automatic speaker ID because primarily the markings are at speaker change\nSpeaker C: But that would be we've got the most channel data we don't have to do it from your signal Good point\nSpeaker G: Yeah, I guess the question is how much time will it really save us versus the time to write all the tools\nSpeaker C: But if we're talking about collecting ten or a hundred hours Which is gonna take a hundred or a thousand hours to transfer it from ten extra five weeks\nSpeaker G: We're gonna need ten to a hundred hours to train the tools and Validate the tools to do the to do all this anyway\nSpeaker C: But if we're just doing silence to tell you we're gonna do that\nSpeaker F: Put it in your sweater\nSpeaker C: It seems like Well, I know you know, I mean it's it's maybe like a Weeks work to get to do something like this so 40 or 50 hours\nSpeaker B: Could you get it so that with so it would detect volume on a channel and insert a marker in the Formants really transparent. It's just a matter of a very clear is XML, isn't it?\nSpeaker B: It's very I mean I looked at the the five or men It's just it has a time a time indication and then something or other and then an end time or something or other\nSpeaker D: So maybe you could try the following experiment take the data that you've already transcribed and\nSpeaker F: And is this already in the past or already in the future? Already in the past you've already you've already done. She's done about half a meeting. Well, I see okay, right about half\nSpeaker B: I'm not sure if it's that much but anyway enough to work right\nSpeaker D: so And and throughout the words but keep the time markings and then go through I mean and go through and try and retranscribe it Given that we had perfect boundary detection and see if it see if it feels easier and forgetting all the words Yeah, I was thinking I'd be cheating a little bit. Yeah, I mean that's part of the problem is that what we really need is\nSpeaker C: When they should do it again from scratch and then do it again with the badge so you did the whole thing three times\nSpeaker F: Since we need some statistics do it three more\nSpeaker B: So you'll get you get down to 1.2x by the time you get that\nSpeaker F: I should have it finished by the end of the day. No, but the thing is the fact that she's she didn't have before just might give a lower bound That's exactly right. Yeah, and if lower bound is 9x then it's a waste of time right well\nSpeaker B: I bet there's an extra problem which is that I didn't really keep accurate. It wasn't a pure task the first time so It's gonna be an upper bound in that case and it's not really strictly comparable So I think though that's a good proposal to be used on new a new batch of texts that I haven't yet done yet in the same meeting\nSpeaker C: Could use it on the next segment. Well, where do we get the the the Oracle boundaries from right?\nSpeaker G: Well, the boundaries yeah one person would have to assign the boundaries and then the other person would have to that's easy enough\nSpeaker B: I could do that but the Oracle boundaries would come from volume on a specific channel, wouldn't they?\nSpeaker D: You want to know given given a perfect human segmentation? Yeah, I mean you want to know how well I mean the question is is it worth giving you the segmentation?\nSpeaker G: I see to me. I mean that's easy enough. I could generate the segmentation that and you could do the words And time yourself on a little double blind here. Yeah, I see okay, so that might be worth doing That's good. I like that. That would at least tell us whether to worth spending a week or two trying to get a tool That will compute the segmentation\nSpeaker F: I think to keep in mind too about this tool guys is that sure you can do the computation for what we're gonna do in the future But if you W is talking about doing two or three or five times as much stuff and they can use the same tool\nSpeaker B: Then right obviously is a real multiplier there the other thing too is with the speaker identification If they can handle speaker identification, that's a big it's pretty easy when we've got separate channels\nSpeaker F: Right use it. That's why we let these bits of microphone\nSpeaker D: It's gonna be in the meeting like the reading group meeting that we had the other day That's it's gonna be a bit of a problem because like I wasn't wearing a microphone Yes, we are the people that weren't working but you didn't see anything worthwhile anyway, right? That's pretty much true\nSpeaker G: It might save 90% of the work though, but yes, so\nSpeaker C: So I need to we need to look at what what the final output is but it seems like We it seems like it's not really not that hard to have an automatic tool to generate the phrase marks and the speaker Speaker at the density without putting in the words. I've already become pretty familiar with the format\nSpeaker G: So it would be easy. We didn't finish you tell me where it is\nSpeaker B: I've already completed on this I mean you you talked a little bit about the transcription conventions Mm-hmm And I guess you've mentioned in your progress report the status report that you had written a script to convert it into So I when I it's quickest for me in terms of the transcription part to say something like you know if Adam spoke to To just say a colon could be you know me at the beginning of the line and E colon instead of entering the interface for a speaker identification and clicking on the thing Indicating the speaker ID. So and then he has a script that will convert it into the the thing that would indicate speaker ID\nSpeaker G: It's pretty cute But at any rate so Right so so I think the guess at 10x seems to be pretty standard Everyone more or less everyone you talk to says about 10 times using more using standard technical transcript using stone age tools You think using stone age tools. I mean I looked at cyber transcriber Which is a service that you send an audiophile they do a first pass speech recognition and then they they do a cleanup But it's gonna be horrible. They're never gonna be able to do a meeting like this Well for cyber transcriber they don't quote a price they want you to call and talk so for other services They were about $30 an hour 30 so yeah for $30 an hour for of their work So so if it's 10 times it's $300 an hour\nSpeaker F: Did you talk to anybody does close captioning for for TV? No Because they usually at the end of the show they'll tell what the company is and captioning companies doing it\nSpeaker G: Yes, so my my search was pretty cursory. It was just a net search and So it was only people who have web pages and are doing stuff\nSpeaker F: Well, the thing think about this is thinking kind of maybe a little more globally than I should here, but Really this could be a big contribution we could make I mean we've been to the STP thing we know what it's like to to manage the manage the process and admittedly They might have been looking for more detail than what we're looking for here But it was a it was a big hassle, right? I mean, you know that constantly because reminding people and going over and Clearly some new stuff needs to be done here, and it's it's only our time where our of course includes Dan So it includes me at all\nSpeaker C: It just seems like yeah, I mean I don't know if we'd be able to do anything to help STP type Problems but suddenly for this problem we can do a lot better why because they wanted a lot more detail No, because they had because they had to speak. It's right. I mean the segmentation problem is\nSpeaker G: Only trivial they had two speakers. Oh, I see what took them so long Mostly because they were doing much lower level time. So they were doing phone and syllable Transcription as well as we're transcript and so we we decided early on that we were not gonna do that\nSpeaker F: But there's still the same issue of managing the process of reviewing and keeping the file straight and all this stuff\nSpeaker G: Which is clearly has right and so so what I'm saying is that if we hire an external service I think we can expect 300 dollars an hour I think that's the ballpark there were several different companies that and the range was very tight for technical documents\nSpeaker D: 28 to 32 dollars an hour who knows if they're gonna be able to manage multiple\nSpeaker C: Yeah, they won't they won't have to mix it. No, but I mean they would they wouldn't they wouldn't they won't they will refuse to transcribe this kind of material That's not what that's quite important for right. Well, they might quote it\nSpeaker G: Several of them say that they'll do meetings and conferences and so on none of them specifically said that they would do speaker ID Or speaker change?\nSpeaker F: There may be this multiplier for five people cost twice as much if you're 10 people cost something like that\nSpeaker G: Well, the way it worked is it was scaled so what they had is if it's an easy task it costs $24 an hour And it will take maybe five or six times real time and what they said is for the hardest tasks Bad acoustics meeting settings. It's $32 an hour and it takes about 10 times real time So I think that we can count on that being about what they would do It would probably be a little more because we're gonna want them to do speaker market a lot of the companies\nSpeaker F: I've worked for the Person leading the meeting the executive whatever sort of go right the room and mentally calculate how many dollars per hour this meeting was costing right in University atmosphere a little different thing He's worth 50 an hour So here we're saying well, let's see if the meeting goes another hour. It's gonna be another thousand dollars We're talking talk really fast\nSpeaker B: But with few words\nSpeaker F: Only talk when you're pointed to\nSpeaker G: Content words only So but anyway, so we we have a ballpark on how much it would cost if we send it out and we're talking about doing how many hours with the meetings 30 or 40 so 30 40 thousand dollars\nSpeaker F: What was 30 times 300?\nSpeaker D: So I 300 right I got an extra factor of 30 hours an hour, but we can pay a graduate student Seven hours an hour and the question is what's the difference?\nSpeaker D: What do you know what the going rate is it's on the order of eight?\nSpeaker B: I think I think I would say this a good estimate I'd say I was gonna say eight you'd say ten. Let's say ten these are not for engineering graduate students right?\nSpeaker G: Right these are linguistics grad students right?\nSpeaker D: So that means that even if it takes some 30 times real time it's cheaper to\nSpeaker G: To do graduate students. I mean that's why I said originally that I couldn't imagine sending it out. It's gonna be cheaper\nSpeaker B: Thing too is that they were linguistics You know in terms of like the post editing Yeah content-wise they might be easier to handle because they might get it more right the first time\nSpeaker G: And also we would have control of I mean we could give them feedback whereas if we do a service It's gonna be limited amount. Yeah, I mean we can't tell them You know for this meeting we really want to mark stress and for this meeting we want And they're not gonna provide they're not gonna provide stress. They're not gonna provide repairs They're not gonna provide they may or may not provide speaker ID\nSpeaker F: So that we would have to do our own tools to do that. So just hypothetically assuming that we go ahead and using graduate students Who's the person in charge who's gonna be the Steve here? I hope it's Jane. Oh, is that all right?\nSpeaker B: Um now would this involve some manner of monetary compensation or would I be the voluntary coordinator of multiple transcribers?\nSpeaker G: Um, I would imagine there would be some monetary involved but we'd have to talk to more than that\nSpeaker F: Yeah, I don't know. It just means you have to stop working for Dave Take that's why Dave should have been here\nSpeaker G: Protect his people Well, I would like you to do it because we have a lot more experience than I do But if that's not feasible, I will do it with you as an advisor\nSpeaker F: We'd like you to do it and we'd like to pay you Not being Morgan though\nSpeaker G: I see we'd like to push that link Yeah, six dollars an hour\nSpeaker B: I see, okay I want to increase my income like it's doing the transcribing You can't be an influencer\nSpeaker F: Be sure and say would you like fries with that?\nSpeaker A: Thank you about your face scale\nSpeaker B: Yeah, no, I would be interested in becoming involved in the project More\nSpeaker G: Um, any more on transcript we want to talk about?\nSpeaker C: What's, so what do you say you've done some portion of the first meeting? Yes What's your plan to carry on doing it?\nSpeaker B: Well, you know what I thought was right now we have So I gave him the proposal for the transcription convention So he made his suggestion It's a good suggestion So as far as I'm concerned those transcription conventions are fixed right now And so my next plan would be\nSpeaker C: What do they call them?\nSpeaker B: They're very minimal so good to just to summarize that So um One of them is the idea of how to indicate speaker change Yeah And this is a way which mesh as well with making it So that you know on the at the point such a nice interface When you when you get the um you get the speech signal You also get down beneath it An indication of If you have two speakers overlapping in a single segment You see them one displayed one above each other And then at the same time the top part of the screen is the actual verbatim thing You can click on individual utterances And it'll take you immediately to that part of the speech signal and play it for you And you can you can work pretty well between those two\nSpeaker F: Is there a limited number of speakers?\nSpeaker G: Um the user interface only allows to And so if if you're Using their interface to specify overlapping speakers you can only do two But my script can handle any and their save format can handle any And so um using this the convention that Jane and I have discussed You can have as many overlapping speakers\nSpeaker F: Is this a university project? This is the French for some great French\nSpeaker G: Yeah and they're they've been quite responsive I've been exchanging emails on the very same If you ask them to change the interface for more speakers?\nSpeaker G: Yes and they said that's on in in the works for the next version Multi channels Multi channels was also well they said they wanted to do it but That the code is really very Organized around single channels so I think that's unlikely to happen\nSpeaker F: Do you know what they're using it for? Why did you develop it?\nSpeaker F: For the language?\nSpeaker F: For the language I ask But I mean are they are they linguists or are they speech recognition people?\nSpeaker G: I think they're linguists\nSpeaker D: They're they have some connection to the LDC because the LDC has been advising them on this process Well linguistic data consortium So but a perfect that\nSpeaker G: It's also all the sources available So great If you if you speak TCLTK And they have they've actually asked if we are willing to do any development And I said well maybe So if we want if we did if we did something like programmed in a delay Which I think is a great idea I'm sure they would want that incorporated\nSpeaker C: They're pretty\nSpeaker B: Pretty Well and they thought about things you know I mean they do have So you have when you when you play it back It is useful to have a break mark to But it wouldn't be strictly necessary because you can use the tab to key to toggle the sound on and off I mean it'll stop the speech and you know if you press a tab And And so That's a nice feature and then also once you put a break in then you have the option of Cycling through the unit you could do it like multiple until you could crazy and decide to stop cycling through the unit\nSpeaker F: You know there's also the user interface. It's missing it's missing from all of our offices And that is some sort of analog input for something like this So what audio people actually use of course is something that when you move your hand further The sound goes faster past it like fast forward. You know like a joy stick or You could wire a mouse or track ball\nSpeaker B: It's not something I wanted to have\nSpeaker F: No, but I'm saying this is what professionals who actually do this kind of thing for For video or for audio where you need to do this and so you get very good at sort of jostling back and forth rather than hitting tab and back space and carriage turn and\nSpeaker G: And things like that. Yeah, we talked about things like foot pedals and other analogs So I mean those are things we could do but I Just don't know how much it's worth. Yeah\nSpeaker B: I agree they they have several options So you know, I mentioned the looping option another option is it'll cause when it reaches the end of the boundary And then to get to the next one you just press tab it goes on the next unit. I mean it's very nicely thought out They thought about and also it'll go around the I want to say cursor, but I'm not sure if that's point Whatever anyway, you can so they thought about different ways of having windows that you work with them and but so in terms of the conventions then basically strictly orthographic which means with some provisions for Uh colloquial forms so if a person said cause instead of because then I put a an apostrophe at the beginning of the word and then and double I think of rackets what the full-excollite would be and this could be something that was handled by table or something But I think to have a convention marking and is a non standard or what not only standard but a non Orthographic whatever Non-continue or want to know the same thing and there would be limits to how much refine what you want Indicating something as non standard\nSpeaker D: How are you handling back channels?\nSpeaker B: Back channels, you know, oh yes, there was some In my view when when you've got it densely overlapping Um, I didn't worry about I didn't worry about specific start times I sort of thought that this is not going to be easily processed anyway And maybe I shouldn't spend too much time getting exactly when the person said no or you know Yeah, immediate And instead just sort of rendered within this time slot There were two people speaking during part of it and if you want more to tail figure it out for yourself\nSpeaker G: I think what way I felt what Eric was talking about was channels other than the direct speech, right?\nSpeaker D: Well, yeah, what I mean is what I mean when somebody says aha in the middle of that happened very seldom Because I was listening to Dan was agreeing a lot to things that you were saying as you were talking\nSpeaker B: Oh, well, thank you Dan, well if it was a more like right, you know, then I would I would indicate That it happened within the same time frame, but wouldn't say exactly what happened\nSpeaker C: I transcribed a minute of this stuff and that was a lot of overlapping A lot of overlapping\nSpeaker G: Well, there's a lot of overlapping at the beginning and end. Yeah, huge Yeah, when when no one is when we're not actually in the meeting and we're all sort of separated then and doing things But even during the meeting there's a lot of overlap, but it's marked pretty clearly I'm some of the back channel stuff Jane had some comments and but I think a lot of them were because you were at the meeting And so I think that often Often you can't tell I mean Jane had that comments like\nSpeaker B: To who the person was speaking to only when it was otherwise gonna be puzzling because he was in the other room talking\nSpeaker G: Yeah, but someone who was just a transcriber wouldn't have known that that's true or when Dan said I wasn't talking to you\nSpeaker F: I know sir you take a bathroom break in the middle and yeah, you turn off your mic\nSpeaker B: Well, he was so so he was checking the meter levels and and we were handling things Well, he was labeling the whatever it was the pd a And and so he was it sort of you were sort of talking you know So saying and I could label this one left right and he and he said I don't see anything And he said he said I wasn't talking to you But when you know he can't hear what he's saying it was a lot funnier if you're there\nSpeaker G: I mean, I know what what what what happens is if you're a transcriber listening to it It sounds like Dan is just being a total Oh, I thought it was you No, but you were asking off the wall question, but but if you knew that that I wasn't actually in the room and that Dan wasn't talking to me It became okay\nSpeaker B: That's where I added comments the rest of the time I didn't bother with who was talking to who but but this was an amusing\nSpeaker F: This is going to go into meeting meeting transcriber bloopers tape right?\nSpeaker B: Yes Well, part of it was funny every so was because it was a mixed signal so you couldn't get any clues from volume That you know he was really far away from this conversation Couldn't do that sufficiently in any case\nSpeaker G: No, I should rewrite the mix tool to put half the people in one channel and half in the other I have an auto gain mixer tool that mixes all the head mount microphones into one signal and that seems to work really well for the transcribers right\nSpeaker B: I thought it would be you know, I didn't want to add more contextual comments that were needed but that seemed to be clarified that the Cont what was going on and\nSpeaker D: Okay, so normalization I was just gonna ask so I just wanted to sort of finish off the question I had about back channels Yeah, okay, which which was so say somebody's talking for a while and somebody goes in the middle of it and and whatnot does the conversation Come out from the person who's speaking for the long time as one segment and then there's this little tiny segment of The southern speaker or does it does the fact that there's a back channel split the\nSpeaker B: it into okay my my focus was to try and maintain content content continuity and To keep it within what he was like wouldn't say breath groups but prismatic or international groups as much as possible So if someone said in the middle of a of someone's International contour I I indicated it as like what you just did Then I indicated it as a segment which contained this utterance plus an overlap\nSpeaker C: Okay, but that's but there's any one there's any one time boundary for both speakers. Yeah, that's right\nSpeaker B: Any you know it could be made more precise in that but I just I see you okay\nSpeaker F: I think whatever you use these speech words we should always do the thing like you're talking about accent\nSpeaker B: Okay, and so then in terms of like words like uh and um I just wrote them because I figured there's a limited number and I keep them to a Limited set because it didn't matter if it was um You know versus um, so I just always wrote it as you am And uh, you know you wait you wait. I mean like a set of like five but in any case\nSpeaker C: No, I didn't notice you H H H Uh-huh is you H H I'd be happy with that that's fine\nSpeaker G: You could have that late in the professions. I did know that there were some segments that had pauses on the beginning and then we should probably mark Areas that have no speakers as no speaker Yeah, that's okay. And so question mark colon is fine for that\nSpeaker F: Yeah, what does that mean you mean no one's talking. Yeah, so I'll explain it to all right. Yeah, yeah, so I had to mark this\nSpeaker C: Then can we sleep them on marked?\nSpeaker G: Well, I want to leave them marked. I don't want them to be part of another utterance. Okay So you just you need to have the boundary the start in the end now that's refinement that maybe it could be handled by\nSpeaker B: Part of the part of the script or something. Yeah, it seems like\nSpeaker C: It seems like the trans- transcription problem would be very different if we had these automatic speaker detection term placing things Because suddenly I mean, I don't know actually sounds like there might be a problem putting into software if the software any handles to parallel channels But see when we can get around that somehow\nSpeaker G: Well, you're saying I think it can read it can read and write as many as you want\nSpeaker C: It's just that it what if you want to edit it All right, I mean the point is we're going to generate this transcript with five five tracks in it But with no words. Someone's going to have to go and type in the words Um, and if that five five people speaking once right it's\nSpeaker G: I didn't explain it well if we use the the little the conventions that Jane has established Oh, yeah, I have a script that will convert from that conventions to their save convention\nSpeaker B: Which allows five and it can be edited after the fact Yes, but they're but they're format if you wanted to indicate the speakers right there instead of doing through this direct route Then the window comes up and it only allows you to enter two speakers, right?\nSpeaker F: But you're saying it by the time you call it back in from their save format it opens up when it was when it was five speakers, right?\nSpeaker F: Oh, that is sort of They didn't quite go the whole user interface. Yeah\nSpeaker G: So go the whole route did they the whole save form but save format and the internal format and all that stuff handles multiple speakers It's just there's no user interface for specifying\nSpeaker F: Right multiple any more than two so you're your your script solves doesn't solve all our problems because we're always going to want to go through this Pre-processing assuming it works. Yeah, that works. I said because this is so quick to enter\nSpeaker B: So I wouldn't want to do it through the interface anyway. Yeah, I see And then see what else oh, yes, I wanted to have so sometimes the person I if in terms of like the continuity of thought for transcriptions It's it isn't just words coming out. It's like there's some purpose for an utterance and sometimes someone will do a back channel in the middle of it But you want to show that it's continued at a later point so I have I have a convention of putting like a dash Arrow just to indicate that this person's utterance continues and then when it Catches back up again, then there's an arrow dash in the opposite direction to indicate continuation of one zone utterance versus Sometimes we had situation which is you know, what you what you get in conversations of someone continuing someone else's utterance And in that case I did a tilde arrow versus a arrow tilde to indicate that it was continuation But it wasn't I guess I did equal arrow for the for the own for yourself Things because the speaker's the same and then tilde arrow if it was a different If a different speaker Continuation but just you know the arrows showing continuation of a thought Mm-hmm, and then you could track whether it was a same speaker not by knowing you know at the end of this unit You know what happened later and that was like this person continued and you'd be able to look for the continuation\nSpeaker C: So the only time That becomes ambiguous is if you have two speakers Like if you if any have one person if any have one thought this continuing across a particular time boundary You just need a one-hourity Jen and then it was picked up by different speakers picked up by different speaker The time it comes ambiguous is if you have More than one speaker and that and they sort of Swap I guess you have more than one track going then They need to know if they were swaps or not Oh, if it does it have a music I play not very much. Yeah, I didn't use it especially for meetings\nSpeaker G: No, if you were just recording someone's day, it would be impossible You know if you're trying to do a remembrance agent, but I think for meetings. It's probably all right But a lot of these issues I think that for uh, from my point of view where I just want to do speech recognition and information retrieval it doesn't really matter I know but other people have other interests\nSpeaker C: It does feel like it's really nice. I you know this description and I marked that I marked it with a lipcess Because it seemed like there was a difference. That's something you wanted to indicate that it This was the end of the phrase. This was the end of that particular transcript, but it was continued later So I picked up a lipcess. Excellent. Yeah. Yeah. I didn't have the equal naughty\nSpeaker B: Boss That's why I didn't do it. I mean, that's why I thought about And it didn't do I didn't do it in ten times\nSpeaker G: Yeah, well, so anyway, are we interested then in writing tools to try to generate any of this stuff automatic\nSpeaker C: So that's something you want to do Dan? No, but it's something that I feel we definitely want to do\nSpeaker B: I also want to ask you if you have a time measurement on the part that you Transcribe do you have a sense of how long yeah, I took me off and I have to transcribe a minute\nSpeaker C: But I didn't have any I didn't even have I was trying to get transcribed it around by good So I was telling it by typing into a text file and trying to think it was horrible\nSpeaker F: So 30 to ones what you got to that's our new upper limit. Yeah\nSpeaker D: I think for first try that\nSpeaker C: It was actually it was quite it was iron infinite number of dance\nSpeaker G: I'm so is a warm up. Are we gonna run out of this space by the way?\nSpeaker E: Okay\nSpeaker D: So just in the beep in the room when you're at this when maybe we should consider also So I'm gonna build up the website around all of these things. That's great\nSpeaker C: Dan sort of started business to business e-commerce as well\nSpeaker D: No, but it will be interesting. Can we sell banner ads? Good paid for click throughs. What a good idea. That's how we pay for transcription. I want to introduce the word we cannot Want to word it won't be recognized\nSpeaker E: I'm so\nSpeaker D: No, all right, so here's here's my thought behind it which is that The stuff that you've been describing Jane One has to of course indicate It's very interesting and I'd like to be able to to pour through you know The types of conventions that you've come up with and stuff like that\nSpeaker B: So I would like to see that kind of stuff on the way. Okay now at the alternative to a website would be to put it in doctor's speech Because because what I have is a song like to my transcription That I have on my\nSpeaker C: We can do it all we can do it all Okay, we can write it We could actually maybe we could use the tsl plugin\nSpeaker B: Yeah, he's committed himself to something\nSpeaker D: Oh\nSpeaker F: See he said the word tsl and that's pretty much such a good job of it. He should be allowed to you know\nSpeaker C: I know it that but I write but I should be allowed to but if you just did a crappy job Nobody would want you to do it I'm like with my priorities all right. Let's look at it anyway So definitely we should we should have some kind of access to yeah\nSpeaker G: We have we have quite a disparate number of web and other sorts of documents on this project sort of spread around\nSpeaker D: I have several and Dan has a few right so we can add in links and other things\nSpeaker C: Well\nSpeaker G: We'll try to try this consolidate me who wants to do that though No one wants to do that\nSpeaker C: Right that's the problem why\nSpeaker D: What's the issue sort of a disorganized sort of group no one I'm not\nSpeaker C: No one what no one owns the project\nSpeaker G: I mean I own the project but I don't want to do it no one wants to end the project Well Mine\nSpeaker C: You have to do the website\nSpeaker F: But what are you talking about for website hack you're talking about writing HTML right?\nSpeaker G: Yeah, I'm talking about putting together all the data in a form that that is legible and pleasant to read and up to date and\nSpeaker F: But it's that easy against a lot of actually use a tool to help your job. Oh, it's your\nSpeaker G: It's a lot of use a tool. I have in family tools that I like. It's just as easy to use to edit the right HTML is anything else\nSpeaker C: No kidding that's obviously not sure\nSpeaker F: Obviously not sure no it's obviously true that he hasn't found that he likes the question is what is what do you look at? You're something old trellis\nSpeaker B: Right, yeah, which produces also sight maps. I guess if I were you're doing more powerful\nSpeaker G: Excuse me more complex websites I might want to But most of the websites I do aren't that complex\nSpeaker B: Well, would this be to document it also for outside people are mainly for in house. You know, I think both\nSpeaker F: Most really internal Well, yeah, but what does internal means I mean you're leaving People UW want to look at it. I mean it's internal until I see we could do an ht access\nSpeaker B: I okay, well\nSpeaker G: Send me links and I will send me pointers rather and I'll put it together\nSpeaker C: I'm not okay. I'm not sure how how important that stinks is I don't think we should say oh, it's internal therefore We don't have to make it pretty good. I mean you can say oh, it's internal therefore we can put data on it that we don't we don't have to worry about releasing I think the point is to try and Be coherent and make a great presentation\nSpeaker B: Yeah, it's true because you can't do the work sooner or later That's right. I mean is even if it's just writing things up\nSpeaker F: Okay, um, let's move on to electronics. Do we we out of tape out of disk? No Great. I was looking for the actual box. I plan to use But I I couldn't find it at the local store, but this is the the technology. It's actually a little bit thinner than this It's two by two by one and it would fit right under the It's there in the ballot lip on the lip. Yeah, there's a lippeny stables Yes, and especially about the bottom along to try and generate some frequencies that you may not already have recorded I need to see what it does to the but this was the just just to review and also about this along rather than the projector so we could put These on the table and and crank all of them diagram That's the six tables that we're looking at these six tables here with with little boxes In the middle here, which would I mean the boxes are pretty much out of the way anyway I'll show you the the car. This is the table cross section\nSpeaker C: I know people realize you're trying to screw up the microphone. Yes, yes, absolutely\nSpeaker F: Why not? I mean because this is what's gonna happen You got plenty of data. I won't come to your next meeting And so this is the box in your paper off my pda Not the record show that this is exhibit to be that's right or not to be yeah The box This a half inch lip here at a box is an inch thick so it hangs down a half an inch and so the two heads that jacks would be in the front And then a little led to indicate that that box is live that the important issue about the led is the fact that we're talking about eight of these total Which would be 16 channels and Even though we have 16 channels back at the capture. They're not all gonna be used for this So there'd be a subset of them used for obviously just use the ones that this end for too many so Excuse me. You'd like a way to tell whether your box is live so the led wouldn't be on so if you plugged in and doesn't work and the LEDs off That's it's a tip off and then the would wire the all the cables in a bundle come through here and obviously collect these cables at the same time So this notion of putting down a pzms and taking them away would somehow have to be turned into leaving them on the table Or right we want to do that that way right so and so the just epochs you down or something big screw into the table And even though there's eight cables they're not really very big around So my model is to get a sleeve piece of yeah that stuff that people put with You slip the wires into it's so to shape like that cross section. I'm not just leaving them all I'm going up and then going down\nSpeaker C: It's like a sleeping placement\nSpeaker A: Spade them who?\nSpeaker F: Yeah, it's like it's a sleeping loop and police and they're actually extruded from plastic. They sort of look like this\nSpeaker C: What does that mean so the wires go through here? Oh, it's a really cruel\nSpeaker G: Okay, so that so it basically goes on diagonal here. So why do we have 16 channels instead of like some fewer numbers?\nSpeaker F: Because\nSpeaker G: Well, okay, let me rephrase that why to each\nSpeaker C: Because then you don't have to just have one each so they have two fields in each other. Yeah, yeah Actually go to the same books\nSpeaker F: So this is really the way people sit on this table Okay Dot dot dot which means two at each station. Well, that's where people sit that's how many chairs in the room In turn, you could do a thing where all 16 were plugged in\nSpeaker G: If you had nothing else. Yeah, that night these and now he's the am's that right right agree only if you had\nSpeaker F: Well, it depends on this box, right true enough and actually my plan is to only bring eight wires out of this box This box is a one-off deal And it's function is just to essentially a wire converter to go from these little blue wires these black wires Plus supply power to the microphones because the the cheap head mount is all required\nSpeaker G: so you imagine some sort of In some sort of patch panel on top to figure out what the mapping was between each of these two and each of those\nSpeaker F: Well, I the simplest thing I can imagine which is really really simple is to quite literally decent plug-in It is a plug-in end of each of these these Eight cables and there's only there's only four slots that are you know in the first version or the version we're planning to build So that that was the whole issue with the LED that you plug it in the LED So I'm in your life now the subtle issue here is that you know, I haven't really figured out a solution for this or You're happy convention what happens if somebody in plugs is because they plug in more of something else well The there's no clever way to let the upstream guys know that you're really not being powered so There will be a certain amount of looking at cables half-married on if people\nSpeaker C: Rewire things. Yeah, I mean we had that last time But there are actually that you know there's an extra there's a mix out on the radio receiver Mm-hmm. So there are actually six XLR outs on the back the radio receiver and Only five cables going in they had the wrong five so I ended up not recording one of the channels and recording the mix\nSpeaker F: Oh, interesting. Did you do any recognition on the mix mix out? No, what were the works? I subtracted the four I did have from the mix and got pretty good cool and it worked. It did sound good\nSpeaker G: Science wonderful So so what's the schedule on your voice?\nSpeaker F: Uh, well, I was wrestling with with literally the number of conductors in the cable and the the Powering system and I I was gonna do this very clever fan and power and I decided a couple days ago not to do it So I'm ready to build it which is to say Averture of weak to get the circuit board done\nSpeaker G: So I think the other thing I'd like to do is do something about the setup so that it's a little more presentable and organized I agree and I'm just not sure what that is. I mean some sort of cabinet\nSpeaker F: Well, I can build the cabinet the difficulty for this kind of projects is the intellectual capital to design the cabinet In other words to figure out exactly what the right thing is that cabinet can go away It would use that for for a kindling or something But if you can imagine what the right form factory is Dan and I sort of gone around and this we were thinking about something Opened up in the top to allow access to the mixer for example But there's these things sticking out of the mixer which are kind of a pain so you end up with this thing that If you stuck to mixer up here in the top open to be fine you wouldn't necessarily well You just say when I'm yeah, you can start sketching it out and I can certainly build it out of oak no problem You know arbitrarily amount of I think the desk at home too\nSpeaker G: Is that gonna be a better solution than just going out and buy one well as we found out with the thing that\nSpeaker F: Jeff bought a long time ago to hold our stereosystem the stuff you buy is total crap And this is something you buy and and it's total crap Well, it's useless for this function works fine for holding a clean X right\nSpeaker G: Clean X and telephone right So yeah, I guess it's just a question is that something you want to spend your time on oh\nSpeaker F: I'm paid for great. I have no problem. No certainly one of the issues is is the Security I mean we've been been lacks been lacks and lucky yeah really lucky with these things But do not ours so the flat panels\nSpeaker G: Oh, yeah, I'm telling you I'm just going to cart one of them away if they stay there much longer let the record show\nSpeaker F: It's for four thirty five Adam Jane is so\nSpeaker G: Then the other question is do we want to try to do a user interface that's available out here\nSpeaker C: Sorry\nSpeaker G: Almost flip the bike to the interface mean do we want to try to get a monitor Oh something sure And how do we want to do that like CC meter readings just so we see something wow\nSpeaker F: How about use the thing that Series doing Which is to say just laptop with a wireless\nSpeaker C: Which will borrow from them\nSpeaker F: What's wrong with yours if we bought you a\nSpeaker G: Culture yeah, you could use my machine. Well, I have an ira machine. I borrowed I can use it no, I'm I'm serious. It does does the wireless thing work on your isn't that ethernet connection or is that phone?\nSpeaker C: That's an ethernet connection. No, no, I'm I ain't joking here. I'm serious. Yeah, no, absolutely\nSpeaker F: It's very convenient especially if Dan happens to be sitting at that end of the table They're not have to run down here and look at the thing every so often\nSpeaker C: We just haven't given that we've got a wireless\nSpeaker F: We've got it right there. Right? The antenna is right there. I don't know. I mean we obviously need to clear this with a Siri but Uh, how tough can it be there? You only needs web access isn't it?\nSpeaker C: In theory it x-axis, but I mean that's fine. That's what it does. Yeah\nSpeaker G: So right, so it's just a question of getting a laptop and a wireless mother no\nSpeaker F: No, he had my my proposal was you have a laptop, you know, I do yeah\nSpeaker C: Yeah, if we bought you the thing would you mind using it with it? No, I'd love to but I'm not sure if my laptop is compatible with the Waveland thing they're using really To do has the right thing right? I'm sorry for Apple has their own thing\nSpeaker F: And I thought it just came to a cereal or ethernet port. Yeah, I think what I think you I think it just plugs in a PC\nSpeaker C: So you could probably make it run with that the questions is there an apple driver? But Anyway, there are there are there are bunch of machines. I'd exceed it have those cards and so I think If if it doesn't we should be able to find a machine that does that I mean I know that doesn't Then the important people have those little well\nSpeaker F: That to me that's a whole nether. It's a whole nether issue the idea of convincing them that we should use their network It is fairly straightforward. Yeah, the idea of being able to walk into their office and say oh kind of bar your machine for a while Is is is a non starter? I don't think that's gonna work So I mean either we figure out how to use a machine somebody already in the group already owns Yeah, the idea is that it's a perk, you know, it's an advantage time or else we literally buy a machine Exactly for that purpose certainly it solves a lot of the problems with leaving a monitor out here all the time I'm not a big fan of doing things the room that make the room less attract your friend rather people right Which is part of the reason for getting all this stuff out of the way and And monitor sitting here all the time. You know people are gonna walk up to and go how come I can't get You know pong on this. I I borrowed the\nSpeaker G: I ran By EO Sony thingy right yeah, and I don't think they're ever gonna want it back You're kidding. Well the next conference they will sure yeah, but that does mean so we can use that as well\nSpeaker F: Well, the certainly you should give it a shot first see whether you can get compatible stuff Ask them what it costs ask if they have an extra one who knows they might have an extra hard work. I tried them at flat-femme display\nSpeaker D: Good what is the um, projectors supposed to be hooked up to\nSpeaker F: uh, the It's gonna be hooked up to all sorts of junk This gonna be actually a plug at the front that'll connect to people's laptops So you can walk it and plug it in and it's gonna be connected to the machine at the back So we certainly could use that as as a constant Reminder what the view meters are doing so people are gonna be going\nSpeaker A: Testing one two\nSpeaker D: I mean, that's another that's another possibility Yeah, you know\nSpeaker F: But but but I think the idea of having a control panel it's it's there in front of you. It's really cool. Yeah\nSpeaker C: And I have a\nSpeaker F: As long as as long as you're not tempted sit there and keep filling with the volume controls\nSpeaker G: I'd actually earlier asked if I could borrow one of the cards to do wireless stuff and they said sure whenever you want\nSpeaker F: So I think it won't be a problem. And it's a PCMCA card, right? Yeah, PC car. So you have a slot, right? Yeah, you know machine\nSpeaker G: Is it really concerned with the driver? Yeah, and it has doesn't work as I said we can use the PC it'll work the first time\nSpeaker F: I trust you jobs Good\nSpeaker G: Um, well, so it's like a good installation on my day. I thought so James is gonna be doing wiring and you're gonna give some thought to cabinets\nSpeaker F: Yeah, we we need to figure out what we want We what is green lights doing\nSpeaker C: flashing uh-oh, does that means the minutes can explode?\nSpeaker F: kept the red wire the red wire\nSpeaker G: um When people talk at this\nSpeaker C: So again, Washington wants to equip a system our system we spent $10,000 on equipment not including the PC However, seven and a half thousand of that was the wireless mics uh-uh using the five thousand for the words of I'm gonna do To joke yeah, that's true. Well, we haven't spent that right but once we once we've done the intellectual part of these So we can just knock them out right we can start oh the board. Yeah, yeah, sure, right And then we could Washington can have a system that didn't have any wireless but it had was based on these and it would cost P-nuts a PC and a peanuts PC and $2,000 for the ATD stuff. Yeah, that's bad because you wouldn't even need a mixer if you didn't have the Oh, PCM's PCM's cost a lot but anyway you'd save on the seven seven eight thousand for the wireless system. So actually that Might be attractive Right\nSpeaker B: I think it's nice it's nice to be thinking toward like if we talk software the disc last longer\nSpeaker G: Well actually shorten yeah There's a speech compression program that works great on things like this because if the dynamic range is low it encodes it with fewer bits And so most of the time no one's talking so it shortens it dramatically But if you talk quieter the dynamic range is lower and it will compress better Yeah, yeah\nSpeaker E: It also helps if you talk in a monotone Probably I'm fine all the time\nSpeaker B: And shorter words shorter words\nSpeaker D: Now sure words wouldn't would induce more dynamics right\nSpeaker C: You want to let's a little predictable. How about if you just go\nSpeaker G: Oh How do you spell that Okay, can you do one more round of digits are we done talking?\nSpeaker F: Well, if the choice if we get a choice, let's keep talking\nSpeaker G: Sure, I have more to talk about now I'm done. I'm done. I'm done\nSpeaker F: But you you there's a problem structural problem with this though You really need an incentive at the end if you're gonna do digits again like you know candy bars or something or I'll remember to bring them next time or a little You know toothbrushes like to give you the dentist\nSpeaker G: Saddam Jenner on the wireless lapel my number one transcript 531 550 2 307 9 4 503 8 5 2 3 7 1 0 8 3 2 7 8 6 Excuse me 8 3 2 7 8 2 6 9 4 2 9 1 0 6 2 9 3 0 1 2 0 6 6 4 1 6 7 4 7 8 5 5 1 6 8 2 6 5 7 7 8 8 8 9 6 5 3 9 Oh 1 2 9 2 3 7 8 6\nSpeaker C: This is Dan Ellis transcript 5 5 1 5 70 359 3 4 6 5 6 7 0 6 9 6 7 6 1 6 9 2 0 0 4 5 6 0 4 2 0 2 Sorry 0 4 0 2 1 8 7 0 2 3 0 4 6 1 2 0 7 8 1 7 1 7 1 0 1 4 8 4 0 1 6 3 6 9 7 3 7 0 0 0 0 1 0 0 1 3 3 6 9 7 9\nSpeaker B: This is Jena Edwards on microphone number 5 transcript 6 1 1 6 3 0 5 0 6 4 4 7 1 6 0 4 8 1 8 4 3 6 0 9 7 8 3 0 0 0 1 0 1 0 1 1 2 9 6 2 0 0 4 5 4 1 1 4 9 5 3 4 6 4 9 3 0 7 9 1 8 8 7 6 0 9 0 0 1 2 8 2 3 4 1 1 3 6 5 3 4 9 9 4 5 5 0 5\nSpeaker D: This is Eric Fossilus here on microphone number 3 reading transcript 5 91 to 6 10 5 3 3 7 0 7 7 6 4 2 7 6 1 9 1 8 9 5 0 9 9 0 0 7 0 2 2 3 5 4 4 3 0 5 8 0 6 7 8 0 1 8 0 3 5 7 1 6 9 6 2 8 9 2 4 0 2 3 4 0 7 6\nSpeaker F: Jim Beck transcript 5 7 1 5 9 0 Channel 4 wireless 4 15 45 6 6 7 8 9 0 1 8 0 2 6 5 1 4 6 3 9 2 5 2 39 4 3 0 0 8 4 5 6 0 5 0 8 8 4 9 4 2 3 0 7 3 4 0 9 3 2 1 9 3 2 3 Eric you and I win we didn't make any mistakes it's harder at the end we don't know that for sure do we\nSpeaker G: I should have mentioned that's a to pause between no I know I've just given you a hard time it's only a hard time for the transcriber not for the speech\nSpeaker B: I also think you said channel 4 and I think you met microphone and I think that's a mistake\nSpeaker F: very good so Eric you win but the other thing is that there's a there's a colon for transcripts and there shouldn't be a colon because you have everything else to stuff you fill in\nSpeaker C: yeah that's been filled in for you great right man but in order they start 6 7 8 9 0 1 2 3 4 5 6 8 9 where did it come from and they're in order because they're sorted lexically by the file names which are have the numbers and digits and so they're actually this is like all the all utterances that were generated by speaker and pj or something and then within mpj they're sorted by what he actually said oh that I should have randomized it but it doesn't matter it's like because you said 6 7 8\nSpeaker F: but we think it doesn't matter but the real question I have is why bother with these why don't you just ask people to repeat numbers they already know like phone numbers you know social security numbers we have these written down\nSpeaker G: because if we have it uh\nSpeaker F: you can go ahead and transcribe credit card numbers you don't have it you don't have it as you just say your credit card numbers say your phone numbers say your mother's name\nSpeaker G: here we could do it pass we're do your count people off the street all right actually this I got this directly from another training set from a rrr so you can compare directly\nSpeaker C: looks good looks like the one that was was the one that when they do our driver errors great good okay might stop it yeah okay it's just the\n"}, {"meeting_id": "ES2006a", "summary": "The meeting was mainly about the exchange of the preliminary ideas about a new TV remote control project. They first got used to the whiteboard and drew some animals with it. Project manager wanted to sell this remote control for 25 Euro and expected profit will be around fifty million Euro. Afterwards the team raised different opinions about function and fashion features of the new remote controls . In particular, they emphasized the fashion of the remote control should be combined with the function, and the design of the battery should be original and not far away from the conventional design.", "dialogue": "Speaker C: Okay, so this is our first meeting of this design project.\nSpeaker C: And I'd like to show you the agenda for the meeting.\nSpeaker C: I don't know if it was sent round to all of you.\nSpeaker E: I didn't rush away.\nSpeaker C: This is the plan for today's meeting is firstly just to introduce the project briefly, although I'm sure you've actually got some of the information already.\nSpeaker C: Then the main purpose is to so that we get to know each other a little bit more.\nSpeaker C: Then we want to practice using some of the tools that we'll be using during the course of the design project and the meetings, specifically the whiteboard over there.\nSpeaker C: Then we need to go through the specifics of our project plan and discuss, come up with some preliminary ideas about it.\nSpeaker C: And then that's it.\nSpeaker C: So we've got 25 minutes to do that.\nSpeaker C: That's until 11.25.\nSpeaker C: So any questions?\nSpeaker C: Not at this point.\nSpeaker C: So this is our project.\nSpeaker C: What we're aiming to do is to create a new remote control for a television.\nSpeaker C: We want it to be something original, something trendy and also something user friendly.\nSpeaker C: So it has to be quite intuitive that people are able to use this product.\nSpeaker C: The method that we're going to use to complete the project, it has three components as such.\nSpeaker C: There's the functional design of the remote control.\nSpeaker C: The way we'll do that I think is to work individually and then come together from meetings to work on that.\nSpeaker C: Similarly, with the conceptual design, we'll start off by working individually with our own expertise on our own laptops.\nSpeaker C: Then we'll bring what we've done together.\nSpeaker C: And then the detailed design will come after that.\nSpeaker C: We'll pull it all together.\nSpeaker B: I'm a bit confused about what's the difference between the functional design and conceptual design.\nSpeaker B: Is it just more detail?\nSpeaker C: I think we're talking that the functional design is more your area of things where you'll be, we want to look at what functions we need in the remote control and what specific things it has to do.\nSpeaker C: But the conceptual design is perhaps bigger than that and includes the how people are going to use it and that kind of thing.\nSpeaker B: Where do we identify the components of our product?\nSpeaker B: I think it's in the conceptual design phase that we identify the components of our product.\nSpeaker C: I think we'll start initially with the functional design already.\nSpeaker C: That's just a brief overview of the project itself.\nSpeaker C: What I'd like us to do now is simultaneously introduce ourselves and start using some of the tools that we're using for the project.\nSpeaker C: Specifically, the whiteboard.\nSpeaker C: Each person in turn, I'd like us to go up to the whiteboard that pens just underneath it there and draw your favorite animal and then tell everyone what your favorite characteristics of that animal are.\nSpeaker C: While you're doing that, tell us your name, what your role is and perhaps how your animal relates to the role that you're taking in this project.\nSpeaker C: How are you looking at me?\nSpeaker C: Would you like to go first?\nSpeaker D: Do I have a choice?\nSpeaker D: Okay.\nSpeaker D: Oh, things falling everywhere.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: Put the in-pockets.\nSpeaker C: Cool.\nSpeaker C: You don't have to hurry, we've got plenty of time.\nSpeaker D: So, my name's Kat, and I'm really not very good at this whole drawing, but I'm not sure.\nSpeaker D: Okay.\nSpeaker D: That's got no eyes.\nNone: Oh, good thing.\nNone: Oh, the eyes always ruin it.\nSpeaker D: Like, okay, what do you want?\nNone: Sorry.\nNone: Okay.\nSpeaker D: Cool.\nSpeaker D: This is a rabbit.\nSpeaker E: Yeah, well, originally at first I thought it was going to be Kat.\nSpeaker D: Yeah, I didn't think it's furry enough.\nSpeaker E: No, no, I understand.\nSpeaker E: I can see by the ears.\nSpeaker D: Yeah.\nSpeaker D: Okay, well, it's a fluffy rabbit.\nSpeaker D: Blue, rabbit is saying come in blue, but you need.\nSpeaker D: Okay, and I like it because it's small and it's fluffy.\nSpeaker D: And one day you'll be able to genetically modify them and they will come in pink.\nSpeaker D: Oh.\nSpeaker D: Okay, excellent.\nSpeaker C: And what's your role within the team?\nSpeaker D: I am the...\nSpeaker D: I need my notebook.\nSpeaker D: Oh.\nSpeaker D: Top banana.\nNone: Thank you.\nSpeaker D: Okay, cool.\nSpeaker D: I am the marketing expert.\nSpeaker D: So, like, I'm going to be doing the apparently according to the little guy on the computer that knows everything.\nSpeaker D: The user requirements specification of the functional design, trend watching in the conceptual design and product evaluation in the detailed design.\nSpeaker D: Okay.\nSpeaker D: So, yeah.\nSpeaker D: And more about yourself, you're from...\nSpeaker D: I'm from Leicester.\nSpeaker D: Second year.\nSpeaker D: What else do you want to know?\nSpeaker D: I like sport.\nSpeaker D: Yeah, aerobics, kickboxing, spinning.\nSpeaker D: But not with rabbits.\nSpeaker D: Not with rabbits, no, no.\nSpeaker D: And vets.\nSpeaker D: I like vets as well.\nSpeaker D: And yeah, and I like cocktails, especially pink ones.\nSpeaker D: Okay.\nSpeaker C: Thanks a lot to match the rabbit.\nNone: Okay.\nSpeaker E: So, my name is Marika.\nSpeaker E: Where's the pen?\nSpeaker E: Okay.\nSpeaker E: There's a...\nSpeaker C: And if you have not enough room, there's a new razor there and you can rub it.\nSpeaker C: Yeah.\nSpeaker E: Or I can make it smaller.\nSpeaker E: So, I'm the interface designer in this project.\nSpeaker E: And my favourite animal, I'm not so sure because I'm not so very familiar with all kinds of animals.\nSpeaker E: But I do like dogs.\nSpeaker E: Oh, sorry.\nSpeaker E: Maybe I shouldn't have said it beforehand.\nSpeaker E: Hmm.\nSpeaker E: There are different kinds of dogs.\nSpeaker D: Okay.\nNone: There.\nSpeaker A: That's so powerful.\nSpeaker A: It looks like a dog.\nSpeaker D: Okay.\nSpeaker D: It's a bit more impressive than one rabbit.\nSpeaker D: I think it needs for like, if it's going to walk.\nSpeaker E: Yeah, maybe it has some colorful patches.\nSpeaker E: Yeah.\nSpeaker E: And I do like dogs because they are good friends.\nSpeaker E: Two people and they are loyal.\nSpeaker E: Well, as compared to small animals like cats, they are really much more fun because they are not so independent.\nSpeaker E: Yeah.\nSpeaker E: Maybe the fact that they protect their home as well.\nSpeaker E: What it has to do with my role in the project is hard to say.\nSpeaker E: I hope to be loyal to the project and not to let people doing similar projects know the details of our project or something.\nSpeaker C: Yeah.\nSpeaker C: Where are you from?\nSpeaker E: I'm from Estonia.\nSpeaker E: Yeah.\nSpeaker E: So, is there anything else you'd like to know?\nSpeaker E: Oh, right.\nSpeaker E: My role.\nSpeaker E: So, in the different stages of the design.\nSpeaker E: So, at first, I will be responsible for designing the technical functions of the remote control.\nSpeaker E: Then in the conceptual design stage, I need to come up with interface concept.\nSpeaker E: And then in the last stage, I will be responsible for the interface design.\nSpeaker E: Okay.\nSpeaker E: Thank you.\nSpeaker E: Okay.\nSpeaker A: Do some.\nSpeaker A: I'll rub the features.\nSpeaker A: I'll rub the drawings too.\nSpeaker B: My name is Gorov.\nSpeaker B: My favorite animal.\nSpeaker B: One of my favorite animals is a cow.\nSpeaker B: I've got no idea how to draw a cow.\nSpeaker C: Good luck.\nSpeaker B: This is going to be...\nSpeaker D: They're not just like a big round body and then some really skinny legs and just some horns.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So, let me draw the body first.\nSpeaker B: Big round body.\nSpeaker B: It's like a big legs.\nSpeaker B: And then put a long tail on a long face.\nSpeaker D: It's like eating.\nSpeaker D: It looks like you are.\nSpeaker B: There is some grass there.\nSpeaker B: This is what I like about...\nSpeaker B: It keeps sitting there eating grass that do not disturb anybody.\nSpeaker B: They're kind of put this in a way.\nSpeaker B: So, yeah.\nSpeaker B: I like cows.\nSpeaker B: My role in the project is industrial designer.\nSpeaker B: I'm supposed to design all the details of the product, how it works and whatever it will take during the functional role, what are the various functions that have to be performed by it during the conceptual design, what are the various components of it.\nSpeaker B: And finally, I'm not sure what was the last part.\nSpeaker B: The detailed design, I guess it will again be the identification of the components and how they integrate with each other.\nSpeaker B: I'm from India.\nSpeaker B: I'm doing my PhD in the cycle linguistics.\nSpeaker B: I set up the department of psychology.\nSpeaker B: Yeah.\nSpeaker E: Thank you.\nSpeaker E: Thanks.\nSpeaker C: Right now it's my turn.\nSpeaker C: Obviously.\nNone: Okay.\nNone: This is a little bit of color.\nNone: It looks really, really cute.\nNone: Yeah.\nNone: I like the cow.\nNone: I'm Jen.\nNone: Mm-hmm.\nSpeaker C: I like dogs too, but I can't do that already because I can't do a dog as well as you can.\nSpeaker C: I like...\nSpeaker C: Mm-hmm.\nSpeaker C: I like dogs too, but I can't do that already because I can't do a dog as well as you can.\nSpeaker C: I like...\nSpeaker C: Mm-hmm.\nSpeaker C: I like...\nSpeaker C: Mm-hmm.\nSpeaker A: Is that a laser?\nNone: No way.\nNone: Mm-hmm.\nNone: I like dogs too.\nSpeaker B: I like dogs too.\nSpeaker B: Mm-hmm.\nSpeaker E: Wow.\nSpeaker C: It's a gecko.\nSpeaker C: Oh, a gecko.\nSpeaker C: Okay.\nSpeaker D: Yeah.\nSpeaker D: Is that a difference?\nSpeaker E: Are they also like lizards or are they?\nSpeaker C: Yeah, they are kind of lizard.\nSpeaker C: Mm-hmm.\nSpeaker C: And I like geckos because they remind me of warm places.\nSpeaker C: And where I was living in Cambodia, they used to live in my house and they were on the ceiling and they would make little gecko noises.\nSpeaker E: I hope you don't like snakes, do you?\nSpeaker C: I don't like snakes.\nSpeaker C: Okay.\nSpeaker C: I go straight home and we have nasty snakes.\nSpeaker C: That's where I'm from.\nSpeaker C: Oh, it's really a...\nSpeaker C: I'm from Melbourne.\nSpeaker C: And I'm your project manager for today.\nSpeaker C: And my role is basically to keep things going and make sure that you all work together in a productive way so that by the end of the day we come up with a great product.\nSpeaker C: Wonderful.\nSpeaker A: Okay.\nSpeaker A: Thank you.\nSpeaker C: So let's see what's next in the PowerPoint presentation.\nSpeaker C: Thank you.\nSpeaker C: I just thought about this that we could even put it much more proficient.\nSpeaker C: There we go.\nSpeaker C: Okay.\nSpeaker C: So this is the overall budget for our project.\nSpeaker C: We've got...\nSpeaker C: We're planning to sell these remote controls for...\nSpeaker C: Let's make that go away.\nSpeaker C: That means we've got five minutes.\nSpeaker C: We're planning to sell the remote controls for 25 euros each.\nSpeaker C: And with that we're aiming for a profit of 50 million euros.\nSpeaker C: And that's selling them on the international market, not just in the UK.\nSpeaker C: So to do that our finance people estimate that we need production costs of maximum 12.5 euro so that we can reach that profit target.\nSpeaker C: So that's something to keep in mind while we're designing.\nSpeaker C: Okay.\nSpeaker C: Let me just skip ahead to say that's the last thing.\nSpeaker C: Okay.\nSpeaker C: We've only got a couple of minutes.\nSpeaker C: Does anyone have any first ideas to bounce around about what we're thinking of this remote\nSpeaker E: control? I'm just wondering whether there is like any special feature that we want to have...\nSpeaker E: One, this remote control have as opposed to the already existing ones.\nSpeaker C: I think that's probably something that it's best if we take away with this.\nSpeaker C: But if we all have a think when we go away from the meeting, what specific things could be included in this remote control.\nSpeaker B: I think in the beginning one thing that was mentioned was that it should be trendy, user friendly and original.\nSpeaker B: So I think your point is relevant as far as the originality is concerned that we should provide some features that are quite unique to this.\nSpeaker D: Something new.\nSpeaker D: Yeah, I was looking at the website and the other things that they've made and I put down to inspirational words that I got from looking at the pictures.\nSpeaker D: So the motto is we put the fashion in electronics.\nSpeaker D: And so it's something that is sleek and stylish but is still functional.\nSpeaker D: So I'm kind of thinking, you know, like those phones that they have, the new generation ones, where they don't actually have any buttons on them and stuff like that.\nSpeaker D: It's not right.\nSpeaker D: So it's something heading towards that so it's not overly...\nSpeaker D: I mean, I don't know what almost the button do or my remote controls, but I figure how many do you need, do you need?\nSpeaker C: So perhaps some sort of menu based thing or something that's a little less crowded than this.\nSpeaker D: I mean, theoretically you can do all kinds of things with your TV, right?\nSpeaker D: But what do most people do? They turn it on, they watch certain specified channels, you know, and then they turn it off again.\nSpeaker B: There is a lot of functionality in there that is not used 90% of the time, but will be used 10% of the time.\nSpeaker D: There's no need to have buttons on it to do that.\nSpeaker C: Maybe it could be one button for a menu or something.\nSpeaker C: And then use that.\nSpeaker D: So if you're the kind of side case that knows how your remote control works, then, you know, that's fine and you can do it on the screen rather than everybody else having to have those buttons which is confusing.\nSpeaker E: Excellent.\nSpeaker D: Because if you look at the train, it's just very like, there's no extra bits on it, the train on the website.\nSpeaker D: I don't know if you can put it up on the thing, but it's just like a long thing used to moving people, but it looks really pretty too.\nSpeaker C: Great.\nSpeaker C: Any other immediate thoughts?\nSpeaker C: Good problem.\nSpeaker B: We can aim for, I mean, we can think about all these little things, but we can aim for something that gives a high battery life.\nSpeaker B: Although I don't think that it's a huge problem for remote controls and in the battery life every now and then you need to replace the batteries.\nSpeaker B: Yeah.\nSpeaker E: But I mean, even though it has to be original, we shouldn't go too far away from the usual ones because otherwise the new users will just have a lot of problems with learning.\nSpeaker B: Yeah.\nSpeaker B: So it should kind of fit in as well.\nSpeaker B: It's like this fancy website.\nSpeaker C: It's a teeter type of a...\nSpeaker C: You can't access because you have no idea how to get in, but the design is not very great.\nSpeaker C: Okay. So we need to wrap it up now so that we can go away and get on with some of this.\nSpeaker C: We've got another meeting in 30 minutes.\nSpeaker C: So you'll be getting specific instructions once you go back to your workspace.\nSpeaker C: Basically, you're looking at the working design.\nSpeaker C: You're looking at the technical functions design.\nSpeaker C: And for you, it's the user requirements specification, like I said at the start.\nSpeaker C: Okay.\nSpeaker C: Okay. Thanks for that.\nSpeaker C: Thank you.\nSpeaker C: I'll see you in half an hour.\nSpeaker C: See you.\nSpeaker C: Carry the laptops back again.\nSpeaker C: Do we need to unplug things?\nSpeaker C: Okay.\n"}, {"meeting_id": "ES2007d", "summary": "The meeting began with a presentation on the prototype by Industrial Designer and User Interface and a double curved rubber prototype with anti-RSI buttons was displayed. Next, they discussed the cost of the remote which only cost ten seventy euros a unit and well under the budget of twelve fifty. Then, the team evaluated the remote according to seven criteria and it passed with flying colours as three criteria scored a one and others no more than three. Finally, the meeting ended with team building and a wrap-up, which Project Marketing concluded that the novel remote fit all criteria and the team did a great job.", "dialogue": "Speaker C: Play those edible. Did you know that?\nSpeaker C: It's definitely...\nSpeaker B: Because kids...\nSpeaker D: I used to.\nSpeaker D: I've definitely eaten it before. I didn't know it was edible.\nSpeaker D: But it's... it's shoe proof.\nSpeaker B: Yeah.\nSpeaker B: It's my edible because kids eat it.\nSpeaker B: I don't know if it wasn't edible, but...\nSpeaker B: Well, no more babies.\nSpeaker D: Actually, that makes sense because I remember like...\nSpeaker D: I don't know if my mom ever did it.\nSpeaker D: I remember other people's moms making.\nSpeaker D: I played over. You just like make the...\nSpeaker D: You can use coloring and mix and stuff.\nSpeaker D: Oh yeah.\nSpeaker B: Oh yeah, it is. Yeah.\nSpeaker B: Oh yeah.\nSpeaker C: All right, everybody ready?\nSpeaker C: Yep.\nSpeaker C: Okay, let's have your...\nSpeaker C: Let's get...\nSpeaker C: Let's have the presentation.\nSpeaker D: We've got a cool presentation.\nSpeaker D: It's pretty exciting.\nSpeaker D: So...\nSpeaker D: Everything that we wanted, we wanted it to be ergonomic and to be...\nSpeaker D: Made out of rubber.\nSpeaker D: Very simple and easy to use.\nSpeaker A: Double curved.\nSpeaker D: Yeah, double curved.\nSpeaker D: But also something that was going to jump out at people, something that would be different.\nSpeaker D: Separate it from the other...\nSpeaker D: Remotes out on the market.\nSpeaker D: So...\nSpeaker D: I think if you put this in the palm of your hand...\nSpeaker D: You'll see what a nice thing we have going in.\nSpeaker D: That is cool.\nSpeaker D: So basically...\nSpeaker D: If you hold it like that, the one on your thumb...\nSpeaker D: The thumb button is the power button.\nSpeaker D: Index finger is channel up.\nSpeaker D: Middle finger is channel down.\nSpeaker D: Ring finger is volume up.\nSpeaker D: Pinky is volume down.\nSpeaker B: What's the big blue thing?\nSpeaker D: That's the lock button.\nSpeaker D: And then the M is a mute button.\nSpeaker D: And then it also has...\nSpeaker C: So what button?\nSpeaker D: M for muting the...\nSpeaker D: And also there's a numeric keypad on the top so you can key directly to the channel if you want.\nSpeaker D: So it's really basic functionality as far as what keys are available.\nSpeaker D: But I think it's really comfortable and innovative and it looks different.\nSpeaker C: That certainly does.\nSpeaker D: So all the important keys are right at your convenient place for you to access them.\nSpeaker A: And T-R-S-I-M.\nSpeaker D: It should be...\nSpeaker D: And it's also conformable to the size of your hand.\nSpeaker D: I mean if that's too big, it's a rubber remote.\nSpeaker D: It's so cute.\nSpeaker D: Is that...\nSpeaker D: What... I'm not sure what you guys were hoping for and expecting.\nSpeaker C: I have one thing about it.\nSpeaker C: But it's a small thing but it means we'd have to make a right-handed one and a left-handed one.\nSpeaker C: How about it?\nSpeaker D: That's good thinking.\nSpeaker C: But that's... I don't see why that's not possible.\nSpeaker D: If we build rocket ships, why can't we build rocket ships?\nSpeaker D: Yeah.\nSpeaker C: It might left-handed scissors.\nSpeaker D: Yeah, but then...\nSpeaker D: I didn't think about that.\nSpeaker B: But then you can learn to use your right hand.\nSpeaker B: Like I was just thinking there's left-handed and right-handed in the family.\nSpeaker B: What do they have to do?\nSpeaker A: I know people who have left-handed and right-handed people in the family and they all use the computer for the whole...\nSpeaker A: The same computer for the first family and they have a mouse.\nSpeaker A: Sure.\nSpeaker B: Everybody using right-handed.\nSpeaker B: I know they'll be able to... I mean it's only pressing buttons.\nSpeaker B: You don't have to do anything, you know, extraordinary.\nSpeaker B: I think everybody can press a button within the left-handed and right-handed.\nSpeaker A: Imagine... are you right-handed?\nSpeaker A: Yeah.\nSpeaker A: Imagine doing it with your left hand.\nSpeaker A: I don't think it's too...\nNone: Yeah, it's not.\nSpeaker A: But we can have both...\nSpeaker B: Yeah.\nSpeaker B: Have them in stock.\nSpeaker B: Yeah.\nSpeaker B: They can make them more appealing.\nSpeaker C: But other than that, I only know.\nSpeaker C: That's just something I think.\nSpeaker C: I think it's great. Yeah, great idea.\nSpeaker D: Do you think it says...\nSpeaker C: R-I?\nSpeaker C: R-I think it's a...\nSpeaker C: I think it's... well, if the R-I motto is, we bring fashion to the electronics.\nSpeaker C: We like to make sure that it could be quite fashionable.\nSpeaker D: Yeah.\nSpeaker D: And it's got the black and yellow and blue.\nSpeaker D: Plus red, which is sort of a fruit and vegetable.\nSpeaker D: That's how it is.\nNone: Yeah.\nSpeaker D: So that's our inner things.\nSpeaker D: Yeah, very good.\nSpeaker C: Yeah, it's come up with the thing that's...\nSpeaker C: what we were looking at doing, wasn't it?\nSpeaker C: Yeah.\nSpeaker C: The little seems to be there.\nSpeaker C: Well done.\nSpeaker B: The little playing around is...\nSpeaker C: Before we move on...\nSpeaker B: Yeah, I'm just... do you want a bucket?\nSpeaker C: I need that. Okay.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: One thing I do need to do, we need to look at is the costs.\nSpeaker D: The costs, is that what you said?\nSpeaker A: The product was very cheap.\nSpeaker C: Well, yeah.\nSpeaker C: I don't want to last very long.\nSpeaker B: Everybody will go like, oh, this is gone.\nSpeaker D: But it's edible.\nSpeaker A: Mm-hmm.\nSpeaker A: Cheap poop.\nSpeaker B: Well, buy more of the meat.\nSpeaker D: That was the main criteria from the last meeting.\nSpeaker D: It had to be cheap-proof.\nSpeaker C: Yeah.\nSpeaker C: Right.\nSpeaker C: Okay, now I think we'll do this.\nSpeaker C: Yeah.\nSpeaker C: I could do, you know, I can do this.\nSpeaker C: Oh.\nSpeaker C: Oh, my, I could do it with you.\nSpeaker C: But it's just easy enough to go through it with you.\nSpeaker C: So, we're going for the...\nSpeaker C: Kinetic power.\nSpeaker C: Yep.\nSpeaker C: And the electronics we decided on it.\nSpeaker C: Being just a simple, the easiest thing that's inside it.\nSpeaker C: Oop.\nSpeaker C: The case we've gone for the double-curved.\nSpeaker C: And it's made out of rubber.\nSpeaker C: The interface is push buttons.\nSpeaker C: And buttons supplements.\nSpeaker C: Well, they're different special colors, aren't they?\nSpeaker C: Yeah.\nSpeaker C: So, very special colors.\nSpeaker B: Special form.\nSpeaker C: Special form.\nSpeaker C: Yeah, they're special form.\nSpeaker C: They're in shapes and stuff.\nSpeaker C: Yeah, I mean, these ones on the side are curved.\nSpeaker C: Yeah.\nSpeaker C: They're made out of any special material.\nSpeaker C: No, they're not made out of wood or titanium or rubber.\nSpeaker C: They're just simple.\nSpeaker C: Well, they're rubber.\nSpeaker C: Okay.\nSpeaker C: So, let's see if that comes within budget.\nSpeaker C: And it does.\nSpeaker C: That is going to cost...\nSpeaker C: I don't know about it.\nSpeaker C: Yeah, that's going to cost 10, 10 euros, 70 cents a unit to make.\nSpeaker C: And our target was it had to come in at under 12.50.\nSpeaker C: So...\nSpeaker D: Actually, making a better profit than the expected.\nSpeaker C: This is all very, very good.\nSpeaker C: The bosses will be very pleased.\nSpeaker C: Okay, let's just say this.\nSpeaker C: So, I can email it to me.\nSpeaker C: So, it's very dynamic.\nSpeaker C: No, I want to...\nSpeaker C: It's so...\nSpeaker C: I'm not...\nSpeaker C: I'm not...\nSpeaker C: I'm not...\nSpeaker C: I'm not...\nNone: I'm not...\nSpeaker C: Spended.\nSpeaker C: Okay.\nSpeaker C: So, that's...\nSpeaker C: That's me done with this...\nSpeaker C: with this...\nSpeaker C: do-da.\nSpeaker C: I'm going to do what you were going to do.\nNone: No, evaluation.\nSpeaker A: Oh, yeah.\nSpeaker C: This is where we all get to.\nSpeaker B: I get to write on the board.\nSpeaker B: Oops, on the board.\nSpeaker B: Right.\nSpeaker B: Oh.\nSpeaker B: It's functioned.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: Oh, I love this man.\nNone: I played on...\nNone: Oh, I'm not...\nNone: Oh, I'm not...\nSpeaker C: Oh, I'm not...\nSpeaker C: Oh, I love this man. I played on...\nSpeaker C: I can't...\nSpeaker C: I can spell it.\nSpeaker B: Okay.\nSpeaker B: So, evaluation.\nSpeaker B: I'm going to do it all together.\nSpeaker B: So, we evaluate it.\nSpeaker B: Each criteria I've got the criteria is.\nSpeaker B: And we have to do it on a scale of one to seven.\nSpeaker B: One being true.\nSpeaker B: So, it's more like it fits the criteria.\nSpeaker B: And seven being as in it doesn't fit the criteria.\nSpeaker B: And the criteria is not...\nSpeaker B: And I'll draw this up on the board.\nSpeaker B: So, we have a box.\nSpeaker B: Just so we know true.\nSpeaker B: And this is false.\nSpeaker B: This is just like...\nSpeaker B: Keep you informed.\nSpeaker B: So, seven is here.\nSpeaker B: And one's here.\nSpeaker B: And then you've got in the middle.\nSpeaker B: So, the first criteria...\nSpeaker B: Do you all get what we're doing?\nSpeaker B: Okay, cool.\nSpeaker B: Okay, first criteria.\nSpeaker B: Look and feel.\nSpeaker B: So, it does look and feel fashionable to what we talked about.\nSpeaker B: Is it color-wise?\nSpeaker B: And is it spongy?\nSpeaker B: So, what mark should we give it for that?\nSpeaker D: I would give it a seven.\nSpeaker D: Yeah.\nSpeaker B: As in, it's not.\nSpeaker D: Oh, sorry.\nSpeaker D: One.\nSpeaker B: One.\nSpeaker B: Yeah.\nSpeaker B: So, I'll just write criteria.\nSpeaker B: Criteria one.\nSpeaker B: Criteria one.\nSpeaker B: Second criteria.\nSpeaker B: New technology.\nSpeaker B: How we implemented new technology within the new high tech.\nSpeaker D: Yeah.\nSpeaker D: That was our main technological innovation.\nSpeaker D: Everything else is fairly simple, but the fact that we use the kinetic energy.\nSpeaker D: Also the...\nSpeaker D: It's ergonomic.\nSpeaker D: Yeah, but that's not a technological thing.\nSpeaker C: That's another thing.\nSpeaker C: That's another marketing thing.\nSpeaker C: So, on the technical side of it.\nSpeaker C: I'd say it's about a two-ish.\nSpeaker D: In the middle somewhere, maybe.\nSpeaker B: Three.\nSpeaker D: Yeah.\nSpeaker D: Three, yeah.\nSpeaker B: Got a three.\nSpeaker B: So, criteria three is...\nSpeaker B: It's easy to use.\nSpeaker B: I think it's a one.\nSpeaker C: I'd say it's...\nSpeaker C: I've not, if you're left-handed, it's not.\nSpeaker C: I would give it a two.\nSpeaker C: Two, so...\nSpeaker C: Because it is more geared for right-handed people than left-handed people.\nSpeaker B: But if we make a right-handed and left-handed then...\nSpeaker D: If we're going to have one left-handed and one right-handed, then I would give it a one.\nSpeaker D: Yeah.\nSpeaker D: Give it a two.\nSpeaker C: Yeah, okay.\nSpeaker B: Okay, criteria four is...\nSpeaker B: Cost.\nSpeaker C: I mean, on the budget.\nSpeaker C: Yeah, it's a definite one.\nSpeaker C: That was great.\nSpeaker B: Yeah.\nSpeaker C: The amount of buttons...\nSpeaker B: Like the amount of buttons, because people like a lot of letters.\nSpeaker B: Only the necessary buttons.\nSpeaker B: So it's a one?\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Um...\nSpeaker B: Criteria six.\nSpeaker B: RSI is a good against.\nSpeaker B: Yes.\nSpeaker B: Very good.\nSpeaker B: It's an NC RSI.\nSpeaker B: It's one.\nSpeaker B: And criteria seven, which is the last one.\nSpeaker B: Does it get lost?\nSpeaker B: It's easy to get lost.\nSpeaker C: I don't think it's going to get lost very freely.\nSpeaker B: No.\nSpeaker B: But it is flawless.\nSpeaker D: It's not the kind of thing that's going to slip like between a couch cushion or something.\nSpeaker D: Maybe it was.\nSpeaker D: I think it would be good.\nSpeaker D: It was a two.\nSpeaker B: Yeah.\nSpeaker B: I mean, it's not fully, it's not fully like, can't say.\nSpeaker B: I mean, it's not a one.\nSpeaker C: I mean, you can still flush it down a toilet theoretically.\nSpeaker D: Yeah, anything.\nSpeaker D: I mean, it's bigger than the average mobile, I guess.\nSpeaker B: Yeah.\nSpeaker D: But yeah, it can get lost.\nSpeaker B: The mobile's getting lost all the time.\nSpeaker B: Okay.\nSpeaker B: But then you ring them.\nSpeaker B: So.\nSpeaker B: Okay.\nSpeaker B: So that's that.\nSpeaker B: That's the evaluation.\nSpeaker B: So I'd say.\nSpeaker B: Yeah.\nSpeaker C: Oh, it's like, oh, system goes.\nSpeaker B: Like a number one.\nSpeaker B: Number one product.\nSpeaker C: We can't fail.\nSpeaker B: Fitted all the criteria.\nSpeaker B: Yeah.\nSpeaker C: Okay.\nSpeaker B: Yeah.\nSpeaker C: Well done, Lisa.\nSpeaker B: That's that one.\nSpeaker C: Okay.\nSpeaker C: I think I just want to put in as projects manager.\nSpeaker C: You know, a little bit of praise for everybody here.\nSpeaker C: But how they've worked on it.\nSpeaker C: You know, both individually and as a team.\nSpeaker C: You've everyone's come up with their own individual ideas and their own different departments.\nSpeaker C: And then come together and work in, you know, integrally, you know, at the right times, you know, especially you too.\nSpeaker C: That's all, you know, gone very, very well and and.\nSpeaker C: You know, it's been good communication going on.\nSpeaker D: And during our design, and there was some some heated, heated discussion, but we kept trying to keep it cool and.\nSpeaker C: Did you have to go down to the corporate squash court and bash a few bills about.\nSpeaker D: We just had to squeeze our product a little bit.\nSpeaker D: Fantastic.\nSpeaker D: Now you guys have been a great team.\nSpeaker D: I think we're the envy of all the, of all the other teams.\nSpeaker C: I think so, I think we've come up with something new, something that hasn't been done before.\nSpeaker C: We're not just rehashing an old design.\nSpeaker B: And for meetings, for the old design, the meetings could be this quick.\nSpeaker C: You know, maybe this isn't a simulation.\nSpeaker C: Maybe this is actually, it's like, I think they're so mean or someone like that.\nSpeaker C: They're just, yeah.\nSpeaker C: They get the.\nSpeaker C: Yeah.\nSpeaker C: Yeah, well, two years time, this will be on the market.\nSpeaker C: Exactly that product.\nSpeaker C: And we'll go, yeah, we design that and no one will believe us.\nSpeaker D: So at this stage, I mean, is this the last meeting of the project?\nSpeaker D: We don't have another one after it's gone.\nSpeaker C: No, I think when this meeting's finished, like officially, they'll, we'll get a questionnaire.\nSpeaker C: Oh, really?\nSpeaker B: Just start summarizing now.\nSpeaker B: I can reply to the same list.\nSpeaker C: I'm got the message.\nSpeaker B: See some way.\nSpeaker B: Just reply to that one.\nSpeaker D: So there's no way to like predict what our, because we had it, we originally had it.\nSpeaker D: As far as our financial goals, we had a specific number for profits that we wanted.\nSpeaker D: It was 50.\nSpeaker B: Was it, was it still your five?\nSpeaker D: I don't remember, but there's not a way to compete that.\nSpeaker D: I mean, since we saved on the, on that production cost, do we know how much we're making on profit?\nSpeaker C: It gets handed over to another department.\nSpeaker C: Well, what our project was was to come up with the product.\nSpeaker C: Basically, like the, and just basically is it, can, can, is it within budget?\nSpeaker C: When it, when it comes to all the other things of how to sell it and, you know, the, the profits and all that, that's other departments.\nSpeaker C: It's another team that actually work out the, the, we have a vested interest in that.\nSpeaker D: Oh, yeah.\nSpeaker C: The all that goes in the profit sharing.\nSpeaker C: Yeah.\nSpeaker C: Yeah, that's it.\nSpeaker C: You know, we've, we've, we've made, we've designed the product.\nSpeaker C: We've made, we've got the prototype.\nSpeaker C: It's within budget.\nSpeaker C: It's, does everything that we wanted it to do.\nSpeaker C: It's new.\nSpeaker C: It's, it's something that, that is now there already.\nSpeaker D: I think actually in one advantage of this is that after the, you know, after this bad of fruit and vegetables passes, this will still be a cool remote, you know, we're not, we're not, you know, tying our cart to that one horse.\nSpeaker C: Yeah.\nSpeaker C: But this is very remarkable.\nSpeaker C: And that it is, it's, it's something that's kind of new and looks a bit quirky for people to want that.\nSpeaker C: It's, that it's, I'm bringing it all right down here somewhere.\nSpeaker C: Oh, I put it in one of my emails that I sent off.\nSpeaker C: That, you know, it's marksable in the sense that it's whilst all these other remotes are actually do give you a pest is strain injury.\nSpeaker C: Our one does the complete opposite.\nSpeaker C: You know, so that's something that's new, which is one of the criteria they asked us to come up with something new and makes watching TV healthy.\nSpeaker C: Yeah, and if you're not having a good time with the TV, you can throw it about, you know, it's fine.\nSpeaker C: It's kid proof.\nSpeaker B: Don't throw it at any of the ornaments and break them.\nSpeaker C: Well, you can break the ornaments, but you won't break that.\nSpeaker C: No.\nSpeaker C: Yeah.\nSpeaker C: So all in all, I think we've done very well.\nSpeaker C: Well done everybody.\nSpeaker C: Yeah.\nSpeaker C: You too.\nSpeaker C: In the corporate bar that's next to the corporate swimming pool on the top floor that we're, that we've all just gained access to.\nSpeaker C: That's it.\nSpeaker C: That's it.\nSpeaker C: Well, as far as I know, that's it.\nSpeaker B: I've got five minutes left.\nSpeaker C: No.\nSpeaker C: Everyone want to play us by?\nSpeaker D: I guess it's probably better than we wrap it up and have five minutes of silence.\nSpeaker D: Should we call it a day then?\nSpeaker C: Yeah, I guess.\nSpeaker C: I mean, we're going to get asked to fill out another questionnaire.\nSpeaker C: It's probably going to be the same as one we've done before.\nSpeaker C: Well, I'm not the authority to say that it is.\nSpeaker C: How does everyone feel about the technology that's been used in this?\nSpeaker C: Sort of using the pens and writing on these special pads and all that?\nSpeaker C: I love it.\nSpeaker B: I love it.\nSpeaker B: I think it's cool.\nSpeaker B: You know, wide up vibrating pen and cool.\nSpeaker B: Being watched.\nSpeaker B: Your moment to shame.\nSpeaker A: Well.\nSpeaker C: You know, I didn't, yeah, as we said earlier, I've not never seen that before.\nSpeaker C: Something that the whiteboard thingy.\nSpeaker C: That's great.\nSpeaker C: But a pen with a camera on it, I don't think it's such a new thing.\nSpeaker C: I mean, it's all in such a new idea.\nSpeaker C: It would be different if it was sort of handwriting, recognizing stuff.\nSpeaker C: But as it's not, it's literally that will come up on the computer screen as a picture file rather than actual text.\nSpeaker D: Right.\nSpeaker D: I think they do want to give me an writing mechanism.\nSpeaker D: Yeah.\nSpeaker B: It's just that we're stuck.\nSpeaker B: Logitech.\nSpeaker D: I guess we should end this as well.\nSpeaker D: Off topic.\nSpeaker D: Yeah.\nSpeaker D: Meeting the joint.\nSpeaker D: Right.\nNone: Yeah.\nSpeaker D: Yeah.\n"}, {"meeting_id": "Bmr016", "summary": "The group discussed digits data, recent ASR results, the status of transcriptions, and disk space and storage format issues. Approximately two hours of digits have been recorded, half of which have been extracted. Researchers doing ASR are looking into methods for generating a better channel-based segmentation to improve recognition results for close-talking microphone data. Transcription checking procedures were reviewed, and efforts to coordinate the channelization and presegmention of data with the tightening of time bins were discussed. The group also talked about downsampling and strategies for coping with low disk space.", "dialogue": "Speaker G: And we already got the crash out of the way.\nSpeaker G: It did crash, so I feel much better earlier.\nSpeaker G: Interesting.\nSpeaker H: Get the door.\nSpeaker F: Hey.\nSpeaker F: So I did collect an agenda.\nSpeaker G: I did collect an agenda.\nSpeaker G: So I'm going to go first.\nSpeaker G: Well, it shouldn't take too long.\nSpeaker G: So we're pretty much out of digits.\nSpeaker G: We've gone once through the set.\nSpeaker G: So the only thing I have to do, that's right.\nSpeaker G: So I just have to go through them and pick out the ones that have problems and either correct them or have them reread.\nSpeaker G: So we probably have like four or five more forms to be read to be once through the set.\nSpeaker G: I've also extracted out about an hour's worth.\nSpeaker G: We have about two hours worth.\nSpeaker G: I extracted about an hour's worth, which are the digits for which whose speaker have speaker forms, have filled out speaker forms.\nSpeaker G: Not everyone's filled out a speaker form.\nSpeaker G: So I extracted one for speakers who have speaker forms and for meetings in which the key file and the transcript files are possible.\nSpeaker G: Some of the early key files it looks like were done by hand.\nSpeaker G: And so they're not automatically possible, and I have to go back and fix those.\nSpeaker G: So what that means is we have about an hour of transcribed digits that we can play with.\nSpeaker H: Liz.\nSpeaker H: I think two hours is the total value.\nSpeaker H: Yeah.\nSpeaker H: Yeah.\nSpeaker H: And you can go to the question, all these different things that are not quite right, but you can go to the other three or the other hour.\nSpeaker G: Yes, absolutely.\nSpeaker G: So that's just a question of a little hand editing of some files and then waiting for more people to turn in their speaker forms.\nSpeaker G: I have this web-based speaker form, and I sent mail to everyone who hadn't filled out a speaker form, and they're slowly trickling in.\nSpeaker H: So the relevance of the speaker form here\nSpeaker G: is for labeling the extracted audio files. Oh, OK.\nSpeaker G: By speaker ID and microphone type.\nSpeaker H: Permission to use their digits.\nSpeaker G: No, I spoke with Jane about that, and we sort of decided that it's probably not an issue, that we edit out any of the errors anyway.\nSpeaker G: So there are no errors in the digits.\nSpeaker G: You always read the string correctly.\nSpeaker G: So I can't imagine why anyone would care.\nSpeaker G: So the other topic with digits is, Liz would like to elicit different prosotics.\nSpeaker G: And so we tried last week with them written out in English, and it just didn't work at all, because no one grouped them together.\nSpeaker G: So it just sounded like many, many more lines instead of anything else.\nSpeaker G: So in conversations with Liz and Jane, we decided that if you wrote them out as numbers instead of words, it would elicit more phone number, social security number, like readings.\nSpeaker G: The problem with that is it becomes numbers instead of digits.\nSpeaker G: When I look at this, that first line is 61, 62, 18, 86, 10.\nSpeaker G: And so the question is, does anyone care?\nSpeaker G: I've already spoken with Liz, and she feels that, correct me if I'm wrong, that for her connected numbers is fine, as opposed to connected digits.\nSpeaker G: I think two hours is probably fine for test set, but it may be a little short if we actually want to do training and adaptation and all that other stuff.\nSpeaker H: Yeah.\nSpeaker H: Do you want different prasadics?\nSpeaker H: So if you always had the same group, things you wouldn't like that, is that correct?\nSpeaker E: Well, we actually figured out a way to do the groupings are randomly generated.\nSpeaker H: No, but I was asking if that was something you really cared about, because if it wasn't, it seems to me if you made it really specifically telephone groupings that maybe people wouldn't bring with a number so much.\nSpeaker H: You know, if it is a bit...\nSpeaker E: They may still do it.\nSpeaker E: Maybe some.\nSpeaker I: What about putting a hyphen between the numbers in the group?\nSpeaker H: So if you have six-1, you mean?\nSpeaker H: If you go six-6-6-2931.\nSpeaker E: Well, OK, it might help.\nSpeaker E: I would like to get away from having only one specific grouping.\nSpeaker E: So if that's your question.\nSpeaker E: But it seems to me that at least for us, we can learn to read the math digits.\nSpeaker E: If that's what people want.\nSpeaker E: I don't think that'd be that hard.\nSpeaker E: I agree.\nSpeaker E: To read the math single digits.\nSpeaker E: And it seems like that might be better for you guys, since then you'll have just more digit data.\nSpeaker E: And that's always a good thing.\nSpeaker E: It's a little bit better for me, too, because the digits are easier to recognize.\nSpeaker E: They're better trained than the numbers.\nSpeaker G: So we could just put in the instructions\nSpeaker E: with the math digits. Right.\nSpeaker E: Read the math single digits.\nSpeaker E: So 61 was read as 6-1.\nSpeaker E: And if people make a mistake, we...\nSpeaker G: So versus 0.\nSpeaker H: I mean, the other thing is we could just beg it because it's worrying about it.\nSpeaker H: I mean, because we do have digits training data that we have from OGI.\nSpeaker H: I'm sorry, it's numbers training data that we have from OGI.\nSpeaker H: We've done lots and lots of studies with that.\nSpeaker H: And...\nSpeaker E: But it's nice to get it in this room with the...\nNone: Yeah.\nSpeaker E:...accus.\nSpeaker H: No, no, I guess what I'm saying is that...\nSpeaker G: Just let them read it.\nSpeaker H: How they read it.\nSpeaker H: Just read it.\nSpeaker H: How they read it.\nSpeaker H: And just means we have to expand our vocabulary out to...\nSpeaker E: Well, that's fine with me.\nSpeaker E: As long as it's just that I didn't want to cause the people who would have been collecting digits the other way to not have the digits.\nSpeaker E: So...\nSpeaker H: We can go back to the other thing later.\nSpeaker H: I mean, we...\nSpeaker H: OK.\nSpeaker H: We can do this for a while and then go back to digits for a while.\nSpeaker H: OK.\nSpeaker H: Yeah, I mean, you want this...\nSpeaker H: Do you need training data or adaptation data or this?\nSpeaker H: How much of this do you need with the...\nSpeaker E: It's actually unclear right now.\nSpeaker E: I just thought, well, if we're collecting digits and add them, it's that we were running out of the TI forms.\nSpeaker E: I thought it'd be nice to have them in groups.\nSpeaker E: And probably, all else being equal, it'd be better for me to just have single digits since...\nSpeaker E: OK.\nSpeaker E:...you know, I recognize it's going to do better on those anyway.\nSpeaker E: And it's more predictable.\nSpeaker E: So we can know from the transcript what the person said and the transcriber in general.\nSpeaker E: But if they make mistakes, it's no big deal.\nSpeaker E: If people say 100 instead of 100 and also maybe we can just let them choose 0 versus 0 as they like.\nSpeaker E: Because even the same person sometimes says 0 and sometimes 0 in different contexts and that's sort of interesting.\nSpeaker E: So I don't have a specific need because if I did, I'd probably try to collect it without bothering this group.\nSpeaker E: But if we can try...\nSpeaker G: So I can just add to the instructions to read it as digits.\nSpeaker E: Right.\nSpeaker E: And you can give an example like, you know, 6-61 would be read as 6-1.\nSpeaker E: Right.\nSpeaker A: And actually, it's no more artificial than what we've been doing with words.\nSpeaker A: I'm sure people can adapt to this.\nSpeaker A: Right.\nSpeaker A: It's simple.\nSpeaker A: The space is already biased toward being separated.\nSpeaker A: Right.\nSpeaker A: And I know I'm going to find this easier than words.\nSpeaker G: Oh, yeah, absolutely.\nSpeaker G: Cognitively, it's much easier.\nSpeaker E: I also had a hard time with the words.\nSpeaker E: But then we went back and forth.\nSpeaker E: OK, so let's give that a try.\nSpeaker G: And is the spacing all right or do you think there should be more space between digits and groups?\nSpeaker E: I mean, what do other people think because you guys are reading them?\nSpeaker A: I think that it's fine.\nSpeaker A: To me, it looks like you've got the idea of grouping and you have the idea of separation.\nSpeaker A: And it's just a matter of the instructions.\nSpeaker A: That's right.\nSpeaker A: OK.\nSpeaker G: And I think there are about 10 different grouping patterns.\nSpeaker G: Isn't that right, Liz?\nSpeaker E: Right.\nSpeaker E: Right.\nSpeaker E: And you just, they're randomly generated and randomly assigned to digits.\nSpeaker E: I did.\nSpeaker H: So what?\nSpeaker H: I was just going to say, so we have, we've sent you 40 hours of recordings now.\nSpeaker H: And you're saying two hours is digits.\nSpeaker H: That's for the ratio.\nSpeaker H: Yeah.\nSpeaker H: 20 to 1, which I guess makes sense.\nSpeaker H: So if we did another 40 hours of recordings, then we could get a couple hours.\nSpeaker H: Right.\nSpeaker H: Yeah, like you say, I think a couple hours for a test sense, OK?\nSpeaker H: It'd be nice to get more later because we're going to use this up in some sense.\nSpeaker H: Right?\nSpeaker A: I also would like to argue for that because it seems to me that there's a real strength in having the same test replicated a whole bunch of times and adding to that basic test bank.\nSpeaker A: Right?\nSpeaker A: Because then you have more and more chances to get away from random errors.\nSpeaker A: And I think the other thing too is that right now, we have sort of a stratified sample with reference to dialect groups.\nSpeaker A: And there might be an argument to be made for for replicating all of the digits that we've done, which were done by non-native speakers, so that we have a core that totally replicates the original data set, which is totally American speakers.\nSpeaker A: And then we have these stratified additional language groups overlapping certain aspects of the database.\nSpeaker G: I think that trying to duplicate, spending too much effort trying to duplicate the existing TI digits probably isn't too worthwhile because the recording situation is so different.\nSpeaker G: It's going to be very hard to be comparable.\nSpeaker A: Except that if you have a stimuli comparable, then it says something about the contribution of setting.\nSpeaker H: But the other differences are so major.\nSpeaker H: OK.\nSpeaker H: Yeah.\nSpeaker H: Red versus not.\nSpeaker H: It's a very serious thing.\nSpeaker A: What's an example of some of the other differences?\nSpeaker A: Any other differences?\nSpeaker H: Well, individual human glottis is going to be different.\nSpeaker G: You know, there's so many things.\nSpeaker G: And not just that.\nSpeaker G: I mean, the corpus itself, we're collecting it in a red digit in a particular list.\nSpeaker G: And I'm sure that they're doing more specific stuff.\nSpeaker G: I remember correctly, it was like postman reading zip codes and things like that.\nSpeaker G: The idea did it suppose?\nSpeaker G: I thought so.\nSpeaker H: Was it red?\nSpeaker H: Yeah, I think the reading zip code stuff, you think you're going to do.\nSpeaker H: Oh, I may well be.\nSpeaker H: Yeah, not TI digits was red in red.\nSpeaker G: I haven't ever listened to TI digits.\nSpeaker G: So I don't really know how to compare this.\nSpeaker G: But regardless, it's going to be hard to compare cross-corpus.\nSpeaker H: It's different people.\nSpeaker H: This is a different thing.\nSpeaker H: And they're different circumstances, different recording, and so forth.\nSpeaker H: So it's really pretty different.\nSpeaker H: But I think the idea of using a set thing was just to give you some sort of framework so that even though you couldn't exact comparison, this is what these valid, something like group is doing.\nSpeaker H: Some kind of reference.\nSpeaker I: OK, what do the groupings represent?\nSpeaker I: You said there's like 10 different groupings.\nSpeaker E: Right, just groupings in terms of number of groups in a line and number of digits in a group and the pattern of groupings.\nSpeaker I: Are the patterns based on anything there?\nSpeaker E: I just roughly looked at what kinds of digit strings are out there, and they're usually grouped into either two, three, or four digits at a time.\nSpeaker E: And they can have, I mean, actually, things are getting longer and longer.\nSpeaker E: In the old days, you probably only had three sequences.\nSpeaker E: Telephone numbers were less and so forth.\nSpeaker E: So there's between, well, if you look at it, there are between like three and five groups.\nSpeaker E: And each one has between two and four groupings.\nSpeaker E: I purposely didn't want them to look like they're in any kind of pattern.\nSpeaker G: And which group appears as picked randomly and what the numbers are are picked randomly?\nSpeaker G: So unlike the previous one, which I simply replicated to T.I. digits, this is generated randomly.\nSpeaker I: Oh, OK.\nSpeaker E: But I think it would be great to be able to compare digits, whether it's these digits or T.I. digits to speakers, and compare that to their spontaneous speech.\nSpeaker E: And then we do need a fair amount of digit data because you might be wearing a different microphone.\nSpeaker E: So it's nice to have the digits replicated many times, especially for speakers that don't talk a lot.\nSpeaker E: So for adaptation, no, I'm serious.\nSpeaker E: So we have a problem with acoustic adaptation.\nSpeaker E: And we're not using the digit data now, but not for adaptation.\nSpeaker E: We were running adaptation only on the data that we ran recognition on.\nSpeaker E: And as soon as someone started to read transcript number, that's red speech.\nSpeaker E: And I thought, well, we're going to do better on that.\nSpeaker E: It's not fair to use.\nSpeaker E: Oh, yeah, that's true.\nSpeaker E: Absolutely.\nSpeaker E: It might be fair to use the data for adaptation.\nSpeaker E: So those speakers who are very quiet shy.\nSpeaker E: Right.\nSpeaker E: Like adapting on, yeah.\nSpeaker E: Well, I mean, it's the same microphone.\nSpeaker E: See, the nice thing is we have that in the same meeting.\nSpeaker E: Right.\nSpeaker E: Same acoustic, same channels.\nSpeaker E: Right.\nSpeaker E: And so I still like the idea of having some kind of good.\nSpeaker H: Yeah, I mean, for the acoustic research, for the signal processing, far field stuff, I see it as the place that we started.\nSpeaker H: But I mean, it'd be nice to have 20 hours of digit data.\nSpeaker H: But the truth is, I'm hoping that the stuff that you guys have been doing is continue that.\nSpeaker H: We get the best we can do on this spontaneous stuff.\nSpeaker H: And then we do a lot of the testing of the algorithms on the digits for the far field.\nSpeaker H: At some point, when we feel it's mature, and we understand what's going on, we can then we have to do the spontaneous data with the far field.\nSpeaker E: The only thing that we don't have, I know this sounds weird.\nSpeaker E: And maybe it's completely stupid.\nSpeaker E: But we don't have any overlapping digits.\nSpeaker G: Yeah.\nSpeaker G: We talked about that a couple of times.\nSpeaker E: I know it's weird.\nSpeaker G: But the problem I see with trying to do over lapping digits is the cognitive load.\nSpeaker E: I know everybody's laughing.\nSpeaker E: OK.\nSpeaker E: No, it's stupid.\nSpeaker E: It's just, I'm just talking for the stuff that I can't do.\nSpeaker E: You try to do it.\nSpeaker E: I mean, here, let's try it.\nSpeaker G: You read the line.\nSpeaker G: I'll read the first line.\nSpeaker E: These are all the same forms.\nSpeaker E: OK.\nSpeaker G: So you read the last line.\nSpeaker G: I'll read the first line.\nSpeaker G: You plug your ears.\nSpeaker G: Oh, I guess you plug your ears.\nSpeaker G: You could do it.\nSpeaker G: But then you don't get the same effects.\nSpeaker E: Well, what I mean is actually not the overlaps that are well governed linguistically, but the actual fact that there's speech coming from two people, beam forming stuff, all the acoustic stuff that like Dan Ellison and company want to do, digits are nice and well-behaved.\nSpeaker E: I mean, anyway, it's just a thought.\nSpeaker E: I think so.\nSpeaker E: It would go faster.\nSpeaker E: We take one over and.\nSpeaker I: It's the remake of DigitRea.\nSpeaker E: Well, looking a little strife.\nSpeaker E: I mean, I was sort of serious, but I really, I mean, I don't feel strongly enough that it's a good idea.\nSpeaker G: I do the last line.\nSpeaker G: I'll do the first line.\nSpeaker G: 6-1-6-2-1-8-6-1-0.\nSpeaker G: That's not bad.\nSpeaker G: No, I can do it.\nSpeaker E: And that probably was great.\nSpeaker E: By the way, I think it was numbers, but I'm not sure.\nSpeaker E: It just sort of sounded like a duet or something.\nSpeaker I: Performance, OK.\nSpeaker H: Let's try three.\nSpeaker H: You'll pick one in the middle.\nSpeaker E: OK.\nSpeaker G: 6-1-6-2-1-8-6-1-0.\nSpeaker G: I'm sorry.\nSpeaker G: I mean, I think it was doable.\nSpeaker G: Of course, transcribers.\nSpeaker E: So we could have a round where you do two at a time, and then the next person picks up when the first hit is done\nSpeaker A: or something.\nSpeaker E: Like a, what do you call it? This pair was.\nSpeaker E: Around.\nSpeaker E: Like, yeah.\nSpeaker E: Yeah, like that.\nSpeaker E: Then it would go like twice as fast.\nSpeaker E: A third is fast.\nSpeaker E: Anyway, it's just a thought.\nSpeaker E: I'm actually sort of serious if it would help people do that.\nSpeaker E: But the people who want to work in it, we should talk to them.\nSpeaker H: I don't think we're going to say the best amount of data that way.\nSpeaker H: Having a little bit might at least be fun for somebody like Dan.\nSpeaker G: I think maybe if we wanted to do that, we would do it as a separate session, something like that, rather than doing a real meeting and, you know, do two people at a time, then three people at a time, and things like that.\nSpeaker E: Can try it out.\nSpeaker E: If we have nothing.\nSpeaker E: So you see what Dan says.\nSpeaker E: We have no agenda.\nSpeaker H: Do it some week.\nSpeaker H: OK.\nSpeaker H: It's been the whole time reading.\nSpeaker H: I thought this was going to happen.\nSpeaker A: Another question about this.\nSpeaker A: So there are these digits, which are detached digits.\nSpeaker A: But there are other words that contain the same general phoneme sequences, like wonderful, has one in it.\nSpeaker A: And Victor Borja had a piece on this where he inflated the digits.\nSpeaker A: Well, I wonder if there would be a value in having digits that are in essence embedded in real words to compare in terms of the articulation of one and wonderful versus one as a digit being read.\nSpeaker H: That's too bad.\nSpeaker E: I'm all for it.\nSpeaker B: Not after I ate that.\nSpeaker H: I don't know what work is well, isn't it?\nSpeaker H: How does nine work in it?\nSpeaker F: Nine.\nSpeaker F: You scream in German?\nSpeaker B: Yes, there's a German.\nSpeaker B: Oh.\nSpeaker D: Ellis, I was good when you scream at them.\nSpeaker H: Everybody's a little punchy here.\nSpeaker A: Well, I mean, I just wanted to offer that as a possible task because we were to each read this embedded numbers words in sentences because it's like an entire sketchy does.\nSpeaker A: And I wouldn't take the inflated version.\nSpeaker A: So he talks about the woman being two to fall in.\nSpeaker A: But if it were to be deflated just the normal word, it would be like a little story that we could read.\nSpeaker A: I don't know if that would be useful for comparison, but it's embedded numbers.\nSpeaker G: I think for something like that would be better off doing like Tim, it.\nSpeaker H: Well, I think the question is what the research is.\nSpeaker H: So I mean, I presume that the reason that you wanted to have these digits this way is because you wanted to actually do some research looking at the exotic form here.\nSpeaker E: Right.\nSpeaker H: Yeah.\nSpeaker H: So if somebody wanted to do that, if they wanted to look at the difference in the phones and the digits in the context or a non-digit word versus in the digit word, that would be a good thing to do.\nSpeaker H: But I think someone would have to express interest in that.\nSpeaker H: I see.\nSpeaker H: OK.\nSpeaker H: Maybe you were interested in doing that.\nNone: OK.\nSpeaker A: Thank you.\nSpeaker G: Yeah, we know digits.\nSpeaker G: We have ASR results from Liz transcripts asked from Jane and disk space and storage formats from Don.\nSpeaker G: Do we have any preference on which way we want to go?\nSpeaker E: Well, I was actually going to skip the ASR results part in favor of getting the transcription stuff talked about since I think that's more important to moving forward.\nSpeaker E: But I mean, Morgan has this paper copy.\nSpeaker E: And if people have questions, it's pretty preliminary in terms of ASR results because we didn't do anything fancy.\nSpeaker E: But I think just having the results there and pointing out some main conclusions, like it's not the speaking style that differs.\nSpeaker E: It's the fact that there's overlap, that causes recognition errors.\nSpeaker E: And the fact that it's almost all insertion errors, which you would expect.\nSpeaker E: But you might also think that in the overlap regions, you would get substitutions and so forth, leads us to believe that doing a better segmentation like your channel-based segmentation or some kind of echo cancellation to get basically back down to the individual speaker utterances would be probably all that we would need to be able to do good recognition on the close documents.\nSpeaker G: Why don't you have a hard copy?\nSpeaker G: Why don't you email it to the list?\nSpeaker E: But this is Morgan has this paper.\nSpeaker E: Oh, it's in the paper.\nSpeaker E: It's that paper.\nSpeaker E: Yeah.\nSpeaker E: Everybody's going to know.\nSpeaker E: OK, then it's already done.\nSpeaker E: Basically did a lot of work on that.\nSpeaker E: And it's, let's see, I guess the other neat thing is it shows for sure that the lapel within speaker is bad.\nSpeaker E: And it's bad because it picks up the overlapping speech.\nSpeaker C: So your results were run on the channel synchronized?\nSpeaker E: Yes, because that's all that had been transcribed at the time.\nSpeaker E: But as we, I mean, I wanted to hear more about the transcription, if we can get the channel asynchronous or the closer to that would be very interesting for us.\nSpeaker H: Because we use the part from which you had about the all over the channels or mixed channels.\nSpeaker I: So if there was a segment of speech this long,\nSpeaker G: someone said, oh, in the front. Oh, in the front.\nSpeaker I: Oh, the whole thing was passed to the record.\nSpeaker E: That's right.\nSpeaker E: In fact, I pulled out a couple classic examples in case you want to use them in your talk of Chuck on the lapel.\nSpeaker E: So Chuck wore the lapel three out of four times.\nSpeaker E: I noticed that Chuck was early on.\nSpeaker E: And I wore the lapel once.\nSpeaker E: And for me, the lapel was OK.\nSpeaker E: I mean, I still, and I don't know why.\nSpeaker E: But for you, it was for who was next to me.\nSpeaker D: And where you were sitting probably affected.\nSpeaker E: Right.\nSpeaker E: But when Chuck wore the lapel and Morgan was talking, there are a couple of really long utterances where Chuck is saying a few things inside.\nSpeaker E: And it's picking up all of Morgan's words pretty well.\nSpeaker E: And so the error rates, because of insertions aren't bounded.\nSpeaker E: So with a one word utterance and 10 insertions, you've got huge error rate.\nSpeaker E: And that's where the problems come in.\nSpeaker E: So this is sort of what we expected.\nSpeaker E: But it's nice to be able to show it.\nSpeaker E: And also, I just wanted to mention briefly that Andreas, when I called up Dan, Alice, who's still stuck in Switzerland.\nSpeaker E: We were going to ask him if there was out there in terms of echo cancellation and things like that, not that we were going to do it, but we wanted to know what would need to be done.\nSpeaker E: And we've given him the data we have so far.\nSpeaker E: So these synchronous cases where there are overlap.\nSpeaker E: And he's going to look into trying to run some things that are out there and see how all it can do.\nSpeaker E: Because right now, we're not able to actually report on recognition in a real paper, like a Euro-Speed Paper, because it would look sort of premature.\nSpeaker I: So the idea is that you would take this big hunk where somebody's only speaking a small amount in it, and then try to figure out where they're speaking based on.\nSpeaker E: Right.\nSpeaker E: Or at any point in time, who's the foreground speaker, who's the background speaker?\nSpeaker G: I thought we were just going to move the boundary.\nSpeaker G: Well, that's sort of hand stuff.\nSpeaker G: So there's, how would you do that automatically?\nSpeaker C: Well, there's actually done some experiments with cross correlation.\nSpeaker C: And it seems to work pretty well to get that sort of thing.\nSpeaker E: Yeah, exactly.\nSpeaker I: So why do you want to do echo cancellation?\nSpeaker E: It would be techniques used from adaptive echo cancellation, which I don't know enough about to talk about.\nSpeaker E: It's just a journey to remove cross-down.\nSpeaker E: But right.\nSpeaker E: And that would be similar to what you're also trying to do, but using more than energy.\nSpeaker E: I don't know what exactly would go into it.\nSpeaker E: So the idea is to basically run this on the whole meeting and get the locations, which gives you also the time of dreams.\nSpeaker I: To do sort of what he's already, what he's trying to do.\nSpeaker E: Right.\nSpeaker E: Except that there are many techniques for the kinds of cues that you can use to do that.\nSpeaker E: I see.\nSpeaker E: Yeah.\nSpeaker H: If Dave is also going to be using the foreground with echo cancellation for the near field, far field stuff.\nSpeaker E: And I guess S-pen?\nSpeaker E: Is he here too?\nSpeaker E: May also be.\nSpeaker E: So that's really the next step, because we can't do too much in terms of recognition results, knowing that this is a big problem, until we can do that kind of processing.\nSpeaker E: And so once we have some reviewers and we'll move on.\nSpeaker I: I think this also ties in to one of the things that Jane's going to talk about, too.\nSpeaker G: I also want to say I have done all this chopping up of digits.\nSpeaker G: So I have some naming conventions that we should try to agree on.\nSpeaker E: Oh, right.\nSpeaker E: So let's do that offline.\nSpeaker E: We don't do it during the year.\nSpeaker E: Right.\nSpeaker E: Definitely.\nSpeaker G: And I've scripts that will extract it out from key files and do all the naming automatically.\nSpeaker G: So I'll do it by hand.\nSpeaker D: Great.\nSpeaker D: So let's do this.\nSpeaker D: You compile the list of speaker names.\nSpeaker D: And OK.\nSpeaker D: Not names, but.\nSpeaker D: Yep.\nSpeaker G: Not names.\nSpeaker G: Names to IDs.\nSpeaker G: So you do it.\nSpeaker G: And it does all sorts of matches, because the way people fill out names is different on every single file.\nSpeaker E: So it does a very fuzzy sort of match.\nSpeaker E: So at this point, we can sort of finalize the naming and so forth.\nSpeaker E: And we're going to basically rewrite out these waveforms that we did, because as you notice in the paper, your MO4 and one meeting and MO2 and another meeting, and we just need to standardize the.\nSpeaker E: That was my fault.\nSpeaker E: No.\nSpeaker E: No, I didn't notice that.\nSpeaker E: That's why those comments are in there.\nSpeaker G: So I now have a script that you could just say, basically, look up Morgan, and it will give you a second.\nSpeaker E: Great.\nSpeaker G: Great.\nSpeaker G: Terrific.\nSpeaker G: All right.\nSpeaker G: Do we, Dawn, you had disk space and storage formats?\nSpeaker G: Is that something we need to talk about at the meeting, or should you just talk with Chuck?\nSpeaker D: At some other time.\nSpeaker D: I had some general questions just about the compression algorithms of shortening waveforms.\nSpeaker D: And I don't know exactly who to ask.\nSpeaker D: I thought maybe you would be the person to talk to.\nSpeaker D: So is it a lossless compression when you compress?\nSpeaker D: So entropy coding.\nSpeaker D: It just uses entropy coding.\nSpeaker D: OK.\nSpeaker D: So I mean, I guess my question would be, is I just got this new 18 gig drive installed.\nSpeaker D: Yeah.\nSpeaker G: And I think half of it is scratch and half of it is.\nSpeaker D: I'm not exactly sure how they partitioned it.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I don't know what's typical here, but it's local, though.\nSpeaker G: So that doesn't matter.\nSpeaker G: But you can access it from anywhere in XC.\nSpeaker G: OK.\nSpeaker D: In fact, how do you do that?\nSpeaker E: Drive versus the thing is the 18 gig.\nSpeaker E: It was a spare that day pattern.\nSpeaker G: Slash N slash machine name slash XC.\nSpeaker D: OK.\nSpeaker D: All right.\nSpeaker G: I didn't know.\nSpeaker G: So the only question is how much of it, the distinction between scratch and non-scratch is whether it's backed up or not.\nSpeaker G: Right.\nSpeaker G: So what you want to do is use the scratch for stuff that you can regenerate.\nSpeaker G: OK.\nSpeaker G: So stuff that isn't backed up, it's not a big deal because disks don't crash very frequently as long as you can regenerate it.\nSpeaker D: Right.\nSpeaker D: I mean, all the stuff can be regenerated.\nSpeaker G: It's just going to put it all on scratch because where XC is model next to buy back up.\nSpeaker G: Yeah.\nSpeaker E: So all the transcripts should be backed up.\nSpeaker E: But all the waveforms should not be backed.\nSpeaker D: Right.\nSpeaker D: The ones that you write out.\nSpeaker D: OK.\nSpeaker D: So I mean, I guess the other question was, then, should we shorten them, downsample them, or keep them in their original form?\nSpeaker G: It just depends on your tools.\nSpeaker G: I mean, because it's not backed up and it's just on scratch, if your tools can't take short and format, I would leave them expanded.\nSpeaker G: So you don't have to un-shorten them every single time you want to do anything.\nSpeaker D: OK.\nSpeaker E: We can downsample them.\nSpeaker D: Do you think that would be OK?\nSpeaker D: Yeah.\nSpeaker E: To downsample them.\nSpeaker E: Yeah, we get the same performance.\nSpeaker E: OK.\nSpeaker E: I mean, the front end on the SRI recognize are just downsamples them on the fly.\nSpeaker D: Yeah.\nSpeaker D: I guess the only argument against downsampling is to preserve just the original files in case we want to experiment with different filtering techniques.\nSpeaker H: Yeah.\nSpeaker H: I'm sorry.\nSpeaker H: Yeah.\nSpeaker H: Overall, our data, we want to not downsample.\nSpeaker E: You want to not.\nSpeaker E: OK.\nSpeaker E: So what we're doing is we're writing out, I mean, this is just a question.\nSpeaker E: We're writing out these individual segments that wherever there's a time boundary from T-load or James transcribers, we chop it there.\nSpeaker E: And the reason is that we can feed it to the recognizer and throw out ones that we're not using and so forth.\nSpeaker E: And those are the ones that we're storing.\nSpeaker E: Yeah.\nSpeaker G: So it's regenerateable.\nSpeaker G: Yeah.\nSpeaker G: What I would do is take downsample it and compress it.\nSpeaker G: However, the SRI recognize or wants to take it in.\nSpeaker G: So we can't shorten them.\nSpeaker E: We can downsample them.\nSpeaker E: Yeah, I mean, I'm sorry.\nSpeaker H: As long as there is a form that we can come from again, that's not downsample.\nSpeaker E: Yeah, that's why we need more dis space because we're basically duplicating the originals.\nSpeaker E: That's fine.\nSpeaker H: But for future research, we're doing different microphones.\nSpeaker E: Oh, yeah.\nSpeaker E: No, we always have the original long ones.\nSpeaker I: So the SRI front end won't take a large audio file name and then a list of segments to chop out from that large audio file.\nSpeaker I: They actually have to be chopped out already.\nSpeaker E: It's better if they're chopped out and it will be, yeah.\nSpeaker E: We could probably write something to do that, but it's actually convenient to have them chopped out.\nSpeaker E: Because you can run them in different orders.\nSpeaker E: You can actually move them around.\nSpeaker E: And that's a whole point of opinion.\nSpeaker E: You can get rid of vengeance.\nSpeaker E: It's a lot faster.\nSpeaker G: Yeah, it's a lot faster.\nSpeaker G: English speaking.\nSpeaker G: All the native speakers and all the non-adjustable.\nSpeaker G: You can grab everything with the word that we're in.\nSpeaker E: And it's a lot quicker than actually trying to access the way file each time, find the time boundaries.\nSpeaker E: So in principle, yeah, you could do that.\nSpeaker E: Well, that's really right.\nSpeaker I: But it's not right now.\nSpeaker F: It's just not right now.\nSpeaker F: These are long.\nSpeaker B: These are long.\nSpeaker G: So for example, what if you wanted to run all the native speakers?\nSpeaker G: So if you did it that way, you would have to generate a program that looks in a database somewhere or extracts out the language, finds the time marks for that particular one, do it that way.\nSpeaker G: The way they're doing it, you have that already extracted and it's embedded in the file name.\nSpeaker G: And so you know, you just say.\nSpeaker G: Yeah, that's part of it.\nSpeaker G: So that's part of it.\nSpeaker G: You just say, you know, asterisk eAsterisk.wave and you get what you want.\nSpeaker E: Right.\nSpeaker E: And the other part is just that once they're written out, it is a lot faster to process it.\nSpeaker E: Rather than doing seeks.\nSpeaker G: So.\nSpeaker G: Through the file.\nSpeaker E: Otherwise, you're just accessing.\nSpeaker G: This is all just temporary access.\nSpeaker G: So I don't, I think it's all just, it's fine.\nSpeaker G: You know.\nSpeaker G: You're wanting to do it however is convenient.\nSpeaker G: Right.\nSpeaker H: I mean, it just depends up if the file is file-sits in memory you can do it.\nSpeaker E: The other thing is that believe it or not, I mean, we have some, so we're also looking at these in waves like for the alignment.\nSpeaker E: And so forth.\nSpeaker E: You can't load an hour of speech into x-waves.\nSpeaker E: Yeah.\nSpeaker E: You need to have these small files.\nSpeaker E: And in fact, even for the transcriber program.\nSpeaker I: Yeah, you can give waves a start in an end time.\nSpeaker E: Yeah, if you try to load really long waveform into x-waves, you'll be waiting there.\nSpeaker E: No, I'm not suggesting you load a long waveform.\nSpeaker I: I'm just saying you give it a start in an end time.\nSpeaker I: And it'll just go and pull out that section.\nSpeaker G: The transcribers didn't have any problem with that. Did they, Jane?\nSpeaker A: What's the loading process? They load it to some problem.\nSpeaker G: It takes a very long time.\nSpeaker C: Yeah, just to load the transcription to a little bit.\nSpeaker C: Right. It takes a long time.\nSpeaker C: But not to the waveform.\nSpeaker C: The waveform is there immediately.\nSpeaker C: Yeah.\nSpeaker G: Are you talking about transcriber or x-waves?\nSpeaker G: Yeah.\nSpeaker E: Actually, you're talking about transcriber, right?\nSpeaker E: Yeah.\nSpeaker E: There's also true of the digit test, which is what it's supposed to do.\nSpeaker G: We need x-waves to do the digits.\nSpeaker G: Yeah.\nSpeaker G: And they were loading the form x-waves in and it didn't seem to be any problem.\nSpeaker E: I agree.\nSpeaker E: Well, we have a problem with that time wise.\nSpeaker E: It's a lot slower to load in a long file.\nSpeaker E: It seems really good.\nSpeaker E: And also to check the file.\nSpeaker E: So if you have a transcript.\nSpeaker E: Well, regardless, it's still there.\nSpeaker E: I mean, it's, I think overall you could get everything to work by accessing the same waveform and trying to find two, you know, the beginning and end times.\nSpeaker E: But I think it's more efficient if we have the storage space.\nSpeaker E: To have the small ones.\nSpeaker G: And it's no problem, right?\nSpeaker G: Because it's not backed up.\nSpeaker E: Yeah.\nSpeaker G: So it's just, if we don't have a spare to sitting around, we go out and we buy ourselves an 80 gigabyte drive and make it all scratch space.\nSpeaker G: It's not a big deal.\nSpeaker A: You're right about the backup being a bottleneck.\nSpeaker A: It's good to.\nSpeaker A: Yeah, so these won't be scratch.\nSpeaker E: It's a first batch.\nSpeaker E: Yeah.\nSpeaker G: Right.\nSpeaker G: So remind me afterward and I'll, and we'll look at your disk.\nSpeaker G: You can see where to put stuff.\nSpeaker G: All right.\nSpeaker D: And I can just do it to a, do you on it?\nSpeaker D: Right.\nSpeaker D: And just see which, how much is on each?\nSpeaker D: Yep.\nSpeaker G: Each partition.\nSpeaker G: And you want to use either XA or scratch.\nSpeaker G: Okay.\nSpeaker G: Well, X question mark.\nSpeaker G: Anything starting with X is scratch.\nSpeaker G: Okay.\nSpeaker A: With two digits.\nSpeaker A: Two digits, right?\nSpeaker G: XA, XB, XC.\nSpeaker G: Okay.\nSpeaker A: Jane.\nSpeaker A: Okay. So I got a little print on here.\nSpeaker A: So three on this side.\nSpeaker A: Three on this side.\nSpeaker A: And I stable them.\nNone: Okay.\nSpeaker A: All right.\nSpeaker A: So first of all, there was an interest in the transcribe transcription checking procedures.\nSpeaker A: And I can tell you first to go through the steps, although you've probably seen them.\nSpeaker A: As you might imagine, when you're dealing with, really, a fair number of words and natural speech, which means self-repairs and all these other factors that there are lots of things to be standardized and streamlined and checked on.\nSpeaker A: And so I did a bunch of checks.\nSpeaker A: And the first thing I did was obviously a spell check.\nSpeaker A: And at that point, I discovered certain things like accommodate with one M, that kind of thing.\nSpeaker A: And then in addition to that, I did an exhaustive listing of the forums in the data file, which included detecting things like faulty punctuation and things.\nSpeaker I: I'm sorry to interrupt you.\nSpeaker I: Could I just back up a little bit?\nSpeaker I: Sure, please, please, please.\nSpeaker A: So you're doing these.\nSpeaker I: So the whole process is that the transcribers get the conversation and they do their pass over it.\nSpeaker I: Yes.\nSpeaker I: And then when they're finished with it, it comes to you and you begin these.\nSpeaker I: Exactly.\nSpeaker A: I do these checks.\nSpeaker A: These quality checks.\nSpeaker A: Uh-huh, exactly.\nSpeaker A: Yeah, thank you.\nSpeaker A: And so do an exhaustive listing of the forums.\nSpeaker A: Actually, I will go through this in order.\nSpeaker A: So if we can maybe wait and stick.\nSpeaker A: Keep that for a second because we're not ready for that.\nSpeaker A: So on the fifth page.\nSpeaker A: Exactly, exactly.\nSpeaker A: All right.\nSpeaker A: So, uh, spelling check first.\nSpeaker A: Then an exhaustive listing of the, uh, all the forums in the data with punctuation attached.\nSpeaker A: And at that point, I pick up things like, oh, you know, word followed by two commas.\nSpeaker A: And then, uh, another check involves, uh, being sure that every utterance has an identifiable speaker.\nSpeaker A: And if not, then that gets checked.\nSpeaker A: Then there's this issue of glossing, so-called spoken forums.\nSpeaker A: So they're- most for the most part, we're keeping it standard, we're-we're level transcription.\nSpeaker A: But there's- and that's done with the assumption that pronunciation variants can be handled.\nSpeaker A: So for things like, and the fact that someone doesn't say the D, uh, that's not important enough to capture in the transcription because of a good pronunciation, uh, you know, model, we'll be able to handle that.\nSpeaker A: However, things like, because, where you're lacking an entire very prominent for a syllable.\nSpeaker A: And furthermore, it's a forum that's specific to spoken language.\nSpeaker A: Those are reasons- for- for those reasons, I-I kept that separate and used the convention of using C, U, Z for that forum.\nSpeaker A: However, glossing it so that it's possible with a script to plug in the full orthographic forum for that one.\nSpeaker A: And a couple of others, not many.\nSpeaker A: So, Wana is another one going- Gona is another one with just the assumption again that this- these are things which it's not really fair to consider expected at pronunciation model to handle.\nSpeaker A: And Chuck, you- you indicated that, and this is- is one of those that's handled in a different way also.\nSpeaker A: Didn't you? Did I?\nSpeaker A: I don't remember.\nSpeaker A: Okay, so I was- it might not have been- it might not have been you.\nSpeaker A: But someone told me that in fact, Cuzz has treated differently in, um, in this context because of that reason that, um, it's a little bit farther than a pronunciation variant.\nSpeaker A: Okay, so after that, let's see.\nSpeaker I: Um, so that was part of the spell check or is that- that was after the spell check?\nSpeaker A: Well, so when I get the exhaust- so the spell check picks up those words because they're not in the dictionary, so it gets Cuzz and Wana and that-\nSpeaker G: And then you gloss them.\nSpeaker A: Yeah, I've run it through- I have a said, you know, so I do a side script saying whenever you see Gona- convert it to Gona, you know, gloss equals quote, going to quote, you know.\nSpeaker A: And with all of these things being in curly brackets, so they're always distinctive.\nSpeaker A: Okay, I also wrote a script which will, um, retrieve anything in curly brackets or anything which I've classified as an acronym and it pronounced acronym.\nSpeaker A: And the way I take- pronounced acronyms is that I have underscores between the component.\nSpeaker A: So if it's ACL, then it's A underscores C underscore L.\nSpeaker A: And-\nSpeaker G: So your list here are these ones that actually occurred in the meetings?\nSpeaker A: Yes, huh? Yeah.\nSpeaker A: Okay, so now- We are asking-\nSpeaker E: Can I ask a question about the glossing of the before we go on? So for a word like, because is it that it's always predictably because- I mean, is CUSE always meaning because?\nSpeaker A: Yes, but not the reverse.\nSpeaker A: So sometimes people will say because in the meeting.\nSpeaker A: And if they actually said because, then it's written as because with no- with cause doesn't even figure into the equation.\nSpeaker A: Because-\nSpeaker H: But not eating as people don't say, hey, cause. Right, right, right, right, yeah.\nSpeaker E: Yeah.\nSpeaker E: So, I guess so from the point of view of- The only problem is that with- For the recognition we map it to B cause.\nSpeaker E: And so we know that CUSE is the- Well, don't have the gloss.\nSpeaker G: But you have the gloss forms, you always replace it.\nSpeaker G: Exactly.\nSpeaker G: If that's what you want to do.\nSpeaker A: Uh-huh.\nSpeaker A: And Don knows this, and he's been- Yeah, I replaced the cause with B cause if it's lost.\nSpeaker E: Right, but- If it's okay.\nSpeaker E: But then there are other glosses that we don't replace, right?\nSpeaker A: Because- Yes, and that's why there are different tags on the glosses.\nSpeaker A: Okay, so- On the different types of comments, which we'll see in just a second.\nSpeaker A: So the pronounceable acronyms get underscores.\nSpeaker A: The things in curly brackets are viewed as comments.\nSpeaker A: They're comments of four types.\nSpeaker A: So it's a good time to introduce that.\nSpeaker A: Four types.\nSpeaker A: And maybe we'll expand that.\nSpeaker A: But the- But the comments are- Four types, mainly right now.\nSpeaker A: One of them is- Um, the gloss type we just mentioned.\nSpeaker A: Another type is-\nSpeaker G: So are we done with acronyms? Because I had a question on what- What does this mean?\nSpeaker A: I'm still doing the overview.\nSpeaker A: I haven't actually gotten here yet.\nSpeaker A: Awesome.\nSpeaker A: Okay, so glosses things like- glosses things like replacing the full form with the, um, more abbreviated one to the left.\nSpeaker A: Then you have- If it's- There's a couple different types of elements that can happen that aren't really properly words.\nSpeaker A: And some of them are laughs and breeze.\nSpeaker A: So we have- That's prepended with a tag of VOC.\nSpeaker A: And the non-vulgones are like door slams and tapping.\nSpeaker A: And that's prepended with a- So the non-vulgination- So the end being curly braces or something else?\nSpeaker A: Oh yeah, so this would- Let's just take one example.\nSpeaker A: Oh, oh, oh.\nSpeaker A: And then the non-vulgization would be something like a door slam.\nSpeaker A: They always end.\nSpeaker A: So it's like they're paired curly brackets.\nSpeaker A: And then the third type right now is things that fall in the category of comments about what's happening.\nSpeaker A: So it could be something like, you know, referring to so-and-so, talking about such and such, you know, looking at so-and-so.\nSpeaker A: So on the middle-\nSpeaker I: So in the first case that gloss applies to the word to the left. Yeah.\nSpeaker A: And this gets so- In the middle two, it's not applying anything, right?\nSpeaker A: No, they're events.\nSpeaker A: They're actually- The quality-\nSpeaker G: The quality is applying to the left.\nSpeaker I: Right, I just meant the middle two.\nSpeaker A: Yeah. Well, and actually it is true that with respect to laugh, there's another one which is while laughing.\nSpeaker A: And that is- An argument could be made for this turning that into a qualitative statement because it's talking about the thing that preceded it.\nSpeaker A: But at present we haven't been coding the exact scope of laughing, you know?\nSpeaker A: And so to have while laughing, you know that it happened somewhere in there, which could well mean that it occurred separately and following, or, you know, including some of the utterances to the left.\nSpeaker A: Haven't been awfully precise about that.\nSpeaker A: But I have here- Now we're about to get to this now.\nSpeaker A: I have frequencies.\nSpeaker A: So you'll see how often these different things occur.\nSpeaker A: But the very front page deals with this final aspect of standardization which has to do with the spoken forms like mm-hmm, and ha, and uh-uh, and all these different types.\nSpeaker A: And someone pointed out to me- This might have been Chuck.\nSpeaker A: About how a recognizer if it's looking for mm-hmm with 3M's, and it's transcribed with 2M's, that it might increase the air rate, which should really be a shame, because I personally would not be able to make a claim that those are dramatically different items.\nSpeaker A: So right now I've standardized across all the existing data with these spoken forms.\nSpeaker A: I should say all existing data except 30 minutes which got found today.\nSpeaker A: So I know- I'm going to check.\nSpeaker A: Yeah, actually, yeah, it was stored in a place I didn't expect.\nSpeaker A: So, and we reconstructed how to happen.\nSpeaker A: And this would be great.\nSpeaker A: So I'll be able to get through that tonight and then actually later today, really.\nSpeaker A: And so then we'll have everything following these conventions.\nSpeaker A: But notice it's a really rather small set of these kinds of things.\nSpeaker A: And I made it so that these are- With a couple exceptions, but things that you wouldn't find in the spell checker so that they'll show up really easily.\nSpeaker D: And- Jane, can I ask you a question?\nSpeaker A: Well, does that very last one correspond to?\nSpeaker A: Yeah, yeah.\nSpeaker A: That's only hers once and I'm thinking of changing that.\nSpeaker A: Is that like someone's warning or something?\nSpeaker A: I haven't heard it actually.\nSpeaker A: I need to listen to that.\nSpeaker E: Actually, we gave this to our pronunciation person.\nSpeaker E: She's like, I don't know what that is either.\nSpeaker E: Did she actually hear it?\nSpeaker E: No, we had- We gave her a list of words that weren't in our dictionary.\nSpeaker E: And so of course it picked up stuff like this.\nSpeaker E: And she just didn't listen.\nSpeaker E: So if she didn't know, we're just waiting on it.\nSpeaker A: Yeah, I'm curious to see what it is, but I didn't want to change it to something else until- You can't hear it.\nSpeaker A: Well, you know-\nSpeaker D: But that's not really like- Yeah.\nSpeaker D: No one really says arg.\nSpeaker D: Right, no- It's a- I said the highest ass, that's right.\nSpeaker F: That's a big problem when we talk about that.\nSpeaker F: We're going to never recognize this meeting.\nSpeaker F: Money ties on the art.\nSpeaker G: Well, yeah.\nSpeaker G: Well, or if you're a see programmer.\nSpeaker E: Yeah, that's right.\nSpeaker E: The art, see, and art, see, and art.\nSpeaker E: That's right.\nSpeaker E: It has a different pros.\nSpeaker G: It has arch-max, arch-max.\nSpeaker F: So Jane, what's-\nSpeaker E: Maybe tie- So I have one question about the EH versus like the AH and U-A.\nSpeaker A: That's partly a non-native native thing, but I have found EH and the native speakers too.\nSpeaker A: But it's mostly non-native.\nSpeaker A: Okay.\nSpeaker A: That's A versus A.\nSpeaker E: A.\nSpeaker E: A, yeah, right, because there were some speakers that did definite A's, but right now we- There were the Canadians, right?\nSpeaker E: So it's actually probably good for us to know the difference between the real A and the one that- Exactly.\nSpeaker E: Because in switchboard you would see all of these forms, but they all were like, ah.\nSpeaker G: I mean, just the single letter A has in the particles group.\nSpeaker E: No, no, I mean like the UH or the UH EH were all the same.\nSpeaker E: And then we have this additional non-native version of A.\nSpeaker D: All the EHs I've seen have been like that.\nSpeaker D: They've been like, eh, like that has been transcribed to EH.\nSpeaker D: And sometimes it's stronger, like A, which is like closer to EH, but- Yeah.\nSpeaker G: I'm just these poor transcribers, I know.\nSpeaker A: Well, we're not doing- We're not doing- We're not doing clause for us.\nSpeaker A: That's right.\nSpeaker A: Who knows?\nSpeaker E: And your native German speaker, so not a- Not an issue for-\nSpeaker G: It's only- Thick-thick Canadians.\nSpeaker E: Not only if you don't have black swallows, I guess, right?\nSpeaker A: That's my sense.\nSpeaker A: That's my sense.\nSpeaker A: That's my sense.\nSpeaker A: Yeah, and so, you know, I mean, I have- There are some Americans who are using this A2, and I haven't listened to it systematically.\nSpeaker A: Maybe with some of them, they'd end up being us, but- My spot checking has made me think that we do have A and also American data represented here.\nSpeaker A: But, in any case, this is reduced down from really quite a long- Much longer list.\nSpeaker A: Yeah, this is great.\nSpeaker A: This is really, really helpful.\nSpeaker A: Functionally pretty, you know, also.\nSpeaker A: It was fascinating.\nSpeaker A: I was listening to some of these, I guess, two nights ago, and it's just hilarious to listen to- To do a search for the m-hms, and you get m-hm, and do- Everybody's doing it.\nSpeaker A: Just doing it.\nSpeaker A: I think it would be fun to make a montage of it, because- Performance are just a extract.\nSpeaker A: Right.\nSpeaker A: It's really, it's really fun to listen.\nSpeaker A: All these different vocal tracks, you know, but it's the same item, it's very interesting.\nSpeaker A: Okay.\nSpeaker A: Then the acronyms, and the ones in parentheses are ones which the transgarber wasn't sure of, and I haven't been able to listen to it to clarify.\nSpeaker A: But you can see that the parentheses convention makes it very easy to find them, because it's the only question mark for- The question mark is punctuation, so they said that- Oh.\nSpeaker A: D-C?\nSpeaker A: So it's PLP?\nSpeaker A: Exactly.\nSpeaker A: Exactly.\nSpeaker A: Yeah, so the only- Well, and I do have a stress marker here.\nSpeaker A: Sometimes the contrast to a stress is showing up, and-\nSpeaker H: That's right, I got lost here. What's the difference between the parenthesis acronym and the non-prenthesis?\nSpeaker A: The parenthesis is something that the transgarber thought was A and N, but wasn't entirely sure.\nSpeaker A: So I'd need to go back, or someone needs to go back and say, you know, yes or no, and then get rid of the parentheses.\nSpeaker A: But the parentheses are used only in that context in the transcripts of- I've noticed- Noticing that there's something uncertain.\nSpeaker A: Yeah, I mean, they have no idea right.\nSpeaker E: If you hear CTPD, I mean, they do pretty well, but it's- I don't recognize- You know, how are they gonna know?\nSpeaker G: I think a lot of them are the networks we think.\nSpeaker A: I think that's true.\nSpeaker A: Yeah, absolutely.\nSpeaker A: In fact, a lot of these are coming from them.\nSpeaker A: I listen to some of that.\nSpeaker D: Although we don't have that many acronyms, comparatively.\nSpeaker D: In this meeting, it's not a bad thing.\nSpeaker A: And robustness is a fair amount, but the NSA group is just very, very many.\nSpeaker E: The recognizeer is funny.\nSpeaker E: kept getting PTA for PTA.\nSpeaker E: This is supposed to rain, and the PTA wasn't these topics about children.\nSpeaker E: That's interesting.\nSpeaker A: Is the PTA working?\nSpeaker A: Sometimes I mean, you see a couple of these are actually OK's.\nSpeaker A: So it's- it's- it's maybe that they got to the point where it was low enough, understandable- and stand-ability that they weren't entirely sure the person said OK.\nSpeaker A: You know, so it isn't really necessarily a- an undes ifable acronym, but it just needs to be double-checked.\nSpeaker A: Now we get to the comments.\nSpeaker H: The number to the left is the number of consonants.\nSpeaker A: The number of times out of the entire database, except for that last 30 minutes I haven't checked yet.\nSpeaker G: So CTS is really good.\nSpeaker C: Yeah, I wonder what it is.\nSpeaker C: So what is- I just between papers, rostling and drostling papers.\nSpeaker A: I'd have to listen.\nSpeaker A: I- I agree.\nSpeaker A: I'd like to standardize these down farther, but, um, to me that sounds equivalent.\nSpeaker A: I'm a little hesitant to- to collapse cross categories unless I actually listen to them.\nSpeaker G: Oh, I'm sure we've said XML more than five times.\nSpeaker G: Well, at least now.\nSpeaker B: That's a least six times.\nSpeaker B: OK, well- I'm so preferential.\nSpeaker E: Yes, it's very bad.\nSpeaker E: Well, this is exactly how people will prove that these meetings do differ because we're recording, right?\nSpeaker E: Yes.\nSpeaker E: Normally you don't go around saying, now you've said it six times.\nSpeaker A: But you notice that there were 785 instances of OK.\nSpeaker A: Yeah.\nSpeaker A: And that's not the right one.\nSpeaker A: And that's not the right one.\nSpeaker A: And that's not the right one.\nSpeaker A: The first question.\nSpeaker A: So, yeah.\nSpeaker A: On the page two acronyms.\nSpeaker D: Yeah.\nSpeaker D: Is this after- Like, did you do some replacements for all the different forms of OK to this?\nSpeaker A: OK.\nSpeaker A: So, that's the single existing convention for OK.\nSpeaker I: Wait a minute.\nSpeaker I: It's not worth a 788.\nSpeaker D: Although, there's one with a slash after it.\nSpeaker D: Yeah.\nSpeaker A: That's kind of disturbing.\nSpeaker A: I look for that one.\nSpeaker A: I actually explicitly look for that one.\nSpeaker A: And I think that I'm not exactly sure about that.\nSpeaker I: Is that somewhere where they were going to say new speakers or something?\nSpeaker A: No, I look for that.\nSpeaker A: That doesn't actually exist.\nSpeaker A: And maybe- I can't explain that.\nSpeaker D: That's all right.\nSpeaker A: It's only- It's the only pattern that has a slash after it.\nSpeaker A: And I think it's not-\nSpeaker G: I was just looking at the bottom of page three. Is that 2B or not 2B?\nSpeaker G: Yeah.\nSpeaker B: There's no tilde in front of it.\nSpeaker B: That's funny.\nSpeaker A: Yeah.\nSpeaker A: OK.\nSpeaker A: Anyways.\nSpeaker G: There is one.\nSpeaker G: There is one.\nSpeaker G: It's no on topic, Adam.\nSpeaker A: Well, let's- Let's legitimate- So, now, comments you can see they're listed again.\nSpeaker A: Same deal with the exhaustive listing of everything found and everything except for these final 30 minutes.\nSpeaker G: OK. So, on some of these quals- Yeah.\nSpeaker G: Are they really quals or are they glosses?\nSpeaker G: So, like, there's a- Qual TCL?\nSpeaker A: TCL. Where do you see that?\nSpeaker A: Uh-oh.\nSpeaker A: The reason is because it was said tickle.\nSpeaker G: Oh, I see, I see.\nSpeaker G: So, it's not gloss.\nSpeaker G: OK, I see.\nSpeaker G: Yeah.\nSpeaker D: It wasn't said TCL.\nSpeaker D: Should it be called TIC-KLE or something?\nSpeaker D: Like, it's not-\nSpeaker A: In the actual script, in the actual transcript, I-so this happens in the very first one.\nSpeaker A: I actually wrote it as tickle.\nSpeaker A: OK.\nSpeaker A: Because they didn't say TCL, they said tickle.\nSpeaker A: Yeah.\nSpeaker A: And then, following that is, Qual TCL.\nSpeaker A: Oh, I see.\nSpeaker D: OK.\nSpeaker D: I forget what's Qual.\nSpeaker A: Qualifier.\nSpeaker A: Just comment about what they said.\nSpeaker I: Comment.\nSpeaker A: Yeah.\nSpeaker A: Comment, dirt, text, your comment.\nSpeaker I: So, they didn't mean tickle as in, Elmo, then tickle.\nSpeaker I: But at some point, we probably-\nSpeaker E: We should add it to the dictionary. No, it's the pronunciation model.\nSpeaker E: What did I say?\nSpeaker G: Language.\nSpeaker G: Well, both.\nSpeaker G: We can add it to the language model.\nSpeaker E: Yeah, but it's the pronunciation model.\nSpeaker E: It has to have a pronunciation of TIC-KLE.\nSpeaker G: Well, TIC-KLE was pronounced TIC-KLE.\nSpeaker G: Right.\nSpeaker G: What do you say?\nSpeaker G: It's pronounced the same as the verb.\nSpeaker G: So, I think it's the language model that makes it different.\nSpeaker G: Oh, sorry.\nSpeaker E: What I meant is that there should be a pronunciation tickle for TCL as a word.\nSpeaker E: And that word stays in the language model wherever it was.\nSpeaker E: Right.\nSpeaker E: Yeah, you never would put tickle in the language model in that form.\nSpeaker E: Right.\nSpeaker E: There's actually a bunch of cases like this.\nSpeaker I: So, how would there be a problem for doing the language model and then with our transcripts over here?\nSpeaker E: Yes.\nSpeaker E: Yeah.\nSpeaker E: Yeah, so there's a few cases like that where the word needs to be spelled out in a consistent way as it would appear in the language, but there's not very many of these.\nSpeaker E: Tickles one of them.\nSpeaker E: And you'll have to do it synchronously.\nSpeaker E: Right.\nSpeaker G: So, whoever's creating the new models will have to also go through the transcripts and choose them synchronously.\nSpeaker E: Right.\nSpeaker E: We have this, there is this thing I was going to talk to you at some point about, you know, what do we do with the dictionary as we're updating the dictionary?\nSpeaker E: These changes have to be consistent with what's in the, like spelling people's names and so forth.\nSpeaker E: If we make a spelling correction to their name, like someone had Deborah Tannen's name, Miss Bell, and since we know who that is, you know, we can correct it, but we need to make sure we have the misspell, if it doesn't get corrected, we have to have a pronunciation as a misspell word in the dictionary, it's like that.\nSpeaker A: Well, of course now the, the Tannen, the spelling, I pick those up in the frequency check.\nSpeaker A: Right.\nSpeaker E: Right.\nSpeaker E: So, if there's things that get corrected before we get them, it's not an issue.\nSpeaker E: If there's things that we change later, then we always have to keep our dictionary up to date.\nSpeaker E: And then, yeah, in the case of tickle, I guess we would just have a word TCL, which, which normally would be an acronym, you know, TCL, but just has another pronunciation.\nSpeaker E: Yeah.\nSpeaker A: XCs is one of those that sometimes people pronounce and sometimes they say I see a sign.\nSpeaker A: So, those that are listed in the acronyms, I actually know they were said as letters.\nSpeaker A: The others, those really do need to be listened to, because I haven't been able to go to all the XC things.\nSpeaker A: And until they've been listened to, they stay as ICSI.\nSpeaker A: Right.\nSpeaker H: Don and I were just noticing, love this one over on page three.\nSpeaker H: Vocal gesture mimicking sound, obscuring something to head to whole blanket place.\nSpeaker H: It's great.\nSpeaker H: It was me.\nSpeaker G: It was, in fact, it was.\nSpeaker G: Yeah.\nSpeaker G: A lot of these are me.\nSpeaker G: He said he said he said with a high pitch and lengthening.\nSpeaker G: That was the, I was imitating beeping out.\nSpeaker G: Perfect.\nSpeaker E: Oh, there is something.\nSpeaker E: I spelled out BEEE.\nSpeaker E: Yeah.\nSpeaker E: That's been changed.\nSpeaker E: Thank you.\nSpeaker E: Because he was saying, how many of these do I have to allow?\nSpeaker D: You need a lot of qualification in it.\nSpeaker D: That's been changed.\nSpeaker A: So, exactly.\nSpeaker A: That's where the lengthening comment can't get in.\nSpeaker A: Right.\nSpeaker A: Right.\nSpeaker A: So, the vocalization.\nSpeaker A: And those, of course, get picked up on the frequency check, because CBE, and you know, I mean, it gets kicked out in the spelling and also gets kicked out in the frequency listing.\nSpeaker A: And I have the various things like breathe versus breath versus inhale.\nSpeaker A: And, you know, I don't know.\nSpeaker A: I think they don't have any implications for anything else.\nSpeaker A: So, it's like I'm tempted to leave them for now.\nSpeaker A: And it's easy enough to find them when they're in curly brackets.\nSpeaker A: We can always get an exhaustive listing of these things and find them and change them.\nSpeaker H: Things finale types on.\nSpeaker H: Yeah.\nSpeaker A: This is the first meeting.\nSpeaker A: Yeah, but I don't actually remember what it was, but that was Eric did that.\nSpeaker G: Yeah.\nSpeaker G: So, on.\nSpeaker G: Ta-da.\nSpeaker A: I think maybe something like that.\nSpeaker A: Maybe, yeah.\nSpeaker A: I'll take a call if I want.\nSpeaker G: On the glosses for numbers.\nSpeaker A: Yeah.\nSpeaker G: It seems like they're lost in different ways it's being done.\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker A: Now, first of all, very important.\nSpeaker A: Check, check lead to refinement here, which is to add numbs if these are parts of the red numbers.\nSpeaker A: Now, you already know that I had places where they hadn't transcribed numbers, put numbers in place of any kind of numbers, but there were places where they, this convention came later and at the very first digits task in some transcripts, they actually transcribed numbers and, um, check point out that this is a red speech and it's nice to have the option of ignoring it for certain other pop things.\nSpeaker A: And that's why there's this other tag here which occurs a 105, or 305 times right now, which is just, well, and numbs by itself, which means this is part of the numbers task.\nSpeaker A: I may change it to digits.\nSpeaker A: I mean, with the said command, you can really just change it however you want because it's systematically encoded.\nSpeaker A: Yeah.\nSpeaker A: You have to think about what's the best for the overall purposes, but in any case, numbers and numbs are part of this digits task thing.\nSpeaker A: Now, then I have these numbers that have quotation marks around them.\nSpeaker A: I didn't want to put them in its gloss comments because then you get the substitution.\nSpeaker A: And actually, the reason I did it this way was because I initially started out with the other version.\nSpeaker A: You have the numbers and you have the full form and the print of these.\nSpeaker A: However, sometimes people stumble over these numbers they're saying.\nSpeaker A: So you say, 78.2 or whatever.\nSpeaker A: And there's no way of capturing that if you're putting the numbers off to the side.\nSpeaker A: So what's the whole left of these?\nSpeaker A: The left is, so example, the very first one, it would be spelled out in words.\nSpeaker G: Okay, that's what I was asking.\nSpeaker A: 0.5.\nSpeaker A: Right.\nSpeaker A: Only it's spelled out in words.\nSpeaker A: So this is also spelled out in words.\nSpeaker A: 0.5.\nSpeaker A: Good.\nSpeaker A: And then in here, numbs.\nSpeaker A: So it's not going to be mistaken as a gloss.\nSpeaker A: It comes out as numbs.\nSpeaker A: Quote.5.\nSpeaker G: Okay, now the other example is in the glosses right there.\nSpeaker G: Gloss.1.1-130.\nSpeaker G: Oh, no.\nSpeaker A: What's the left of that?\nSpeaker A: In that case, it's people saying things like 1.1.1-so-so or they're saying 2.0, whatever.\nSpeaker A: And in that case, it's part of the numbers task and it's not going to be included in the red digits anyway.\nSpeaker A: So there will be a numbs tag on those lines?\nSpeaker I: There is.\nSpeaker A: Yeah.\nSpeaker A: I've added that all now too.\nSpeaker A: There's a numbers tag.\nSpeaker D: I'm sorry, I didn't follow that last thing.\nSpeaker A: So gloss.\nSpeaker A: In the same line that would have a gloss, quote, 1.1.1-130.\nSpeaker A: Right.\nSpeaker A: You'd have a gloss at the end of the line saying curly bracket, numbs curly bracket.\nSpeaker A: So if you did a graph minus V numbs, so you get rid of anything that was red.\nSpeaker E: So there wouldn't be something like, if somebody said something like, boy, I'm really tired, OK, and then started reading.\nSpeaker E: That would be in a separate line.\nSpeaker E: Yes.\nSpeaker E: OK, great.\nSpeaker E: Because I was doing the graph minus V quick and dirty and looked like that was working OK.\nSpeaker E: Good.\nSpeaker E: Great.\nSpeaker E: Now, why do we, what's the reason for having like the 0.5 have the numbs on it?\nSpeaker E: Is it just like when they're talking about their data or something?\nSpeaker A: This is more because, yeah, oh, these are all these, the numbs point side.\nSpeaker A: These are all where they're saying point.\nSpeaker A: It's something or other.\nSpeaker A: And the other thing too is for readability, the transcript.\nSpeaker A: I mean, if you're trying to follow this while you're reading it, it's really hard to read.\nSpeaker A: So in the data column 5 has 1.5 compared to 79.6.\nSpeaker A: It's like, when you see the words, it's really hard to follow the argument.\nSpeaker A: And this is just really a way of someone who would handle the data in a more discoracy way to be able to follow what's being said.\nSpeaker A: So this is where it chucks overall architecture comes in, where we're going to have a master file of the channelized data.\nSpeaker A: There will be scripts that are written to convert it into these main two uses.\nSpeaker A: And some scripts will take it down into a format that's usable for the recognizer.\nSpeaker A: Other scripts will take it to a form that's usable for linguistics and discourse analysis.\nSpeaker A: And the implication that I have is that the master copy will stay unchanged.\nSpeaker A: These will just be things that are generated.\nSpeaker A: And by using scripts, when things change, then the script will change.\nSpeaker A: But there won't be stored copies in different versions of things.\nSpeaker E: So I guess I'd have one request here, which is just maybe to make it more robust, that the tag, whatever you would choose for this type of numbs, where it's inside the spontaneous speech is different than the tag that you use for the red speech.\nSpeaker E: That would argue for changing the other ones.\nSpeaker E: That way, if we make a mistake parsing or something, we don't see the.5 or it's not there, then we just, and actually for things like 7-8th, or people do fractions too, I guess, maybe you want one overall tag for sort of that would be similar to that.\nSpeaker E: Or as long as there's different strings that will make our processing more robust, because we really will get rid of everything that has the numbs string in it.\nSpeaker I: I suppose what you could do is just make sure that you get rid of everything that has curly brace numbs, curly brace.\nSpeaker I: Exactly. That would be that. That was my motivation.\nSpeaker A: And these can be changed, like I said.\nSpeaker A: As I said, I was considering changing it to digits.\nSpeaker A: And it's just a matter of deciding on whatever it is and being sure the scripts know.\nSpeaker E: It would probably be safer if you're willing to have a separate tag, just because then we know for sure, and we can also do counts on them without having to do the processing.\nSpeaker E: But you're right, we could do it this way. It should work.\nSpeaker E: Yeah, and it makes it, I guess, the thing about... Probably not hard for a person to tell the difference, because one's in the context of a transcribed word.\nSpeaker A: The thing is you can get really so minute with these things and increase the size of the files and decrease of readability such an extent by simply something like percent.\nSpeaker A: Now, I could have adopted a similar convention for percent, but somehow percent is not so hard.\nSpeaker A: When you have these points and you're trying to figure out where the decimal voices are, percent's easy to detect. Point, however, is a word that has a couple different meanings, and you'll find both of those in one of these meetings.\nSpeaker A: We're saying the first point I want to make is so-and-so on.\nSpeaker A: Those two four points and also has all these decimals.\nSpeaker I: So, Liz, what does the recognizer do?\nSpeaker I: What is the SRA recognizer output for things like that?\nSpeaker I: 7.5. Does it output the word...\nSpeaker I: 7.5.\nSpeaker E: Right, the word 7?\nSpeaker I: The number 7?\nSpeaker I: The word 7.\nSpeaker I: Yeah.\nSpeaker E: And actually, you know, the language...\nSpeaker E: What we talk about point...\nSpeaker E: The same point, actually, the word 2 and the word...\nSpeaker E: They're going to and to go to.\nSpeaker E: Those are two different twos.\nSpeaker E: So, there's no distinction there.\nSpeaker E: It's just the word point.\nSpeaker E: Every word has only one version, even if it's...\nSpeaker E: Actually, even like the word read and read, those are two different words.\nSpeaker E: They're spelled the same way.\nSpeaker E: They're still going to be transcribed as R-E-A-D.\nSpeaker E: So, yeah, I like the idea of having this in there.\nSpeaker E: I was a little bit worried that the tag for removing the red speech.\nSpeaker E: Because what if we have like red letters or red...\nSpeaker E: We might want to just send them to the tag.\nSpeaker B: It says it's red.\nSpeaker E: Yeah, basically.\nSpeaker E: But other than that, it sounds great.\nSpeaker G: Okay.\nSpeaker G: Are we done?\nSpeaker A: Well, I wanted to say also regarding the channelized data that...\nSpeaker A: Yeah.\nSpeaker A: We requested that we get some segments done by hand to reduce the size of the time.\nSpeaker A: It's what was Chuck was mentioning earlier.\nSpeaker A: That if you said, oh, and it was in part of a really long complex overlapping segment, that the same starting end times would be held for that one, for the longer utterances.\nSpeaker G: We did that for one meeting, right?\nSpeaker G: So, you have that data, don't you?\nSpeaker G: Yeah, that's a training data.\nSpeaker A: He requested that there be similar samples done for five minute stretches, involving a variety of speakers and overlapping sections.\nSpeaker A: He gave me...\nSpeaker A: He did the very nice...\nSpeaker A: He did some shopping through the data and found segments that would be useful.\nSpeaker A: And at this point, all four of the ones that he specified have been done in addition.\nSpeaker A: I have the transcribers expanding the amount that they're doing, actually.\nSpeaker A: So, right now, I know that as of today, we got an extra 15 minutes of that type.\nSpeaker A: And I'm having them expand the realm on either side of these places where they've already started.\nSpeaker C: Okay.\nSpeaker A: But if...\nSpeaker A: And he's going to give me some more sections that he thinks would be useful for this purpose.\nSpeaker A: Because it's true.\nSpeaker A: I mean, if we could do the more fine grain tuning of this using an algorithm, that would be so much more efficient.\nSpeaker A: And so, this is going to be...\nSpeaker C: I thought we should perhaps try to start with those channelized versions, just to try to give one transcriber the channelized version of my speech and on speech detection.\nSpeaker C: And look, if that's helpful for them or just let them try, if that's better.\nSpeaker C: You mean to start from scratch and brand new transcriber?\nSpeaker A: That'd be excellent.\nSpeaker A: Yeah, that'd be really great.\nSpeaker A: As it stands, we're still on the phase of sort of cleaning up the existing data, getting things in more tightly, time in the line.\nSpeaker A: I also want to tell...I also want to bring the issue that...\nSpeaker A: Okay, so there's this idea we're going to have this master copy, the transcript, it's going to be modified by scripts into these two different functions.\nSpeaker A: And actually the master...\nSpeaker A: Two or more to...\nSpeaker A:...to our more.\nSpeaker A: And the master is going to be the channelized version.\nSpeaker A: Right.\nSpeaker A: So right now, we've taken this initial one.\nSpeaker A: It was a single channel, basically, the way it was input.\nSpeaker A: And now, thanks to the advances made in the interface, we can, from now on, use the channelized part and any changes that are made in the channelized version, anything.\nSpeaker A: But I wanted to get all the finishes and the checks.\nSpeaker A: So that has implications for your scripts.\nSpeaker D: So have those...the 10 hours that have been transcribed already, have those been channelized?\nSpeaker D: Yes they have.\nSpeaker A: And I've seen they've been channelized.\nSpeaker A: Except for the missing 30 minutes.\nSpeaker D: Right.\nSpeaker D: And they've been...has the time...have the time marking been adjusted?\nSpeaker A: For a total of like 20...for the total of the C-Pros, total of about 30 minutes, that's been the case.\nSpeaker A: And plus the training.\nSpeaker D: I guess, I mean, I don't know if we should talk about this now or now.\nSpeaker D: But that's just where...\nSpeaker D: Missing T.\nSpeaker D: Yeah, I know.\nSpeaker D: No, but I mean my question is, like, should I wait until all of those are processed and channelized like the time markings are adjusted before I do all the processing?\nSpeaker D: And we start like branching off into the...into our layer of...\nSpeaker A: Well, you know, the problem is that some...\nSpeaker A: some of the adjustments that they're making are to bring...\nSpeaker A: are to combine bins that were...\nSpeaker A: time bins, which were previously separate.\nSpeaker A: And the reason they do that is sometimes there's a word that's cut off.\nSpeaker A: Right.\nSpeaker A: And so it's true that it's likely to be adjusted in the way that the words are more complete.\nSpeaker A: And...\nSpeaker A: Okay.\nSpeaker A: No, I know that adjusting those things is going to make it better.\nSpeaker D: Yeah.\nSpeaker D: I'm going to be more reliable in that.\nSpeaker D: I'm not sure.\nSpeaker D: I'm not sure.\nSpeaker D: Yeah.\nSpeaker D: I'm going to be more reliable in that.\nSpeaker A: It's going to make it better.\nSpeaker A: Yeah.\nSpeaker A: I'm going to be more reliable in that.\nSpeaker A: Yeah.\nSpeaker A: And it will be to apply an algorithm because...\nSpeaker A: Yeah.\nSpeaker A: This takes time.\nSpeaker A: You know, it takes a couple hours to do 10 minutes.\nSpeaker D: Yeah, I don't doubt it.\nSpeaker D: So...\nSpeaker I: So right now, what you're doing is you're taking the...\nSpeaker I: the original version and you're sort of channelizing yourself.\nSpeaker I: Yeah.\nSpeaker D: I'm doing it myself.\nSpeaker D: I mean, if the time markings aren't different across channels, like the channelized version really doesn't have any more information.\nSpeaker D: So I was just... I mean, originally I had done it before, like the channelized versions were coming out.\nSpeaker D: Right.\nSpeaker D: And so it's a question of...\nSpeaker I: I think probably the way it'll go is that, you know, when we make this first general version and then start working on the script, that script that will be made, you know, primarily come from what you've done, we'll need to work on a channelized version of those originals.\nSpeaker I: And so it should be pretty much identical to what you have to...\nSpeaker I: except for the one that they've already tightened the boundaries.\nSpeaker I: Right.\nSpeaker I: Yeah.\nSpeaker I: So...\nSpeaker I: And then probably what will happen is as the transcribers finish tightening more and more, you know, that original version will get updated and we'll rerun the script and produce better versions.\nSpeaker I: Okay.\nSpeaker I: But I guess the effect for you guys, because you're pulling out the little waveforms into separate ones, that would mean these boundaries are constantly changing.\nSpeaker I: You definitely constantly rerun that.\nSpeaker I: Right.\nSpeaker E: But that's not hard.\nSpeaker E: I think the harder part is making sure that the transcription...\nSpeaker E: So if you merge two things, then you know that it's a sum of the transcripts.\nSpeaker E: But if you split inside something, you don't know where the word, which words moved.\nSpeaker E: And that's where it becomes a little bit having to rerun the processing.\nSpeaker E: The cutting of the waveform is pretty trivial.\nSpeaker D: I mean, as long as it can all be done automatically, I mean, then that's not a concern.\nSpeaker D: So if I just have to run three scripts to extract it, I'll let it run on my computer for an hour and a half, or however long it takes to parse and create all the reference files, it's not a problem.\nSpeaker D: So yeah, as long as we're at that point, and I know exactly what steps, what work, what's going on in the editing process.\nSpeaker D: Okay.\nSpeaker A: So that's...\nSpeaker A: I mean, I could...\nSpeaker A: There were other checks that I did, but it's...\nSpeaker A: I think that we...\nSpeaker A: Unless you think there's anything else I think that I covered it.\nSpeaker A: I can't give...\nSpeaker C: Any other...\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: Oh.\n"}, {"meeting_id": "IS1007d", "summary": "Industrial Designer and User Interface gave a presentation about two versions of the prototype, one with and one without LCD. They both had a speech recognition system, microphone, a few number of buttons, rechargeable batteries and different modules. The prototype with LCD was too expensive, while the other one was quite cheap. Marketing made evaluation criteria. The scale included fancy look and feel, innovativeness and easy to use. Then Marketing asked Industrial Designer and User Interface to give scores towards two versions of the prototype according to the scale. Both Industrial Designer and User Interface gave average points to look and feel, high points to innovativeness and to easy to use. At the end of the meeting, group members did project evaluation. They felt good about leadership and teamwork.", "dialogue": "Speaker C: Last meeting we have discussed the conceptual design and we ask a new to prepare a prototype for the remote control.\nSpeaker C: At the end.\nSpeaker C: Okay.\nSpeaker C: So let's see what.\nSpeaker B: Yeah. So can you go out to the shed folder?\nSpeaker B: The shed folder?\nSpeaker C: The shed folder for your presentation.\nSpeaker B: We have a presentation because I have a...\nSpeaker B: The palm.\nSpeaker B: So I got about this open three.\nSpeaker B: Three.\nSpeaker B: It's a final design.\nSpeaker B: Yeah.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker B: So I discussed the view.\nSpeaker B: Yeah.\nSpeaker B: And so we have...\nSpeaker B: And we are both agreed on some two versions of the prototypes because we were not decided whether we wanted to have an LCD or not because it's too expensive.\nSpeaker B: So we come out with two versions.\nSpeaker B: One with and one with our LCDs.\nSpeaker B: But both comes with a charger and then detachable or unparental control module and detachable big buttons for all people.\nSpeaker B: So...\nSpeaker D: Okay.\nSpeaker D: So I'll show you the two prototypes.\nSpeaker D: Here are the first ones with the beautiful LCD and display.\nSpeaker D: Here.\nSpeaker D: And you can just browse into the navigation menu by just the joystick-like button.\nSpeaker D: You can choose the direction and if you just push on it, it's considered like an enter function.\nSpeaker D: You have on the side here the microphone for the speech recognition system here.\nSpeaker D: And here the switch that controls if you want.\nSpeaker C: But why you put it in the side.\nSpeaker D: Well, I think it's not a good thing to do that.\nSpeaker B: No, it's all around camera. The microphone picks up the speech from only a single microphone.\nSpeaker D: So it's a microphone array.\nSpeaker C: Oh, it's very costly.\nSpeaker D: It's just a single microphone.\nSpeaker D: I think we put it here because I think when you are browsing your LCD, you will be close to...\nSpeaker D: Well, it's better to place it here than here, for instance.\nSpeaker D: And here is a little switch that controls if you want the speaker recognition system to be on or off.\nSpeaker D: And so this remote control comes up with it's shorter.\nSpeaker D: How much that is cost this one?\nSpeaker D: Well, this prototype is made for about, well, 15 dollars.\nSpeaker D: Yeah, 15 dollars.\nSpeaker D: 15 dollars.\nSpeaker D: But, well, it's not a...\nSpeaker C: It's not a...\nSpeaker D: It's not a...\nSpeaker D: Yeah, but it's just a prototype and if we optimize the...\nSpeaker D: The cost will build up.\nSpeaker D: Yeah, look at the current and the prediction costs.\nSpeaker D: We can achieve about 10 dollars.\nSpeaker C: How many...\nSpeaker C: Who's there?\nSpeaker D: How many, excuse me, but a...\nSpeaker D: Well, battery...\nSpeaker D: We use about...\nSpeaker B: The two AAs batteries, isn't it?\nSpeaker B: Yeah.\nSpeaker B: AA rechargeable batteries.\nSpeaker B: Retouchable, of course.\nSpeaker D: Of course we have the charger.\nSpeaker B: We have the charger, so it's no problem.\nSpeaker C: You just have one battery?\nSpeaker D: Yeah, one battery.\nSpeaker E: Is it two or one?\nSpeaker C: It's kitty-tick.\nSpeaker B: Actually, it's a flexible thing.\nSpeaker E: No, what is the voltage rating for that?\nSpeaker E: The voltage rating.\nSpeaker E: What type of battery?\nSpeaker D: Oh, yeah, it's just a simple battery, a rechargeable...\nSpeaker D: Yeah.\nSpeaker D: If you like it, it's just...\nSpeaker E: Like two A3 size batteries.\nSpeaker D: Yeah.\nSpeaker D: So, and if you want to charge the battery, you just put the remote control like that, to plug in the charger and leave it alone.\nSpeaker D: It's all right.\nSpeaker D: Then the next time you pick it, oh...\nSpeaker D: It works.\nSpeaker B: Yeah.\nSpeaker B: I forgot to tell you there is only a single button there.\nSpeaker B: Just explain the button.\nSpeaker B: Yeah, all right.\nSpeaker B: This button is like the mouse.\nSpeaker B: It's like a joystick.\nSpeaker B: You can move in 360 degrees of direction.\nSpeaker B: So, you can make a turn, and it...\nSpeaker B: There are functions associated to all the actions you make.\nSpeaker B: You have a click, you have a double click.\nSpeaker B: Oh, and a single button.\nSpeaker B: You can move it up, down, left, right, or you can do a swing.\nSpeaker B: So, a swing to the left, a swing to the right, defines other functions.\nSpeaker B: So, even though it's a single button, it's pretty flexible because of the 360 degree movement.\nSpeaker E: And the LCD is this one, on the remote.\nSpeaker B: Yeah.\nSpeaker B: This is the version that comes with LCD.\nSpeaker B: Yeah.\nSpeaker B: Here, present another version without the LCD.\nSpeaker C: I think it's not a good idea, because after maybe one or two months of function, this button gets destroyed.\nSpeaker C: Oh, okay.\nSpeaker B: This is a new prototype.\nSpeaker D: So, we have the second version, and also with the different button configuration.\nSpeaker D: The second version is also simpler.\nSpeaker D: We just have to put the microphone.\nSpeaker D: So, basically, it's the same version.\nSpeaker D: Yeah.\nSpeaker D: Basically, it's the same thing as I presented before.\nSpeaker D: But here we have four buttons for navigation, plus one button acting as an enter button.\nSpeaker D: So, according to what you said, it's more robust to the user.\nSpeaker D: And it's also cheaper to produce.\nSpeaker D: You can produce set remote control for about four dollars.\nSpeaker E: And I think you forgot the point to have a button to find the charger.\nSpeaker E: Oh, no.\nSpeaker D: Because there's a measure.\nSpeaker D: Actually, it's embedded in the...\nSpeaker D: Yeah.\nSpeaker D: Speech recognition system.\nSpeaker E: Okay.\nSpeaker E: And if you disable speech recognition system, then?\nSpeaker B: I'll come to that point later.\nSpeaker D: So, no one will explain to you.\nSpeaker C: Maybe we will share the charger with this.\nSpeaker B: Yeah, yeah.\nSpeaker B: Of course.\nSpeaker B: Either this, the charger, in either versions, they use the same charger.\nSpeaker C: Okay.\nSpeaker C: So, the price of the charger included in the...\nSpeaker B: It's a standard module, so you should get for a cheap price.\nSpeaker B: Thank you.\nSpeaker D: It's the same charger that you can use for the hand form.\nSpeaker B: But again, it depends on the types.\nSpeaker B: I think we have to investigate more on that.\nSpeaker C: The price should be below 12 and 12.\nSpeaker C: Yeah.\nSpeaker B: But as the marketing manager says, people are willing to pay more for good design.\nSpeaker E: Yeah, people are willing to pay more, but the company is not willing to invest more.\nSpeaker C: The price of the money is 25 euros.\nSpeaker C: And the price of...\nSpeaker D: All right.\nSpeaker D: So, go on now, man, with the special features.\nSpeaker B: All right.\nSpeaker B: Well, come out to that.\nSpeaker B: The thing that makes the controller cheap is that it is modular.\nSpeaker B: So, you want more function, you pay more.\nSpeaker B: If you want less function, if you want a reasonable price, you pay for the function that you add on to the system.\nSpeaker B: So, it's modular.\nSpeaker C: So, there's basis function, and if you want more, you pay.\nSpeaker C: Yeah.\nSpeaker B: For example, the LCC can put it back in, or you can use the other one.\nSpeaker B: Oh.\nSpeaker B: The speaker can raise the microphone.\nSpeaker B: Yeah.\nSpeaker B: Then the microphone.\nSpeaker B: Then it's actually...\nSpeaker B: You don't want...\nSpeaker B: You pay less for the system, you see.\nSpeaker B: So...\nSpeaker E: Okay.\nSpeaker B: It's pretty flexible in the...\nSpeaker D: You also have the two other modules for the parental control.\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker D: This is the other one.\nSpeaker D: You can add that to your remote control.\nSpeaker D: If you want to have more power on what you do, and with channel you want to choose, and so on, here is just the module for the kids, and if the parents want to watch TV, they come up with their modules, they just plug in it, and they can have all the control they want here.\nSpeaker D: We also have this module for old people, with big buttons, clearly labeled, and it acts like the previous one, you just plug in, and it works.\nSpeaker B: Yeah.\nSpeaker B: The idea, the conceptual idea is that simplicity and powerful simplicity, meaning that we have few buttons.\nSpeaker B: Powerful is that all the controls, the one you saw earlier in the meetings, with lots of buttons, but here you only have few buttons, but you retain them, the same powerful functionality.\nSpeaker B: But in addition, with simplicity.\nSpeaker B: That's the best idea.\nSpeaker B: That's our innovation in this design.\nSpeaker B: Okay, now we come to other important features that I did not tell you.\nSpeaker B: The first one is the speech organizer.\nSpeaker B: Again, it's detachable, or add-on.\nSpeaker B: And then we also have security feature, for example, we see, oh!\nSpeaker B: It's very robust, basically.\nSpeaker B: It doesn't break in the material, what's the material?\nSpeaker B: It's very rare.\nSpeaker B: Again, add-on battery life is an end-yourable one.\nSpeaker B: Again, because it's AA batteries, you can choose the types of batteries you want.\nSpeaker B: Lithium, you're maybe a good one, but you can replace with a cheaper one.\nSpeaker B: Again, you pay for what you get.\nSpeaker B: And then the other point is a robust one, it's a plasma reminder.\nSpeaker B: So, when you finish watching your TV, and you turn off the TV, and then there's this message coming out, please put that back to the charger.\nSpeaker B: Okay.\nSpeaker B: So, that's the reminder part.\nSpeaker B: Yeah.\nSpeaker B: And also, user programmable channel and vocabulary.\nSpeaker B: So, we didn't define the vocabulary, so it's after the user defined.\nSpeaker B: So, sorry.\nSpeaker B: And then, because children are using the device, we have also a pre-filter of TV programs by Jean.\nSpeaker B: This can be used by the adult or by the children.\nSpeaker B: So, instead of choosing the channel, you are choosing the TV contents of the night.\nSpeaker B: So, it's pretty powerful.\nSpeaker B: And that's why the knob buttons are reduced because of this feature.\nSpeaker B: Yeah.\nSpeaker B: And for the materials that is cheap to produce, I think it's quite clear from...\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker B: And also lightweight and offensive designs.\nSpeaker B: Maybe you can improve more on the design.\nSpeaker B: Maybe, yeah.\nSpeaker C: What's the price to produce?\nSpeaker A: This is...\nSpeaker D: Well, so the price to produce for the simplest one, say, we start from four dollars to produce such a device.\nSpeaker D: It's about...\nSpeaker C: Without the charger?\nSpeaker D: Yeah.\nSpeaker D: It's about, well, two dollars to produce the microphone and speech recognition systems.\nSpeaker D: Two dollars for the... for the... I don't know.\nSpeaker D: And finally, three dollars for the charger.\nSpeaker D: So, if you sum up everything, we don't have charger with the head CD, which gets two dollars.\nSpeaker D: You have two plus four plus four plus two plus two.\nSpeaker E: I think you're the excel.\nSpeaker C: We don't have all the options here.\nSpeaker D: So, the total cost if you want to hold the functionality, the piece would be about 15 dollars.\nSpeaker D: Yeah.\nSpeaker D: But it's just... if you want all the functionality...\nSpeaker C: Excuse me.\nSpeaker C: It's below the... the...\nSpeaker E: Yeah, it's a nice input, but we have an...\nSpeaker E: other input from the public demands.\nSpeaker E: I think we would just have a rough look and then we can make all the statements and we can finalize the product based on this discount.\nSpeaker E: I think we can just go to my presentation and we can wind up...\nSpeaker E: or we could come to some final conclusions.\nSpeaker E: So, the marketing made an evaluation criteria and this was some of the findings.\nSpeaker E: And in the market, the people are not really interested with LCD, without LCD, with speech recognition interface or without speech recognition interface.\nSpeaker E: But most of the people, what they are interested in is...\nSpeaker E: first thing is they want to have a fancy look and feel.\nSpeaker E: It should be very fancy with colourful and very handy to hold.\nSpeaker E: And the second thing is it should be much more technologically innovative.\nSpeaker E: Of course, in that we could put LCD or recognition to be more technologically innovative.\nSpeaker E: And the third one is easy to use.\nSpeaker D: Well, I think that the technical innovation is in the predicted set of scenes.\nSpeaker D: We haven't yet a remote who are Halloween to choose interactively with the TV screen.\nSpeaker D: We just for buttons.\nSpeaker E: If you go to the next slide, here you can find these are the latest fashion updates.\nSpeaker E: And so this is Paris and Milan devs for their detector, these trends.\nSpeaker E: That this year it's going to be an ear of fruit and vegetable.\nSpeaker E: So people are really interested to see if they have a remote in the shape of the fruit or vegetable or whatever they like.\nSpeaker E: So I think it should be much more customised to make different shapes.\nSpeaker E: And the second thing is, and if the material, they really do not want it to be very hard.\nSpeaker E: As in the case of very plastic or titanium, it should be somewhat spongy.\nSpeaker B: But the problem is that... is it robust to mishandling?\nSpeaker B: We should find a material that is robust at the same time as spongy.\nSpeaker E: So finally we have these three criteria.\nSpeaker E: One is fancy look.\nSpeaker E: The second is...\nSpeaker E: In a way too.\nSpeaker E: And third is easy to use.\nSpeaker E: So these are the three criteria on which we are going to build our remote.\nSpeaker E: And we have an evaluation criteria for each one of these.\nSpeaker E: Say that we have a seven point scale from one to seven.\nSpeaker E: And for each of the product, you could just give me the scale according to this.\nSpeaker E: So now you have with the LCD and without LCD.\nSpeaker E: So on this scale, if it is true, if it is... if it has a fancy look, you could just tell me one.\nSpeaker E: Or if it is false, it doesn't... it looks... it doesn't look much fancy.\nSpeaker E: Then you could just tell me seven.\nSpeaker E: So we can make our study on this.\nSpeaker B: So is the evaluation depending on us or other users?\nSpeaker E: No, no.\nSpeaker E: We have... you have designed two products now.\nSpeaker E: One is with LCD and without LCD.\nSpeaker B: But who will give the scoring?\nSpeaker B: We ourselves or...\nSpeaker B: You.\nSpeaker B: According to you...\nSpeaker E: What's the party?\nSpeaker E: No, according to you designers.\nSpeaker E: How you feel does it with LCD and without LCD?\nSpeaker E: Okay.\nSpeaker E: Let us make this L and without LCD.\nSpeaker E: On this scale, the LCD remote control.\nSpeaker E: How do you look? How does it look?\nSpeaker E: Is it very fancy or doesn't look much fancy?\nSpeaker D: Well, it doesn't look much fancy.\nSpeaker D: I'll say three or four.\nSpeaker E: And you both agree for that?\nSpeaker C: And you?\nSpeaker B: I think we can improve on the design.\nSpeaker B: Yeah.\nSpeaker A: Some are commented.\nSpeaker B: This is like the stone edge, frankly.\nSpeaker B: Design.\nSpeaker B: We have been focused all this time on the...\nSpeaker B: Technical aspect.\nSpeaker B: On the technical aspect, functional aspect.\nSpeaker B: But also the simplicity.\nSpeaker B: As for the design, maybe we should hire a designer.\nSpeaker E: So I take three on the LCD.\nSpeaker E: So without LCD, how would you rate it?\nSpeaker D: Four.\nSpeaker D: I think it's more easy to make fancy things when it's not complicated.\nSpeaker D: So removing the LCD gives us more liberty to have a fancy look.\nSpeaker E: And in the sense of innovativeness with LCD?\nSpeaker D: Well, it's the same for both.\nSpeaker D: So I will give a five, six.\nSpeaker E: Five with LCD and...\nSpeaker D: Five or so.\nSpeaker D: Norman, please.\nSpeaker B: From the innovation aspect, I feel that what is most innovative of our prototype compared to the existing prototype that we saw in the last year.\nSpeaker B: So in the last meeting, is that the retrieval of TV programs by Content.\nSpeaker B: Because that is really innovative.\nSpeaker B: And for now, another innovation we have is simplicity and simplicity.\nSpeaker B: Few buttons with...\nSpeaker D: With a lot of functionalities.\nSpeaker B: A lot of functionalities.\nSpeaker B: So without LCD...\nSpeaker B: I don't think that our product has the best, you know, well, this is a bias, judgment.\nSpeaker B: No, no.\nSpeaker E: According to design aspect, we want to know how do you feel the innovativeness?\nSpeaker B: And the innovation is very high.\nSpeaker B: With LCD.\nSpeaker E: And without LCD?\nSpeaker D: For both.\nSpeaker D: For both.\nSpeaker D: The same innovation.\nSpeaker B: It's the same innovation.\nSpeaker B: Okay.\nSpeaker B: So you have to make it with LCD.\nSpeaker D: No, it's 6.\nSpeaker B: No, it's 6.\nSpeaker E: Without LCD.\nSpeaker E: Both.\nSpeaker E: No, actually, to make it with LCD, you make more efforts.\nSpeaker E: So there is high innovativeness included if you make it with LCD.\nSpeaker E: Yeah.\nSpeaker E: Then when it is without LCD, there is not much innovativeness.\nSpeaker B: So you can't go...\nSpeaker B: There was an issue here.\nSpeaker B: What is display on LCD?\nSpeaker B: We haven't really...\nSpeaker B: That's what it is.\nSpeaker B: That's what the design is.\nSpeaker B: Actually, it's good to have an LCD, but what are we going to display on LCD?\nSpeaker B: No, it's like this.\nSpeaker E: I mean, on the LCD, according to what I understand from your model is...\nSpeaker E: So let's remove it.\nSpeaker E: You have a joystick here.\nSpeaker E: Yeah.\nSpeaker E: And you have LCD.\nSpeaker E: Yeah.\nSpeaker E: Just for us, you are joystick.\nSpeaker E: You get here a program.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But you have the same program on the TV screen.\nSpeaker D: Yes.\nSpeaker D: Exactly.\nSpeaker E: The TV is the problem.\nSpeaker E: No, on the TV, it doesn't display on the TV now.\nSpeaker E: If you have the LCD.\nSpeaker E: LCD, then there is no meaning in having on the TV.\nSpeaker E: Yeah.\nSpeaker E: Then again, you have a channels, volume, and all this stuff.\nSpeaker B: But what do you want to display all on the LCD?\nSpeaker E: I mean, that depends upon your design.\nSpeaker E: So...\nSpeaker B: Anyway, for the users who opt for LCD, we are giving them the LCD.\nSpeaker B: We give them the customer ones, right?\nSpeaker E: Yeah.\nSpeaker E: So without LCD, you want to put it on the same level of you know, virtualness.\nSpeaker E: And which word do you recommend?\nSpeaker E: Is it to use?\nSpeaker E: With LCD or without LCD?\nSpeaker D: Well, I think both are really easy to use because there are a few buttons.\nSpeaker D: Yeah.\nSpeaker D: And the menu are clear.\nSpeaker D: Yeah.\nSpeaker D: And it's nice, so...\nSpeaker D: Well, with little knowing of the technology, you can easily use the product.\nSpeaker D: So I would give a six for the user to use.\nSpeaker B: I think...\nSpeaker C: Sorry.\nSpeaker C: No, it's just if I...\nSpeaker C: I see the...\nSpeaker C: the rate...\nSpeaker C: I think it's better to do it without the LCD.\nSpeaker C: Because with LCD, just we increase the price.\nSpeaker C: And I think they have the same nature.\nSpeaker E: Or you can just go back to the previous slide where we have a few updates.\nSpeaker E: So, I mean, in the product design, you could just amend it to make some future changes, according to the fruit or on the stable or as a spongey touch, without LCD and without speech recognition.\nSpeaker E: Even then, our product is going to be very good.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And if we have less technology, we have more liberty to...\nSpeaker D: Yeah, that's right.\nSpeaker D:...for the shape and things like that.\nSpeaker B: I have another thing to say about the easy to use and spend.\nSpeaker B: I think the easy to use and spend is different for different people.\nSpeaker B: For the young generation, easy to use may be very complicated because they want a lot of controls with lots of buttons.\nSpeaker B: For kids, they want simplicity, a city.\nSpeaker B: Similarly, for the old people, they want simplicity.\nSpeaker B: So, that's why we have the parental module.\nSpeaker B: We can have a bit of more complex design if they want it.\nSpeaker B: But what I'm trying to say here is that we have different sets of buttons for different kinds of people.\nSpeaker B: So, in terms of easy to use, I think it's very user-customizable.\nSpeaker B: User-customization is very important.\nSpeaker B: Yeah.\nSpeaker E: So, without LCD, I just take it 5 or you want it to be 6.\nSpeaker D: I think it's the same.\nSpeaker D: Yes, as you say, we'd better...\nSpeaker D: Without LCD.\nSpeaker E: I think it's better to have this without LCD.\nSpeaker D: And to improve the...\nSpeaker E: As our project manager says that without LCD, it is going to reduce the cost also.\nSpeaker E: It's going to be much simpler to use.\nSpeaker D: It's more on e-beady to have a fancy look.\nSpeaker A: Yeah.\nSpeaker D: So, let's go back to our laboratory.\nSpeaker C: It's true for what's about this is a special permission.\nSpeaker C: And you want to buy the integration.\nSpeaker B: The special recognizer is an on-module.\nSpeaker E: Yeah. That's going to be an optional.\nSpeaker E: If someone wants to buy it, they can have it otherwise.\nSpeaker D: So, it's an optional...\nSpeaker D: Yeah.\nSpeaker D: Since, well, according to this study, people are more...\nSpeaker C: I think it's more...\nSpeaker C: That was pongy.\nSpeaker C: I'm really in control there.\nSpeaker C: I think it should be better if we have all these modules in the same remote container.\nSpeaker C: Because maybe it's balanced.\nSpeaker E: Yeah, I mean, if you have an add-ons, the kids may just hide the...\nSpeaker E: Hide the parental module.\nSpeaker C: And so that the parents can't use it.\nSpeaker C: Oh, yeah.\nSpeaker C: Can't forget where they put it.\nSpeaker C: So, maybe better if you have all these in the same.\nSpeaker C: In the same sets, yeah.\nSpeaker E: And individual buttons to make them work.\nSpeaker B: The problem is if we are black, if a customer wants a certain component and doesn't want the printer control because it's a couple who do not have children.\nSpeaker B: When you sell the product on the market, you could meet any kind of people.\nSpeaker B: And not all people will need all the functionalities you propose to them.\nSpeaker B: So, we might lose the customer because of these.\nSpeaker B: I don't know what do you think...\nSpeaker C: So, you mean that even if these modules will be functional, they will be optional?\nSpeaker B: Yes.\nSpeaker B: What do you think?\nSpeaker B: I don't know.\nSpeaker B: Yeah, I mean...\nSpeaker E: I have to make a marketing survey that...\nSpeaker E: Yeah.\nSpeaker E: How many married couples or how many couples have a TV or to the number of singles who have a television?\nSpeaker E: Yeah.\nSpeaker E: So, based on that, I think...\nSpeaker B: Yeah.\nSpeaker B: But the question is, does those modularity increase the production cost or decrease the production cost?\nSpeaker B: Well, it's about the same because if...\nSpeaker D: Well, you can...\nSpeaker D: If you had something like...\nSpeaker D: I think it's the easiest.\nSpeaker D: Oh, no.\nSpeaker D: Because if you...\nSpeaker D: If you had something...\nSpeaker D: Well, yeah, maybe because it's three different pieces to build.\nSpeaker E: But you can make it on a single piece with three different options.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I think it's cheaper if you already built the whole functionalities on the same material.\nSpeaker D: Yeah.\nSpeaker E: On the same piece.\nSpeaker B: Well, the other aspect is that if in the... when they have the existing product, you're gonna maintain the...\nSpeaker B: You're gonna keep the same customer because the same customer will come back to buy other ads on...\nSpeaker E: Yeah, that's right.\nSpeaker B: So...\nSpeaker E: How many good is it, provide with the options and this?\nSpeaker B: Yeah.\nSpeaker B: So, if you don't don't buy it now, they can buy it in the future.\nSpeaker B: So, in that way, you're actually keeping the same customer, attracting them to come back again and again.\nSpeaker B: So, something like a Microsoft product.\nSpeaker E: Yeah, that's fine.\nSpeaker E: We don't want to...\nSpeaker C: And we don't want to...\nSpeaker B: Updates and we sell it.\nSpeaker E: We make updates and sell it for this.\nSpeaker E: Yeah.\nSpeaker B: Well, if the buyers can be updated, it can be upgraded.\nSpeaker B: It'd be a good thing.\nSpeaker B: Yeah.\nSpeaker B: They had the depends on the product.\nSpeaker E: And what's the idea about the shape of the remote controls?\nSpeaker E: Well, sure.\nSpeaker E: Can they be made into a fruit and vegetable types?\nSpeaker D: Yeah.\nSpeaker D: They require different types of PCBs.\nSpeaker D: Yeah, but, well, so the components will have to change because we have to deal with the shape.\nSpeaker D: But I don't think it will increase the cost much.\nSpeaker E: Or it would be like this.\nSpeaker E: The PCB would be the same for all, but only the exterior shape is different for all.\nSpeaker D: Well, but it's a bit complicated to...\nSpeaker D: If you need different machines to...\nSpeaker B: What we can change is to propose the customers with skins.\nSpeaker B: For example, the mobile phones nowadays, there's different kinds of skin.\nSpeaker B: And then people just feel that, oh, I have a new skin and it looks better.\nSpeaker D: Yeah.\nSpeaker D: We can have the same global shape and then add on other skins.\nSpeaker D: Yeah.\nSpeaker D: And with this technology texture, things like that.\nSpeaker B: You can't fly.\nSpeaker B: In what material would that be?\nSpeaker B: Spong.\nSpeaker E: You need to look into the material.\nSpeaker B: Yeah.\nSpeaker B: A flari, like a pet dog kind of thing.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker D: So we finished a prototype presentation, if you have something else to say.\nSpeaker E: I think the project manager would be interested to look about the financial issues that make these models.\nSpeaker C: Project evaluation project.\nSpeaker C: Project.\nSpeaker C: Project.\nNone: I don't know.\nSpeaker C: We have to ask this question.\nSpeaker C: We have to give answers.\nSpeaker B: I suppose if we let this criteria...\nSpeaker C: Project evaluation.\nSpeaker C: Let's go.\nSpeaker B: Room for creativity.\nSpeaker B: This room is a bit smaller.\nSpeaker B: But I think it's okay for us to work with.\nSpeaker E: I feel it's fine.\nSpeaker E: We don't.\nSpeaker E: I think it's okay.\nSpeaker B: Yeah.\nSpeaker E: It's fine for making a presentation, for project presentations.\nSpeaker B: Yeah.\nSpeaker B: Anyway, we have the laboratory nearby to build our models.\nSpeaker B: It's okay.\nSpeaker E: And the leadership was excellent.\nSpeaker C: Yeah.\nSpeaker C: I think so.\nSpeaker C: Good job.\nSpeaker C: Right.\nSpeaker C: He gave you the liberty to go.\nSpeaker C: Yeah, that's right.\nSpeaker C: Yeah, that's right.\nSpeaker C: The teamwork was very, very good.\nSpeaker C: Very dynamic.\nSpeaker C: I am very satisfied to work with.\nSpeaker D: Thank you, Abbas.\nSpeaker B: Thank you, Mumau.\nSpeaker D: I'm not sweet.\nSpeaker D: All right.\nSpeaker A: All right.\nSpeaker E: New ideas for them.\nSpeaker E: Any new ideas to make this presentation more interactive and more interesting?\nSpeaker D: Come up with new product.\nSpeaker E: What type of product do you think could we make this more interesting?\nSpeaker C: Less fancy.\nSpeaker B: I know something.\nSpeaker B: We need more kegs, more bits, goods on the table where we have meetings.\nSpeaker D: Yeah.\nSpeaker D: And also more, well, more seriously, I think it's a good idea to have clear points whenever we do a meeting, pre-order the meeting to have a, well, at least with the points, we'll be discussed or things like that.\nSpeaker D: But I think it's all right.\nSpeaker D: We achieved project goal.\nSpeaker E: I think that is within the budget.\nSpeaker E: Yes.\nSpeaker C: And the evaluation.\nSpeaker C: That was without ICD, without special copies.\nSpeaker C: Yeah, but we seem to.\nSpeaker E: All right.\nSpeaker E: And then I do celebration.\nSpeaker D: So, coffee machine.\nSpeaker C: Coffee.\nSpeaker B: All right, then we finish.\nSpeaker B: Yeah.\nSpeaker E: Thank you very much.\nSpeaker C: All right.\nSpeaker C: All right.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: All right.\n"}, {"meeting_id": "TS3003d", "summary": "The whole meeting was the final meeting of the project for discussion about the final design and project evaluation. Firstly, though Project Manager came late, he efficiently started the prototype presentation, during which Industrial Designer and User Interface presented the final design they worked out together. However, part of the features contained in their design, particularly too many buttons, were effectively rejected in the finance phase to reduce cost. Then, Marketing hosted the product evaluation and calculated a total score for their new remote control. After that, Project Manager guided the process evaluation and led the project to the closing phase.", "dialogue": "None: Hello.\nNone: Good afternoon.\nSpeaker D: Sorry I'm a little late.\nNone: Oh, probably.\nNone: It got stuck in the traffic.\nSpeaker C: It's possible this time of day.\nSpeaker C: It starts at 3 o'clock.\nSpeaker B: Yeah.\nSpeaker B: Let me see. Our fourth and last meeting.\nSpeaker B: There he is.\nSpeaker B: Yes.\nSpeaker B: Okay, this is our last meeting.\nSpeaker B: In this meeting we will discuss our final design.\nSpeaker B: And we will do some evaluation about not only the product, but also the project.\nSpeaker B: And then we're going to close the project today as well.\nSpeaker B: So, after that you'll be free to go and spend all your money.\nSpeaker B: This design, detailed design meeting, we will discuss the look and feel of the design, the usability interface design, and we will do the product evaluation.\nSpeaker B: In order to do that we have this agenda.\nSpeaker B: We will have the prototype presentation first.\nSpeaker B: Then we will set up some evaluation criteria.\nSpeaker B: Then we will look at the finance.\nSpeaker B: We will have to see if everything we want to do is also possible within our budget.\nSpeaker B: Because everything costs money and the more functions you want to have, the more money it will cost.\nSpeaker B: So, we will see if it fits within the 12 and a half euro per remote.\nSpeaker B: We will see that later.\nSpeaker B: Then we will do the project evaluation and the closing after that.\nSpeaker B: We've got 40 minutes.\nSpeaker B: So, we will be finished at half past three.\nSpeaker B: But first we will do the prototype presentation.\nSpeaker B: So, if somebody wants to sit forward.\nSpeaker A: Well, this is what me and Richard came up with.\nSpeaker A: The default spot for the on-off button, the mute button just below that.\nSpeaker A: Then there's the volume and channel selectors.\nSpeaker A: Simple plus minus button.\nSpeaker A: We thought of a help button.\nSpeaker A: If you hold it and you press another button, the help goes to the LCD screen.\nSpeaker A: Then there's the 0 to 1 to 0 buttons, a button for teletext, a button for the subtitles and the company logo.\nSpeaker A: So, it's rather simple prototype.\nSpeaker A: We'll have to see from testing how do users take it.\nSpeaker C: Okay.\nSpeaker C: Is this a moment where we can ask questions about the functionality?\nSpeaker A: Sure. Just pop in at any time.\nSpeaker C: When you're in teletext, usually your buttons where you just press it and you go to the next teletext page.\nSpeaker E: That's just the normal, as the normal remote controls.\nSpeaker E: So, you press it, you come in page 100.\nSpeaker E: You can use the normal 1 to 0 buttons.\nSpeaker E: You can also use the button for changing the channel, so the shifting button.\nSpeaker E: For shifting up in the teletext and shifting down.\nSpeaker C: Okay.\nSpeaker B: I thought in a few meetings earlier, we tried to give it simple.\nSpeaker B: Just a few buttons and large buttons.\nSpeaker B: But I think these are all together.\nSpeaker B: Quite a few buttons.\nSpeaker B: I'm wondering if we really need all of those buttons.\nSpeaker A: I think if you look at most controls, they've got more buttons than this.\nSpeaker A: And, well, the on-off button, it's a necessity.\nSpeaker A: You can't drop that one.\nSpeaker A: The volume and channel buttons, you obviously need those.\nSpeaker A: The mute button could be replaced by pressing the volume down button twice.\nSpeaker A: So, we could cancel that one.\nSpeaker A: I think the help button really is necessary because there's no other way to know when someone wants to know what a button does.\nSpeaker E: Or you can build in when you press one button.\nSpeaker E: Long time.\nSpeaker E: Yeah.\nSpeaker E: We just discussed that already.\nSpeaker E: But we think all people don't know that option.\nSpeaker E: They just put in a press button.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker A: Well, you can't leave out the number buttons, I guess.\nSpeaker A: Yeah.\nSpeaker B: And it takes some subtitles.\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, it's rather basic already.\nSpeaker C: Yeah.\nSpeaker C: I think so.\nSpeaker C: That's what I pointed out earlier.\nSpeaker C: You just make a control for just the TV.\nSpeaker C: There's just not much to gain here.\nSpeaker D: Yeah.\nSpeaker B: Okay.\nSpeaker B: So, yeah, somebody, but, you can do something about the design of the buttons there.\nSpeaker A: Yeah.\nSpeaker A: It's not a lot of hard to draw on the whiteboard, but they're supposed to be equal sized round with a little logo on it for the volume, the dry angle and stuff.\nSpeaker A: Just for recognition.\nSpeaker E: Yeah.\nSpeaker E: And we don't need a letter or...\nSpeaker E: No.\nSpeaker E: Yeah.\nSpeaker E: So, because we use the little display, we don't have to use it.\nSpeaker B: Yeah.\nSpeaker A: Okay.\nSpeaker A: So, and for the materials we've just chosen for rubber buttons with a different color than the case, so they jump out.\nSpeaker A: Okay.\nSpeaker A: That's about it.\nSpeaker B: That's nice.\nSpeaker B: Then, because we only have 30 minutes left.\nSpeaker B: I will move on to the finance part, which is pretty exciting to see if it's all possible what we want to do.\nSpeaker B: And I can tell you that we're going to have a pretty hard time producing this for 12 and half euros.\nSpeaker B: If we see...\nSpeaker B: I don't know if I'm filming correctly, so just correct me if you see something wrong.\nSpeaker B: I counted two batteries, but maybe we can also use one.\nSpeaker B: I don't know if it's possible.\nSpeaker B: It's rechargeable.\nSpeaker B: It's rechargeable.\nSpeaker B: That's right.\nSpeaker E: Okay.\nSpeaker B: That's two euros off.\nSpeaker B: We need the advanced chip, so there's not much to say there.\nSpeaker B: Here we have the single curve.\nSpeaker B: We can see that the difference between N-curved and the single curve is one euro.\nSpeaker B: I don't know, but I think the single curve is good for the design and also for the display to have a prominent place on the remote control.\nSpeaker B: I think we have to keep that.\nSpeaker B: Then we have the case material supplements.\nSpeaker B: It's plastic.\nSpeaker B: It's the cheapest one we need.\nSpeaker B: That's not much to say either.\nSpeaker B: The biggest cost are the buttons.\nSpeaker B: Maybe we really should try to discard some buttons to keep our cost though, because we will have to get 12 and half euros at the end.\nSpeaker B: Let's first count the buttons we have now, because I'm 16.\nSpeaker B: That will be even more than 18 euros.\nSpeaker B: So 17.\nSpeaker B: Okay, including the help?\nSpeaker B: Yeah, 17.\nSpeaker B: I think we can discard the help and the mute button by pressing down volume long or pressing down a number.\nSpeaker B: Long.\nSpeaker B: That saves us one euro already.\nSpeaker B: Then we have got 15, I think.\nSpeaker B: Yeah.\nSpeaker B: 15.\nSpeaker B: And this is...\nSpeaker C: I assume you count the volume and channel things for two buttons each, right?\nSpeaker B: Those are one, I think.\nSpeaker E: I think actually there are two buttons.\nSpeaker E: It's just one bit.\nSpeaker E: Two, one big button, but they are more expensive than the small ones I hear.\nSpeaker B: Yeah, but it's not stated in this class.\nSpeaker B: Maybe we can even make one button with the volume and the channel in one by pressing...\nSpeaker C: Well, I'm thinking maybe you could just integrate three of those numbers to one button.\nSpeaker B: That's a possibility as well.\nSpeaker C: That would cut the cost.\nSpeaker B: So, and it's good for the design as well, so you can make...\nSpeaker B: Let's see.\nSpeaker B: Like this.\nSpeaker B: Looks a bit like a...\nSpeaker B: A cross.\nSpeaker B: Min?\nSpeaker E: But I don't know if it's here.\nSpeaker E: Yeah, I still got four buttons, but just...\nSpeaker E: So...\nSpeaker B: Yeah, I think they count the materials.\nSpeaker E: Not a better one, itself, but...\nSpeaker E: On the chip there.\nSpeaker B: Yeah, on the chip you got still four as well.\nSpeaker B: That's right.\nSpeaker B: But I think because we have the other ones, chip, we can just count this as one button.\nSpeaker C: But now, but I think...\nSpeaker C: I think this really is four buttons, anyhow.\nSpeaker C: Yes.\nSpeaker B: Yeah.\nSpeaker B: Maybe it is, but then it seems to me that it's impossible to get the 12 and a half euros.\nSpeaker B: Also, the LC display.\nSpeaker B: I think it's too expensive for the display we use.\nSpeaker A: I think they try...\nSpeaker A: If we leave out the display, we can also save money on the chip.\nSpeaker B: That's right, but what's the biggest advantage of our remote then?\nSpeaker C: Only the docking station, I guess?\nSpeaker A: Yeah.\nSpeaker A: The docking station isn't even in this schematic.\nSpeaker A: No, it's not even taking into the price.\nSpeaker B: That's extra.\nSpeaker B: That's right.\nSpeaker C: Maybe we should move to a different supplier.\nSpeaker B: That's an option.\nSpeaker B: Poland, something.\nSpeaker B: Polish, a player.\nSpeaker B: Don't you think we can...\nSpeaker B: If we can count this as one button and integrate these buttons in three,\nSpeaker A: then we save a lot of money as well. We could save money on it, but would it make the remote more usable for elderly people?\nSpeaker C: Yeah, that's what I'm wondering.\nSpeaker A: My motor can't even send an SMS message.\nSpeaker B: Yeah.\nSpeaker B: That's a point.\nSpeaker E: We kind of buy a remote control when you can use it.\nSpeaker E: We missed...\nSpeaker E: Stay below 1250 or...\nSpeaker E: Yeah.\nSpeaker E: Can't...\nSpeaker D: No.\nSpeaker C: Well, since the market research indicated that older people spend their money easier, more easy, maybe it's feasible to just put the price of the remote up a little, especially since we have those nice features.\nSpeaker B: Yeah, but we have to stick to the 12 and a half years.\nSpeaker B: We don't have any more budget to the market.\nSpeaker E: It's possible to make one for 1250.\nSpeaker B: I don't think so.\nSpeaker B: And if you leave out the LC display and if you use less buttons, say...\nSpeaker B: You can't use the single chip.\nSpeaker E: There it is.\nSpeaker E: Do you want to make a...\nSpeaker E: So fancy and...\nSpeaker E: Quick looking.\nSpeaker E: Yeah, objective option.\nSpeaker E: So you can stay below 1250.\nSpeaker B: I think it's difficult as well.\nSpeaker C: Basically becoming a choice between either a good remote and a higher production cost and just any other remote control.\nSpeaker B: Or we can leave out the 10 buttons and take one scroll wheel for the program numbers.\nSpeaker B: That was one.\nSpeaker B: Because then we save 10 buttons and we have five and one.\nSpeaker B: And let's see if we have this one and we've got the other ones.\nSpeaker A: We're getting closer.\nSpeaker C: But how would the scroll wheel work here?\nSpeaker B: Then you will...\nSpeaker B: Maybe you can...\nSpeaker B: If you scroll, you will see the numbers on the LCD display.\nSpeaker B: Until you've got the right number and then you push it.\nSpeaker C: That would bring up the price of the scroll wheel also.\nSpeaker B: Alright, so you scroll and push.\nSpeaker B: But then you can push some other buttons as well.\nSpeaker A: You could just not scroll for half a second.\nSpeaker B: Yeah, so if you scroll to a number and then you wait a half second, then it turns to that channel.\nSpeaker C: I think that would be the end of our usability.\nSpeaker B: It would definitely drop cost a lot.\nSpeaker B: And we need battery.\nSpeaker B: And the regular chip is not possible.\nSpeaker A: It has to be advanced.\nSpeaker A: If you want to use an LCD screen, you need an advanced chip.\nSpeaker B: Now we can save a euro by a flat design.\nSpeaker B: That's an option we can...\nSpeaker B: Then we're almost there.\nSpeaker A: If you could just leave out one more button.\nSpeaker E: Or we have to skip the subtitle button.\nSpeaker B: But I think that is a big advantage for us.\nSpeaker B: But can we integrate the data text and the subtitles in one button?\nSpeaker B: I think so.\nSpeaker C: If you push it three times.\nSpeaker C: The teletext subtitles are right now.\nSpeaker C: You just push the teletext button.\nSpeaker C: Go to page 888.\nSpeaker C: And the teletext disappears.\nSpeaker C: But the subtitles stay there.\nSpeaker E: But if you push the data twice.\nSpeaker A: What if you have to scroll to page 888?\nSpeaker C: That could be like the scroll to page 8.\nSpeaker C: Scroll to page 8.\nSpeaker C: But then again, there would be just an option for older people.\nSpeaker C: You don't even know what a scroll wheel is.\nSpeaker C: Holding a remote with the respect to have 10 buttons for the numbers 1 to 0.\nSpeaker C: There are only 5 buttons on it.\nSpeaker C: And a scroll wheel.\nSpeaker B: I think if you make a good advertisement on television and in the guide, you can explain to the people how to use the scroll wheel.\nSpeaker B: You can just make it real simple.\nSpeaker B: It saves a lot of money and we can keep our LCD screen which can provide extra information how to use the scroll wheel, how to use the other buttons as subtitles.\nSpeaker B: And it's good for the innovative design as well.\nSpeaker B: If you would erase these.\nSpeaker B: If you would erase them.\nSpeaker B: If you would put the...\nSpeaker B: It looks a bit odd maybe.\nSpeaker C: It's a pretty big scroll wheel.\nSpeaker D: Something like that.\nSpeaker B: Then you've got the scroll wheel.\nSpeaker B: If you erase this one.\nSpeaker B: And these are two buttons then.\nSpeaker A: You could make two buttons out of that and just if you press the volume button you can control the volume with the scroll wheel.\nSpeaker A: So that would save two buttons if you do the same for the channel.\nSpeaker B: That's really a good idea I think.\nSpeaker B: And that will make the use of the scroll wheel more obvious.\nSpeaker B: So we make one for the volume, one for the channel.\nSpeaker B: So if we've got one, two, three.\nSpeaker A: If we leave out all those buttons perhaps you can go with the flat case and make it smaller overall.\nSpeaker A: So if you put the volume and channel buttons on the same height as the enough button, the screen right under that then the scroll button.\nSpeaker A: You get a much smaller remote.\nSpeaker B: So you can decrease this one to four buttons.\nSpeaker E: You can make a big remote control for just the same price as a small one.\nSpeaker E: You have to pay for the double curved or single curved.\nSpeaker B: It's more the moles in which they are being made I think single curved is really easy to fill in and cases come right out of the machine and single curves you have to have a different machine than a flat one.\nSpeaker B: I think this is a really easy one.\nSpeaker B: Does everybody think about the flat or single curved design?\nSpeaker C: I still think we should go for the single curved design.\nSpeaker C: Rich at argument was very good.\nSpeaker C: Since we have to cut costs.\nSpeaker C: I'm agreeing with the use of research and see what the options are.\nSpeaker B: I think we can keep the single curved one just to express our LCD screen a bit more.\nSpeaker B: So people will use it more and especially for the help functions will be good if we have scroll button.\nSpeaker B: It's necessary to have a good help function as well.\nSpeaker B: I think it was better to have this price this bit earlier before we went on to the whole design.\nSpeaker B: But I'm glad we could make a bit.\nSpeaker B: That's pretty different.\nSpeaker C: This is not really like for the group we were making or remote.\nSpeaker C: This will really require a lot of marketing to get this to sell.\nSpeaker C: If other people like familiar firms and familiar stuff, this is not familiar for them.\nSpeaker C: No, but I think they have to do a lot of convincing them.\nSpeaker E: I think the most, but they know the scroll buttons from old radio.\nSpeaker E: They also have buttons like that for changing.\nSpeaker B: We can stick it out.\nSpeaker B: It's like an old button.\nSpeaker B: Maybe that's recognizable for them as well.\nSpeaker B: That's a good one.\nSpeaker B: So I'm afraid we have to move on.\nSpeaker B: It's less frightening, I think, if they see this design, they think there are only five buttons.\nSpeaker C: It might be confusing too.\nSpeaker E: That's very special.\nSpeaker B: I would buy it.\nSpeaker B: Yeah, but you're not 60.\nSpeaker B: I would buy it if I was 60.\nSpeaker B: No, I didn't know.\nSpeaker B: Articos and the 12 and 12 euros?\nSpeaker B: No.\nSpeaker B: Yes, they are.\nSpeaker B: No, but we can go on with the project evaluation.\nSpeaker B: Let's see.\nSpeaker B: I can sit down.\nSpeaker B: I think we still have 15 minutes left, so nice on schedule.\nSpeaker B: The project process.\nSpeaker C: So I like to set up all these criteria.\nSpeaker C: Evaluation criteria.\nSpeaker B: That's right.\nSpeaker B: That was the point I forgot.\nSpeaker B: There we are.\nSpeaker B: Evaluation criteria.\nSpeaker B: Thank you.\nSpeaker B: Go ahead.\nSpeaker B: We've got five minutes, I think.\nSpeaker B: Five more minutes?\nSpeaker B: No, we've got 15 minutes, but okay.\nSpeaker B: Yes.\nSpeaker C: Okay, so I did some literature study, analysis of the requirements we set up earlier.\nSpeaker C: Translated it into criteria, which would be these.\nSpeaker C: It's a remote fancy, a shaped look and feel.\nSpeaker C: Innovative, what new functions are there.\nSpeaker C: Easy to use.\nSpeaker C: Is it functional?\nSpeaker C: Are there not too many functions?\nSpeaker C: Those functions are there useful?\nSpeaker C: The cost.\nSpeaker C: The target group is the remote really for the group we're making it for.\nSpeaker C: And great cognitive ability.\nSpeaker C: If our company is easily recognizable, their company made it.\nSpeaker C: So we're supposed to evaluate it right now.\nSpeaker C: Okay.\nSpeaker C: Let's see.\nSpeaker C: It's hard because the minutes of our last meeting were not here.\nSpeaker B: That's because my pen failed to upload his data.\nSpeaker B: Okay.\nSpeaker B: I tried to, but it's giving errors by downloading software.\nSpeaker C: Okay.\nSpeaker C: It's the design fancy on a scale of one to seven.\nSpeaker E: I think it's fancy.\nSpeaker E: Six.\nSpeaker C: Good.\nSpeaker C: Is the design innovative?\nSpeaker C: Very.\nSpeaker C: So with our LCD-\nSpeaker B: Set that up station. Let's go to buttons.\nSpeaker B: Yeah.\nSpeaker B: Seven.\nSpeaker B: Seven?\nSpeaker B: Yeah.\nSpeaker B: So we've got 20 cents for our docking station.\nSpeaker C: Should be dual.\nSpeaker C: Yeah.\nSpeaker C: Is the design easy to use?\nSpeaker B: Yeah.\nSpeaker C: It's a bit dodgy.\nSpeaker B: Well, would be for us, but for all of the...\nSpeaker B: I would say four or five.\nSpeaker B: Four or five.\nSpeaker B: It's between four.\nSpeaker C: I'd go for four or two.\nSpeaker B: Is it functional?\nSpeaker C: Is the design functional?\nSpeaker C: Do we have all functions that we want to include?\nSpeaker C: Yes.\nSpeaker C: I think we do.\nSpeaker C: Do we have too many functions?\nSpeaker C: No.\nSpeaker C: I think so.\nSpeaker B: It's pretty slow.\nSpeaker B: I think it's seven.\nSpeaker C: Seven.\nSpeaker C: Oh, and do we take care of the biggest frustrations of the remote control, like it getting lost?\nSpeaker C: And...\nSpeaker C: Oh, sorry.\nSpeaker A: I haven't thought of that one.\nSpeaker B: It was...\nSpeaker B: I think we do.\nSpeaker B: I had them work there.\nSpeaker B: They are ugly, not very functional, and getting lost.\nSpeaker C: Are the production costs within the preset limits?\nSpeaker C: Well, they are now.\nSpeaker D: Yeah.\nSpeaker C: Does the design fit the group of focus?\nSpeaker A: I think that's a free.\nSpeaker A: Yeah.\nSpeaker B: I think with a new radio button, I think it's better.\nSpeaker E: We have to test this.\nSpeaker A: Oh, no.\nSpeaker A: I still...\nSpeaker A: I think it's too fancy to fleshy.\nSpeaker C: I think your button is not exactly what older people expect of remote control.\nSpeaker C: Yeah.\nSpeaker C: True.\nSpeaker B: I will give you the four.\nSpeaker C: I'm not sure.\nSpeaker C: I think I'd go for two.\nSpeaker C: I'd go for three.\nSpeaker E: So...\nSpeaker D: I think I'll do three.\nSpeaker C: Is the company recognizable?\nSpeaker C: Yes, it is.\nSpeaker C: Or is it the logo there?\nSpeaker C: Yeah.\nSpeaker C: So...\nSpeaker B: And they want to put fashion in their products in the slogan of the company.\nSpeaker B: We have the removable front cases.\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker B: I think it's very recognizable.\nNone: Yep.\nSpeaker C: So, that would be this.\nSpeaker C: I was also supposed to calculate the score, but thought we would have another private...\nSpeaker C: We've got the company after this.\nSpeaker A: Is this...\nSpeaker C: Like after this?\nSpeaker B: Are we done?\nSpeaker B: We've got to evaluate it.\nSpeaker C: Anyhow, these are seven factors.\nSpeaker C: Time seven is forty...\nSpeaker C: Something?\nSpeaker C: Two?\nSpeaker C: No.\nSpeaker B: Nine?\nSpeaker B: No, 49.\nSpeaker C: Minus one.\nSpeaker C: Minus three, minus four.\nSpeaker C: Minus four.\nSpeaker C: It will be minus eight.\nSpeaker C: Forty-one, out of forty-nine.\nNone: That's...\nSpeaker D: I'm around eighty percent.\nSpeaker C: I guess I'll just type that in on the bottom here.\nSpeaker B: That's eighty-four percent.\nSpeaker B: I think that's a pretty nice score.\nSpeaker C: So that would be the evaluation.\nSpeaker B: Okay, thank you.\nSpeaker B: So that brings us to the project evaluation.\nSpeaker B: Yep.\nSpeaker B: Thank you.\nSpeaker B: Project process.\nSpeaker B: Did we move through the right faces, you think, along the process?\nSpeaker E: Yes.\nSpeaker C: I think we should have the finance thing a lot earlier.\nSpeaker B: That's right.\nSpeaker C: In the design phase.\nSpeaker B: So a lack of information about prices.\nSpeaker B: Yeah.\nSpeaker B: That's true.\nSpeaker B: Okay.\nSpeaker D: Okay.\nSpeaker B: Okay.\nSpeaker B: So that's the evaluation.\nSpeaker C: Was there room for creativity?\nSpeaker B: No.\nSpeaker B: No.\nSpeaker A: Not too much.\nSpeaker A: No.\nSpeaker B: It's because of the finance sheet.\nNone: Yeah.\nSpeaker C: It wasn't enough room, but the finance.\nSpeaker C: The room was...\nSpeaker C: Yeah.\nSpeaker C: And I would have liked to go for the younger users also.\nSpeaker C: But I was able to access internet and get something of your own.\nSpeaker B: Strictions.\nSpeaker B: Internet access.\nSpeaker B: Okay.\nSpeaker B: Leadership.\nSpeaker B: Was it clear what everybody had to do?\nSpeaker B: According to their roles and functions.\nSpeaker E: Yes.\nSpeaker B: Teamwork.\nSpeaker B: Did it work out?\nSpeaker B: Working together.\nSpeaker B: Also, you too, here with the...\nSpeaker B: Yeah, it was okay.\nSpeaker B: Last phase.\nSpeaker B: Nice.\nSpeaker A: The tasks are very strictured, so you can just do step by step.\nSpeaker A: It was very easy.\nSpeaker B: Strictions.\nSpeaker B: Okay.\nSpeaker B: Where the means efficient for the tasks we had to do?\nSpeaker B: Or where they...\nSpeaker B: Too much.\nSpeaker A: Well, the smart board, the drawing is just a pain in the ass, but...\nSpeaker A: Smart board.\nSpeaker A: The digital pen is very nice.\nSpeaker B: If it wants to download this, very data.\nSpeaker B: Smart board was irritating.\nSpeaker B: So you rather wanted to have a flip over or something else?\nSpeaker A: A flip over or a more precise...\nSpeaker B: Faster as well.\nSpeaker D: Yes.\nSpeaker D: Yeah.\nSpeaker B: Flip over.\nSpeaker E: But if you want to share of when you make a picture like this on the presentation, it's easier to share it with other people.\nSpeaker E: So you can save it and put it on the internet or...\nSpeaker E: In shared folder or shared.\nSpeaker C: Yeah, I tried to open the file on my laptop, but...\nSpeaker B: No, it didn't work.\nSpeaker B: No, you need the smart board application, I think.\nSpeaker B: But I think you can...\nSpeaker B: Yeah.\nSpeaker B: It should be possible.\nSpeaker B: Yeah, you can export this image.\nSpeaker B: Yeah, it must be.\nSpeaker C: You should have done that, man.\nSpeaker B: So the sharing of information was okay.\nNone: Turn this here.\nSpeaker B: Yeah, it's working.\nSpeaker B: Yeah, it's been in here on the table.\nSpeaker B: Yeah.\nSpeaker B: Yeah, it's possible.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: That's nice.\nSpeaker B: We've found new IDs for having this kind of meeting.\nSpeaker B: So...\nSpeaker E: Yeah, maybe...\nSpeaker E: Just give the numbers of...\nSpeaker E: One of our...\nSpeaker E: Yeah.\nSpeaker E: So it's easier to lose things like that on a computer.\nSpeaker E: Yeah.\nSpeaker E: So it's like a university.\nSpeaker E: Yeah.\nSpeaker B: So everybody puts his own score and then it mediates the score and you can get one.\nSpeaker B: Yeah, that's better.\nSpeaker B: So it's...\nSpeaker B: How do we go?\nNone: Okay.\nSpeaker B: Any further action?\nSpeaker B: Right, there we are.\nSpeaker B: Okay.\nSpeaker B: Any more ideas or questions about the project or about the product?\nSpeaker B: It goes, I think...\nSpeaker B: Then that's it.\nSpeaker B: Get to our last few articles within the budget.\nSpeaker E: Yeah.\nSpeaker C: Yeah.\nSpeaker A: So we are considering we are not going to make a docking station.\nSpeaker B: That's a good point.\nSpeaker B: But maybe because of the docking station, the price of the remote can also be a little bit higher.\nSpeaker C: Yeah.\nSpeaker B: And I think you can compensate that as well.\nSpeaker B: So I think that shouldn't be the biggest problem because it's very cheap to make as well.\nSpeaker B: Do we think we are going to get 12 and a half million?\nSpeaker B: What was it?\nSpeaker B: 15 million.\nSpeaker B: 15 million.\nSpeaker B: I think we will.\nSpeaker B: If we are going to export this product, it's innovative.\nSpeaker B: Especially in America.\nSpeaker B: People are pretty.\nSpeaker B: Take a dent.\nSpeaker B: Sometimes.\nSpeaker C: Yeah, still nothing.\nSpeaker C: I mean if we are going to make a scroll button anyway, we can do more.\nSpeaker C: You can do more for the younger people and for the older people.\nSpeaker C: Yeah.\nSpeaker C: So maybe we should just refocus.\nSpeaker B: Just put it on the market for everybody.\nSpeaker B: Specifically for everybody.\nSpeaker B: You can change markets by changing the front cover.\nSpeaker B: That's one thing you can change it with.\nSpeaker B: So you can also try to reach the younger people by putting fancy covers on the market.\nSpeaker B: Just as a test.\nSpeaker B: See how it works?\nSpeaker B: That's I think a good advantage.\nSpeaker C: Yeah, I think then it would be way more popular with younger people than it is.\nSpeaker C: Then we would be with older people.\nSpeaker B: Yeah.\nSpeaker C: Even if it were their covers.\nSpeaker B: Then I think we can have our little celebration right now.\nSpeaker B: That's it.\nSpeaker B: I think we still have to fill in one more questionnaire.\nSpeaker B: Okay.\nSpeaker B: I think, I don't know, I didn't get a message from the ending of the...\nSpeaker B: Oh, I think I have one.\nSpeaker B: Five more minutes.\nSpeaker B: Oh, that's nice.\nSpeaker B: Then we still have some questions.\nSpeaker B: If somebody has some questions, they can ask them now.\nSpeaker B: We can put these in the project folder as well.\nSpeaker B: This one.\nSpeaker B: Export it.\nSpeaker B: Picture, I think.\nSpeaker E: Yeah.\nSpeaker B: This is another, that's where I guess.\nSpeaker B: Let's see, project folder.\nSpeaker D: I don't.\nSpeaker B: My documents.\nSpeaker E: Maybe they will save it.\nSpeaker B: I just put it in my documents and that's okay.\nSpeaker B: I think I'm going to make our final presentation now.\nSpeaker B: You're supposed to make it.\nSpeaker B: I will be back in about 10 minutes to show you the final presentation.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Summary of the project.\nSpeaker C: Are we supposed to go into our rooms again?\nSpeaker E: Maybe we have to fill in our questionnaire.\nSpeaker B: I think you'll get the questionnaire in your room.\nSpeaker B: Yeah.\nSpeaker B: And then I'll meet you in, think about 10 minutes.\nSpeaker B: Okay.\nSpeaker B: Okay.\nNone: Thank you.\nSpeaker B: Hopefully my panel download is software.\nSpeaker B: Yeah.\nSpeaker B: Or data.\nSpeaker C: That would be nice.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "ES2011d", "summary": "Project Manager initiated the meeting topic on the detailed design. Specifically, the prototype presentation was given by Industrial Designer, and calculation on each evaluation criteria was led by Marketing. Next, group members discussed changing aspects of the remote. Finally, they decided on changing the shape, instead of batteries. Then, Project Manager had a discussion on the process of the project with other teammates, including room for creativity, leadership, and convenience of materials. The group decided to write a final report in the end.", "dialogue": "None: Okay.\nSpeaker C: My what this.\nNone: Sorry.\nNone: Yep.\nSpeaker C: Okay.\nNone: Okay.\nNone: Is that okay now?\nSpeaker C: Okay.\nSpeaker B: Okay.\nSpeaker B: Hey everybody.\nSpeaker B: Welcome to the detailed design meeting.\nSpeaker B: Um, let's see.\nSpeaker B: I'm going to talk about the agenda.\nSpeaker B: Last time we discussed the squishy foodiness of our remote controls and how we might pursue that.\nSpeaker B: Um, and I think looks like we've come up with some ideas.\nSpeaker B: Um, and we also talked about materials we'd use and what kind of chip would be necessary.\nSpeaker B: Um, so for later in the meeting, I've done a spreadsheet of.\nSpeaker B: I'm going to look at that and see if it's changed at all from what we last discussed it.\nSpeaker B: Um, soon.\nSpeaker B: Actually, yeah.\nSpeaker B: Um, you guys want to give a prototype presentation?\nNone: Okay.\nSpeaker C: Well, um, so our design looks something like this.\nSpeaker C: This being the wheel that you use to change channels or volume or whatever.\nSpeaker C: There's a button serves as the power button.\nSpeaker C: If you hold it down and just tap on it, I think it brings up the menu.\nSpeaker C: And the base of the remote control, which has a squishy, spun, you grab what you feel is interchangeable.\nSpeaker C: So you can change the color according to, you know, the suit, your living room or whatever.\nSpeaker C: I can change the vegetable.\nSpeaker C: Yeah.\nSpeaker C: So this snaps off and you can put on whichever one you want.\nSpeaker C: This is not to scale because it would have the battery inside it.\nSpeaker C: This is a mango.\nSpeaker C: It's trendy fruit.\nSpeaker C: It's not just ordinary fruits.\nSpeaker C: You don't have orange.\nSpeaker C: You have mango.\nSpeaker C: Um, I guess strawberries, not just tran very bright strawberry.\nSpeaker C: It's been a little bit difficult to make sure that it's hand holdable and that the user can use it.\nSpeaker C: You know, it's not too big, but we think that this size will be okay.\nSpeaker C: And we will have to fit the battery case in there somehow.\nSpeaker C: And I guess the only other thing that we really didn't talk about was the battery.\nSpeaker C: It would be the thing, the locator.\nSpeaker C: So locator is just chiplets inside there.\nSpeaker C: Okay.\nSpeaker C: So that's just the beepers also inside there too.\nSpeaker C: So you have to have a button on your, on your, you have to attach the button to the.\nSpeaker C: Yeah, we didn't design that.\nSpeaker C: Yeah, we have that, that is, yes.\nSpeaker C: But it would be coordinating with that.\nSpeaker A: Yeah.\nSpeaker C: Like, hey, that, that, you know, match the handset.\nSpeaker C: Get out the broccoli.\nSpeaker C: There you go.\nSpeaker C: Ta-da.\nSpeaker B: Um, if you want to look in your project documents folder, there's an Excel spreadsheet.\nSpeaker B: The only one that's in there, production costs.\nSpeaker B: And if you open it up, I've just stuck the numbers in.\nSpeaker B: It's a real challenge there.\nSpeaker B: But if I missed anything that we've gone over, or if you see something that has changed, I mean, we decided on batteries and the regular chip would be necessary for the more advanced iPod-like button.\nSpeaker B: Um, I said, uncurve your flat.\nSpeaker B: I think that's what you have there.\nSpeaker B: Is that right?\nSpeaker B: For the, for the plastic part would be.\nSpeaker C: My impression was that flat meant like, like one of those square remotes.\nSpeaker C: But it's really not very clear because you got single curve and double curve and.\nSpeaker C: Right.\nSpeaker C: Yeah, I don't know what that means.\nSpeaker C: One side is curved and then the other side is curved.\nSpeaker B: Well, if we're talking about the area, just.\nSpeaker B: I don't know.\nSpeaker B: I guess we'd have to contact the company that makes them and see.\nSpeaker B: Um, so what else?\nSpeaker B: There's plastic for that area around the button.\nSpeaker B: Um, and then rubber would be the squishy, like, thing, right?\nSpeaker B: Um, and lots of special colors, actually.\nSpeaker B: Uh, school real.\nSpeaker B: Do you see anything that I've missed?\nSpeaker C: No, I think that's all right.\nSpeaker B: Okay, so that would make our total of 11.9, which is less than 12.5, which means we have to make you more of a profit.\nSpeaker B: And if we saw a lot of squishy things.\nSpeaker B: Oh, yeah.\nSpeaker B: Um, okay.\nSpeaker A: So.\nSpeaker A: Um,\nSpeaker B: what did you work on?\nSpeaker C: Um, evaluation, quite.\nSpeaker B: Okay.\nSpeaker C: I've got the presentation. Okay.\nSpeaker C: All right.\nSpeaker C: What happens is we have to, um, decide whether this, this whole, this whole project we've been working on actually, um, meets the standards we were set at the start.\nSpeaker C: Um, Okay.\nSpeaker C: Um, the method is we, well, I've analyzed the user requirements and integrated them to the trans found in marketing reports and in our company's strategy marketing.\nSpeaker C: And, um, the findings were that we needed to look in a certain way, feel in a certain way, and this is everything's listed down.\nSpeaker C: Um, looking at certain way, feeling a certain way, it has to be technologically innovative.\nSpeaker C: And it has to be easy to use. These are all things we looked at at the start.\nSpeaker C: Um, and criteria that have to be met. We have to use a table. I'll show you that later together to decide whether it meets the standards.\nSpeaker C: And we, we have therefore in total, um, we have five.\nSpeaker C: We have 11 points according to which this should be evaluated.\nSpeaker C: And, um, the quite, well, basically the findings are the same as the evaluation criteria.\nSpeaker C: I would like to show you the table we have to use.\nSpeaker C: Um, Nope.\nSpeaker C: This is the table. Can you see this here?\nSpeaker C: Um, so the questions I've given you, could you write that down?\nSpeaker C: Um, Santa is one and, and forces seven.\nSpeaker C: And we'll just go through each point together, hopefully.\nSpeaker C: Um, I think if each of us gives an opinion, then they can be mixed somehow. I don't know how it works exactly.\nSpeaker C: I'm really told.\nSpeaker C: Is it possible that we can bring this up on our own?\nSpeaker C: Yeah, it's in the, um, it's in the project documents. Is it meeting three minutes? No, it's, it's called evaluation criteria.\nSpeaker C: And it's under evaluation.\nSpeaker B: The PowerPoint one. Yeah. Okay.\nSpeaker C: So it was, um, yeah, choose one.\nSpeaker C: Choose one and forces seven. Do you want us to discuss this together? Do you want us to do it soon, do you?\nSpeaker B: Um, we can do it separately and then discuss it.\nSpeaker B: If that's what people want to do.\nSpeaker C: So it's actually a scale. Right. One is true. And, um, so these are the questions we're answering. Yes. And one is, if it's fancy, you put one, right? If it's really unfancy.\nSpeaker B: I've said somewhere in between you put four. Okay.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: Does it feel fancy?\nSpeaker C: There's already big questions.\nSpeaker C: Okay.\nSpeaker C: So it's easy to insult. I'm going to say yes. Yes. Very, very true.\nSpeaker C: I imagine there's somewhere on the front, we have a little case that you slip them in. Go.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker B: Do we, um, is it necessary? I don't think so. It's just the marker. I kind of stopped working.\nSpeaker B: Unless you feel you need it.\nSpeaker C: Right. Right. So 1.1. We're just going to circle one. Five. Oh, I don't know. Right. One. Five. Five. Five. Two. Okay. So do we just start it up and divide it by four? Is that the company?\nSpeaker C: I think we should. It's four if you want to do that. Yeah. Yeah. That's the 16. That's four. Oh, no, that's the 13. 1552. Oh, I thought she said five. 1552 13 over four for now.\nSpeaker C: So do we have the next three? Six. Six? Six.\nSpeaker C: It wasn't cheating. Right. 1.3. Is.\nSpeaker C: Sorry. Wait. So one was true and seven was false.\nSpeaker C: Yeah. Okay.\nSpeaker B: So we do have them about the same thing.\nSpeaker B: We just have it in the free.\nSpeaker B: Yeah, yeah, yeah.\nSpeaker B: I was thinking one means no points.\nSpeaker A: Full of self.\nSpeaker C: I should have cut the table up.\nSpeaker A: Oh, gosh.\nSpeaker A: OK.\nSpeaker A: Well, first of all, it's no problem.\nSpeaker A: Yeah.\nNone: All right.\nSpeaker C: Well, I'm glad this came out.\nSpeaker C: I was like, why did you guys design it that way?\nSpeaker A: I thought you guys hated it.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker C: It's quite funny.\nSpeaker C: OK.\nSpeaker C: OK.\nSpeaker C: Yeah.\nSpeaker C: OK.\nSpeaker C: So, sorting again 1.1.\nSpeaker C: So 2.3.\nNone: OK.\nSpeaker C: 1.2.\nSpeaker C: 3.2.\nSpeaker C: 2.2.\nSpeaker C: OK.\nSpeaker C: 1.3.\nSpeaker B: 1.1.1.\nSpeaker C: 2.1.\nSpeaker C: 2.2. 2.2.\nSpeaker B: I think I missed 2.\nSpeaker B: Wait, is that 2.1?\nSpeaker C: Yeah, I put it down to 0.4.\nSpeaker C: 1.4. 1.5.\nSpeaker B: OK, right.\nSpeaker B: That's I have to.\nSpeaker B: I can use all kinds of problems.\nSpeaker C: 1.\nSpeaker C: Sorry about that.\nSpeaker C: 2.2, which is 1.5.\nSpeaker B: 1.\nSpeaker C: 3.\nSpeaker C: Wait, wait, wait, wait.\nSpeaker C: I put 3.\nSpeaker C: 1.\nSpeaker C: Then 1 on my 2.3.1.\nSpeaker C: Is that correct on my slide?\nSpeaker C: Yeah.\nSpeaker C: 1.\nSpeaker C: 3.1.\nSpeaker C: I have 4.\nSpeaker C: 3.1.4.3.3.3.3.3.3.3.3.3.3.3.1.3.3.1.3.1.3.1.3.1.3.1.1.2.\nSpeaker C: 2.4.1.2.1.2.1.2.4.3.4.2.4.3.4.2.4.2.1.2.1.2.\nSpeaker C: Right, so I put 1 on that.\nSpeaker C: Okay, I'll just do the calculations now if you want to continue.\nSpeaker C: Or is it tedious? I'm sorry this was so tedious.\nSpeaker B: No, no, that's...\nSpeaker B: I think we should look at the ones that, like, where people set four, where it looks like we might want to discuss changing an aspect of the remote.\nSpeaker C: Well, the worst ones were 3.1. Does everyone have a slide? 3.1?\nSpeaker B: That was material.\nSpeaker C: Slide show.\nSpeaker C: Material, technologically innovative.\nSpeaker C: Okay.\nSpeaker C: Do you want to change it? What's the suggestion?\nSpeaker C: Which one is that again, sorry? 3.1?\nSpeaker C: Yeah, it's 3.1 was not that good. 4.1.\nSpeaker B: Does the shape?\nSpeaker C: Shape. I think 4.2.\nSpeaker B: See, I'm having trouble imagining that it's going to be the size, like the controller.\nSpeaker B: Or bigger.\nSpeaker C: What if we just smashed all the vegetables down flat?\nSpeaker C: And, like, then you could hold it in your hand better.\nSpeaker C: I think the base would definitely be larger, because some of these are not as easy to hold.\nSpeaker C: Yeah.\nSpeaker C: Kind of smallish.\nSpeaker B: No, but I imagine even if it was bigger, like, if it's round and it's big, then you can't get...\nSpeaker B: That's why remote controls are not long, because you have that thumb kind of...\nSpeaker B: So they could all be bananas in cucumbers.\nSpeaker B: Yeah.\nSpeaker B: But I don't...\nSpeaker B: I personally don't think this is comfortable to sit there.\nSpeaker C: But, like, if you just squashed them flat, like, you made it flat.\nSpeaker C: If they're that stress ball stuff, they would be pretty squishable.\nSpeaker C: But it's still too big, I think.\nSpeaker C: Yeah.\nSpeaker B: And would it even resemble fruit that way?\nSpeaker B: I mean...\nSpeaker C: Yeah, like certain ones you'd have to limit the fruit selection.\nSpeaker C: Like, you could probably do a strawberry still.\nSpeaker C: I think the broccoli would be out.\nSpeaker C: You could do a little bit of that. Although the broccoli is quite comfortable, I have to say.\nSpeaker C: Yeah, but when you were holding that before, it actually looked...\nSpeaker C: Yeah, I looked really good.\nNone: Yeah, so...\nSpeaker C: And then you thought it looked like broccoli, no?\nSpeaker C: Eh...\nSpeaker C: I thought I could think of rhubarb.\nSpeaker C: Rhubarb.\nSpeaker C: I think it's a mistake.\nSpeaker C: I think broccoli is my favourite, actually.\nSpeaker C: It's quite fair.\nSpeaker B: I think we need...\nSpeaker B: So, it was just patterns on...\nSpeaker B: Like, we chose the shape, or the shape could be whatever we wanted, and then it would just be like a design on the rubber.\nSpeaker C: Huh?\nSpeaker C: Oh, okay.\nSpeaker B: Like, just the printed.\nSpeaker B: So, just colour it, not just the shape of it.\nSpeaker C: Yeah, that could work.\nSpeaker C: Yeah, that could work.\nSpeaker C: Or...\nSpeaker B: I mean, we could even have fruit, like, around...\nSpeaker B: Yeah.\nSpeaker B: I mean...\nSpeaker C: Yeah, I don't know.\nSpeaker B: But if we need...\nSpeaker B: If we need...\nSpeaker B: If we want to incorporate the fruit thing somehow, there might be...\nSpeaker B: I mean, if it conflicts with the comfort of actually holding me.\nSpeaker C: Yeah, because no one wants to hold a remote that's uncomfortable, obviously.\nSpeaker C: Like, I don't know, some of them you can kind of think...\nSpeaker C: See is like...\nSpeaker C: Like, you could...\nSpeaker C: If it was only this, you know, if it was shaped like that, and it just had that, but the problem is you have to attach it.\nSpeaker C: And this has to be detachable.\nSpeaker C: Yeah.\nSpeaker C: So, like, maybe that's just too big because...\nSpeaker B: Well, some of it...\nSpeaker B: The reason the broccoli works is you can kind of hold it like that.\nSpeaker B: Yeah, it's kind of...\nSpeaker B: Joystick.\nNone: Yeah.\nNone: But...\nSpeaker C: Yeah, I don't know.\nSpeaker B: I mean, is there some way we could make it this kind of shape?\nSpeaker B: Because like...\nSpeaker C: We could make it that shape, but just have different colours.\nSpeaker B: Yeah.\nSpeaker B: And like, even with a shape, because we were having fun with that.\nSpeaker B: And like...\nSpeaker B: Yeah, like you said like a joystick like that, you know?\nSpeaker C: Yeah.\nSpeaker C: Like, we could do...\nSpeaker C: I'm trying to think of other shapes, like fruits that are...\nSpeaker B: Just that, I think.\nSpeaker B: I mean, that fits the whole round iPod idea.\nSpeaker B: Yeah.\nSpeaker B: And you still have the comfort of holding it like that.\nNone: Mm.\nSpeaker B: And you could...\nSpeaker B: Like, if it's like this, you could put fruit designs and stuff on...\nSpeaker B: Yeah.\nSpeaker B: But yeah.\nSpeaker B: Alright.\nSpeaker B: But I mean, to really have any other ideas about that?\nSpeaker A: Um...\nSpeaker A: We could...\nNone: Shh.\nNone: Yeah.\nSpeaker C: The critical ones came out to be...\nSpeaker C: Yeah, that one...\nSpeaker C: The...\nSpeaker C: Bats are easier to then suck, somewhere easy.\nSpeaker C: Which could be easily...\nSpeaker C: They're not fun to problem them.\nSpeaker C: The better is they're going...\nSpeaker C: I thought everyone gave that a one or two.\nSpeaker C: In the back?\nSpeaker C: Yeah, they'd probably be either on the front or the side.\nSpeaker C: The reason I didn't give it a one...\nSpeaker B: I think I gave it a three, because I thought you'd have to like unclip.\nSpeaker B: No.\nSpeaker B: No, you'd...\nSpeaker C: I imagine there'd be sort of a...\nSpeaker C: Catch...\nSpeaker B: Just like any other one.\nSpeaker C: I don't want the normal enough.\nNone: Alright.\nSpeaker C: So it'd probably be either...\nSpeaker C: No, it can't be in the front.\nSpeaker C: Because the IRR is right there, but it'd be on one of the sides, probably.\nSpeaker C: I think everyone's in the end of three, anyway.\nSpeaker C: So I think it's...\nSpeaker C: That's the end of two points.\nNone: Cool.\nSpeaker C: Well...\nSpeaker C: Yeah, the broccoli...\nSpeaker C: I guess, when this...\nSpeaker C: Yeah, I'd agree with changing the shape.\nSpeaker C: Why would you just having fun making strawberries and stuff?\nSpeaker C: We were a bit off task.\nSpeaker C: So...\nSpeaker C: I can't think of any...\nSpeaker C: So we'll have to like...\nSpeaker C: Yeah, I don't know.\nSpeaker C: You could make the touchpad in different shapes, but then that kind of redesigns the whole project.\nSpeaker C: But you could do...\nSpeaker C: Like, some sort of annoying people who get used to having the buttons.\nSpeaker C: Yeah, that's a shame.\nSpeaker B: Yeah.\nSpeaker B: And it would probably cost more to produce.\nSpeaker B: Yeah, that's just that irregular.\nSpeaker C: How about having different colors and what you've been having different shapes too?\nSpeaker C: Yeah.\nSpeaker B: Which is why printing might be, like, just printing the...\nSpeaker B:...through...\nSpeaker B:...on.\nSpeaker B: Okay.\nSpeaker B: Well, we've done finance evaluation criteria.\nSpeaker B: Production evaluation.\nSpeaker B: So project evaluation.\nSpeaker C: You want this and it can...\nSpeaker B: I guess we're supposed to discuss the process of the project and how satisfied.\nSpeaker B: Oh, it's alright.\nSpeaker B: That's a lot.\nSpeaker B: Yeah.\nSpeaker B: Did you feel there was a lot of room for creativity?\nSpeaker B: Sort of.\nSpeaker B: Sure.\nSpeaker B: I did.\nSpeaker B: I mean, fruit and squishiness are more creative.\nSpeaker B: Bunchiness.\nSpeaker C: The prototype making was very creatively stimulating.\nSpeaker C: And I think we've come up with a product that's fun.\nSpeaker C: And it's all the criteria.\nSpeaker B: And how is our leadership and teamwork?\nSpeaker C: I think it was good.\nSpeaker C: We knew what we were doing.\nSpeaker C: It was a very democratic process.\nSpeaker C: Everyone got to contribute.\nSpeaker B: Well, I thought my leadership was crap personally.\nSpeaker B: He told us when to start and end and that's all the tips.\nSpeaker B: I think you were fine.\nSpeaker C: I'm just going to jump.\nSpeaker B: I'm never going to do a management position.\nSpeaker B: I know that now.\nSpeaker B: Yeah, I thought we all worked very well together.\nSpeaker C: It all sort of blended.\nSpeaker C: Yeah, I think it more than anything.\nSpeaker C: We didn't really have our set role so much as we just would be like, you know, all had ideas.\nSpeaker C: Very democratic.\nSpeaker C: Yeah.\nSpeaker B: No spats.\nSpeaker B: That was good.\nSpeaker B: And the means for like the materials we used, how convenient were they?\nSpeaker B: Like the pens, the whiteboard.\nSpeaker C: I mean, well, I'm not a big fan of any Microsoft PowerPoint or any Mac person.\nSpeaker C: No, no, I never touch Macs either.\nSpeaker C: I just use the Unix or the off market sort of word perfect and all these other things.\nSpeaker C: Which is a lot of you need to be able to.\nSpeaker C: Well, the problem is if you don't like my new computer never has word perfect.\nSpeaker C: So I have to go track someone down who has an old disk and then I have to reinstall it.\nSpeaker C: So I have all these documents.\nSpeaker C: I can't use them.\nSpeaker C: No.\nSpeaker C: But yeah, I mean, I guess it's okay.\nSpeaker C: I don't know if it was just my role, but I thought that the information that was available to me was kind of just like, or maybe it was just the idea that we had, but there's kind of, it was kind of like, okay, I don't really think, I don't know what I'm doing here.\nSpeaker C: So I didn't really think it was helpful.\nSpeaker C: So I kind of just made up my own stuff.\nSpeaker C: And I didn't really like the PowerPoint presentations.\nSpeaker C: So I kind of wrote a lot of notes instead.\nSpeaker C: But yeah, I don't really like PowerPoint, personally.\nSpeaker C: It's kind of stupid.\nSpeaker C: I never use it.\nSpeaker B: Yeah, but I can't say I found everything particularly helpful.\nSpeaker B: Yeah, I didn't really.\nSpeaker B: Yeah.\nSpeaker C: I first bit of information was like, the child's drawn.\nSpeaker A: Actually, how it would work.\nSpeaker C: Mine was really helpful.\nSpeaker C: I don't know.\nSpeaker B: My problem was you guys had access to, like, they'd send you to sites and stuff, right?\nSpeaker B: Yeah.\nSpeaker B: So I didn't really know what you guys were doing.\nSpeaker B: And when you were talking about it, it's just like, that's why it seems so ignorant when you're explaining things.\nSpeaker C: Yeah, mine was mostly made up, except when they told me, like, you know, titanium costs more than it would to make a remote control.\nSpeaker C: But it must have been quite difficult for them to build the whole system.\nSpeaker C: The whole system, because of course they can't give you anything comparable to the internet.\nSpeaker C: Yeah, I mean, I think it's interesting how it all went together.\nSpeaker C: Like, I had the stuff about how rubber is cheap and you have how people want it to be spongy.\nSpeaker C: Yeah.\nSpeaker C: It seems planned.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I kind of thought that I felt like I would go and, like, try to use my information.\nSpeaker C: Or, like, I'd make this stupid little presentation and then, like, I just would end up talking about something completely unrelated because I felt like I was off task all the time.\nSpeaker C: But, um...\nSpeaker B: Well, I mean, we really got into talking about, like, personal practicality.\nSpeaker B: Like, it wasn't necessarily what was, like, we would never have thought of fruit or sponginess, you know?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker C: So...\nSpeaker C: So...\nSpeaker C: Yeah.\nSpeaker C: And it might be to see whether people actually all come up with the same.\nSpeaker B: Oh, right.\nSpeaker B: Given certain information or...\nSpeaker C: Yeah, like, if everyone's given the same input, I don't know.\nSpeaker C: I don't know.\nSpeaker C: Um...\nSpeaker C: What's next?\nSpeaker C: I know that's not...\nSpeaker C: Um...\nSpeaker B: What did you guys think of the pens that asked about?\nSpeaker B: It's pretty cool.\nSpeaker C: I want to start to write with them.\nSpeaker B: Yeah.\nSpeaker B: And I...\nSpeaker B: I forgot one to try to check the box.\nSpeaker C: I'd like to see what it looks like.\nSpeaker C: Oh, wow.\nSpeaker C: They're nicer than the pen that I'm using.\nSpeaker C: Because, like, your stuff actually shows up here rather than having to look at the screen and write.\nSpeaker A: Mm-hmm.\nSpeaker C: That even so, how did I...\nSpeaker B: And...\nSpeaker B: New ideas found?\nSpeaker C: That's all very new.\nSpeaker C: Yeah.\nSpeaker C: I don't know.\nSpeaker C: Yeah, I think I'd like to...\nSpeaker C: I don't know.\nSpeaker C: Like, it was that I think the microphones are okay when you're sitting down, but, like, they're kind of clumsy.\nSpeaker C: I guess I'm just going up to the light, probably.\nSpeaker C: Well, I drop off if you like moved too much.\nSpeaker A: Yeah, I don't know, but there.\nSpeaker C: I don't think we're supposed to be testing these microphones.\nSpeaker C: Maybe we are.\nSpeaker C: I don't know.\nSpeaker C: I think, and I think that, uh...\nSpeaker C: All this technology, like, I guess some people must be interested in using it, but I can't imagine finding it any more useful than, like, looking at someone's notes.\nSpeaker C: Or, like, I actually worked in a company and I had a role and I had to go to meetings.\nSpeaker C: And, like, the last thing I would have wanted would be to have to watch a video because I missed a meeting.\nSpeaker C: Like, usually I miss meetings deliberately.\nSpeaker C: There's just...\nSpeaker C: There's really not that much information that actually goes through a real meeting.\nSpeaker C: Like, when you're actually at a real meeting in a real company.\nSpeaker C: Especially, like, rehashing old stuff.\nSpeaker C: And just sort of going over general stuff that anybody who sort of wants tasks should already know.\nSpeaker C: It's like, there's just really not a lot of information.\nSpeaker C: It's kind of, like, an excessive iteration.\nSpeaker B: It seems like way overkill.\nSpeaker C: Like, if I missed a meeting, I could probably get it summed up in, like, one sentence.\nSpeaker C: Yeah.\nSpeaker C: I wouldn't need to have to watch, like, a two-hour video with a sound and a transcript.\nSpeaker C: Mm-hmm.\nSpeaker C: I guess it would be...\nSpeaker C: It's got to be worth it.\nSpeaker C: I guess I'm just mostly curious about, like, what kind of person or, like, company would really find it useful?\nSpeaker C: Can't really imagine.\nSpeaker C: I don't know.\nSpeaker C: How about a...\nSpeaker C: A pita?\nSpeaker C: Er...\nSpeaker A: Er...\nSpeaker B: Well, that's anyone want to talk about fruit more while we still have time?\nSpeaker B: When are you other ideas?\nSpeaker C: Well, what we... we supposed to... we supposed to watch report.\nSpeaker C: We end there.\nSpeaker C: Erm...\nSpeaker B: We still have time to face any other input.\nSpeaker B: I mean, I think we did really well, personally, which is why we've, you know, gone through this so quickly.\nSpeaker B: Because, I mean, we've all, we all kind of agree that our product is satisfactory.\nSpeaker B: It fits the budget, and it's trendy.\nSpeaker B: It's amazing.\nSpeaker B: So...\nSpeaker B: Erm...\nSpeaker C: I'm with my dad.\nSpeaker B: So I think that's all for today.\nSpeaker B: Okay.\nSpeaker C: We'll have to put in all this stuff.\nSpeaker C: Meeting adjourned.\nSpeaker C: Meeting adjourned.\nSpeaker C: I think I've learned not to bring, play, note to meetings.\nSpeaker C: I think it would be a good idea.\nSpeaker C: I like it.\nSpeaker C: I'm probably not enough to get people actually paying attention, especially if you have food.\nSpeaker B: So I guess we're supposed to write final reports.\nSpeaker B: All of this?\nSpeaker B: I don't know.\nSpeaker C: Well, as well, I have nine...\nSpeaker A: Mmm...\nSpeaker B: Is that just me?\n"}, {"meeting_id": "Bed014", "summary": "Several technical matters are being resolved: a conversion program is being written for data to be translated between XML and the Java Embedded-Bayes notation; the language generation templates are now available for the english version of the SmartKom system; SmartKom now works on three different machines at ICSI. On the other hand, future collaboration on belief-nets has already been agreed with another research group. The construction analysis and formalism are also progressing. Several issues that have been dealt with were mentioned during the meeting: indefinite pronouns and wh-questions, noun-phrase structure, etc. This analysis is being done with the help of a linguist, who often provides different perspectives to methods and terminology.", "dialogue": "Speaker E: see\nSpeaker F: Mental mental Rom pilot\nSpeaker D: testing\nSpeaker E: Suppolins. Suppolins.\nSpeaker E: Three.\nSpeaker E: I am three.\nSpeaker E: See, that matches the seat up there.\nSpeaker C: Yeah, well, I guess it's coming up then.\nSpeaker E: Because it's, that starts counting from zero and you start counting from one.\nSpeaker E: There it goes, the classic off by one error.\nSpeaker F: But mine is correct.\nSpeaker F: Is it?\nSpeaker F: It's one.\nSpeaker F: Channel one.\nSpeaker F: Your mic.\nSpeaker F: Oh, oh, oh.\nSpeaker E: So it would.\nSpeaker E: Oh.\nSpeaker E: Oh.\nSpeaker E: So.\nSpeaker E: I've bested you again, Nancy.\nSpeaker E: But your paper is correct.\nSpeaker F: Paper is correct.\nSpeaker E: Look at the paper.\nSpeaker E: I didn't, I was saying the microphone, not the paper.\nSpeaker E: It's always offset.\nSpeaker F: Yes, you bested me again.\nSpeaker F: That's how I think of our continuing interaction.\nSpeaker F: Damn, I boiled again.\nSpeaker F: So is Keith showing up?\nSpeaker E: He's talking with George right now.\nSpeaker E: Is he going to get a rip himself away from, from that?\nSpeaker B: Probably not.\nSpeaker B: Is my guess.\nSpeaker E: Oh, then it's just going to be the five of us.\nSpeaker D: Well, he was very affirmative in his mail saying he will be here at four.\nSpeaker D: But, you know, there was before he knew about the George Vector problem.\nSpeaker D: Right.\nSpeaker B: This is not, it's not bad for the project if Keith is talking to George.\nSpeaker B: Okay.\nSpeaker B: So my suggestion is we just.\nSpeaker B: For the head.\nSpeaker B: For the head.\nSpeaker B: Oh.\nSpeaker F: Are you in charge?\nSpeaker F: Sure.\nSpeaker D: Well, I sort of had informal talks with most of you.\nSpeaker D: So, either just the report, he's really happy about the CBT's being in the same order in the XML as in the.\nSpeaker D: The Java declaration formats.\nSpeaker D: So you don't have to do too much in the star sheet.\nSpeaker D: Yeah.\nSpeaker D: Transversion.\nSpeaker D: The Java, the embedded base wants to take input a base net in some Java notation.\nSpeaker D: And Eva is using the XLR and star sheet processor to convert the XML that's output by the Java base.\nSpeaker D: With it into the ebase input.\nSpeaker F: I should maybe try like emailing the guy and see if he has something already.\nSpeaker F: Sure.\nSpeaker C: He has both the Java base and the embedded base.\nSpeaker E: But that's some sort of conversion program.\nSpeaker F: Yeah.\nSpeaker F: And put them into different formats.\nSpeaker E: I think you should demand things from him.\nSpeaker B: He charges so much.\nSpeaker B: Yeah. No, I think it's a good idea that I'm just going to ask.\nSpeaker D: And well, pretty much on the top of my list, I would have asked Keith how the where is X hand parts is standing.\nSpeaker D: Let's skip that.\nSpeaker D: There's good news from Drone Outer generation templates I've run.\nSpeaker E: So the trees for the XML trees for the for the genet for the synthesizer are written.\nSpeaker E: So I just need to do the red and new set of tree combining rules.\nSpeaker E: But I think those are pretty similar to the old ones.\nSpeaker B: So just going to be, you know, you were going to send me a note about hiring.\nSpeaker B: Yes.\nSpeaker B: I didn't finish the sentence, but he understood it.\nSpeaker B: I know what he's talking about.\nSpeaker B: OK.\nSpeaker B: The names he doesn't.\nSpeaker F: Hiring somebody.\nSpeaker D: The guy.\nSpeaker D: OK, so natural language generation produces not just a surface string that is fed into a Texas speech.\nSpeaker D: Yeah.\nSpeaker D: But surface string with a syntax tree that's fed into a concept of speech.\nSpeaker D: Now, this concept of speech module has certain rules on how if you get the following syntactic structure, how to map this into the present rules.\nSpeaker D: And Faye has full heartedly agreed to rewrite the German concept syntax to the present rules.\nSpeaker F: I know she spoke German.\nSpeaker D: No, she doesn't.\nSpeaker D: Oh, OK.\nSpeaker F: She speaks English.\nSpeaker F: Oh, we write the German ones in English.\nSpeaker D: In English.\nSpeaker D: And therefore, if it's OK that we give her a couple more hours a week, then she'll do that.\nSpeaker E: What languages are written?\nSpeaker E: Is that that scheme thing you showed me?\nSpeaker D: That's the list type scheme.\nSpeaker D: She knows how to program in scheme.\nSpeaker E: I hope.\nSpeaker D: No, I...\nSpeaker D: My guess is I ask for a commented version of that file.\nSpeaker D: If we get that, then it's doable even without getting into it.\nSpeaker D: Even though the scheme stuff is really well documented in the...\nSpeaker E: Well, I guess a few not used to functional programming scheme can be completely incomprehensible, because there's no...\nSpeaker E: Like, there's lots of unnamed functions and, you know...\nSpeaker B: Anyway, it will sort this out.\nSpeaker B: But anyway, send me the note and then I'll check with Morgan on the money.\nSpeaker B: I don't anticipate any problem, but we have to ask.\nSpeaker B: Oh, so this is, you know, on the generation thing, if she's really going to do that, then we should be able to get prosody as well.\nSpeaker B: So it'll say it's nonsense with...\nSpeaker A: Perfect.\nSpeaker A: We should.\nSpeaker E: Are we going to...\nSpeaker E: Can we change the voice of the thing?\nSpeaker E: Because right now the voice sounds like a murderer.\nSpeaker D: Yep, we have to change the voice.\nSpeaker E: Which one?\nSpeaker E: The little smarticus.\nSpeaker E: Smarticus sounds like a murderer.\nSpeaker E: Oh.\nSpeaker E: I have your reservations.\nSpeaker C: But I will not give a key on this, you come into my land.\nSpeaker D: We have the choice between the usual festival voices, which I already told the smart people.\nSpeaker D: We aren't going to use because they're really bad.\nSpeaker D: Festival?\nSpeaker D: Yes, program.\nSpeaker B: Oh, okay, no, since the song.\nSpeaker D: But...\nSpeaker D: Oh, gee, I has crafted a couple of iPhone-type voices that are really nice, and we're going to use that.\nSpeaker D: We can still agree on a gender if we want.\nSpeaker D: We still have male or female.\nSpeaker F: Well, it's like whatever sounds best.\nSpeaker F: Whatever sounds best.\nSpeaker E: There's O.G.A. Stanford.\nSpeaker E: Fortunately, probably male original German institute.\nSpeaker D: Oregon.\nSpeaker B: Oregon, graduate institute.\nSpeaker B: Turns out this, the longstanding links with these guys in the speech group.\nSpeaker B: In fact, there's this guy who's basically got a joint appointment.\nSpeaker B: Heenick Hermanski.\nSpeaker B: He spends a fair amount of time here.\nSpeaker B: Anyway.\nSpeaker D: Won't be a problem.\nSpeaker D: Okay.\nSpeaker D: And it's probably also absolutely uninteresting for all of you to learn that as a group, as of 20 minutes ago, David and I per accident managed to get the whole smart-com system running on the XC Linux machines with the XC and T machines, thereby increasing the number of running smart-com systems in this house from one on my laptop to three.\nSpeaker D: I was just by accident.\nSpeaker D: Yeah, I know, but it's very nice.\nSpeaker D: I suggested to try something that was really kind of, you know, against better knowledge, shouldn't have worked, but it worked.\nSpeaker D: Intuition.\nSpeaker D: Maybe a bit for the intuition thing.\nSpeaker D: Okay.\nSpeaker D: And we'll never find out why.\nSpeaker D: It's just like why the generation for the presentation manager is now working.\nSpeaker C: Which is something you get used to as a programmer, right?\nSpeaker C: It's cool it works that way.\nSpeaker D: So the people in Zabrik and I decided not to touch it ever again.\nSpeaker D: Okay.\nSpeaker D: I was going to ask you where something is.\nSpeaker D: What we know about that.\nSpeaker D: Where is that?\nSpeaker D: Where is the junction?\nSpeaker D: Oh, but we can ask, did you get to read off 400 words?\nSpeaker D: Was it a day?\nSpeaker E: I was looking at it, it doesn't follow logically.\nSpeaker E: The first paragraph doesn't seem to have any link to the second paragraph.\nSpeaker C: Each paragraph is good though.\nSpeaker C: It's fine.\nSpeaker C: It was written by committee.\nSpeaker B: Anyway.\nSpeaker B: But the meeting looks like it's going to be good.\nSpeaker F: I didn't know about it until I remember telling you.\nSpeaker B: I ran across it and I didn't even know where.\nSpeaker B: It's just some weird place.\nSpeaker B: I'm surprised I didn't know about it.\nSpeaker B: We know all the invited speakers.\nSpeaker B: I didn't tell me.\nSpeaker B: Anyway.\nSpeaker B: But anyway.\nSpeaker B: Before we were sort of.\nSpeaker B: I also had a nice email correspondence with Daphne Kohler.\nSpeaker B: I'd love to work with us on using these structured belief nets and stuff.\nSpeaker B: But starting in August.\nSpeaker B: She's also got a new student working on this.\nSpeaker B: We should get in touch with them again in August.\nSpeaker B: We'll figure out a way for you to get seriously connected with their group.\nSpeaker B: That looks pretty good.\nSpeaker B: Yeah, I'll say it now.\nSpeaker B: It looks to me like we're now at a good point to do something.\nSpeaker B: Start working on something really hard.\nSpeaker B: And so far working on things that are easy.\nSpeaker B: Which is mental spaces and or.\nSpeaker B: But the other part of it is the way they connect to these.\nSpeaker B: So there's all the problems that the linguists know about mental spaces.\nSpeaker B: And the cognitive linguists know about.\nSpeaker B: But then there's this problem of the.\nSpeaker B: Belief net people have only done a moderately good job of dealing with temporal belief nets.\nSpeaker B: Which they call dynamic.\nSpeaker B: They incorrectly call dynamic belief nets.\nSpeaker B: So there's a term dynamic belief net.\nSpeaker B: It doesn't mean that.\nSpeaker B: It means time slices.\nSpeaker B: And it's really used those people using.\nSpeaker B: But one of the things I would like to do over the next month and make take more.\nSpeaker B: Is to understand to what extent we can not only figure out the constructions for them for multiple world.\nSpeaker B: But the formalism will look like and where the slots and fillers will be.\nSpeaker B: But also what that would translate into in terms of belief net and the inferences.\nSpeaker B: So the story is that if you have these probabilistic relational models, they're set up in principle.\nSpeaker B: So that you can make new instances and instances connect to each other and all sort of stuff.\nSpeaker B: So it should be feasible to set them up in such a way that if you've got the past tense and the present tense, and each of those is a separate belief structure that they do their inferences with just the couplings that are appropriate.\nSpeaker B: But that's as far as I can tell is putting together two real hard problems.\nSpeaker B: One is the linguistic part of what are the couplings.\nSpeaker B: And when you have a certain construction that implies certain couplings and other couplings.\nSpeaker B: Between, let's say the past and the present or any other one of these things.\nSpeaker B: And then we have this inference problem of exactly technically how does the belief net work?\nSpeaker B: If it's got, let's say one in different tenses or my beliefs in your beliefs or any of these other ones of multiple models.\nSpeaker B: So, you know, in the long run we need to solve both of those and my suggestion is that we start digging into them both in a way that hopefully turns out to be consistent.\nSpeaker B: And sometimes it's actually easier to solve two hard problems than one because they constrain each other. I mean, you've got huge, huge drain of possible choices.\nSpeaker B: Anyway, so that's...\nSpeaker C: Well, yeah, like I solved the problem of... you were talking about how do you... various issues of how come a plural noun gets to quote counters and noun phrase, you know, occurs an argument about the construction. But a bear singular stem doesn't get to act that way.\nSpeaker C: And it would take a really long time to explain it now, but I'm about to write it up this evening.\nSpeaker C: I solved that at the same time as how do we keep adjectives from floating to the left of determiners and how do we keep all of that from floating outside the noun phrase to get something like I, the kick dog.\nSpeaker C: That's great.\nSpeaker C: Did it at once. So maybe it will be a second.\nSpeaker B: No, I know. I think that is going to be sort of the key to this whole... to the big project of the summer of getting the instructions right is that...\nSpeaker B: People do manage to do this, so there probably are some relatively clean rules. They're just not context-free trees.\nSpeaker B: And if the formalism is good, then we should be able to have a sort of moderate scale thing.\nSpeaker B: And that, by the way, is key to what I encourage George to be talking with you about not the formalism yet, but the phenomena.\nSpeaker B: Oh, another thing.\nSpeaker B: There was this thing that Nancy agreed to in a weak moment this morning that...\nSpeaker B: That was very strong.\nSpeaker B: Sorry, in a friendly moment.\nSpeaker B: That's the same thing.\nSpeaker B: Anyway, that we were going to try to get a first cut at the revised formalism by the end of next week.\nSpeaker B: Okay, probably skipping the mental spaces part.\nSpeaker B: Just trying to write up essentially what you guys have worked out so that everybody has something to look at.\nSpeaker B: We've talked about it, but only the intermost-intergroup currently knows...\nSpeaker F: Yeah, and not even all of them really do.\nSpeaker C: The group has a whole no. It's about no individual members.\nSpeaker B: This one of the advantages of document, right, is that it actually transfers from head to head.\nSpeaker B: So, communication.\nSpeaker B: Communication.\nSpeaker B: Anyway, so with a little luck, let's head that as a goal anyway.\nSpeaker C: What was the date there of Monday or...\nSpeaker B: No, no, no, no.\nSpeaker B: We've talked about it at the end of next week.\nSpeaker F: But the two of us will probably talk to it well before...\nSpeaker C: Anyway, let's talk separately.\nSpeaker C: It's a week after that.\nSpeaker B: So sometime next week, it turns out that that effort leads us into some big hole.\nSpeaker B: That's fine.\nSpeaker B: You know, if you say, here's a really hard problem, we haven't solved it.\nSpeaker C: That's just fine.\nSpeaker C: But at least sort of trying to work out what the state of the art is.\nSpeaker B: Right, so to the extent that we have it, let's write it.\nSpeaker B: And to the extent that we don't, let's find out what we need to do.\nSpeaker D: Can we... is it worth thinking of an example out of our tourism thing domain that involves a decent mental space shift?\nSpeaker B: I think it is.\nSpeaker B: But I interrupted before Keith got to tell us what happened with where is the powder tower or whatever.\nSpeaker C: Well, what was supposed to happen?\nSpeaker C: I sort of actually caught up in some other ones.\nSpeaker C: You know, I don't have a write-up of... or have it elaborated on the ideas that we were already talking about.\nSpeaker D: Yeah, I think we already came to the conclusion that we have two alternative paths that we... two alternative ways of representing it.\nSpeaker D: When it sort of has...\nSpeaker D: It's gone.\nSpeaker C: The question of whether the polysamy is sort of like in the construction or the frag management.\nSpeaker D: It comes... this resolved later here.\nSpeaker C: I think it has to be the second case.\nSpeaker C: Yeah.\nSpeaker C: So you... isn't clear what we're talking about here.\nSpeaker C: The question is whether the construction is semantic or ambiguous between asking for location and asking for direction.\nSpeaker C: Should we have... or whether the construction semantically is clearly only asking for location but pragmatically that's construed as meaning...\nSpeaker D: And these are two notes that we can observe in the base net.\nSpeaker D: So these are either true or false, and it's also just true or false.\nSpeaker D: If we encounter a phrase such as where is X, should that set this to true and this to true and the base net figures out which under the situation is more likely?\nSpeaker D: Or should it just activate this, have this be false and the base net figures out whether this actually no means?\nSpeaker B: Oh, that's a separate issue.\nSpeaker B: So I agree with you that it's a disaster to try to make separate constructions for every pragmatic reading.\nSpeaker B: Although there are some that will need to be there.\nSpeaker B: Right.\nSpeaker F: Or have every construction list all the possible pragmatic implications...\nSpeaker F: You can't do that in the same way.\nSpeaker B: Almost certainly, can you pass the solve?\nSpeaker B: Is the construction worth noting that there is this?\nSpeaker C: This one is maybe in the gray area.\nSpeaker C: Is it like that or is it just sort of obvious from the world knowledge that no one...\nSpeaker C: You wouldn't want to know the location without wanting to know how to get there.\nSpeaker D: Or in some cases it's quite definite.\nSpeaker C: So that you just want to know what the question is basically is this conventional or conversational?\nSpeaker C: Exactly.\nSpeaker B: I guess the more important thing at this stage is that we should be able to know how we would handle it in the short run.\nSpeaker B: It's more important to know how we would treat technically what we would do if we decided A and what we would do if we decided B.\nSpeaker B: Then it is to decide A or B right now.\nSpeaker F: There will be other examples that are one way.\nSpeaker B: We know for sure that we have to be able to do both.\nSpeaker B: So I guess in the short run let's be real clear on what the two alternatives would be.\nSpeaker D: And then we had another idea floating around which we wanted to get your input on and that concerns the...\nSpeaker D: But the next thing is we would have a person that would like to work on it.\nSpeaker D: And that's Irina Gurriavitch from EML who is going to be visiting us the week before August and a little bit into August.\nSpeaker D: And she would like to apply the ontology that is being crafted at EML.\nSpeaker D: That's not the one I sent you. The one I sent you was from GMD out of the European Crumpet Sterbil.\nSpeaker D: Agreed.\nSpeaker D: And one of the reasons... one of those ideas was...\nSpeaker D: So back to the old journal observation that if you have a dialogue history and it said the word admission fee was mentioned.\nSpeaker D: It's more likely that the person actually wants to enter than just take a picture of it from the outside.\nSpeaker D: Now what could imagine to have a list for each construction of things that one should look up in the discourse history?\nSpeaker D: That's the really stupid way.\nSpeaker D: Then there is the really clever way that was suggested by Keith.\nSpeaker D: And then there is the middle way that I'm suggesting.\nSpeaker D: That is you get X, which is whatever the castle. The ontology will tell us that castles have opening hours, that they have admission fees, they have whatever.\nSpeaker D: And then this is... we go via a Thessaurous and look up certain linguistic surface structures that are related to these concepts and feed those through the dialogue history and check dynamically for each entity.\nSpeaker D: We look it up, check whether any of these were mentioned and then activate the corresponding nodes on the discourse side.\nSpeaker D: But Keith suggested that a much cleaner way would be to keep track of the discourse in such a way that you... if you know that something like that has been mentioned before, this just continues to add up.\nSpeaker C: If someone mentions admission fees, that activates an enter schema which sticks around for a little while in your representation of what's being talked about.\nSpeaker C: And then when someone asks where is Axe, you've already got the end of a priming.\nSpeaker F: Yeah.\nSpeaker B: Priming, spreading the information.\nSpeaker B: So that's certainly more realistic, psychologically.\nSpeaker B: Now it's technically...\nSpeaker E: Well, does it seem like if you just manage the dialogue history with a thread that kept track of the activity of...\nSpeaker E: I mean, because the thread would know what nodes needed to be activated so we could just keep track of how long it's been since something's been mentioned and automatically loaded in.\nSpeaker B: You could do that. But here's a way... in the bullet base network, you could think about it this way.\nSpeaker B: That if at the time admissions fee was mentioned, you could increase the probability that someone wanted to enter.\nSpeaker E: Well, that's what I wasn't thinking in terms of enter scheme as I was just...\nSpeaker B: Fair enough. Okay. But in terms of the current implementation, right?\nSpeaker B: So that...\nSpeaker F: It would be higher in the...\nSpeaker B: The conditional probability that someone... so at the time you mentioned it, this is essentially the base net equivalent of the spreading activation.\nSpeaker B: In some ways, it's not as good, but it's the implementation we got. We don't have a connection with the implementation.\nSpeaker B: Now, my guess is that it's not a question of time, but it is a question of whether another intervening object has been mentioned.\nSpeaker B: I mean, we could look at dialogue. This is... of course, the other thing we do is we have this data coming, which probably will blow all our theories.\nSpeaker B: But skipping that. So... but my guess is what will probably happen... here's a proposed design.\nSpeaker B: Is that there are certain constructions which for our purposes do change the probabilities of EFA decisions and various other kinds.\nSpeaker B: And that the standard way that these contexts work is sort of stacked like or whatever. But that's sort of the most recent thing.\nSpeaker B: And so it could be that when another... tourist entity gets mentioned, you...\nSpeaker B: Re-initialize the state.\nSpeaker B: And of course, if we had a fancy one with multiple worlds, you could have... you could keep track of what someone was saying about this and that.\nSpeaker B: You know, I want to go... in the morning, I want...\nSpeaker C: What's my plan for today? Here's my plan for today.\nSpeaker B: Yeah, in the morning. I'm planning to go shopping the afternoon, to the powder tower. I talk to some talking about shopping.\nSpeaker B: And then you say, you know, well, what's it cost? Or something? Anyway, so one could well imagine, but not yet.\nSpeaker B: But I do think that the... it'll turn out that it's going to be... pretty much on whether there's been an override.\nSpeaker D: Yeah, I mean, if you ask how much I said train ride in Sonoma, around the Winiards cost and then somebody tells you it's $60, and then you say, okay, how much is...\nSpeaker D: I would like to visit the... whatever, something completely different. Then I go to... you know, poor Drase. It's not more likely that you want to enter anything, but it's...\nSpeaker F: It's not a matter of fact, I think, the rejection of entering and doing that. So when you admit... have admission fee and it changes something, it's only for that particular...\nSpeaker F: It's relational, right? Yeah, I think that...\nSpeaker B: Yeah, well, and the simple idea is that it's only... it's only for the current tourist entity of interest.\nSpeaker D: But that's... I mean, this... this function... so has the current object being mentioned in... with...\nSpeaker B: No, no, no. It goes in the... it goes in the direction. Is... when the... this is mentioned the... probability of... let's say entering...\nSpeaker B: A change of objects.\nSpeaker B: Changes.\nSpeaker E: You could just... basically, it... it observes and... that's the... a node for...\nSpeaker B: Enter to... but I think... Robert's right that to determine that...\nSpeaker B: Okay, you may well want to go through a... the... the... so... if the issue is... if... so now this... this construction has been matched and you say, okay, does this actually have any implications for our decisions? Then there's another piece of code that presumably... does that computation?\nSpeaker F: So sort of forward-changing in a way... rather than... yeah.\nSpeaker B: But what Robert's saying is... is... and I think he's right... is you don't want to try to build into the construction itself...\nSpeaker B: all the synonyms and all the... maybe. I have to think about that.\nSpeaker B: I don't know. I mean, I can think... I can think of arguments in either direction on that. But somehow you want to do it.\nSpeaker D: It's just another sort of construction side. It's... how to get the possible inferences we can draw from the discourse history...\nSpeaker D: or changing of the probabilities and... or...\nSpeaker F: I guess it's like... the other thing is whether you have a user model that has, you know, whatever, a current plan...\nSpeaker F: whatever... plans I have been discussed and... I don't think.\nSpeaker E: What's the argument for putting it in the construction? Is it just that the synonyms selection is better?\nSpeaker B: Well, the argument is that... you're going to have the... if you recognize the word...\nSpeaker B: you recognize the word which means you have a electrical construction for it...\nSpeaker B: so you could just as well tag the electrical construction with the fact that it's a...\nSpeaker B: you know, 30% increase in probability of entering. So you could... you could invert the whole thing...\nSpeaker B: so you... you tagged that information onto the lexicon since you... had to recognize it anyway.\nSpeaker B: That's the argument in the other direction.\nSpeaker D: Even though the lexical construction itself or the other context won't do it.\nSpeaker D: You have to keep track of what the person is. But I'm not interested in the opening times...\nSpeaker D: is sort of more a V time.\nSpeaker B: Yeah, there's that as well.\nSpeaker D: But we'll... we have time to... just a sidetrack.\nSpeaker D: But I think it's also something that people have not done before is...\nSpeaker D: sort of abuse and autology for these kinds of inferences on whether anything relevant to the current...\nSpeaker D: something has been... has crept up in the data history already or not.\nSpeaker D: And I have the... if we wanted to have that function in the dialog...\nSpeaker D: dialog module of SmartCom, I have the written consent of Jan to put it in there.\nSpeaker B: Good. Okay. Well, this is highly relevant to someone's thesis.\nSpeaker D: Yes. You've noticed that.\nSpeaker D: That's... I'm keeping on good terms with Jan.\nSpeaker B: So the point is, it's very likely that Robert's thesis is going to be along these lines.\nSpeaker B: And the local rules are... if it's your thesis, you get to decide how it's done.\nSpeaker B: Okay. So if you know... if this is... seriously, if this becomes part of your thesis, you say, hey, we're going to do it this way. That's the way it's done.\nSpeaker F: Yeah, it's not me. It's always me when it's someone's thesis.\nSpeaker B: No, no, no. No, no. We got a lot of thesis.\nSpeaker B: It's going to be nice right now. Yeah.\nSpeaker F: So nice.\nSpeaker D: Well, let's talk after Friday the 29th, then we'll see you all.\nSpeaker D: Right. So he's got a...\nSpeaker D: He's got a... meeting a company with a thesis advisor.\nSpeaker D: I should try to finish it by then, yeah.\nSpeaker B: Yeah. So I think, in fact, that's the other thing.\nSpeaker B: This is... speaking of heart problems, it's a very good time to start trying to make explicit where construal comes in and where the construction per se ends and where construal comes in.\nSpeaker B: Because this is clearly part of...\nSpeaker B: I've never been to that.\nSpeaker B: We've been doing quite a bit of that.\nSpeaker B: That's part of what the...\nSpeaker B: Many jobs for you, Robert.\nSpeaker B: Yeah. Well, he's going to need you.\nSpeaker B: He's always landing your camera in your line.\nSpeaker B: Yeah. Right. So...\nSpeaker B: So that's why we want the formalism.\nSpeaker B: It's because it is going to have implicit in it.\nSpeaker F: No, you weren't there on purpose.\nSpeaker F: Maybe my thesis decision.\nSpeaker B: Right. Well, that's...\nSpeaker B: Ten of... they aren't decisions. They're just proposals.\nSpeaker B: Decisions.\nSpeaker B: Okay.\nSpeaker B: That's the point.\nSpeaker D: It's constraints.\nSpeaker D: That's kind of constraints that Robert's going to do.\nSpeaker B: Yeah.\nSpeaker E: Yeah, but the decisions I made were headed to with my thesis.\nSpeaker E: So consequently, don't I get to decide then that it's Robert's job?\nSpeaker E: No.\nSpeaker F: I'll just pick a piece of the problem and then push the hard stuff into the center and say it's Robert's.\nSpeaker D: I've always been completely in favor of consensus decisions.\nSpeaker D: So we'll...\nSpeaker D: We'll find a way.\nSpeaker B: We will.\nSpeaker B: But...\nSpeaker B: I haven't.\nSpeaker D: It might even be interesting then to say that I should be forced to sort of pull some of the ideas that if we're floating in my head out of the top head.\nSpeaker D: Yes.\nSpeaker D: And...\nSpeaker D: That's better for us, that's good.\nSpeaker D: Anyway.\nSpeaker B: No, absolutely.\nSpeaker B: So you had done one draft.\nSpeaker D: Yes.\nSpeaker D: And...\nSpeaker B: And another draft.\nSpeaker D: None of that is basically still around, but...\nSpeaker D: I guess it's not around.\nSpeaker B: I didn't read it.\nSpeaker B: I'm shocked.\nSpeaker B: This is the first time I've seen a thesis proposal change.\nSpeaker B: Anyway.\nSpeaker B: But yeah, second...\nSpeaker B: That would be great.\nSpeaker B: I mean, you're going to need it anyway.\nSpeaker D: Yeah, and I would like to discuss it and get your guys' input and make it sort of bomb-proof.\nSpeaker B: Yep.\nSpeaker B: Bomb-proof.\nSpeaker B: Put it proof.\nSpeaker B: That's what I was searching for.\nSpeaker B: Anyway.\nSpeaker B: Good luck.\nNone: So that...\nSpeaker B: I mean, so this is the point is we're going to have to cycle through this.\nSpeaker B: But the draft of the proposal on construction is going to tell us a lot about what we think needs to be done by Construro.\nSpeaker B: And we've got to be doing it.\nSpeaker D: Yeah, we need some...\nSpeaker D: Then we need to make some dates.\nSpeaker D: Meeting...\nSpeaker D: Regular meeting time for the summer.\nSpeaker D: We really haven't found what we did.\nSpeaker D: Thursday is one for a while.\nSpeaker D: I just talked to ARMY.\nSpeaker D: It's a coincidence that he can't do...\nSpeaker D: Couldn't do today here.\nSpeaker D: Usually he can.\nSpeaker D: Usually he has no real constraints.\nSpeaker B: And the...\nSpeaker B: The NTL meeting moved to Wednesday because of...\nSpeaker B: There was just...\nSpeaker B: Yeah, you wanted to put...\nSpeaker B: But...\nSpeaker B: So if that's okay with you, you...\nSpeaker C: Is it staying basically at that Wednesday noon?\nSpeaker C: Yeah, it was...\nSpeaker C: I thought it was staying.\nSpeaker F: Right.\nSpeaker F: I thought it was just a week that we were changing.\nSpeaker D: Okay.\nSpeaker D: And how do we feel about doing it Wednesday?\nSpeaker D: Because it seems to me that this is sort of a time where when we...\nSpeaker D: Have things to discuss with other people, they seem to be tons of people around.\nSpeaker B: The only disadvantage is that it may interfere with other...\nSpeaker B: Subgroup meetings.\nSpeaker B: You know, other...\nSpeaker B: No, people in this group connecting with...\nSpeaker B: Those people...\nSpeaker B: Those people...\nSpeaker B: Who might not be around so much.\nSpeaker B: I don't care.\nSpeaker B: You know, I have no fix.\nSpeaker C: I would like to avoid more than one ICSI meeting per day.\nSpeaker C: And passing over.\nSpeaker B: Okay.\nSpeaker C: I mean, I don't know.\nSpeaker D: No, that's fine.\nSpeaker D: I mean...\nSpeaker D: I'd like to have them all in one day.\nSpeaker D: Yeah.\nSpeaker B: I get to them often then.\nSpeaker B: People different their tastes in this matter.\nSpeaker B: I'm neutral.\nSpeaker B: Yeah.\nSpeaker A: Who's here anyways?\nSpeaker B: Yeah, but...\nSpeaker B: That's me too.\nSpeaker D: I'm basically I'm here.\nSpeaker D: One sort of thing is this room is taken after 3.30, pretty much every day by the data collection.\nSpeaker D: So we have subjects in here except for this week we have subjects in here.\nSpeaker D: That's why it was 1.\nSpeaker D: Okay.\nSpeaker D: So we just knew...\nSpeaker F: Do you say that Ami can't make 1 o'clock?\nSpeaker D: No, he can.\nSpeaker D: Okay.\nSpeaker D: So let's say Thursday 1.\nSpeaker D: But for next week this is a bit late.\nSpeaker D: So I would suggest that we need to talk...\nSpeaker D: Oh, okay.\nSpeaker F: We do Thursday at 1.30 with that...\nSpeaker F: No.\nSpeaker D: No.\nSpeaker D: Yes.\nSpeaker D: Oh, really?\nSpeaker D: Because this room is again taken at 2.30 by Morgan.\nSpeaker F: Okay.\nSpeaker F: Okay, that's fine.\nSpeaker D: And the meeting recording meeting...\nSpeaker D: Okay.\nSpeaker D: Meeting recording on meeting meetings.\nSpeaker D: So...\nSpeaker B: So you propose that we meet Tuesday.\nSpeaker B: How about that?\nSpeaker F: I think...\nSpeaker F: We are meeting Tuesday.\nSpeaker F: I mean we usually meet Tuesday for...\nSpeaker F: I think we have a link list at 2.\nSpeaker F: So...\nSpeaker F: Do you want to meet again here?\nSpeaker E: Is the speech gen meeting still at on Tuesdays?\nSpeaker D: Well, actually we did scrap our Monday time just because...\nSpeaker D: Maybe I didn't need a scrap.\nSpeaker D: Come Monday.\nSpeaker D: So there is nothing synceding Monday anymore either.\nSpeaker E: Although I thought you wanted to go camping on Monday...\nSpeaker E: Or take off Mondays a lot so you could go camping.\nSpeaker D: That's another thing.\nSpeaker D: But I mean those are usually the holidays anyways.\nSpeaker D: Like usually.\nSpeaker D: Sometimes.\nSpeaker D: It works out that way.\nSpeaker D: So...\nSpeaker F: Well, I mean the linguist meeting happens to be at 2.\nSpeaker F: But I think that's...\nSpeaker C: It should be relatively flexible.\nSpeaker F: It's pretty flexible I think.\nSpeaker C: Yeah.\nSpeaker C: There's just sort of the...\nSpeaker C: The multiple-error meetings.\nSpeaker C: Yeah.\nSpeaker C: And you know, of course Nancy and I are just sort of always talking anyway.\nSpeaker C: Yeah.\nSpeaker C: You do it in that room.\nSpeaker D: You know, I mean...\nSpeaker D: Okay, so forget about the camping thing.\nSpeaker D: So...\nSpeaker D: Any other problems?\nSpeaker D: I suggest that Monday.\nSpeaker D: If that's a problem for me then I shouldn't suggest it.\nSpeaker D: So...\nSpeaker C: All of the proposed times I'm trying to...\nSpeaker B: One day...\nSpeaker B: Okay, what I mean...\nSpeaker B: What I think we're just saying is that...\nSpeaker B: Early...\nSpeaker B: At least for next week there's a lot of stuff we want to get done.\nSpeaker B: So why don't we find the meet Monday and...\nSpeaker B: We'll see if we want to meet any more than that.\nSpeaker A: Okay.\nSpeaker A: What time?\nSpeaker A: At...\nSpeaker D: One, two, three.\nSpeaker D: Three is too late.\nSpeaker D: Two third time.\nSpeaker B: Oh, I...\nSpeaker B: Yeah, I actually...\nSpeaker B: Two is the earliest I can meet on Monday.\nNone: Okay, two.\nSpeaker B: You're I'm blissfully agreeing to things in realizing that I actually do have some scuffle schedule on Monday's.\nSpeaker F: You guys look streamlined, right?\nSpeaker E: No way.\nSpeaker F: You'll come and take all the headphone first.\nSpeaker F: I'm a dead memory mind.\nSpeaker D: And...\nSpeaker D: Thank you.\nSpeaker D: Why do I have this and those I'm getting?\nSpeaker D: Do I get to see your formalism before that?\nSpeaker F: Uh...\nSpeaker F: Would you like to?\nNone: Mm-hmm.\nNone: Okay.\nSpeaker D: I was actually going to work on it for tomorrow.\nSpeaker D: Get a notion of what you guys have been told for me.\nSpeaker B: Maybe we can put this part of what we can do Monday if we want.\nSpeaker B: Okay.\nSpeaker B: I mean, I...\nSpeaker F: Some version.\nSpeaker F: Yeah, so there was like, you know, in my head the goal to have like an intermediate version, like, everything I know.\nSpeaker F: And then I would talk to you and figure out everything you know.\nSpeaker F: Okay.\nSpeaker C: But, you know, see if they're consistent.\nSpeaker C: Maybe you and I should meet sort of formalized first thing Monday.\nSpeaker C: Yeah.\nSpeaker F: That's fine with me, sir.\nSpeaker F: You said you're busy over until the weekend, right?\nSpeaker C: Yeah, sort of through the weekend because Kate has a photography.\nSpeaker C: That's fine.\nSpeaker F: So we might continue our email thing that might be fine to you, sir.\nSpeaker C: Yeah, if there's time after this, I'll show you the number.\nSpeaker F: Okay, that would be cool.\nSpeaker F: Okay.\nSpeaker F: And we'll...\nSpeaker D: So the ideas on Monday and Tuesday will see an intermediate version of the formalism for the construction.\nSpeaker D: So that's okay for you.\nSpeaker D: And do an online merging with my Constoral ideas.\nSpeaker D: Sure.\nSpeaker D: Okay.\nSpeaker D: Sure.\nSpeaker D: So it won't be like a, say my formal presentation of my proposal, it'll be more like towards finalizing that proposal.\nSpeaker D: Okay.\nSpeaker F: Because then you'll find out more about what we're making you do.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: And then we'll make a presentation of your proposal.\nSpeaker D: Perfect.\nSpeaker D: Can you also write it up?\nSpeaker D: So what we're doing and the complement is Robert.\nSpeaker D: I send you a staff file, right?\nSpeaker D: It's just...\nSpeaker D: I always send you my, my, my new files.\nSpeaker D: Okay.\nSpeaker D: And...\nSpeaker D: So it's good.\nSpeaker C: So we should probably talk about the other side of the, where is X construction, which is the issue of, how do you simulate questions?\nSpeaker C: Yeah.\nSpeaker C: Does SimSpec look like for questions?\nSpeaker C: Mm-hmm.\nSpeaker C: Because...\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Now we had an idea for this, which seems like a problem.\nSpeaker B: Great.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: SimSpec may need, we may need to rename that.\nSpeaker B: Oh yeah.\nSpeaker B: Okay.\nSpeaker B: So let's think of a name for whatever the, his intermediate structure is.\nSpeaker B: Oh, we talked about SimSpec for some unexpected specification.\nSpeaker B: And that seems...\nSpeaker B: Uh...\nSpeaker B: You know...\nSpeaker B: You know, it's a minimal change.\nSpeaker F: When you change one vowel, it's great.\nSpeaker F: All the old like...\nSpeaker F: Right.\nSpeaker F:...crabs just changed the...\nSpeaker B: Right.\nSpeaker B: You're a little substance to substance.\nSpeaker B: That's what text substitutes in Macros or for.\nSpeaker B: Anyway.\nSpeaker B: So let's, let's, for the moment, call it that until we think of something better.\nSpeaker B: And yeah, we absolutely need to find part of what was missing were markings of all sorts that weren't in there.\nSpeaker B: Including questions.\nSpeaker B: We never did figure out how we were going to do emphasis in...\nSpeaker B: In the SimSpec.\nSpeaker F: Yeah, we talked a little bit about that too, which...\nSpeaker F: It's hard for me to figure out that sort of general linguistic issues, how they map onto this particular one, but...\nSpeaker F: Okay.\nSpeaker B: That's part of the formalism has got to be...\nSpeaker B: How things like that get marked.\nSpeaker F: Do you have data like the... you have preliminary data?\nSpeaker F: Because I know, you know, we've been using this one, you see, sentence, and I'm sure you guys have...\nSpeaker F: Maybe you're the one who's been looking at the rest of it.\nSpeaker F: It'd be unusual for me if we wanted to have it a little bit more.\nSpeaker C: What I've been looking at...\nSpeaker C: No, it's not been the data so far.\nSpeaker C: I just sort of said, all right, let's see if I can get noun phrases and...\nSpeaker C: Major verb constructions out of the way first.\nSpeaker C: And I have not gotten them out of the way yet.\nSpeaker C: Yeah, I've not really approached a lot of the data, but I mean, obviously, like the question one, since we have this idea about the indefinite pronouncing and all that, you know, like, can try and run with that, you know, try and do some of the question constructions now.\nSpeaker D: You want to run the indefinite pronoun idea past Jerry?\nSpeaker C: Oh, yeah.\nSpeaker C: The basic idea is that, you know, I would see...\nSpeaker D: So, very fixed the car with a wrench. So you're following me.\nSpeaker D: And then, who fixed the car with a wrench?\nSpeaker D: You basically are told to do this.\nSpeaker D: And elegantly, to the way you would do someone fixed the car with a wrench.\nSpeaker D: And then you hand it back to your hippocampus and find out what that means.\nSpeaker D: And then come up with that, who that someone was.\nSpeaker C: So, the WHO question has this sort of extra thing which says, and when you're done, tell me who fills that slot.\nSpeaker C: And, you know, this is sort of a nice way to do it.\nSpeaker C: The idea of sort of saying that you treat from the simulation point of view or whatever, you treat WHO constructions similarly to indefinite pronouns like someone fixed the car because lots of languages have WHO questions with an indefinite pronoun in CQ or whatever, and you just get an intonation to tell you that it's a question.\nSpeaker B: So, which is, school amazation.\nSpeaker B: In logic, it's, yeah.\nSpeaker F: It's a problem.\nSpeaker F: It's a problem constant in, yeah.\nSpeaker B: Sure.\nSpeaker B: Okay.\nSpeaker B: There's nothing that's bad.\nSpeaker B: Right, no.\nSpeaker B: The logicians have...\nSpeaker B: That's right.\nSpeaker C: It makes sense from that point of view too, which is actually better.\nSpeaker C: Anyway, but just that kind of thing, and we'll figure out exactly how to write that up and so on.\nSpeaker C: I know all the focus stuff.\nSpeaker C: You sort of just dropped that because it was too weird, and we didn't even know like when we were talking about it.\nSpeaker C: That's the top of it.\nSpeaker C: Yeah.\nSpeaker B: Well, part of what the exercise is, but in the next week, you say, what are the things that we just don't have answers for yet?\nSpeaker B: That's fine.\nSpeaker D: Well, if you do want to discuss focus background, then get me into that because I mean, I've scientifically worked on that for almost two years.\nSpeaker C: Okay.\nSpeaker C: And certainly...\nSpeaker F: Yeah, you should definitely be on that.\nSpeaker F: Maybe by after Monday, you could see what things we...\nSpeaker C: I think we'll figure out what our questions are, for example.\nSpeaker C: Yeah.\nSpeaker C: To ask you.\nSpeaker C: Okay.\nSpeaker B: Then, Hans.\nSpeaker B: I haven't seen Hans, but...\nSpeaker B: It's been around.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So has he been involved with this or...?\nSpeaker B: With us?\nSpeaker F: Yeah.\nSpeaker F: I would say that those discussions have been primarily Keith and me, but...\nSpeaker F: I mean, he sort of... the last meeting we had, I think we were all very much part of it.\nSpeaker C: But...\nSpeaker C: Sometimes, Hans has been sort of coming in there as sort of like a devil's advocate, like role or something like this.\nSpeaker C: I'm going to pretend I'm a linguist who has nothing to do with this.\nSpeaker C: This makes no sense, and he'll just go off on parts of it, which definitely need fixing, but aren't where we're at right now.\nSpeaker F: Like, what you call certain things, which we decided along, go, we don't care that much right now.\nSpeaker F: But in the sense, it's good to know that he, of all people, like maybe a lot of people would have...\nSpeaker F: He's a stronger reaction, so he's like a relatively friendly linguist, and yet, word like constraint, because there are a lot of problems.\nSpeaker B: Okay.\nSpeaker B: This is consistent with the role I had suggested that he play, which was that...\nSpeaker B: One of the things I would like to see happen is a paper that was tentatively called, towards a formal cognitive semantics, which was addressed to these linguists who haven't been following this stuff at all.\nSpeaker B: So, it could be that he's actually, at some level, thinking about how am I going to communicate this story?\nSpeaker B: So, internally, we should just do whatever works, because it's hard enough.\nSpeaker B: But if he is really going to turn around and help to write this version that does connect with as many as possible of the other linguists in the world, then it becomes important to use terminology that doesn't...\nSpeaker B: Make it hard... I mean, it's going to be plenty hard for people to understand that his idea is...\nSpeaker B: But you don't want to make it worse.\nSpeaker C: Right, I mean, that role is indispensable, but that's not where sort of our heads were at in this meeting.\nSpeaker B: Yeah, yeah, no, that's fine. I just wanted to...\nSpeaker B: I have to catch up with him, and I wanted to get a feeling from that.\nSpeaker C: Okay.\nSpeaker C: So, I don't know what his take will be on these meetings.\nSpeaker C: Exactly.\nSpeaker A: Okay.\nSpeaker C: I think it's good to know that we're talking about your govily.\nSpeaker F: When we're into data and looking at the some specific linguistic phenomenon in English, or in German, particular word effort, that's great.\nSpeaker F: And Hans, if anything, they have more say than I'd say I would about these things.\nSpeaker F: But when it's like, well, how do we capture these things?\nSpeaker F: I think it's definitely been Keith and I have...\nSpeaker F: Well, that's good.\nSpeaker B: I think that should be the court group, and that's, you know, I think very close to the maximum number of people working together that you've got.\nSpeaker F: We actually have... I think we have been making progress.\nSpeaker F: I definitely get the impression.\nSpeaker B: Yeah, that's great.\nSpeaker B: Yeah.\nSpeaker F: So, anyone else?\nSpeaker F: We're in the balance, anyway.\nSpeaker B: Well, then we have to come back to the bigger group.\nSpeaker B: Right.\nSpeaker B: Right.\nSpeaker B: And we're going to...\nSpeaker B: Because this other big thing we haven't talked about is actually implementing this stuff.\nSpeaker B: So they guess the three of us are going to connect tomorrow.\nSpeaker F: I was just going to say that, for instance, there was, you know, out of a meeting with John O.\nSpeaker F: Came the suggestion that, oh, could it be that the meaning constraints really aren't used for selection, which has sort of been implicit in the parsing strategy we talked about.\nSpeaker F: In which case, we can just say that they're the effects or the bindings, which, so far, in terms of like putting up all the constraints as, you know, pushing them into type constraints, when I then propose it to the linguist who haven't yet given me, you know, we haven't yet thought of a reason why that wouldn't work.\nSpeaker F: Right.\nSpeaker F: As long as we allow our type constraints to be reasonably complex.\nSpeaker F: Well, anyway.\nSpeaker B: Yeah, it has to, in the sense that you're going to use them eventually. So, you know, it's sort of a generating test kind of thing.\nSpeaker B: And if you over-generate, then you'll have to do more.\nSpeaker B: I mean, if there are some constraints that you hold back and don't use in your initial matching, and you'll match some things.\nSpeaker B: I mean, I don't think there's any way that it could completely fail. It could be that you wind up many.\nSpeaker B: The original bad idea of purely context free grammars died because there were just vastly too many parses, you know, exponentially number many parses.\nSpeaker B: And so that the concern might be that nothing would totally fail that that would still generate too many.\nSpeaker F: Right. So, Joe, by just having semantic, bringing semantics in, for matching just in the form of semantic types.\nSpeaker F: Like, conceptually, these have to be construed as this, this, and this.\nSpeaker F: Yeah.\nSpeaker F: Might still give us quite a few possibilities.\nSpeaker F: We don't know.\nSpeaker F: And it certainly helps a lot.\nSpeaker B: No question.\nSpeaker B: And I think it's a perfectly fine place to start, you know, and say, let's see how far we can go this way.\nSpeaker B: And, uh, well, it definitely makes the problem easier.\nSpeaker B: I'm, I'm, I'm, I'm, I'm into that.\nSpeaker B: Because I think, I think it's, as you know, I think it's real hard and if we, if, all right.\nSpeaker F: So, Friday, Monday, Monday.\nSpeaker F: Yeah.\nSpeaker F: Okay. That's Tuesday.\nSpeaker F: That's the, here, question.\nSpeaker D: So, you, your dance card is, you did it.\nSpeaker F: Yeah. And I've nothing to do this weekend, but work.\nSpeaker E: Yeah.\nSpeaker F: What about, what about DDR?\nSpeaker F: Uh, I don't have it this weekend, so don't have to worry about that.\nSpeaker F: DDR.\nSpeaker F: Yeah.\nSpeaker F: Dancing, dance revolution.\nSpeaker F: It's a, it's like a game, but it's for like dancing.\nSpeaker F: It's hard to, it's like karaoke, but for dancing.\nSpeaker F: And they tell you what.\nSpeaker F: It's amazing.\nSpeaker F: It's so much fun.\nSpeaker F: Oh, yeah.\nSpeaker F: Yeah.\nSpeaker F: So, my friend has a home version.\nSpeaker F: He brought it over and we are so into it.\nSpeaker F: It's so amazing.\nSpeaker F: Oh, yeah.\nSpeaker F: Now, this, what you, you know of it?\nSpeaker F: It's, it's one of your hobbies.\nSpeaker B: Although, I, God knows I could do something.\nSpeaker F: It's great exercise, I must say.\nSpeaker B: But, uh, no, I, I can't wait to hear.\nSpeaker B: Yeah.\nSpeaker B: Apparently, always the hottest thing in Tokyo.\nSpeaker F: Oh, definitely.\nSpeaker F: They have like places instead of like, instead of karaoke bars, now they have like DDR.\nSpeaker F: Yeah, I didn't until I started hanging out with this friend who was like, oh, I can bring over the DDR if you want.\nSpeaker F: Oh, dance, dance revolution.\nSpeaker F: Okay.\nSpeaker F: He actually brought a clone called stepping selection, but it's just as good.\nSpeaker F: Hey.\nNone: You You You You You You You You You You You You\n"}, {"meeting_id": "Bmr027", "summary": "The meeting was mostly about the logistics of covering the legal bases around releasing meeting data. The team wanted to make sure that meeting participants would not sue for libel or releasing unwanted information. The team also went over the transcriptions that IBM had done as well as storage space, which was finally looking up. The meeting ended with a general discussion about the progress of the group and future directions.", "dialogue": "Speaker G: Okay, Adam, what is the mic that Jeremy's wearing?\nSpeaker G: It's the earplug mic.\nSpeaker C: It's the earplug.\nSpeaker G: Is that a wireless or no?\nSpeaker G: No, it's wired.\nSpeaker C: Is that old school?\nSpeaker C: You can't hear anything during the meeting.\nSpeaker H: What?\nSpeaker H: What?\nSpeaker H: Should we close the door?\nSpeaker F: It's a fairly good mic actually.\nSpeaker F: Well, I shouldn't say it's a good mic.\nSpeaker F: All I really know is that the signal level is okay.\nSpeaker F: It's a quality.\nSpeaker H: That's a...\nSpeaker H: Oh, necessary.\nSpeaker F: So, I can send out agenda items because until five minutes ago we only had one agenda item and now we have two.\nSpeaker F: Okay.\nSpeaker H: So, just to repeat the thing that we said last week, there was this suggestion of alternating weeks on more automatic speed recognition related or not.\nSpeaker H: Right.\nSpeaker H: So, what sort of division?\nSpeaker F: So, which week are we in?\nSpeaker F: Well, we haven't really started, but I thought we were more or less did meeting recorder stuff last week, so I thought we could do...\nSpeaker H: I thought we had a thing about speed recognition last week too.\nSpeaker F: But I think you're also, if they're short agenda items, we could also do a little bit of each.\nSpeaker F: Yeah.\nSpeaker F: It seemed to be having difficulty getting this adjusted.\nSpeaker F: Here we go.\nSpeaker F: So, as most of you should know, I did send out the consent form thingies.\nSpeaker F: And so far no one has made any comments on them.\nSpeaker F: So, no one has bleeped out anything.\nSpeaker F: So, yeah.\nSpeaker F: I don't expect anyone to, but...\nSpeaker H: So, what follows at some point you go around and get people to sign something?\nSpeaker F: No. We had spoken about this before and we had decided that they only needed to sign once.\nSpeaker F: And the agreement that they already signed simply said that we would give them an opportunity.\nSpeaker F: So, as long as we do that, we're covered.\nSpeaker H: And how long of an opportunity did you tell them?\nSpeaker F: July 15th.\nSpeaker H: July 15th.\nSpeaker H: Oh, so they have a plenty of time.\nSpeaker H: Yeah. Given that it's that long, why was that day chosen?\nSpeaker H: You just felt like...\nSpeaker F: Jane told me July 15th.\nSpeaker F: So, that's what I said.\nSpeaker C: I just meant that that was the release date that you had on the data.\nSpeaker F: Oh, I didn't understand that there was something specific.\nSpeaker F: I had heard July 15th, so that's what I put.\nSpeaker H: The only mention I recall about that was just that July 15th or so is when this meeting starts.\nSpeaker H: That's why.\nSpeaker H: Oh, I see.\nSpeaker C: You said you wanted it to be available then.\nSpeaker C: You mean it to be the hard deadline.\nSpeaker C: It's fine with me if it is or a week.\nSpeaker C: But I thought it might be good to remind people two weeks prior to that.\nSpeaker C: So, by the way, what's your last?\nSpeaker H: We probably should have talked about it.\nSpeaker H: Because if we want to be able to give it to people July 15th, if somebody is going to come back and say, okay, I don't want this and this and this and this used.\nSpeaker H: Clearly, we need some time to respond to that.\nSpeaker F: Yeah, as I said, I just got one date and that's the one I used.\nSpeaker F: So, but I can send a follow-up.\nSpeaker F: I mean, it's almost all us.\nSpeaker F: I mean, the people who are in this meeting was...\nSpeaker F: These meetings that were in set one.\nSpeaker G: Was my response okay?\nSpeaker G: I just replied to the email saying they were all fine.\nSpeaker F: That's fine.\nSpeaker F: My understanding of what we had agreed upon when we had spoken about this months ago was that we don't actually need a reply.\nSpeaker F: We just need to tell them that they can do it if they want.\nSpeaker F: I just didn't remember.\nSpeaker F: And so, no reply is no changes.\nSpeaker C: And he's got it.\nSpeaker C: So, the default thing you see when you look at the page is okay.\nSpeaker C: So, that's very clear all the way down the page.\nSpeaker C: And they have two options.\nSpeaker C: They can change it to one of them is sensor.\nSpeaker C: And the other one is incorrect.\nSpeaker C: Which means also we get feedback on if there's something that needs to be adjusted.\nSpeaker C: Because these are very highly technical things.\nSpeaker C: I mean, it's an added level of checking on the accuracy of the transcription as I see it.\nSpeaker C: But in any case, people can agree to things that are wrong.\nSpeaker F: So, the reason I did that was just so that people would not ask to have stuff removed because it was transcribed incorrectly.\nSpeaker C: And the reason I liked it was because it gives them the option of being able to correct it.\nSpeaker C: Very.\nSpeaker C: Approved and corrected.\nSpeaker C: And so, you have it nicely set up so they email you and...\nSpeaker F: When they submit the form, it gets processed and emailed to me.\nSpeaker C: And I wanted to say the meetings that are involved in that set are robustness and meeting recorder.\nSpeaker C: The German ones will be ready for next week.\nSpeaker C: Those are three of those different set of people.\nSpeaker C: And we can move on.\nSpeaker C: German ones?\nSpeaker C: Well.\nSpeaker H: NSA.\nSpeaker C: Okay, I spoke loosely.\nSpeaker C: The German French... sorry, the German Dutch and Spanish ones.\nSpeaker G: All those are the NSA's.\nSpeaker C: The German Dutch and Spanish.\nSpeaker F: That's a month, which is...\nSpeaker F: The only thing I said in the email is that the data is going to be released on the 15th.\nSpeaker F: I didn't give any other deadline.\nSpeaker F: So, my feeling is if someone after the 15th says, wow, I suddenly found something, we'll delete it from our record.\nSpeaker F: We just won't delete it from whatever has already been released.\nSpeaker C: That's a little bit difficult.\nSpeaker F: What else can we do?\nSpeaker F: If someone says, hey, look, I found something in this meeting and it's libelous and I want it removed.\nSpeaker F: What can we do?\nSpeaker C: We have to remove it.\nSpeaker C: It's true.\nSpeaker C: I agree with that part.\nSpeaker C: But I think that we need to have a message to them very clearly that beyond this date, you can't make additional changes.\nSpeaker H: I mean, I think that somebody might request something even though we say that, but I think it's good to at least start someplace like that.\nSpeaker H: So, if we agreed, okay, how long is a reasonable amount of time for people to have if we say two weeks or if we say a month?\nSpeaker H: I think we should just say that, say that, you know, as...\nSpeaker H: or the page you signed, you have the ability to look over this stuff, so forth.\nSpeaker H: And because we...\nSpeaker H: I would imagine it's a certain generic thing, I would say, because we will continually be making these things available through other researchers.\nSpeaker H: This can't be open-ended, and so please give us back your response within this, you know, within this amount of time.\nSpeaker F: Well, did you rep-email and look at the pages, I sent?\nSpeaker F: Did I know?\nSpeaker F: I haven't.\nSpeaker F: Okay, well, why don't you do that and then make comments on what you want me to change?\nSpeaker H: No, no, I'm not saying that you should change anything.\nSpeaker H: But I'm trying to spark a discussion, hopefully, with people who have ever had it, so that you can decide on something.\nSpeaker H: So, I'm not telling you what to decide.\nSpeaker H: I'm just saying you should decide something.\nSpeaker F: I already did decide something, and that's what's in the email.\nSpeaker F: And if you disagree with it, why don't you read it and give me comments on it?\nSpeaker C: I think that there's one missing line.\nSpeaker H: Well, the one thing that I did read that you just repeated to me was that you gave the specific date of July 15th.\nSpeaker H: And you also just said that the reason you said that was because someone said it to you.\nSpeaker H: Right. So, what I'm telling you is that what you should do is come up with a length of time that you guys think is enough, and you should use that rather than the state that you just got from somewhere.\nSpeaker H: That's all I'm saying.\nSpeaker H: Okay.\nSpeaker C: I have one question.\nSpeaker C: This isn't the summer period, and presumably people may be out of town, but we can make the assumption can't we that they will be receiving email most of the month, right?\nSpeaker H: Well, you're right. Sometimes somebody will be away, and for any length of time that you choose, there is some person, sometimes who will not end up reading it.\nSpeaker H: It's just a certain risk to take.\nSpeaker B: So maybe when am I on by the way?\nSpeaker B: I don't know.\nSpeaker B: You should be.\nSpeaker F: Hello.\nSpeaker F: You should be Channel B.\nSpeaker B: Okay. All right. So maybe we should say in when the whole thing starts, when they sign the agreement that specified exactly how they will be contacted, and they can be asked to give a phone number and an email address or both.\nSpeaker B: And then say very clearly that if they don't hear from them, as more suggested by a certain time or after a certain period, after we contact them, that is a bliss of the giving their agreement.\nSpeaker F: Well, they've already signed a form.\nSpeaker F: And the form was approved by human subjects, so if that's going to be a little hard to modify.\nSpeaker C: Well, the form doesn't say if you don't respond by X number of days or X number of weeks.\nSpeaker B: So what does it say about the process of the review process?\nSpeaker C: That you'll be provided access to the transcripts and then allowed to remove things that you'd like to remove before it goes to the general larger audience.\nSpeaker F: Yeah, you can read what you already signed.\nSpeaker D: I guess when I read it, I'm not as diligent as Chuck, but I had the feeling I should probably respond until Adam, like, I got this and I will do it by this date.\nSpeaker D: And if you don't hear from me by then, in other words, responding to your email once, right away, saying, as soon as you get this, could you please respond.\nSpeaker D: And then if the person thinks they'll need more time because they're out of town or whatever, they can tell you at that point because...\nSpeaker F: Oh, I just didn't want to do that because I don't want to have a discussion with every person if I can avoid it.\nSpeaker F: So what I wanted to do was just send it out and say on the 15th of the date is released if you want to do something about it, do something about it, but that's it.\nSpeaker D: I kind of like this.\nSpeaker D: Okay, so we're assuming that...\nSpeaker B: That would be great, but you should probably have a legal person look at this and make sure it's okay.\nSpeaker B: Because if you do this and then there's this put later and someone who understands these matters concludes that they didn't have enough opportunity to actually exercise their right.\nSpeaker D: Or they might never have gotten the email because although they sign this, they don't know by which date to expect your email.\nSpeaker D: So someone who's machine is down or whatever, I mean we have no internally that people...\nSpeaker F: Well, okay, let me reverse this.\nSpeaker F: So let's say someone I send this out and someone doesn't respond.\nSpeaker F: Do we delete every meeting that they were in?\nSpeaker D: I don't think so.\nSpeaker F: We're hoping that doesn't happen, but that's why there's such a dangerous, dirt mail.\nSpeaker F: That will absolutely happen because people don't read their email or they'll read and say I don't care about that, I'm not going to delete anything and they just won't reply to it.\nSpeaker B: So we have mailing addresses for these people?\nSpeaker F: No. We have what they put on the speaker form which was just generic contact information.\nSpeaker C: But the ones that we're dealing with now are all local except the ones who...\nSpeaker C: We're totally in contact with all the ones in those two years.\nSpeaker C: So maybe...\nSpeaker C: That's not that many people and if there is an advantage to having them admit and if I can help with processing that, I will.\nSpeaker C: There is an advantage having them be on record as having received the mail and indicated.\nSpeaker F: I thought we had discussed this like a year ago.\nSpeaker F: And so it seems like this is a little odd for it to be coming up yet again.\nSpeaker H: Well, we haven't experienced it before.\nSpeaker D: You'll either wonder at the beginning or you'll wonder at the end.\nSpeaker D: I mean there's no way to get around.\nSpeaker D: It's pretty much the same amount of work except for an additional email just saying they got the email.\nSpeaker D: And maybe it's better legally to wonder before a little bit earlier than.\nSpeaker C: It's much easier to explain.\nSpeaker F: Morgan can talk to our lawyer about it and find out what the status is on this because I don't want to do something that we don't need to.\nSpeaker F: Because what I'm telling you, people won't respond to the email.\nSpeaker F: No matter what you do, they're going to be people who you're going to have to make a lot of effort to get in contact with.\nSpeaker F: And do we want to spend that?\nSpeaker E: It's kind of like signing up for a mailing list. They have opt-in and opt out.\nSpeaker E: And there are two different ways.\nSpeaker E: I mean in either way works probably.\nSpeaker C: Except I really think in this case, I agree with Liz that we need to be in the clear and not have to after the fact say, oh but I assumed.\nSpeaker C: And oh I'm sorry that your email address was just accumulating mail without notifying you.\nSpeaker H: If this is a purely administrative task, we can actually have administration do it.\nSpeaker H: But the thing is that I think without going through a whole expensive thing with our lawyers from my previous conversations within my sense very much is that we would want something on record.\nSpeaker H: Yes.\nSpeaker H: Is indicating that they actually were aware of this.\nSpeaker F: Well we had talked about this before and I thought that we had even gone by the lawyers asking about that and they said, you have to, they've already signed away with that form.\nSpeaker F: They've already signed once.\nSpeaker C: I don't remember that this issue of the time period allowed for response was ever covered.\nSpeaker H: Okay. We never really talked about that.\nSpeaker D: Or the data which they would be receiving the email.\nSpeaker D: Yeah.\nSpeaker D: Or how they would indicate.\nSpeaker H: We certainly didn't talk a bit with them at all about the manner of them being.\nSpeaker H: We do it like with the materials available.\nSpeaker H: That was something that was sort of just within our implementation.\nSpeaker B: We can use a, we can use a, like they use to, you know, when they serve like, you know, like deadbeat, bats, they make it look like they want something in the lottery and then they open that love with that.\nSpeaker B: Right. And then the thing is served. So you just make it, you know, you want, you know, go to this website and you've.\nSpeaker D: That's why you never open these things.\nSpeaker F: Well, it's just we've gone from one extreme to the other where at one point a few months ago, Morgan was you were saying, let's not do anything.\nSpeaker F: And now we're, we're saying we have to follow up each person and get a signature. I mean, what are we doing here?\nSpeaker B: And then this is perfectly, you know, this is enough to give us the basis to just assume their consent if they don't reply.\nSpeaker B: But I'm not, you know, me not being a lawyer, I wouldn't just want to do that without having the the expert.\nSpeaker F: I think we had better find out so that we can find out.\nSpeaker C: These people are either email almost all the time.\nSpeaker C: Right.\nSpeaker C: I really don't see that it's a problem. I think that it's a common courtesy to ask them to expect for them to be able to have.\nSpeaker C: Try to contact them in just in case they had.\nSpeaker C: I think they appreciate it.\nSpeaker H: Yeah, my, my view before was about the nature of what was of the presentation of how my, the things that were questioning were along the lines of how easy.\nSpeaker H: How, how much implication would there be that it's likely you're going to be changing something as opposed to that was the kind of dispute I was making before.\nSpeaker H: But the attorneys, I guarantee you the attorneys will always come back with, we have to decide how stringent we want to be in these things, but they will always come back saying that you need to, you want to have something, some paper trail, or which includes electronic trail, that they have in fact, OK, did.\nSpeaker H: So I think that if you, if we send the email as you have it, if there's half the people say who don't respond at all by some period of time, we can just make a list of these people and hand it to, you know, just give it to me, you know, hand to the minister, or whatever, and they'll just call them up and say, you know, have you, is this OK? Would you please mail, mail, Adam that it is if it is or not.\nSpeaker D: So the other thing that there's a psychological effect that leads to most people that if they've responded to your email saying, yes, I will do it or yes, I got your email, they're more likely to actually do it later, then to just ignore it.\nSpeaker D: And of course, we don't want them to bleep things out, but it's a little bit better for getting their final response once they've answered you once, then if they never answer you at all, at all.\nSpeaker D: So I mean, it's not completely lost work because it might benefit us in terms of getting responses, you know, an official OK from somebody is better than no answer, even if they've responded that they got your email, and they're probably more likely to do that once they've responded that they got the email.\nSpeaker C: I also think they just simply appreciate it. I think it's a good way of fostering good will on our subjects.\nSpeaker H: I think the main thing is, I mean, what lawyers do is they always look at worst cases. So that's what they're paid to do. And so it is certainly possible that somebody's server would be down or something and they wouldn't actually hear from us and then they find this thing is in there and we've already distributed someone.\nSpeaker H: So what it says in there, in fact, is that they will be given an opportunity to blah, blah, blah.\nSpeaker H: But in fact, if we sent them something, we thought we sent them something, but they didn't actually receive it for some reason, then we haven't given them.\nSpeaker F: So how far do we have to go? Do we need to get someone's signature or is email enough? Do we have to have it noterized? I mean.\nSpeaker H: Email is now.\nSpeaker H: I mean, I've been through this, I mean, not a lawyer, but I've been through these things. Things are like this a few times with lawyers now. So I'm pretty comfortable with that.\nSpeaker G: Do you track when people log in?\nSpeaker F: If they submit the form, I get it. If they don't submit the form, it goes in the general web log, but that's not sufficient.\nSpeaker F: Because if someone just visits the website, that doesn't imply anything in particular.\nSpeaker G: Except that you know they got the mail, right?\nSpeaker C: That's right. I could give to you on the number five list if you want me to.\nSpeaker F: I'm already on it.\nSpeaker C: For that directory?\nSpeaker C: Okay, great.\nSpeaker H: So again, hopefully this shouldn't be quite as odious a problem either way.\nSpeaker H: Any of the extremes we've talked about because we're talking a pretty small-\nSpeaker F: For this set, I'm not worried because we basically know everyone on it. They're all more or less here or it's Eric and Dan and so on.\nSpeaker F: But for some of the others, you're talking about visitors who are gone from XC, who email addresses may or may not work.\nSpeaker F: So what are we going to do when we run into someone that we can't get in touch with?\nSpeaker C: I don't think there's so recent these visitors.\nSpeaker C: And they're also so prominent enough they're easy to find through.\nSpeaker C: Other methods?\nSpeaker C: I'll be able to, if you have any trouble finding them, I really think I could find them.\nSpeaker C: Okay.\nSpeaker H: Yeah, because what it really does promise here is that we will ask their permission.\nSpeaker H: And I think if you go into a room and close the door and ask their permission, they're not there.\nSpeaker H: It doesn't seem that's the intent of meaning here.\nSpeaker F: Well, the question is just whether how active it has to be.\nSpeaker F: I mean because they filled out a contact information and that's where I'm sending the information.\nSpeaker F: And so far everyone has done email.\nSpeaker F: There isn't anyone who did any other contact method.\nSpeaker H: Well, the way XC goes, people who were here 10 years ago still have forwards to other counts and so on.\nSpeaker H: So it's unusual.\nSpeaker F: So my original impression was that that was sufficient.\nSpeaker F: That if they give us contact information, that contact information isn't accurate.\nSpeaker F: That we've fulfilled our bird.\nSpeaker H: So if we get to a boundary case like that, then maybe I will call the attorney about it.\nSpeaker H: But hopefully we won't need to.\nSpeaker C: I just don't think we will.\nSpeaker C: All right.\nSpeaker C: For all the reasons that we've discussed.\nSpeaker C: So we'll see.\nSpeaker F: We'll see how many people respond to that email.\nSpeaker F: So far two people have.\nSpeaker H: Yeah, I think very few people will.\nSpeaker H: And people see long emails about things that they don't think.\nSpeaker H: It's going to be high priority.\nSpeaker H: They typically don't.\nSpeaker H: Don't read it or have read it.\nSpeaker H: Right.\nSpeaker H: Because people I swamped.\nSpeaker C: And actually, I didn't anticipate this.\nSpeaker C: So that's why I didn't give this comment.\nSpeaker C: And this discussion has made me think it might be nice to have a follow-up email within the next couple of days saying, by the way, we want to hear back from you by X date.\nSpeaker C: And then add what Liz said.\nSpeaker C: Please respond.\nSpeaker C: Please indicate you received this mail.\nSpeaker H: Or maybe even additionally.\nSpeaker H: Even if you've decided you have no changes you'd like to make if you could respond to that.\nSpeaker D: That would definitely work on me.\nSpeaker D: It makes you feel like if you're predicting that you might not answer.\nSpeaker D: You have a chance now to say that.\nSpeaker D: Whereas, I mean, I would be much more likely myself.\nSpeaker D: And yet, all my email to respond at that point saying, you know what, I'm probably not going to get to it or whatever.\nSpeaker D: Rather than just having seen the email thinking I might get to it and never really pushing myself to actually do it until it's too late.\nSpeaker G: I was thinking that it also lets them know that they don't have to go to the page to accept this.\nSpeaker D: Right.\nSpeaker G: That's true.\nSpeaker G: Yeah, so that way they can see from that email that if they just write back and say I got it, no changes.\nSpeaker G: They're off the hook.\nSpeaker G: They don't have to go to the webpage.\nSpeaker H: The other thing I've learned from dealing with people sending in reviews and so forth is if you say you've got three months to do this review, people do it two and seven eighths months from now.\nSpeaker H: Right.\nSpeaker H: So you've got three weeks to do this review.\nSpeaker H: They do two and seven eighths weeks from now.\nSpeaker H: They do the review.\nSpeaker H: And so if we make it a little less time, I don't think it will be that.\nSpeaker F: Well, and also if we want it ready by the 15th, that means we better give them deadline of the first.\nSpeaker F: If we have any prayer of actually getting everyone to respond in time.\nSpeaker H: There's the responding part in there is also what if, I mean, hopefully this doesn't happen.\nSpeaker H: What if there are a bunch of deletions that have to get put in the changes, then we actually have to deal with that if we want it to.\nSpeaker F: Diskspace.\nSpeaker F: Oh my god.\nSpeaker F: I hadn't thought about that for every meeting, any meeting which has any bleeps in it, we need yet another copy of.\nSpeaker F: Just get to that channel.\nSpeaker F: Oh no, we need all the channels.\nSpeaker F: Do you have to do the other close talk?\nSpeaker F: Yes, absolutely.\nSpeaker F: You could just log across talk.\nSpeaker F: Well, you have to copy the whole file, right, because we're going to be releasing the whole file.\nSpeaker C: I think at a certain point that copy that has the deletions will become the master copy.\nSpeaker F: Yeah, it's just I hate deleting any data.\nSpeaker F: So I don't want, I really would rather make a copy of it rather than bleep it out.\nSpeaker H: Are you bleeping it by adding?\nSpeaker F: Overlapping.\nSpeaker F: So it's exactly a sensor bleep.\nSpeaker F: So what I really think is bleep.\nSpeaker H: Is it summing signals or do you delete the old one and put the new one?\nSpeaker H: I delete the old one and put the new one.\nSpeaker F: There's nothing left of the original.\nSpeaker H: Because if you're summing, it could, anyway.\nSpeaker F: Yeah, it's easy to get it back again.\nSpeaker C: And then I was going to say also they, they don't have to say on the system as you know, because once it's been successfully bleeped.\nSpeaker C: We'll tell people to free.\nSpeaker C: You can hide it, yeah.\nSpeaker C: Can't you rely on the art heart?\nSpeaker E: We got that.\nSpeaker E: Yeah, that's true.\nSpeaker E: To hide it.\nSpeaker E: You can crypt it.\nSpeaker H: If you sponsor it with it, too hide it.\nSpeaker H: Here we go.\nSpeaker B: There you go.\nSpeaker B: There you go.\nSpeaker B: Like people say what they said.\nSpeaker E: And you have like subliminal.\nSpeaker F: You've seen the speech recognition system that reversed very short segments.\nSpeaker F: Did you read that paper?\nSpeaker F: No.\nSpeaker F: It wouldn't work.\nSpeaker F: The speech recognizer still works.\nSpeaker F: Yeah.\nSpeaker D: And if you do it backwards.\nSpeaker G: Yeah.\nSpeaker G: That's because they use forward backwards.\nSpeaker G: Forward backwards.\nSpeaker F: Forward backwards.\nSpeaker F: That's right.\nSpeaker F: Good point.\nSpeaker F: A point.\nSpeaker F: Well, I'm sorry if I sound a little peeved about this whole thing.\nSpeaker F: It's just we've had meeting after meeting after meeting on this.\nSpeaker F: And it seems like we've never gotten it resolved.\nSpeaker H: Well, we've never also, we've also never done it.\nSpeaker H: It's the first time we've been releasing widgets the first time through.\nSpeaker H: And I'm sorry responding without having much knowledge.\nSpeaker H: But the thing is I am like one of these people who gets the gazillion males.\nSpeaker H: Well, that's exactly why I did it the way I did it.\nSpeaker F: Which is the default is if you do nothing, we're going to release it.\nSpeaker F: Because I have my stack of emails to be done.\nSpeaker F: That 50 or 60 long.\nSpeaker F: And the ones at the top, I'm never going to get to.\nSpeaker F: Right.\nSpeaker F: So the only thing we're missing is wait and then you respond easily to say,\nSpeaker H: okay, you're going to have a mail-in deer.\nSpeaker F: Yeah, that's actually definitely a good point. The email doesn't specify that you can just reply to the email as opposed to going to the forum.\nSpeaker C: And it also doesn't give a specific, I didn't think of it.\nSpeaker C: So I think it's a good idea.\nSpeaker C: And explicit time by which this will be considered.\nSpeaker C: Yeah, release.\nSpeaker C: And it has to be time earlier than that endpoint.\nSpeaker C: Yeah.\nSpeaker H: It's converging.\nSpeaker B: I've seen this recently.\nSpeaker B: I got the email.\nSpeaker B: And if I use a mind-capable mail reader, it actually says, you know, click on this button to confirm receipt of the mail.\nSpeaker B: Oh, that's interesting.\nSpeaker F: You can, a lot of mailers support return receipt.\nSpeaker B: Right.\nSpeaker F: But it doesn't confirm that they've read it.\nSpeaker B: No, no, this is different.\nSpeaker B: This is not, so I know you can tell, you know, the mail delivery agent to confirm that the mail was delivered to your mailbox.\nSpeaker B: But no, this was different.\nSpeaker B: In the mail, there was a...\nSpeaker B: Oh, just a button.\nSpeaker B: There was a button that when you clicked on it, it would send, you know, actual acknowledgments to the center that you had actually looked at the mail.\nSpeaker B: Yeah, we could do that, but I hate that.\nSpeaker B: But it only works for, you know, mind-capable, you know, if you use a landscape or something.\nSpeaker H: It does respond to the mail.\nSpeaker H: And we actually need a third thing.\nSpeaker H: It's not that you've looked at it.\nSpeaker H: And you've looked at it and agree with one of the possible actions, right?\nSpeaker B: No, no, you can do that.\nSpeaker B: You can put this button anywhere you want.\nSpeaker B: Oh, I see.\nSpeaker B: And say here by clicking on this, I agree.\nSpeaker H: I acknowledge my first born children are yours.\nSpeaker H: Yeah.\nSpeaker H: Yeah.\nSpeaker D: Quick question.\nSpeaker F: Well, I could put a URL in there without any difficulty.\nSpeaker F: And even pretty simple mind readers can do that.\nSpeaker C: But why shouldn't they just email back?\nSpeaker C: Yeah, reply.\nSpeaker C: It's very nice. I like the high tech aspect.\nSpeaker B: I actually don't.\nSpeaker B: Appreciate it.\nSpeaker B: I'm not.\nSpeaker F: I use a text mail.\nSpeaker D: You use Vi?\nSpeaker H: Oh, that's my guy.\nSpeaker F: You use the VI?\nSpeaker B: Yeah.\nSpeaker B: There's these logos that you can put at the bottom of your webpage, like powered by Vi.\nSpeaker B: Yeah.\nSpeaker B: Wow.\nSpeaker F: You can put web bugs in the email.\nSpeaker D: There were three meetings this time or six, but of different people.\nSpeaker D: So I guess if you're in both these types of meetings, you'd have a lot.\nSpeaker D: But how, I mean, it also depends on how many, like, if we release this time, it's a fairly small number of meetings.\nSpeaker D: But what if we release like 25 meetings to people?\nSpeaker F: Well, what my expectation is is that we'll send out one of these emails every time a meeting has been checked and is ready.\nSpeaker F: Oh, okay.\nSpeaker F: So that was my intention.\nSpeaker F: It's just, yeah, that we just happen to have a bunch all at once.\nSpeaker F: That's a good idea.\nSpeaker F: I mean, maybe is that the way it's going to be, Jane?\nSpeaker C: We could do it. I'd be happy with either way.\nSpeaker C: Batch wise, what I was thinking, so this one, that was exactly right, that we had a...\nSpeaker C: I had wanted to get the entire set of 12 hours ready.\nSpeaker C: Don't have it.\nSpeaker C: But this was the biggest clump I could do by a time where I thought it was reasonable.\nSpeaker C: People would be able to check it and still have it ready by then.\nSpeaker C: I was thinking that with the NSA meetings, I liked there are three of them.\nSpeaker C: And there...\nSpeaker C: I will have them done by Monday.\nSpeaker C: Unfortunately, the time is later and I don't know how that's going to work out.\nSpeaker C: But I thought it'd be good to have that release as a clump too, because then, you know, they have a...\nSpeaker C: It's in a category.\nSpeaker C: It's not quite so distracting to them is what I was thinking and so on and once.\nSpeaker C: But after that, when we're caught up a bit on this process, then I could imagine sending them out periodically as they become available.\nSpeaker C: I could do it either way.\nSpeaker C: I mean, it's a question of how distracting these to the people who have to do the checking.\nSpeaker H: You hear anything?\nSpeaker G: Maybe I'm...\nSpeaker G: Let's see.\nSpeaker G: Right.\nSpeaker G: So we got the transcript back from that one meeting.\nSpeaker G: Everything seemed fine.\nSpeaker G: Adam had a script that will put everything back together and there was one small problem, but it was a simple thing to fix.\nSpeaker G: And then, I sent him pointer to three more.\nSpeaker G: And so he's off and working...\nSpeaker F: Now we haven't actually had anyone go through that meeting to see whether the transcript is correct and to see how much was missed and all that sort of stuff.\nSpeaker F: So at some point, you need to do that.\nSpeaker C: Oh, that's on my list.\nSpeaker G: Yeah, it's going to have to go through our regular...\nSpeaker F: I mean, the one thing I noticed is it did miss a lot of back channels.\nSpeaker F: There are a fair number of yes and a-huzz that it's just that aren't in there.\nSpeaker H: But I think, yeah, you see, that's going to be our standard process.\nSpeaker H: That's what the transcript was doing.\nSpeaker H: I would imagine once.\nSpeaker H: Yes, absolutely.\nSpeaker C: One question about the back channels.\nSpeaker C: Do you suppose that was because they weren't caught by the pre-segmenter?\nSpeaker C: Yes, absolutely.\nSpeaker F: Oh, interesting.\nSpeaker F: Yeah, they're not in the segment.\nSpeaker F: It's not that the IBM people didn't do it.\nSpeaker F: Okay.\nSpeaker C: Just they didn't get marked.\nSpeaker C: Okay, so maybe when the detector for that, it's better or something.\nSpeaker C: There's another issue, which is this...\nSpeaker C: We've been contacted by University of Washington now, of course, to...\nSpeaker C: We sent them the transcripts that correspond to those six meetings.\nSpeaker C: And they're downloading the audio files.\nSpeaker C: So I'll be doing that.\nSpeaker C: Check...\nSpeaker C: Check that...\nSpeaker C: Put that in.\nSpeaker G: I pointed them to the set that Andreas put on the web so that they want to compare directly with his results they can.\nSpeaker G: And then once...\nSpeaker G: We can also point them at the original meetings.\nSpeaker G: They can grab those too with...\nSpeaker D: Do you put the reference files?\nSpeaker G: No, no.\nSpeaker G: They wanted the audio.\nSpeaker G: Jane sent them the transcripts.\nSpeaker D: I mean, I have the transcripts.\nSpeaker D: Well, we can talk about offline.\nSpeaker F: There's another meeting in here, what, at four.\nSpeaker F: Right?\nSpeaker F: Yes, so we have to finish by 345.\nSpeaker B: So the...\nSpeaker B: Does UW want to use the state of for recognition or something else?\nSpeaker G: For recognition.\nSpeaker D: I think they're doing what...\nSpeaker D: Didn't they want to do language modeling on, you know, recognition compatible transcripts?\nSpeaker C: This is to show you some of the things that turn up during the checking procedure.\nSpeaker C: So this is from one of the NSA meetings.\nSpeaker C: And if you're familiar with the DIF format, the arrow to the left is what it was.\nSpeaker C: And the arrow to the right is what it was changed to.\nSpeaker C: So, now the first one.\nSpeaker C: Okay, so then we start at weekly meeting.\nSpeaker C: The last time on the transcriber thought little too much, but really it was we learned too much, which makes more sense syntactically as well.\nSpeaker C: And these departancies were...\nSpeaker C: Oh, that's the convention for indicating uncertain.\nSpeaker C: So the transcriber was right.\nSpeaker C: You know, she was uncertain about that.\nSpeaker C: Okay.\nSpeaker C: She's right to be uncertain.\nSpeaker C: And it's also a good indication of the next one.\nSpeaker C: This was about Claudia, and she'd been really busy with stuff such as waivers.\nSpeaker C: Okay.\nSpeaker C: Next one.\nSpeaker C: This was an interesting one.\nSpeaker C: So the original was...\nSpeaker C: So Claudia is not the bad master here.\nSpeaker C: And then he'll ask, but it's really a lot of master.\nSpeaker C: And he's another type of uncertainty, which is, you know, they just didn't know what to make out of that.\nSpeaker C: So instead of split upon unknown, it's split in principle.\nSpeaker E: So I'm saying these are from IBM, the top line.\nSpeaker C: No, no, these are local transcriptions of the NSA meetings.\nSpeaker C: Transcribers version versus the checked version.\nSpeaker C: Oh, I see.\nSpeaker C: My checked version, which I go through it.\nSpeaker C: Then you get down here.\nSpeaker C: Sometimes some speakers will insert foreign language terms.\nSpeaker C: That's the next example, the next one.\nSpeaker C: The version beyond this is so instead of saying, or especially those words, also in older and some other ones, those sneak in.\nSpeaker C: The next one.\nSpeaker C: This is called...\nSpeaker C: This is called...\nSpeaker C: Sorry, what?\nSpeaker B: Discourse on...\nSpeaker B: This is called...\nSpeaker C: And it makes sense because it's like below this...\nSpeaker C: It's a little subliminal there.\nSpeaker C: Okay, next one.\nSpeaker C: This is a term, a problem with terminology.\nSpeaker C: Description with...\nSpeaker C: The transcriber is X as an advance, but really it's QS in advance.\nSpeaker C: I mean, I've benefited from some of these cross-group meetings.\nSpeaker C: Okay, then you got...\nSpeaker C: Instead of from something or other cards, it's for a multi-cast.\nSpeaker C: Instead of...\nSpeaker C: And system related, it's in system related.\nSpeaker C: This was changed to an acronym initially and it shouldn't have been.\nSpeaker C: And then you can see here, GPS was misinterpreted.\nSpeaker C: It's just totally understanding.\nSpeaker C: This is a lot of jargon.\nSpeaker C: And the final one, the transcriber in the core network itself, or the exit unknown, not the internet unknown, and it comes through as in the core network itself of the access provider, not the internet backbone core.\nSpeaker C: This is a lot of terminology.\nSpeaker C: And they're generally extremely good.\nSpeaker C: But, you know, in this area, it really does pay to double check.\nSpeaker C: And I'm hoping that when the checked versions are run through the recognizer that you'll see substantial improvements in performance, because there are a lot of these in there.\nSpeaker F: Yeah, but I bet they're acoustically challenging parts anyway, though.\nSpeaker F: I actually know.\nSpeaker C: Oh, really?\nSpeaker F: Oh, so it's just jargon.\nSpeaker C: It's jargon.\nSpeaker C: Yeah, I mean, this is because, you know, you don't realize in daily life how much you have top-down influences in what you're hearing.\nSpeaker C: And it's jargon coupled with a foreign accent.\nSpeaker B: But we don't, I mean, our language model right now doesn't know about these words anyhow.\nSpeaker B: So, you know, until you actually get a decent language model, add up a certain one to your problem.\nSpeaker B: It's a different one.\nSpeaker B: But it's definitely good that these are fixed, I mean, obviously.\nSpeaker C: Well, also from the standpoint of getting people's approval, because if someone sees a page full of barely desirable sentences, and then it's asked to prove it or not, it's not.\nSpeaker C: Yeah.\nSpeaker H: Yeah, that would be the same with people.\nSpeaker H: So, well, I don't approve it because it's not what I said.\nSpeaker F: Well, that's exactly right.\nSpeaker F: The extra option in.\nSpeaker F: It's that I was afraid people would say, let's censor that because it's wrong, and I don't want them to do that.\nSpeaker C: Yeah.\nSpeaker C: And then I also, the final thing I have for transcription is that I made a purchase of some other headphones because of the problem of low gain and the originals.\nSpeaker C: And they very much prefer the new ones.\nSpeaker C: Actually, I think that there will be fewer things to correct because of the choice.\nSpeaker C: We originally chosen very expensive headsets, but they're just not as good as these with respect to this particular test.\nSpeaker F: It's probably impeding, smatching problems.\nSpeaker C: I don't know, but we chose them because that's what's been used here by prominent projects in transcription.\nSpeaker C: So, every reason they think they would work.\nSpeaker C: Yes, spare headsets.\nSpeaker F: They're just earphones.\nSpeaker F: They're not headsets.\nSpeaker F: They're not microphones.\nSpeaker B: No, no, I mean earphones.\nSpeaker B: Because I could use one of my workstations just to do something I have to listen to audio files and I don't have to go borrow it from someone.\nSpeaker C: Actually, I have...\nSpeaker C: The thing is that if we have four people come to work for a day, I was hanging on to the others for spare spare.\nSpeaker C: No, but if you knew it, just get it.\nSpeaker H: Yeah, we should get it.\nSpeaker H: Yeah, I need it.\nSpeaker H: It's just something to get a pair too.\nSpeaker E: They're pretty inexpensive.\nSpeaker E: I'm using one of these.\nSpeaker E: Two or three or four X-ray.\nSpeaker B: I think I haven't had that I brought from home, but it's just from you.\nSpeaker B: Just buy them.\nSpeaker H: Where do you buy these from?\nSpeaker C: Cambridge Soundworks.\nSpeaker C: Just down the street.\nSpeaker C: You just go and...\nSpeaker C: Oh, yeah.\nSpeaker C: They want to have them stock.\nSpeaker F: Good idea.\nSpeaker F: Do you email out the brand?\nSpeaker F: Oh, sure.\nSpeaker F: Yeah.\nSpeaker F: It sounds like people are interested.\nSpeaker C: Yeah, it's made a difference in...\nSpeaker H: I realize something I should talk about.\nSpeaker H: So, what's the other thing on the agenda?\nSpeaker F: The only one was Don wanted to talk about disc space yet again.\nSpeaker E: Yeah.\nSpeaker E: Yeah, it's short.\nSpeaker E: I mean, if you want to go, we just want to go.\nSpeaker E: No, no, no, no, no, no.\nSpeaker E: It's short.\nSpeaker E: Well, you met the disc space.\nSpeaker F: No, we know the disc space.\nSpeaker D: That's a great ambiguity.\nSpeaker D: Yeah.\nSpeaker D: It's one of the...\nSpeaker D: It's social and it's level.\nSpeaker D: Yeah.\nSpeaker D: It's great.\nSpeaker D: Yeah.\nSpeaker D: It was really cool.\nSpeaker D: See if I had that little crack head, I would have made an X there.\nSpeaker F: Well, we'll give you one, man.\nSpeaker E: Yeah.\nSpeaker E: So, without thinking about it, when I offered up my hard drive last week, oh no.\nSpeaker E: That's not a big surprise.\nSpeaker E: No, I realized that we're going to be doing a lot of experiments for this paper we're writing.\nSpeaker E: So, we're probably going to need a lot more.\nSpeaker E: We're probably going to need that disc space that we had on the 18 gig hard drive.\nSpeaker E: But we also have someone else coming in that's going to help us out with some stuff.\nSpeaker H: We just ordered a hundred gigabytes.\nSpeaker D: Okay.\nSpeaker D: We just need to...\nSpeaker D: We just need to...\nSpeaker H: We're getting 336s.\nSpeaker H: Okay.\nSpeaker H: That are going into the main file.\nSpeaker E: Mark, well, you're so...\nSpeaker E: Yeah, I mean, I guess the thing is, is all I need is to hang it off like the person who's coming in, so now all these computer...\nSpeaker B: You mean the...\nSpeaker B: The disc...\nSpeaker B: Or we can move them.\nSpeaker B: Or machines that we just got?\nSpeaker B: No.\nSpeaker B: Or...\nSpeaker E: These are going to go on to discs.\nSpeaker E: So, are we going to move the stuff off of my hard drive onto that when those come in?\nSpeaker E: Let's take a minute.\nSpeaker G: Sure.\nSpeaker G: Okay.\nSpeaker G: Do you find it one of those plan for?\nSpeaker G: They should be...\nSpeaker G: I imagine next week or something.\nSpeaker G: Okay.\nSpeaker F: So, for desperate, I have some space on my drive, but I...\nSpeaker F: I think if I...\nSpeaker E: I vacillate between no space free and a few gigs.\nSpeaker E: Yeah, I think I can find something if I'm desperate.\nSpeaker E: And in the meantime, I'll just hold out.\nSpeaker E: That was the only thing I wanted to bring out.\nSpeaker E: It should be soon.\nSpeaker H: Okay.\nSpeaker H: So, there's another hundred gigs.\nSpeaker E: All right. Great.\nSpeaker H: That's it.\nSpeaker H: It's great to be able to do that.\nSpeaker H: I just say, oh yeah.\nSpeaker F: Yeah.\nSpeaker D: A hundred gig year, 100 gig there.\nSpeaker D: It's eventually.\nSpeaker D: Yeah.\nSpeaker F: Real disc space.\nSpeaker F: Yeah.\nNone: Yeah.\nSpeaker H: Yeah.\nSpeaker H: I was just going to comment that I'm going to be on the phone with Mark.\nSpeaker H: I'm going to be on the phone with Marie tomorrow.\nSpeaker H: Good afternoon.\nSpeaker H: We're supposed to get together and talk about where we are on things.\nSpeaker H: There's this meeting coming up.\nSpeaker H: And there's also an annual report.\nSpeaker H: And I never actually...\nSpeaker H: I was asking about this.\nSpeaker H: I don't really quite understand this.\nSpeaker H: She was referring to it as...\nSpeaker H: I think this actually didn't just come from her, but this is what the ARPA had asked for.\nSpeaker H: She was referring to it as an annual report for the fiscal year.\nSpeaker H: Of course, the fiscal year starts in October.\nSpeaker H: So I don't quite understand how we do an annual report that...\nSpeaker H: She's either related really early.\nSpeaker F: Or she's getting good early start.\nSpeaker H: I think basically it's none of those meetings in July.\nSpeaker H: So the so-dabr to said do an annual report.\nSpeaker H: So anyway, I'll be putting together stuff.\nSpeaker H: I'll do it as much as I can without bothering people just by looking at papers and status reports.\nSpeaker H: I mean, the status reports you do are very helpful.\nSpeaker H: So I can grab stuff there.\nSpeaker H: If I have some questions.\nSpeaker H: Yeah, if people could do it as soon as you can if you haven't done one.\nSpeaker H: Recently.\nSpeaker H: But, you know, I'm sure before it's all done on debugging people for more clarification about stuff.\nSpeaker H: But I don't know.\nSpeaker H: I guess I know pretty much what people have been doing.\nSpeaker H: We have these meetings and there's the status reports.\nSpeaker H: But, yeah, so that wasn't a long one just to tell you that.\nSpeaker H: And if something hasn't...\nSpeaker H: I'll be talking to you later tomorrow afternoon.\nSpeaker H: If something hasn't been in the status report and you think it's an important thing to mention.\nSpeaker H: This kind of thing just popped me online.\nSpeaker H: And I'll have it in front of me for the phone conversation.\nSpeaker H: I guess you're still packing away at the demos and all that.\nSpeaker F: And Donna's gonna be helping out with that.\nSpeaker F: Do you want to talk about that this afternoon?\nSpeaker E: I'm here, but later today.\nSpeaker E: We should probably talk offline about when we're gonna talk offline.\nSpeaker H: Okay.\nSpeaker H: Yeah, I might want to get updated about it in about a week.\nSpeaker H: Because I'm actually gonna have a few days off the following week after the picnic.\nSpeaker H: Okay.\nSpeaker H: So I had...\nSpeaker F: So we were gonna do sort of status of speech transcription, automatic transcription, but we're kind of running late.\nSpeaker D: How long does it take you to save the data?\nSpeaker F: 15 minutes.\nSpeaker F: Yeah.\nSpeaker F: So, if you want to do a quick 10 minute.\nSpeaker D: It's really should stop like 20 oh, that the latest.\nSpeaker D: We have never a meeting coming in that they want to record.\nSpeaker H: And there's the digits to do.\nSpeaker H: Well, we can skip the digits.\nSpeaker H: Maybe we could five minute reporters.\nSpeaker D: Up to you.\nSpeaker D: Whatever you want.\nSpeaker H: Yeah.\nSpeaker F: Oh, we'd love to hear about this message.\nSpeaker H: I'm interested.\nSpeaker H: Yeah, well, I'm gonna be on the phone tomorrow.\nSpeaker H: So this is just a good example of sort of thing I'd like to hear about.\nSpeaker H: Wait, why is everybody looking at me?\nSpeaker H: Sorry.\nSpeaker H: Because he looked at you and she was just catching.\nSpeaker B: I'm not sure what you're doing.\nSpeaker B: Are we supposed to have done something?\nSpeaker F: We're just talking before about alternating the subject of the meeting.\nSpeaker F: And this week we were going to try to do automatic transcription status.\nSpeaker F: Oh, we sort of failed.\nSpeaker F: Right?\nSpeaker F: No.\nSpeaker F: Did we?\nSpeaker F: Did we?\nSpeaker F: Okay.\nSpeaker F: Okay, so now we have the schedule.\nSpeaker F: So next week we'll do automatic transcription status plus anything that's real timely.\nSpeaker H: Okay.\nSpeaker H: Okay, shoot update.\nSpeaker H: I was going to stop that bullet.\nSpeaker H: Nicely done, there's.\nSpeaker C: I'll move on a few words.\nSpeaker H: But lots of pros.\nSpeaker H: Okay.\nSpeaker B: We really haven't done anything.\nSpeaker B: Excuse me?\nSpeaker B: Sorry.\nSpeaker B: The next thing on our agenda is to go back and look at the automatic alignments because I got some.\nSpeaker B: I learned from Tilo what data we can use as a benchmark to see how well we're doing automatic alignments.\nSpeaker B: So after that, I have the foreground speech with background speech.\nSpeaker B: Yeah.\nSpeaker D: And then I guess the new data that Dunn will start to process that when you can get these, you know, before we were working with these segments that were all synchronized and that caused a lot of problems.\nSpeaker D: Oh, right.\nSpeaker D: So sort of a stage two of trying the same kinds of alignments with the tighter boundaries.\nSpeaker D: Right.\nSpeaker D: Really the next step.\nSpeaker D: We did get our, I guess, good news.\nSpeaker D: We got our abstract accepted for this conference workshop, isco workshop in New Jersey.\nSpeaker D: But we're hoping to have a paper for that as well, which should be an interesting.\nSpeaker D: What's it to?\nSpeaker D: The tapers and do until August, the abstracts were already due, so it's that kind of workshop.\nSpeaker D: But I mean, the good news is that that will have sort of the European experts in pros.\nSpeaker D: It's sort of a different crowd.\nSpeaker D: I think we're the only people working on pros in meeting so far, so that should be interesting.\nSpeaker D: What's the name of the meeting?\nSpeaker D: It's, it's go workshop on pros in speech recognition and understanding or something like that, some generic.\nSpeaker D: So it's focused on using pros in automatic systems and there's a web page.\nSpeaker H: You're going to your speech?\nSpeaker F: I don't have a paper, but I kind of like to go if I could.\nSpeaker F: Is that all right?\nSpeaker F: We'll discuss it.\nSpeaker G: Okay.\nSpeaker G: My car needs a good wash.\nSpeaker H: Hey, that's what it takes.\nSpeaker I: I'll pick up your dry cleaning tape.\nSpeaker F: All right.\nSpeaker F: 890-14495-1-4147-9206-888-893-407-0537-326-587-997-9670-3944-33-267-05671-265-06-647161-20229401-correction-0544.\nSpeaker B: Okay, I'm going to exit because I have to be there.\nSpeaker B: Go for it.\nSpeaker B: Transcript L-184-686-686-8-2261-213295-30817519-78111410-9627372719-6402-0860-4473.\nSpeaker B: 733-448075-25404-9726537-0373-614260.\nSpeaker E: Transcript L-185-804-1609371-377747.\nSpeaker E: 1-3-7-4-9-509 9-7-6-0-19-7-9-7-5 8-9-7-9-9-3-0-09 2-7-4-7-2-1-9-1-4-1 9-1-7-1-6-286 1-583-19-364-4\nSpeaker A: Transcript L-187 9-369-6-0-7-5-0-0 8-9-5-3-6-997-8-3-90 0-784-29-0-555-1 6-8-7-9-4-246 1049-7-9-8-6 317-47-87-36 1-8-80-76-43-84 9-271-6-446-6-128\nSpeaker D: Transcript L-188-2-30-383-317 8-8-8-8-092-5-25-066-459-1 00421-7-336-915-03-384-6 7-8-95-34-57-30 574-237-435 5923-46-04-53\nSpeaker G: Transcript L-180 807-39-3837 1988-4996 1916-0229-2294 07-436-727 747-327-7-914 765-07-1443 890-858-285 413-425-9572\nSpeaker H: Transcript L-181-116123-8588 6624-99997-204 4963-57-2507 6470-63-1107 908-880-959 533-711-485-4 619-07-180-2442 2238-6338-8114\nSpeaker C: Transcript L-182-0066-3352 442-116-315 933-05-728-0 632-04-8202 7205-7205-544-853 1667-745-563-674-61-8159-8025-92-7316\nSpeaker H: Thank you. So you get to be the one who has all the paper wrestling.\n"}, {"meeting_id": "IS1009b", "summary": "The meeting started off with Marketing leading the discussion on product positioning of the remote, which was going to be an impulse purchase and certain requirements from management, such as the remote only controlling the television. Then, Industrial Designer shared the working design and main components of the remote. Also, they discussed functions that they would add to the remote such as a child lock and the appearance of the remote, such as having a comfortable shape and different colours. User Interface then raised some features that the remote ought to have such as having a timer. Finally, Project Manager stated their target group, which was people aged ten to forty, and they discussed how to cater to their needs, such as having large buttons for the elderly.", "dialogue": "None: I'm okay.\nSpeaker A: Okay, is everybody ready?\nSpeaker E: Yeah.\nSpeaker E: Just put on my microphone here and I'll be right with you.\nNone: Okay.\nSpeaker D: I take it you all have received instructions as to what you were supposed to do.\nSpeaker D: And I think the marketing manager probably should go first addressing the needs and desires.\nSpeaker E: Okay, you want me to start right now?\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: Could you put my slides up because I think it might be helpful if we looked at the slides at the same time.\nNone: Okay.\nSpeaker E: You're a participant.\nSpeaker E: I'm a participant for I believe yes.\nSpeaker D: Okay, and now I can full screen open.\nNone: Okay.\nSpeaker E: There we go.\nSpeaker E: Okay, well, I think we have introduced ourselves.\nSpeaker E: So the functional requirements is part of my goal.\nSpeaker E: But why don't we pass it right to the second slide?\nSpeaker E: Because that's where my discussion starts.\nSpeaker E: All right.\nSpeaker E: Well, since I'm in charge of trying to figure out what we should put on this thing since I have to try to sell it.\nSpeaker E: I thought that the method I should follow would be gather suggestions from everybody.\nSpeaker E: And the reason I just put that there like that is that in the initial stage, I think I should just be open to lots of suggestions.\nSpeaker E: You know, you can say anything you want, no matter how silly it sounds.\nSpeaker E: You know, it should run your car.\nSpeaker E: It should heat up your motor.\nSpeaker E: It should turn on your seat either whatever you want it to do.\nSpeaker E: And then as we go on, what we'll have to do is accept and eliminate these suggestions according to design and budget feasibility.\nSpeaker E: So I'll be coming to you frequently as the industrial expert to tell me how hard it's going to be to add a feature or how expensive it's going to be.\nSpeaker E: Or if your time takes five years to develop this, it's just something we can't do.\nSpeaker E: So in the beginning, just have a big puddle of things that we, anybody can throw anything in and then just weed things out that can't be done for one reason or another.\nSpeaker E: And then the things that seem the most attractive to a customer will try to then prioritize those.\nSpeaker E: So that was what I meant there.\nSpeaker E: And as I said on the slide, they're consulting the industrial engineer about that.\nSpeaker E: And the other thing is time is really going to be as important as money because if we're going to sell this thing, I think the best time to sell this is Christmas present.\nSpeaker E: Twenty five euros makes a nice little present and we want it to be an impulse purchase.\nSpeaker E: We want somebody to see it and think it's, I just got to have that and take out their wallet and buy it.\nSpeaker E: So it's got to be really attractive.\nSpeaker E: But it's got to go to market by September because anything that you don't have already out there in September showing it around isn't going to sell for Christmas.\nSpeaker E: And then I'll be coming to you as the user interface person to try to tell me from your point of view what are the most friendly features that we could put on it and try to help me with that prioritizing of the features and of the look and the color.\nSpeaker E: And I'll be coming back to you to help weed out those suggestions from that point of view. So we're coming to you for how much is it going to cost and how long is it going to take you?\nSpeaker E: And I'll be coming to you to tell me what's going to make somebody take out their wallet.\nSpeaker E: You know, what's going to really be what they call a sizzle because we've got to sell the sizzle.\nSpeaker E: A lot of times the thing that works the best from an engineering point of view isn't the thing that somebody's really just going to take out their wallet and buy for Christmas for their child or for their husband or whatever.\nSpeaker E: Okay, can we go to the next slide please?\nSpeaker E: Alright, I already did a little bit of research after our first meeting where we threw out some ideas.\nSpeaker E: And it looks to me that within the budget that we're looking at the whole house idea really isn't going to be possible.\nSpeaker E: So I'd like your suggestions to come back to the other slide where I was saying we could suggest anything.\nSpeaker E: I'd like the suggestions to be really specific so that we'll have a list of things we can cross off, not something like, you know, whole house control.\nSpeaker E: And then I found on the internet from my research that some extended electronic entertainment control should be possible.\nSpeaker E: At the budget that we're looking at, and at the price point we're looking at, we should be able to make it work the TV, the VCR, the stereo set, maybe something else cute like a coffee pot or one other appliance or maybe a lamp.\nSpeaker D: Okay, can I at this point interject something?\nSpeaker D: We have received instruction from higher up that certain things should not be considered.\nSpeaker D: The one thing for example, something to eliminate maybe that's the teletext because that's sort of outdated with the internet.\nSpeaker D: And according to higher management, it should only control the TV mostly because they feel that it's too complex a task to include other things and they're concerned with the time to market.\nSpeaker D: And the third thing that they want to make sure that the corporate image is being maintained and that the corporate color and design are being used on the product so that it's easy, that they can be easily identified as a product of the company and that there's no mistake that it could be somebody else who is bringing this out.\nSpeaker D: So I just wanted to interject this here so we're not getting too much of track here with the things we want to look at.\nSpeaker D: These were instructions from higher up so we have to eliminate these things.\nSpeaker D: It's only going to be TV but the one thing maybe that could be eliminated is the teletext idea.\nSpeaker D: Okay, okay.\nSpeaker D: All right, thanks for that.\nSpeaker E: All right, now other things that I found out in my research is that the complaints that people have about the remote controls that are out there now, there's a lot of them take too much time to learn how to use and that was 34% and even more important.\nSpeaker E: The thing that we did address in our last meeting that frequently it's lost somewhere in the room.\nSpeaker E: So those are two things that we definitely do want to address.\nSpeaker E: Now we want to make it as simple as possible. We want to make it obvious and intuitive to use.\nSpeaker E: And then the things about finding it.\nSpeaker E: We talked about the light emitting thing as well as maybe a beep.\nSpeaker E: And I think that those are things after my research that we definitely want to try to incorporate.\nSpeaker E: Okay, can we go to the next slide please?\nSpeaker E: Okay, so my personal preference is in this project.\nSpeaker E: We really have to concentrate on the sizzle.\nSpeaker E: That is the selling point, the thing that's going to make it an impulse purchase.\nSpeaker E: Because once there's no be back in sales they always say, you know, be back, don't come back.\nSpeaker E: If somebody says, oh, I'll come back and get it next week, you're dead.\nSpeaker E: They're never going to come back and buy it.\nSpeaker E: You've got to make it attractive enough so they buy it now, now, now as a big word in my book for selling this thing.\nSpeaker E: And in order to make it really saleable, we've got to shorten the learning curve, make sure it's really intuitive and easy to use.\nSpeaker E: We have to have as few buttons as possible because more buttons is more confusion.\nSpeaker E: So that's why I'm saying simplicity is good.\nSpeaker E: Finding it's important, obviously you can't use it if you can't find it.\nSpeaker E: So we've got to concentrate on the features that help you find it.\nSpeaker E: And I've already said this several times but I put it down in writing here.\nSpeaker E: It should be an attractive impulse purchase at 25 euros.\nSpeaker E: So it has to have enough value that when somebody looks at it, this 25 euros, I'm not going to take that.\nSpeaker E: It has to be so great that they're going to say, oh, 25 euros isn't much.\nSpeaker E: And then maybe a motto, like we put fashion in electronics might be something we can use in our marketing campaign.\nSpeaker E: Okay, that's about it for me right now.\nSpeaker E: Okay.\nSpeaker D: And who would be next?\nSpeaker D: I guess that would be you.\nSpeaker D: You want me to get your slideshow up?\nSpeaker D: Thank you.\nSpeaker D: And you are number three?\nSpeaker D: Number two.\nSpeaker A: Can you make it a full screen, please?\nSpeaker A: No, that's like, yeah, you have to press here.\nNone: Cup shape.\nSpeaker A: Yeah, yeah.\nSpeaker B: So today I'm going to talk about the working design of the remote controller.\nSpeaker B: Okay.\nSpeaker B: Can you go to the next slide, please?\nSpeaker B: Mm-hmm.\nSpeaker B: The method is like in a remote controller, you have a chip integrated circuit, which is like a brain of the remote controller.\nSpeaker B: It takes the power from a battery, say a battery.\nSpeaker B: It can be an electric supply, like you have to switch connect your remote controller to power supply from the electricity or something like that.\nSpeaker B: It should be a battery, because remote controller should be like you should take it to wherever you want.\nSpeaker B: Mm-hmm.\nSpeaker B: And then this integrated circuit takes energy from the power source and whatever, like if you press a button, it's like an input for the remote controller.\nSpeaker B: And it takes the input and it transforms into infrared bits and it sends it to the device, like a TV or a conditioner, something like that.\nSpeaker B: So a remote controller is specifically designed to a single device.\nSpeaker B: If you want to design it for multiple devices, then you should make all the devices compatible with the frequency.\nSpeaker B: Like the remote controller, it sends some bits, some waves, like with a particular frequency.\nSpeaker B: The device should know what the frequency is.\nSpeaker B: It should recognize the waves, which are coming from the remote controller and it should take the action.\nSpeaker B: Like if you press a button channel or something like that, then the remote controller will send a signal.\nSpeaker B: And the TV, it should translate that into like change the channel or something like that, change the volume control.\nSpeaker B: And so I think it's hard to design a remote controller for multiple devices.\nSpeaker B: Yeah, well that's already been eliminated by management.\nSpeaker E: We're off the hook.\nSpeaker B: So user interface controls the chip and accordingly the messages.\nSpeaker B: Like there should be a user interface like a switchpad or something like that, buttons should be there.\nSpeaker B: So you can control whatever you want, you want to change the channel, you want to control the volume, you want to mute the TV or you want to have a child lock or you want to do some operations.\nSpeaker B: There should be some device to tell what to do to the integrated circuit so that integrated circuit can send the signals and TV can perform the actions.\nSpeaker B: So can you go to the next slide please?\nSpeaker B: I just would like to add some extra features to the remote controller.\nSpeaker B: I think these are the very simple features and they don't take much of the investment also.\nSpeaker B: It's like the text or buttons which are there on the remote controller.\nSpeaker B: Those we can make like florocent.\nSpeaker B: They'll be like light emitting if it is dark so that you can find your remote control if it is dark.\nSpeaker B: And there should be a beep if many buttons are pressed.\nSpeaker B: If suppose a child is playing with the remote controller and she or he is pressing the buttons all at the time then there should be a beep saying that this is not a action, there can be no action taken to that.\nSpeaker B: And there should be a child lock like you should be able to lock your remote controller so that whatever buttons are pressed by a child they can't be like if you have kids and all then they will be playing with remote controllers so you can lock the remote controller.\nSpeaker B: Make it usable for more than one device. It's hard but I think it's possible.\nSpeaker D: Yeah well that has been eliminated so that's unfortunately a mute point now.\nSpeaker B: And different shapes that we can do like we can have you know all animal shapes or you know comfortable which can fit into your hands.\nSpeaker E: That's good from a marketing point of view the fun shape.\nSpeaker E: You say that won't add too much to the budget to the shape.\nSpeaker B: I don't think it will be like you can have you know if you want to.\nSpeaker B: It just builds them all basically and you know it's just a second shape so it doesn't matter.\nSpeaker D: As a budget we're looking at if you build one mold I don't think that's going to make a big difference whether it's going to be square or.\nSpeaker E: Do you think there's any chance of having so having basically the same machine with the same buttons but maybe several different shapes.\nSpeaker E: Yeah that is all different shapes.\nSpeaker D: Is that going to be possible?\nSpeaker D: I think we'll have to look at the budget on that but I think in principle that that will be that will be kind of fun.\nSpeaker D: Because we had something sort of sexy for adults.\nSpeaker E: Yeah.\nSpeaker E: So we could have something sort of silly for children.\nSpeaker C: Like an animal.\nNone: An Or an animal.\nSpeaker E: Or an elephant so they can remember where it is.\nSpeaker B: And the button buttons also I think if you want to have more features in your remote controller then there should be more buttons.\nSpeaker B: If there are more buttons then it will be more complicated.\nSpeaker B: If you have less features then your remote controller won't be attractive.\nSpeaker B: So I think we need to make some buttons which are which are like in in it for two or three operations.\nSpeaker B: Like if you press one button in one mode then it will change the channel.\nSpeaker B: If you press the other button in another mode it will change the color.\nSpeaker B: So if you want to have less buttons we can have that option but I think it will complicate the matter more.\nSpeaker E: I think that that's something that we'll have to discuss with the user interface person.\nSpeaker E: Because I think there's a lot of arguments we made for one button for one feature.\nSpeaker E: I think one of the things people were complaining about when I found out in my research is when they complained about how hard it is to learn a new one.\nSpeaker E: The changing modes were something.\nSpeaker E: You and I all four of us work with computers all the time.\nSpeaker E: Changing modes is nothing for us.\nSpeaker D: But people who...\nSpeaker D: A little arthritic hand you know and it's a small button and they don't press it exactly.\nSpeaker D: Something else happens and not their favorite channel comes up with something else and they're very frustrated.\nSpeaker E: It's the kind of thing people learn by feel.\nSpeaker E: You don't feel the mode change.\nSpeaker E: So maybe having buttons be various shapes might be helped.\nSpeaker E: Yeah shapes also different.\nSpeaker E: You know like the triangle is for the volume and a square is for changing channels.\nSpeaker E: So that people can develop a tactile sense of it.\nSpeaker E: But we'll get to that with you.\nSpeaker B: Also text should be very clear so that there won't be any ambiguities.\nSpeaker B: That's right.\nSpeaker B: That's a good point.\nSpeaker B: And display clock if you want more features then we can display a clock.\nSpeaker B: I don't think it will take any extra money because we have a little bit of a circuit.\nSpeaker B: I think we can definitely fit that feature into the circuit.\nSpeaker E: Yeah that's a good one.\nSpeaker E: Because the clock could be really friendly.\nSpeaker E: Yes.\nSpeaker E: When is your favorite channel coming off?\nSpeaker D: You see there on the radio maybe you have no other...\nSpeaker E: Yeah a lot of time there's not a clock on the screen and you have to go somewhere but you just want to look at the news for a minute.\nSpeaker E: Yeah that's good.\nSpeaker E: The clock is good.\nSpeaker E: Okay.\nSpeaker B: Can you go to the next slide please.\nSpeaker B: The main components may need for buttons and underneath that there should be switch and bulbs.\nSpeaker B: We can have a bulb like whenever we are operating a remote controller or whenever the TV is on suppose.\nSpeaker B: Then the remote controller should automatically have a light.\nSpeaker B: Then it will be like if you switch on your TV through another source not from through a remote controller and you lost your remote controller.\nSpeaker B: Maybe you can find with the light.\nSpeaker B: Maybe we can have that.\nSpeaker B: Whenever the TV is on remote controller will have a light.\nSpeaker B: And then infrared bulbs.\nSpeaker B: This is like when it is dark.\nSpeaker B: And battery there should be a battery for power supply.\nSpeaker B: And a chip which is like a brain to the controller which does all the operations.\nSpeaker B: And why it's connecting all chips which is lights everything.\nSpeaker B: So there should be wires.\nSpeaker B: And of course there should be a case where we can keep all the things and you know different shapes or whatever it is.\nSpeaker B: There should be a case to give a shape.\nSpeaker A: To keep the remote.\nSpeaker A: Yeah.\nSpeaker C: Case.\nSpeaker C: Case holder.\nSpeaker C: A holder.\nSpeaker A: Yeah.\nSpeaker A: Next slide please.\nSpeaker B: I have referred the site the whole page of the website from where I have a few points.\nSpeaker B: That's it.\nSpeaker B: That's it from me now.\nSpeaker D: Okay.\nSpeaker E: So we should visit this site and have a look at what's up there.\nSpeaker E: Oh, you don't have to.\nSpeaker B: It's not like that.\nSpeaker B: I have referred the page to get new ideas or like what can be the working design to how it works and all the things.\nSpeaker E: Okay.\nSpeaker E: So it might be helpful if we had a look at that.\nSpeaker E: Yeah, if you want.\nSpeaker E: Okay.\nNone: Thank you.\nSpeaker E: Did you go back to that slide where she had that side up Betsy?\nSpeaker B: It's actually been on the screen.\nSpeaker D: This one?\nSpeaker D: Yeah.\nNone: Okay.\nSpeaker D: Thank you.\nSpeaker D: And Frances.\nSpeaker C: I'm just going to.\nSpeaker C: You are number two.\nSpeaker C: No, no, I'm three.\nSpeaker D: Would you want the tool for screen?\nSpeaker C: Yes.\nSpeaker C: Okay.\nSpeaker C: As a user interface designer, I did a little research to find out what are the features which a user would like to have on their remote.\nSpeaker C: Yeah, can you please go on to the next slide?\nSpeaker C: So I found out that the main purpose of a remote is to control the function of a television at a far off distance at remote distance.\nSpeaker C: Now for that a remote controller should have a switch on of button by which a user can sit anywhere in the room in front of the TV and can control the functionality, different functionalities of the TV.\nSpeaker C: There should be a signal, something like a radio wave or an infrared light or a lead which can be used to change the different functionalities in the television.\nSpeaker C: If the user wants to change the channels or increase the volume, he can change it.\nSpeaker C: Now there should be some timer to set for viewing a particular program or switching on and off a particular program according to the user choice.\nSpeaker D: On the remote.\nSpeaker D: Yes.\nSpeaker D: Yes.\nSpeaker D: Yes.\nSpeaker C: So if you want to view a particular program at say 9 o'clock, he can set the time and the TV will automatically switch on at that particular time.\nSpeaker C: So he can use that kind of properties of features and then there should be a child lock system.\nSpeaker C: If a particular channel is not to be viewed by a particular for a certain age, then the parents can lock that particular channel so that the children cannot change.\nSpeaker C: Children cannot view that channel.\nSpeaker C: And the one of the features a user would like to have is the compactness of the remote.\nSpeaker C: The remote should be compact and it should have as many buttons as possible for controlling different functionalities of the TV, television.\nSpeaker C: And as this is my personal preference that it should be in the shape of a T, an alphabet for more compactability.\nSpeaker C: And there is one more point which I noted it down.\nSpeaker C: Like the material which is used for remote should be human friendly.\nSpeaker C: It should not cause any skin disease or something.\nSpeaker C: Some energy by the children not to adult person.\nSpeaker C: And it should have an alarm clock.\nSpeaker C: A person if somebody wants to get a put around 8 p.m. and he can set the time and it can be used as an alarm clock.\nSpeaker C: And I don't think it will cost much to set an alarm clock inside a remote.\nSpeaker C: If there is a clock then there can be a test.\nSpeaker C: And as John we said the buttons can be, we can use the fluorescence to light up the buttons.\nSpeaker C: So different buttons will glow differently.\nSpeaker C: So even in the dark the user can know what buttons to use to set a particular channel.\nSpeaker C: And the design of the remote should be in such a way that there should not be sharp projections.\nSpeaker C: So that if a child plays with a remote he should not be harmed in any way.\nSpeaker C: And findings.\nSpeaker C: I found out in different sites that there are different remotes which can be used.\nSpeaker C: There are remotes which use the infrared for controlling the different functionalities.\nSpeaker C: There are remotes which use the radio waves to control the functions.\nSpeaker C: And there are, so you have different types of remotes with different light source.\nSpeaker C: Which can be used for controlling the different functionalities of a television.\nSpeaker C: Next thing.\nSpeaker C: And these are my few personal preferences.\nSpeaker C: Like it can be used for a multipurpose use.\nSpeaker C: Not, no.\nSpeaker C: Like it can, it can be used as a TV control as plus an alarm clock to set an alarm timing.\nSpeaker C: And it should have a child lock.\nSpeaker C: And then to save electricity there should not be much more, lots of circuits and all that.\nSpeaker C: And if a person, if parents wants the television to be switched off by ten o'clock, then it should be switched off at ten o'clock automatically.\nSpeaker C: So that nobody else comes later and use it.\nSpeaker C: Okay, it's like a timer here.\nSpeaker C: And then you can use a timer as well.\nSpeaker C: Yes.\nSpeaker A: Yeah, that's it.\nSpeaker A: Okay.\nSpeaker A: Now I have...\nSpeaker D: On my slideshow, basically we have already done the agenda.\nSpeaker D: And on the closing, basically there would be lunch break and all that.\nSpeaker D: However, the decision we have to take in this meeting is who is our target group.\nSpeaker D: And what function, working design, how does it work.\nSpeaker D: So these are the two things we should come up with.\nSpeaker D: So the first thing is maybe a little easier on who is our target group.\nSpeaker D: I guess in many ways everybody.\nSpeaker D: Everybody who has a TV.\nSpeaker C: TV television, yes.\nSpeaker D: And I think today there are probably not many people who don't have a TV.\nSpeaker D: They are a few, but in general not.\nSpeaker D: Now, talking about the target group, which is in a sense everybody, but I think within the target group we have subgroups.\nSpeaker D: We have earlier it was mentioned about, for example, elderly people who have limited function with their fingers in hands.\nSpeaker D: So I think that's one group that's certainly important.\nSpeaker D: Then we have just the nervous people who can never press a little button on anything unless it's really very clear.\nSpeaker D: And I don't know how much we want to cater to children's use.\nSpeaker D: I mean that's a question.\nSpeaker D: Whether that's important that children can really use it or not.\nSpeaker D: So these are just some thoughts I have on it.\nSpeaker D: And I don't know what you feel about whether we can just say we have one target group and for this one target group we're going to design this one thing.\nSpeaker D: Or whether we're looking at what we like, we're talking about different shapes.\nSpeaker D: Whether that different shape also includes maybe different buttons for different groups.\nSpeaker E: Well, you know there's the old motto, children under six never shop alone.\nSpeaker E: So if you design something that's very attractive to children, the mommy please mommy please.\nSpeaker E: You know we want it now, we want to go to the store and see it.\nSpeaker E: That has a lot of marketing pull.\nSpeaker D: That has a lot of appeal but I think I'm talking about the functionality now on it.\nSpeaker D: Whether we're looking at different groups.\nSpeaker D: Children's issue can be addressed with the shape and with colors.\nSpeaker D: That's right.\nSpeaker D: You know like you make it nice and pink, fluorescent, banana, color or whatever.\nSpeaker D: You can make a banana shaped one.\nSpeaker D: Yeah, for example, you know.\nSpeaker D: But the question is really who is our target group?\nSpeaker D: Do we look at one target group and with the 25 euros, you know, can we can we afford to have I'm asking the technical people here whether.\nSpeaker D: To look at the sort of subgroups with maybe different buttons for each group.\nSpeaker D: How much would that throw us out of the cost we are supposed to respect.\nSpeaker D: Respect.\nSpeaker C: Okay.\nSpeaker C: Generally we can we can design a remote which is mainly for people with age from 10 to 40.\nSpeaker C: Right.\nSpeaker C: And then we can add on specific functional buttons for children as well as the elderly people or the people with.\nSpeaker C: With.\nSpeaker C: With.\nSpeaker C: Who have nervous problems.\nSpeaker B: Maybe we can design different remotes for different people.\nSpeaker B: Like for aged people there will be big buttons and you know.\nSpeaker C: Well that's.\nSpeaker C: In a family there will be a aged person, children and a middle age.\nSpeaker C: But they cannot buy three different remotes.\nSpeaker C: They would like to buy just one and just one which can be used by all the three.\nSpeaker B: Or that we can do but specifically if like elderly people want a big buttons then you can't really make a big remote controller.\nSpeaker B: So maybe specifically you can design a big remote controller elderly and for children.\nSpeaker E: What about the electronics that's not really going to change much.\nSpeaker C: It doesn't cost.\nSpeaker E: No that will that won't change much will it.\nSpeaker D: The question to be addressed here is only who is a target group and how will it function.\nSpeaker D: And I think how will it function is probably the question of the buttons.\nSpeaker D: You know.\nSpeaker D: Within the target group or subgroups.\nSpeaker D: The question is only whether our budget will allow to have more than one design in a sense.\nSpeaker D: Yeah.\nSpeaker D: I mean the basic design I understand will be the same but the question is how much will that set us back if they say.\nSpeaker D: 40% we make large buttons and the rest we make regular buttons for example.\nSpeaker D: Yeah.\nSpeaker D: So that remains to be seen but target group as a whole is basically everybody with a TV.\nSpeaker D: That's right.\nSpeaker E: And can we can we is it even feasible to make one remote control with something for everyone.\nSpeaker E: Or we have to try to use the same electronics and make three different shapes.\nSpeaker E: Yeah well that's that's the question.\nSpeaker E: The same electronics and basically all that's going to be different is the plastic case.\nSpeaker E: And in that case we could probably molded plastic isn't all that expensive is it.\nSpeaker E: Here is an industrial person.\nSpeaker D: Well maybe there's there's an idea you know the new for example new portable phones.\nSpeaker D: They have like removable plastic cases.\nSpeaker D: Yeah.\nSpeaker D: So you can have a striped one or you can have a fluorescent blue one or whatever.\nSpeaker D: Maybe something like that that in each package you know you have.\nSpeaker D: There's a real idea.\nSpeaker D: You have you have for example you have let's say a fluorescent blue on it.\nSpeaker D: But then you have another color or stripe whatever that already comes in a package.\nSpeaker D: And then if people really want more colors they can buy it separately.\nSpeaker D: Right.\nSpeaker B: Like mobile phones we have different you know covers.\nSpeaker D: That's right and but like if then everybody is tired of the fluorescent blue and of the stripe to whatever they have as another one in the package then they can go to the store and for a few Frank's or euros whatever they can buy an alternate package.\nSpeaker D: An alternate package with another two colors in it you know for example.\nSpeaker C: But yeah I have a doubt like will it be cause twice effective if we if we design a remote having all the different features for different people or designing three different remotes for three different categories of people.\nSpeaker D: Well I think I think the idea here is to design one remote.\nSpeaker D: Okay.\nSpeaker D: And what the only change is going to be the functional cosmetics if you want to put it that way.\nSpeaker D: Having maybe larger buttons or buttons that light up or buttons that are slightly differently shaped for people.\nSpeaker D: Either for children.\nSpeaker D: Five minutes I guess that's the old message.\nNone: Morning.\nSpeaker D: Finish me now.\nSpeaker D: Well we may have to come back to our next meeting.\nSpeaker D: But I think to sum it up the target group is basically everybody.\nSpeaker D: If money permits we can address some features for some subgroups within that.\nSpeaker B: Okay.\nSpeaker D: Do we agree on that in principle like money will tell whether we will be able to do that or not.\nSpeaker D: Okay.\nSpeaker D: So now I understand it's lunch break.\nSpeaker E: Okay.\nSpeaker A: So thank you very much.\nSpeaker A: Thank you.\nSpeaker A: And we'll see you after lunch.\n"}, {"meeting_id": "ES2009b", "summary": "The meeting was about a second-stage idea of the new television remote. The meeting began with individual presentations in which the members shared their personal views towards the design, the target consumers and the functions and so on. Later they also paid attention to the marketing method like how to limit the cost within 12.5 euros and include the multi functions in only one microchip. What is also important, they allocated some time to the issue of putting fashion into the design, for example, using plastic material and carefully choosing the company colour and the slogan on the screen.", "dialogue": "Speaker C: I'll set.\nSpeaker C: Okay, cool.\nSpeaker C: Right, so basically I'm just going to go over real quickly some news I just got from the board on how we're supposed to do with this remote control, and then I'm going to turn it over to you guys to make brief presentations on what you found, and then we'll have a bit of discussion.\nSpeaker C: So basically, I've just found out from the board I don't know if you guys got this email as well, but it needs to be television only.\nSpeaker C: So we're not doing DVD, we're not doing anything else, it's just going to be a television remote.\nSpeaker C: It also needs to have the company colors included in it.\nSpeaker C: So that's red and black, and it has to have the slogan.\nSpeaker C: In case you guys forget the slogan, we put fashion in electronics.\nSpeaker C: And no teletext.\nSpeaker C: I'm not sure what teletext is, but I'm assuming you guys do, so we don't want to include that in this particular design for reasons that I don't really know.\nSpeaker C: But it's the board, so there you go.\nSpeaker C: So basically, given those guidelines, which we'll have some effect on how we design, we'll discuss it later, I mean, because it's television only, we'll be able to change our well, we can sacrifice more function for a better television remote.\nSpeaker C: Anyway, so I'm going to turn it over to the industrial designer to go ahead and make a\nSpeaker B: presentation. Okay.\nSpeaker B: So do I implode this bit here?\nNone: Yep.\nSpeaker C: Might have to hit function F8, but it looks like it's going to come up.\nSpeaker B: Okay.\nSpeaker B: Right.\nSpeaker B: That's page one of my presentation.\nSpeaker D: Very nice.\nSpeaker D: Pre-first PowerPoint is lovely.\nSpeaker B: So the method, we're going to have to understand how remote controls work and we're going to successfully complete this project.\nSpeaker B: Remote control works as follows.\nSpeaker B: It's all pretty basic stuff, you guys.\nSpeaker B: It sends a message to another system, so there's an energy source involved in that, like battery or solar power, something along those lines.\nSpeaker B: There's an integrated circuit, just the microchip, and that actually composed the messages, and usually the way a remote control works is it sends infrared bits to another system.\nSpeaker B: User interface controls the chip, basically that's the casing and the buttons, and accordingly, the messages as well.\nSpeaker B: So my findings, I just did a preliminary study here, and I found that too much metal in remote design can potentially cause interference with the ability of the remote to send commands.\nSpeaker B: Too much metal can cause remote to behave unexpectedly by receiving false signals.\nSpeaker B: Too much metal is used, sometimes people pick up radio signals and the like.\nSpeaker B: And there's also the possibility of the remote catching on fire and injuring the customer.\nSpeaker B: I just think of those lawsuits, that would be really bad.\nSpeaker B: Therefore, I suggest primarily plastic construction.\nSpeaker B: Components, just some ideas that I had, energy source.\nSpeaker B: It's kind of hip to be eco-friendly, so I thought maybe we could do something with solar power with an alkaline battery backup.\nSpeaker B: User interface, since we can't use metal, I was thinking maybe a high-grade recycled plastic.\nSpeaker B: The chip, silicon-based chip, I don't really see any way around that, we can't really be different in that respect.\nSpeaker B: The sender, while I'm thinking infrared, because it is the industry standard, multi-channel, that's the word I made up, I don't really know what it means.\nSpeaker B: Pal and NTSC compatible, and probably a 200-foot range.\nSpeaker B: And the receiver, of course, is a number of electronic devices, but in this case, only one of these.\nSpeaker B: Personal preferences, I really think that we should use plastic as opposed to metal.\nSpeaker B: The company simply can't afford this kind of lawsuits, which admittedly is going to come at the cost of a certain aesthetic value, because we were thinking.\nSpeaker C: Is there a way that we can use modern types of polymers or modern types of plastics that maybe do have some kind of aesthetic value?\nSpeaker C: Like if we talk about, like, on the laptop, and the other ones we use, that's pretty nice, and you can do, is there some kind of nice color there in quality of plastic that we can work with?\nSpeaker B: Yeah, that shouldn't be a problem.\nSpeaker B: For example, the plastic they have in their laptop there is something that's perfectly possible for us to do.\nSpeaker C: Great, thank you very much, Nathan.\nSpeaker C: If next we can have the user interface developer, go ahead and make a brief presentation, that would be great as well.\nSpeaker C: Just plug yourself in here.\nNone: Hit function if they're real quickly, hold down.\nSpeaker B: This is plugged in well.\nSpeaker B: There it goes, computer adjusting.\nSpeaker C: There you go, sweet.\nSpeaker A: Well, so here we have my technical function design presentation.\nSpeaker A: So a few of the requirements we need here, we basically need to offer an electronic device, it needs to be universal, and it possibly operates several different types of devices, although we now find that.\nSpeaker A: Sorry, I couldn't get that get to use before.\nSpeaker A: So some of my findings basically want to send messages to a television set.\nSpeaker A: That would be any number of different things, such as switch on the television, switch to the next channel, that sort of thing, I think we're all quite intelligent and no what a normal remote control does.\nSpeaker A: Now, some of the other things I found is a complicated remote control, I'm sorry that we can't quite see my red there very well, but this remote control has many functions.\nSpeaker A: So it can do a lot of things, but it is quite complicated and most users will find that they won't use most of the functions because they don't know how to use them and don't want to take the time to learn how to do it.\nSpeaker A: As you also notice, it's quite a boring design.\nSpeaker A: Another remote control, slightly different, it's a simple remote control, many less buttons, but has many fewer functions, much easier for the user to manipulate and use.\nSpeaker A: It also has a bit of a cheap look and is also quite boring.\nSpeaker A: So my personal perspective revolutionized the idea of a remote control.\nSpeaker A: So attaining the functionality of the complicated device but use a simple format of display for the user to work with.\nSpeaker A: I was going to add another slide here, but I didn't quite have time there.\nSpeaker A: Just incorporating some of the ideas that we had previously, like having multiple faceplates.\nSpeaker C: Great, thanks for that Ron.\nSpeaker D: I think so.\nSpeaker D: I can plug in.\nSpeaker D: That would be perfect, thank you.\nSpeaker D: Slide your open ready.\nSpeaker D: Or not.\nSpeaker C: Give it a little bit.\nSpeaker C: There we go.\nSpeaker D: Okay, so this is me.\nSpeaker D: Basically, I was looking through some marketing reports that we've got.\nSpeaker D: We had a usability test where we were actually watching 100 people use TV remotes and see what it is that they're using and then they feel that questionnaire about what they like and what they don't about their general TV remote control practices.\nSpeaker D: Pretty much through testing we were finding out that most of the time everybody's just using changing the channel, turning it on, using the volume, the majority of the time that's all that's going on.\nSpeaker D: The other functions happen for some people, they're important, but the primary uses are really, really basic.\nSpeaker D: And so, big complicated remotes, like the one we saw in the last presentation are really not the general public's use.\nSpeaker D: They're not using a lot of it.\nSpeaker D: They don't need it.\nSpeaker D: They even find it frustrating when there are all those buttons they don't know what to do with.\nSpeaker D: And we also found out that 50% of our people, the worst thing about a remote is how often they lose it and then they can't find it in the room.\nSpeaker D: So, I think what we were talking about with a pager or something will really come into play with a lot of these people.\nSpeaker D: There's also a survey about what they like about the remote and pretty much they all think they're hideous and not very useful.\nSpeaker D: And the younger demographics are all really interested in voice recognition options.\nSpeaker D: I don't know if that's something we're ready to look into technically, that's up to the design people, but it is something we're thinking about, especially since the younger demographics is obviously the one that's going to keep growing.\nSpeaker D: So, if that's the direction we're headed into something to think about.\nSpeaker D: But basically it really is the primary functions and getting it to look nice, which are the standards.\nSpeaker D: So, it's a good start for us.\nSpeaker C: That's great.\nSpeaker C: Thank you, Sarah.\nSpeaker C: Right.\nSpeaker C: So, yeah, I'll just switch that back here.\nSpeaker C: I'll finish up, which is a bit of discussion.\nSpeaker C: I'll just go ahead and put that on the panel for the next phase.\nSpeaker C: Right.\nSpeaker C: So, I think we've covered most of these important questions through this, through you guys's presentations.\nSpeaker C: We've got the industrial designers suggest, or pretty much emphatically suggested that we go with plastic.\nSpeaker C: Sarah, she's recommended that we go for simpler functions, so fewer functions.\nSpeaker C: But we need to decide who are we selling this to.\nSpeaker C: Your stats suggested that 75% of people under 35 want to be thought about voice control.\nSpeaker C: So, do we want to go for that, or do we want to go for an older demographic?\nSpeaker C: And my thought is, we've got, if we're going to go for a sleek look, I mean, we are putting the fashion and electronics.\nSpeaker D: We're not catering to the pensioners of the world, but we don't think so.\nSpeaker C: So, maybe this, we should look into this younger demographic.\nSpeaker C: So, we need to wonder how we can make it better and smaller and faster.\nSpeaker C: We're constrained to plastic very well.\nSpeaker C: We've got this idea, Ron, was saying we need to think about revolutionizing the way it's looking, which might be easier given that we're going for simpler function and that we're only going for tele.\nSpeaker C: So, this voice operation thing is a good idea, assuming that it's doable, at least for the basic controls.\nSpeaker C: Maybe we can balance it that way.\nSpeaker C: You know, we can see.\nSpeaker C: Okay, you can't say record alias tonight, except we might be able to say volume up.\nSpeaker C: Yeah.\nSpeaker B: Right.\nSpeaker B: I think it would be possible to combine the locator device and the voice recognition technology.\nSpeaker D: Oh, that could work.\nSpeaker D: I like that.\nSpeaker B: With a simple command like locate.\nSpeaker B: Yeah, something very basic.\nSpeaker B: It could start to be right.\nSpeaker B: Or be found.\nSpeaker D: Is that only going to be within our 200 foot range then?\nSpeaker B: Oh, yeah, I think that's right.\nSpeaker B: Doable.\nSpeaker C: The difficulty, what would be in, I think, you can't speak into the remote that you're trying to find.\nSpeaker C: Can you have something that picks up a voice from far away?\nSpeaker C: It's a good point.\nSpeaker C: It's hidden under the couch.\nSpeaker C: But then again, you have this we thing.\nSpeaker C: You know, just a little chip or whatever that has the page button.\nSpeaker C: Maybe that could be voice activated too.\nSpeaker A: A little sticky pad to stick on top of your television.\nSpeaker A: I need to say slinging to that.\nSpeaker A: Yeah.\nSpeaker A: I don't know if you can see the lines here.\nSpeaker C: Or an isolated magnet or something like, you know, something that wouldn't interfere.\nSpeaker C: I don't know if that's the technical thing.\nSpeaker C: Yeah, I like that.\nSpeaker C: I like that.\nSpeaker C: The voice recognition to the paging system.\nSpeaker A: The other thing is we might be able to handle the simplicity of the remote control and kind of put the more complicated things into a voice control.\nSpeaker A: So it could be sold to both the younger market and the older market.\nSpeaker A: The younger market could use kind of the voice control method.\nSpeaker A: The older market might get it just an option.\nSpeaker A: Exactly.\nSpeaker A: You might consider the older market could use this simpler design for the traditional buttons.\nSpeaker A: I was thinking about the screen.\nSpeaker D: Sorry.\nSpeaker D: Oh, go ahead.\nSpeaker D: The if we're going to do this touchpad screen thing, are we still do we know if that's an option technically right now?\nSpeaker A: Definitely an option.\nSpeaker A: Okay.\nSpeaker A: I've looked into costs of touch screen methods and whatnot.\nSpeaker A: They seem to be almost as cheap as a button method at this point.\nSpeaker A: Okay.\nSpeaker D: It seems like an interesting option, especially because then you could have like your primary screen just be these, you know, four or five basic functions and you have menu options or something to have all these other complicated voice recognition settings, things that you're not going to use every day and that a lot of people aren't going to use.\nSpeaker D: But it is an option there for this high tech market that sort of be is this leak thing we're going for.\nSpeaker B: It's kind of wonder though, for adding so much technology to this one remote, are we still going to be able to meet a 12 or 1250 euro goal for selling these things.\nSpeaker B: It seems like you're not going to be able to handle all those functions with just one microchip.\nSpeaker B: The microchip is probably the most expensive part of the whole mechanism.\nSpeaker B: Yeah.\nSpeaker B: So it's just something to consider.\nSpeaker C: Absolutely.\nSpeaker C: Okay.\nSpeaker C: Well, yeah, I guess we'll cross that bridge and the later stages of development.\nSpeaker C: But yeah, that's a perfectly viable question.\nSpeaker C: Okay.\nSpeaker C: So I'm seeing that we're going to just basically focus on this young demographic, aim it at them, but then in the sense that it's bells and whistles are available for anybody who wants them.\nSpeaker C: But basically we'll make a sleek, simple function, remote control.\nSpeaker C: I think this voice recognition thing is we've got a market for it.\nSpeaker C: I don't think there's too many, more or less be cornering the market on as well.\nSpeaker C: We don't have many.\nSpeaker C: Yeah.\nSpeaker C: Compared to a blast-man microphone.\nSpeaker C: Right.\nSpeaker C: We don't have many people.\nSpeaker C: There's not very many competitors out there that do that.\nSpeaker C: So cool.\nSpeaker C: Right.\nSpeaker C: I guess we've touched on most of this, the idea of a paging function, a touch-screen, faceplates.\nSpeaker C: The thing with, I see, would there not be a, we'd have to maybe sacrifice the faceplates for a touch-screen?\nSpeaker A: I'm not sure that's sincerely correct.\nSpeaker A: I think if you kind of take the example of a mobile phone that kind of has a portion of the device is not interchangeable, whereas the surrounding portions are interchangeable.\nSpeaker A: Okay.\nSpeaker A: The case.\nSpeaker A: We could have the casing.\nSpeaker A: Okay.\nSpeaker B: Back to the costs, the material, we have to ask whether we're going to include a certain number of faceplates with the package.\nSpeaker B: That's something that we're, say we're including three or four faceplates, we can drive the costs up.\nSpeaker B: The other question is if we don't include them, are we really in a position to evaluate that market?\nSpeaker B: We haven't done any tests on faceplates and whether there's even interest out there.\nSpeaker B: Okay.\nSpeaker B: So we're going to go ahead, sounds kind of like a gimmick that wouldn't really go anywhere.\nSpeaker C: Yeah, because it would have to, it's not like with cell phones, like we have a Nokia model X and then 10 people make faceplates for it.\nSpeaker C: We just are a model of remote control.\nSpeaker D: One publicity of a faceplate on a phone is you have it out and around it is sort of emblematic, whereas just sitting at home, so unless somebody comes over to watch TV.\nSpeaker A: Yeah.\nSpeaker A: Well, hopefully some people have people coming over to watch TV.\nSpeaker A: Sure.\nSpeaker A: And most people have their televisions in the living room.\nSpeaker A: All right.\nSpeaker C: Well, we can, we can discuss that one further when we think about whether when we do costs and so forth, true plastic is dead cheap and if we're making the whole thing out of plastic anyway.\nSpeaker C: Yeah, cross-apriach later.\nSpeaker C: But yeah, we will have to evaluate what's most important.\nSpeaker C: I think we've had a bit of discussion already on these things.\nSpeaker C: Any other questions, comments that came up in these presentations?\nSpeaker A: Well, have we confirmed that we're going to go ahead with a touchscreen?\nSpeaker C: Yeah, yeah.\nSpeaker C: Okay.\nSpeaker C: Yeah, I think that would be best.\nSpeaker C: Let's based on what you guys have all said to me, let's go for a plastic case because that this is the on the cost.\nSpeaker C: Try to look for some kind of high quality recycled plastic as you recommended.\nSpeaker C: That's a great idea.\nSpeaker C: With a touchscreen for the basic functions.\nSpeaker C: And we'll, yeah, provisionally let's go for a touchscreen room with several sub-manues for possible extra stuff that one basically put the channel and the on and off switch on the touchscreen.\nSpeaker C: Do we have, wait a minute, it occurs to me that if we have a touchscreen, people are going to have to recharge their remote controls.\nSpeaker C: Yet at the same time, that might help for this whole complaint of it being lost.\nSpeaker D: Because it would have a docking base.\nSpeaker C: Mm-hmm.\nSpeaker C: Then again, that costs as well.\nSpeaker C: Yeah.\nSpeaker A: So those new within that even the last 20 years, even with the touchscreen.\nSpeaker A: Do they?\nSpeaker A: Okay.\nSpeaker B: So we have three reports that include one of those.\nSpeaker D: And will somebody buy it if we tell them to?\nSpeaker C: Well, I don't think, yeah, I can't see anybody buying a remote control that we have to plug in.\nSpeaker C: So we'd have to see some kind of new battery technology.\nSpeaker C: Okay.\nSpeaker C: So let's go with a touchscreen with some kind of, you know, with some cutting edge battery technology.\nSpeaker A: For 12 years.\nSpeaker A: Yeah.\nSpeaker C: Well, hey, you know, it's worth looking into.\nSpeaker C: If not, we can always default to just doing a well-presented plastic.\nSpeaker C: Simple.\nSpeaker C: You know, it's, you know, well, yeah, I mean, you can put the, we could, I don't know.\nSpeaker C: I mean, I suppose we could put the basics on the center easiest, you know, you know, people know the channel and volume function make them large and easy to get at.\nSpeaker C: And then the other bits and bobs, you know, go through a menu.\nSpeaker C: We'll do the aesthetics.\nSpeaker C: Okay.\nSpeaker C: So look for the touchscreen in the battery.\nSpeaker C: Focus on presentation.\nSpeaker C: With this voice recognition option as well.\nSpeaker C: Just for the simple functions, the on, off, channels, volume, great.\nSpeaker C: And a small paging function, even if you can't do voice recognition for the paging, you know, just some kind of simple button that's just, I guess, another infrared signal to the remote control and well to emit some kind of paging.\nSpeaker C: Just okay.\nNone: Right.\nSpeaker C: So any comments, thoughts before we break into going to the next round of individual work on this?\nSpeaker B: Since we're doing a touchscreen, we want to look into the possibility of people being able to input different types of skins for the, you know, the actual interface part of it and things like that.\nSpeaker B: Or is it just going to be one touchscreen for everybody?\nSpeaker A: Interesting.\nSpeaker B: What would be on that touchscreen?\nSpeaker B: Because you said earlier that we have to think about company colors and logo or something or motto.\nSpeaker B: I can't remember exactly what you said.\nSpeaker A: We put fashion into electronics.\nNone: Yeah.\nSpeaker C: It's my understanding that if you were going to do a skin, you'd need to have some way for people to download or import skins into the remote control.\nSpeaker C: I think perhaps it's too much.\nSpeaker C: Good idea.\nSpeaker C: But yeah, I think that that one might just be, and just, yeah, I think that one might just be out of the range for this particular.\nSpeaker C: A PDA would, it makes a lot of sense for a PDA because you're going to be using it to connect up to things anyway.\nSpeaker C: I don't know what you guys think.\nSpeaker B: I think we just need to come up with a nice black and red interface on the touchscreen.\nSpeaker B: Yeah.\nSpeaker A: It'd be okay.\nSpeaker A: I'm in agreement with that.\nSpeaker A: I'm wondering how we're going to get, we put fashion into electronics.\nSpeaker D: Well, but if we're going to use a touchscreen where it's going to come on, like on your cell phone, it'll have your carrier provider name come up first like while it's loading and then it goes away.\nSpeaker D: Perhaps it could be like a temporary comes on every time you turn it on and then that's it because it is a bit much to have it like engraved on the back.\nSpeaker A: Yeah, true.\nSpeaker A: I'm hoping for subliminal maybe half of a second as it turns.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It seems like it would suffice to just have the RR on there.\nSpeaker D: Yeah, you would think.\nSpeaker C: But apparently not.\nSpeaker B: People are going to want their remote to boot up and see flashing things.\nSpeaker B: Come on.\nSpeaker B: It's going to be on and ready to go.\nNone: Yeah.\nSpeaker C: Well, fair enough.\nSpeaker C: Any of that would help the battery life too.\nSpeaker C: And if the remote, they do have to press a button for the remote to turn on.\nSpeaker C: Then again, who wants to turn on or remote control?\nSpeaker A: Well, all you have to do is touch the screen and it automatically.\nSpeaker A: Oh, it'll wake up.\nSpeaker A: Okay.\nSpeaker C: I'll go into it back to the remote.\nSpeaker C: Go to the sleep mode.\nSpeaker C: Okay.\nSpeaker C: Oh, yeah.\nSpeaker C: I like that.\nSpeaker C: I like the idea of putting the logo in the boot up screen.\nSpeaker C: Nice.\nNone: Cool.\nSpeaker C: So, any last things before we break?\nSpeaker C: All right.\nSpeaker C: Fair enough.\nSpeaker C: Sounds good.\nSpeaker C: I'm going to save the copy of this case you guys need reminders.\nSpeaker C: I'm going to save a copy of this and the main set I'll do in a second and put them in the shared folder for a later reference.\nSpeaker A: I've put my files in the shared folder.\nSpeaker C: Brilliant.\nSpeaker C: That's fab guys.\nNone: Cool.\nNone: All right.\n"}, {"meeting_id": "ES2009c", "summary": "The whole meeting was about a detailed prototype of the remote control. As before, this meeting also began with personal presentations. The Industrial Design focused on the components of the remote control, including the energy source, the buttons and the materials, while the User Interface suggested to use voice recognition technique and add secured or hidden programming onto the touch screen. Then the problem came that if they would like to make it so complex, they would go beyond the budget and have to raise the cost. Thus they decided to simplify some functions and buttons. Before the meeting ended, they also discussed the shape of the remote control.", "dialogue": "Speaker C: That's great.\nSpeaker D: Okay, all right.\nSpeaker D: Let me just PowerPoint this up.\nSpeaker D: Right, so this meeting will be about the conceptual design.\nSpeaker D: Don't ask me precisely what conceptual design is.\nSpeaker D: It's just something important that we need to do.\nSpeaker D: I think it's kind of a turning the abstract into slightly more concrete, and this meeting ideally will come to some final decisions on what we're going to do for the prototype.\nSpeaker D: Right, so apologies for the last meeting.\nSpeaker D: It was brought to my attention that I did not make the rules clear enough, so I will attempt to do so more accurately in this particular meeting.\nSpeaker D: Fair enough. Thanks for the input. It's always good.\nSpeaker D: So basically all we're going to do is have some presentations again, much like last time.\nSpeaker D: I'm going to go through you. Whoever wants to go first is fine by me, and we'll collate what we know about what we discussed in the last meeting, possible directions. And then we'll make some more decisions on basic, firm up our idea on how we want this remote control to look and work.\nSpeaker D: Perfect. So without further ado, whoever wants to go first is free too.\nSpeaker D: I'll go first. All right, Nathan. Take it away.\nSpeaker D: It is Nathan. I'm not calling you the wrong name over.\nSpeaker A: No, Nathan's fine. Good.\nSpeaker A: It's either Nathan or participant too.\nSpeaker B: Mr. and participant too, that is.\nSpeaker B: Nice. Okay. Nice.\nSpeaker A: So basically what I'm going to have to talk to you about today is component design.\nSpeaker A: And it's been brought to my attention that we may be somewhat limited as to what we can do because of what our manufacturer offers.\nSpeaker A: So basically what I'm going to be doing is talking to you about that.\nSpeaker A: Components of remote control. Okay. We've already kind of gone over this, but we're going to have to get into more detail and probably have to reach some conclusions sometimes soon. Energy source.\nSpeaker A: Our manufacturer offers a variety of energy sources.\nSpeaker A: Your standard battery, solar cells are manufactured in to anything about lithium.\nSpeaker A: So we might have to look, if we do go that route, we might have to look elsewhere.\nSpeaker A: And also there's a kinetic energy possibility.\nSpeaker A: Basically it's like a idea of moving the remote would create enough energy to keep it running.\nSpeaker A: So that's one possibility, but I don't know whether that would be powerful enough to illuminate a touch screen. So we'll have to look into that.\nSpeaker A: The case, we have a few options.\nSpeaker A: Plastic, rubber, or wood.\nSpeaker A: And then as far as the way it's shaped, we can do standard-borne flat, which we probably don't want to do. Curved or very sexy, double-curved.\nSpeaker D: What kind of thickness are we looking at?\nSpeaker A: I imagine that we could specify.\nSpeaker A: I don't see any reason to go outside of the convention of three or four millimeters.\nSpeaker A: Okay.\nSpeaker A: The buttons, there are multiple scope buttons available for a manufacturer, but to use those we would have to use more chips.\nSpeaker A: And that would cost us more.\nSpeaker A: And if we do go with rubber, double-curved case, we'll have to use rubber push buttons.\nSpeaker A: Because the other buttons aren't compatible with that.\nSpeaker A: And just a little note there, touch screen equals many chips, which equals many euro.\nSpeaker A: Right.\nSpeaker A: One thing that I noticed is that most remotes operate on the infrared part of the spectrum.\nSpeaker A: So you notice when you push a button on the remote, you can't see anything coming out of it.\nSpeaker A: But in fact, there is light coming out of the remote and the television conduct that.\nSpeaker A: And if you were to record, if you were to make a video recording, you could actually see the light.\nSpeaker A: One thing that I thought might be interesting was to use visible light coming out of the remote, just kind of as a fun gimmick.\nSpeaker A: So you could actually see something coming out of the remote when you pushed it.\nSpeaker A: Of course, it would have to be a part of the spectrum that wouldn't damage the human eye.\nSpeaker D: Is there an option that we could have that offer on so a person could select like...\nSpeaker A: Yeah. I am sure that we could do that.\nSpeaker A: Of course, yeah, yeah, it's a good idea.\nSpeaker A: Yeah, just as a fun gimmick.\nSpeaker A: Just to set us apart a little bit.\nSpeaker A: And then onto the circuit board that we're going to use, also, notice the chip.\nSpeaker A: We really don't have any way around the TA-11835.\nSpeaker A: Findings. Okay.\nSpeaker A: We're very limited by what our current manufacturers can offer.\nSpeaker A: And my question to all of you is, should we look to other manufacturers, or just make two with what we have available?\nSpeaker A: Interesting question.\nSpeaker A: It's a bit of a challenge question.\nSpeaker C: Well, I'd say shop around, but with our time constraints, is that really a feasible option?\nSpeaker A: Right. That's my concern too.\nSpeaker A: If we do go to the lithium battery route, then we'll have to go outside current manufacturing.\nSpeaker A: My personal preference is, I'll just throw my cards on the table.\nSpeaker A: I think we should probably go to the solar battery route, just to kind of keep with the environmentally friendly theme that we have going on.\nSpeaker A: I like the idea that visible light signaling is something to set us apart.\nSpeaker A: And I was thinking about...\nSpeaker A: I was thinking of ways that we could produce the remote in a variety of different case materials to suit different tastes.\nSpeaker A: So not so confined by one style and say, say, if we just go with one, and it doesn't go over well, then we're in a bad situation.\nSpeaker D: Right.\nSpeaker D: Can we do marketing piloting too?\nSpeaker D: Try to see what kind of before we launch. Can we see how they're received?\nSpeaker C: It's an option. Actually, I've got some research already on what we're looking at and trends in casing right now, which actually might even come into play beforehand.\nSpeaker C: Okay, perfect.\nSpeaker C: I'm coming up with a side for now.\nSpeaker C: Great. Thank you very much.\nSpeaker D: It's perfect. So I guess that makes sense for you to take it from here.\nSpeaker C: I guess so.\nSpeaker C: Because I found some interesting things.\nSpeaker C: Fascinating.\nSpeaker C: I know.\nSpeaker C: I wanted to hear.\nSpeaker C: Right.\nSpeaker C: So current market trends.\nSpeaker C: Screen.\nSpeaker C: Basically, I was looking at what's going on in the remote control market right now, and what's going on in other design fields to see sort of what's trendy, what's new, what's happening.\nSpeaker C: Remote control right now, basically, everybody says they want newer, fancier, more exciting.\nSpeaker C: They're sick of this boring, normal, functional, that we need innovative design options, and that needs to be an easy user interface.\nSpeaker C: The challenge is that current trends right now across the board in fashion, in furniture, in technology is a very organic, fruit and vegetable kind of thing.\nSpeaker C: Now, I'm not saying we should have, you know, tomato shaped remote controls or anything, but I think it is possible maybe to use natural colors, like if wood is an option that whole organic sleek clean, lined thing, maybe something we can look into, different skin options, or if we can't afford this touch plate thing or touch face screen interface, maybe having the images be specific like you could choose your menu bullets to be different.\nSpeaker C: Okay, not the example I would choose, but you know what I mean to sort of, and apparently the feel of the next couple of years is spongy.\nSpeaker C: Not something I come up with a definite answer on, though, if we can get around to getting piloting, I thought maybe a casing option, like not like a skin, but like a holder almost, if you could do like,\nSpeaker A: leather options or wood options or something. I should have mentioned this, as far as the rubber that we can use, we can use a rubber as part of the case, it has a consistency of those stress balls.\nSpeaker C: Might be an issue way to go. Yeah, so something to sit down for now. So overall, I think we should stick with what we're finding. Every's looking for easy use, technologically innovative and this fancy new, I think perhaps the double curve thing and maybe this rubber option is our best way to go for right now.\nSpeaker D: Interface. Oh, the interface graphics for the.\nSpeaker C: Well, but then if the touch screen thing isn't going to work out for us, that's really not an issue.\nSpeaker A: I like the idea of rubber at sea because it's tenset, be associated with being durable, something that you can drop and it doesn't matter.\nSpeaker A: So many, you go to so many houses these days and you see broken remote control.\nSpeaker A: Very true.\nSpeaker A: Yeah, it's like with duct tape.\nSpeaker A: Very much so.\nSpeaker A: You would have a problem.\nSpeaker B: You could have a duct tape casing.\nSpeaker C: I think that goes against the whole fancy something.\nSpeaker B: But worth a shot.\nSpeaker B: The granola crowd.\nSpeaker C: Yeah, that's what I know.\nSpeaker D: Great. Thanks for that, Sarah.\nSpeaker D: No problem.\nNone: Run.\nSpeaker B: Computer is adjusting.\nSpeaker B: One moment, please.\nSpeaker B: So, interface concept by your faithful user interface.\nSpeaker B: So, user interface guys is basically aspects of a computer system that we can see.\nSpeaker B: Or here.\nSpeaker B: Or otherwise, perceive commands and mechanisms that basically user uses to control the operating system.\nSpeaker B: Here's a series of different remote controls that are out on the market today.\nSpeaker B: I think we're definitely trying to get away from this kind of a look.\nSpeaker B: Yeah.\nSpeaker B: So, the following are a bunch of different interface concepts.\nSpeaker B: Voice recognition.\nSpeaker B: We actually have some new information from our research design team.\nSpeaker B: But I'll get to that in a moment.\nSpeaker B: So, current voice recognition starts up to about 80 speech samples.\nSpeaker B: And basically record your own verbal labels and connect them to the remote control.\nSpeaker B: Now, our design team, research team has been able to set up a system in which you can teach the remote control voice recognition system to respond to with standard responses like you could say good morning remote control.\nSpeaker B: And it will say in a sexy female voice.\nSpeaker B: Good morning, too.\nSpeaker B: In fact, we already have this for our copy maker line.\nSpeaker D: A lot of single people on the remote control.\nSpeaker B: Yeah.\nSpeaker B: Another concept is what Apple has come up with.\nSpeaker B: The spinning wheel with LC display like on the iPods which I'm sure most of you know.\nSpeaker B: Then we have this scroll button with the integrated push button.\nSpeaker B: Like a modern loss of both of you.\nSpeaker B: Crazy. I don't think that's what we're necessarily going for.\nSpeaker B: And some special components.\nSpeaker B: Ideas like blocking having the ability to block channels from your children.\nSpeaker B: And dedicated buttons for commonly used channels.\nSpeaker B: And even ideas like secured or hidden programming.\nSpeaker B: But again, if we go with touch screen, I don't think that's a big issue.\nSpeaker B: And this is kind of the big daddy.\nSpeaker B: The jumble universal remote control is almost impossible to miss place.\nSpeaker B: Again, probably not what we're going for.\nSpeaker B: So I mean, my ideas here and kind of where I think we're heading is something slightly larger than a regular iPod.\nSpeaker B: With a hard classic plastic casing, I think some of the suggestions we've come up with are definitely very good ideas.\nSpeaker B: Changeable casings of our design team was possibly talking about including one extra faceplate with the package to kind of set the idea that you can change it.\nSpeaker B: And you can try changing and kind of get used to thinking about maybe buying another one which can add value to our bottom line.\nSpeaker B: Touch screen interface, possibly having go to buttons being stuck into the system so those don't move away from the screen.\nSpeaker B: The important ones like power volume and jump between channels.\nSpeaker B: And of course our voice commands system which I've talked a little bit about already.\nSpeaker B: And the use of recognizable colors in shave state recognition of the features that are around.\nSpeaker B: Red for power arrows for different volume ups and downs and channels ups and downs and whatnot.\nSpeaker B: And perhaps you've been adding in some stupid little jokes with the voice recognition idea.\nSpeaker B: For instance, my toasty maker that I got from my bank has jokes when it's ready.\nSpeaker B: Nice.\nSpeaker B: Great and that is about it.\nSpeaker D: Great wonderful run. Cool.\nSpeaker D: A lot of good ideas, good facts to have.\nSpeaker D: That's what they need is like a little dongle that just sticks up this furthest.\nSpeaker D: You don't have to stand up every time.\nSpeaker D: Right.\nSpeaker D: So good to know all that stuff. Thanks guys.\nSpeaker D: I know we kind of have to come to some decisions.\nSpeaker D: I figure we can just go down the line and all three of us can have a chat about it.\nSpeaker D: Based on what Nathan presented as far as the various costs and benefits.\nSpeaker D: I think, I don't know, what do you guys think about the touch screen at this point?\nSpeaker C: I think it's our most marketable feature just because it's so new and it's something that is showing up in other places.\nSpeaker D: Okay.\nSpeaker C: But can we really afford it because it looks like they would be that would be a really main cost source then.\nSpeaker C: Right.\nSpeaker A: My estimate is that in order to incorporate touch screen technology, it's going to cost those upwards of 1750 euro.\nSpeaker A: It's pretty easy to watch.\nSpeaker B: It's just an estimate that.\nSpeaker B: You guys are always the dampers on these hands.\nSpeaker B: You're industrial designer.\nSpeaker B: I know.\nSpeaker C: And our goal was to be under 1250 or we have to be under 1250.\nSpeaker A: Well, I thought there was some flexibility with that.\nSpeaker D: There is.\nSpeaker D: It's just it is a question of.\nSpeaker D: To justify it.\nSpeaker D: And how much does that mean we're going to have to increase the price to make money.\nSpeaker D: From 1250 if we want to get a 100% profit margin.\nSpeaker D: That would mean selling it for 25.\nSpeaker D: If you multiply 1750 by two, that's 35.\nSpeaker D: Right.\nSpeaker D: So.\nSpeaker B: Where do you guys come up with these numbers?\nSpeaker A: There's just stuff in the head.\nSpeaker A: It is pending for their emails.\nSpeaker D: From the board.\nSpeaker D: Well.\nSpeaker C: Do I think that's what people would pay for.\nSpeaker C: I mean, if you're going to pay for an expensive high-class remote, you're going to expect to do something.\nSpeaker A: That's true.\nSpeaker D: I would be in a class of a term.\nSpeaker D: Yeah.\nSpeaker D: And that's to be fair.\nSpeaker D: The percent of the market, we're not going for mass, you know, mass sales anyway.\nSpeaker D: We're going to make.\nSpeaker D: I mean, we're not talking about selling eight zillion of these things.\nSpeaker D: We just couldn't not for 25 euros.\nSpeaker D: So.\nSpeaker D: We could probably maybe shrink the profit margin rather than selling for 25, sell them for 30.\nSpeaker D: But that's something we can have finance deal with.\nSpeaker D: I say that we provision would go with the touchscreen.\nSpeaker B: What was your thought of the matter on?\nSpeaker B: It's a really good idea.\nSpeaker B: And I also think that we could probably come up with some sort of a cheaper means to go about this kind of production.\nSpeaker B: My team on the third floor suggested that.\nSpeaker C: Well, and we can look into this other manufacturing option.\nSpeaker C: I mean, you can get them somewhere else cheaper.\nSpeaker A: Yeah.\nSpeaker A: It's true.\nSpeaker A: We could initially go with what we have and if we can find the system.\nSpeaker D: Starting point anyway.\nSpeaker D: So.\nSpeaker D: We could have a very simple touchscreen.\nSpeaker D: You know, there's always the opportunity that's going to be about the size of the iPod or whatever.\nSpeaker D: Yeah, I guess we can play around with it.\nSpeaker D: All right.\nSpeaker D: Let's say that.\nSpeaker D: Okay.\nSpeaker D: So the touchscreen will be our main selling point here.\nSpeaker B: I mean, I think that we really have two main selling points.\nSpeaker B: I think that our casing.\nSpeaker B: And the voice record.\nSpeaker C: I mean, really, this is pretty bells and whistles kind of remote.\nSpeaker D: Yeah.\nSpeaker D: The voice rack thing.\nSpeaker D: I mean, if we're looking at bottom line.\nSpeaker D: Now we're looking at opening the cost of 17 to get the touchscreen on.\nSpeaker D: I think we might have to drop the voice rack.\nSpeaker C: I think we'd have to decide between them.\nSpeaker C: Definitely.\nSpeaker B: Yeah.\nSpeaker B: To be honest, we have the key.\nSpeaker B: We have the design in house.\nSpeaker B: I mean, we've, we've come up with this, which is new voice.\nSpeaker B: We're using it for coffee machines already.\nSpeaker B: I can pass you on that email from my guy and guy down the hall.\nSpeaker B: Sounds good.\nSpeaker D: What do you think on anything about the voice rack?\nSpeaker A: I think if we, we do both the obviously production costs are going to go way up.\nSpeaker A: But it does put it into become the Rolls Royce of remote controls.\nSpeaker A: Pretty much.\nSpeaker A: Very nice.\nSpeaker B: I mean, we have to reflect back on what our market research did say.\nSpeaker C: Right.\nSpeaker C: And they said they won voice recognition.\nSpeaker C: Of course, maybe they hadn't thought of this whole touchscreen option.\nSpeaker C: But definitely we know the market is there for voice recognition.\nSpeaker C: So to say we have the technology and we're not going to use it, even though we know it will sell.\nSpeaker C: There's a call I don't think I can give the high-ups.\nSpeaker C: Like really, I can't go in and say no.\nSpeaker C: We're going to stick nor everything we know.\nSpeaker D: Yeah.\nSpeaker B: Does having both really up our costs?\nSpeaker D: I can't see how it wouldn't.\nSpeaker D: I mean, there's, you know, the old aphorism.\nSpeaker D: You can have it fast.\nSpeaker D: You can have it cheap.\nSpeaker D: You can have it quality.\nSpeaker D: Pick two of three.\nSpeaker D: You know, you can't have all three.\nSpeaker A: Because you, it's just a number of chips that you need to deal with each different function.\nSpeaker D: Well, if we're going to pick it to, all right.\nSpeaker D: So we have to pick between one of the two.\nSpeaker D: Otherwise, yeah, we just, it just becomes cost prohibitive.\nSpeaker D: What, which, which do we suspect we should hold on?\nSpeaker D: We should hold on to.\nSpeaker C: Well, we already have research backing voice recognition as, you know, viscerally solvent.\nSpeaker C: But I, I, I personally would tend to the direction, but if that's what's going to sell, I think that's what we need to go with.\nSpeaker C: And maybe we can table this touchscreen for our next model.\nSpeaker A: I would have to side with that.\nSpeaker A: I think the voice recognition is simpler.\nSpeaker A: We have the, all the technology in house that's ready to go.\nSpeaker A: It's packaged.\nSpeaker D: What does the cost look like?\nSpeaker D: Well, it's a cheaper to do the VR to do the touchscreen.\nSpeaker A: Well, I, um, it's just off the top of my head.\nSpeaker A: Keep in mind.\nSpeaker A: But I think the voice recognition would, they're both it.\nSpeaker A: They're both going to push the cost up.\nSpeaker A: But, um, since we already have the technology in house for the voice recognition, we're not going to have to do as much design work.\nSpeaker A: And sometimes the design work is what pushed the cost up.\nSpeaker A: Right.\nNone: If you know what I mean.\nSpeaker B: I definitely have to agree with that last.\nSpeaker C: And we're still not, then we don't have to deal with this battery issue really as much either.\nSpeaker C: We can stick with what we've already got in a lot of other ways too.\nSpeaker D: Okay.\nSpeaker D: So I'm getting.\nSpeaker D: So more or less, you guys think that of the two of them, the voice recognition will be better.\nSpeaker C: I think it's our lower risk option, which right now we can have it on the market sooner, which is all in all our best option.\nSpeaker D: Okay.\nSpeaker D: Sorted.\nSpeaker D: We will omit the touchscreen in favor of voice recognition.\nSpeaker C: So, and when are we going to have basic prototypes coming up next?\nSpeaker C: That's you guys next step, right?\nSpeaker D: Yeah.\nSpeaker D: So, I'm going to sort out what what what else we're going to talk about for the prototype.\nSpeaker D: But yeah, that's our next step.\nSpeaker A: Are we going to talk now about the materials we're going to use for the case.\nSpeaker A: We'll just run through it.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: You discussed either a lithium or a solar power with the solar power be enough to fuel a voice recognition of this kinesthetic one.\nSpeaker D: Would that be enough to fuel a voice recognition remote control?\nSpeaker A: The solar power definitely would be, but I think just to keep people from getting annoyed because sometimes solar power fails just not around that.\nSpeaker A: We should install a small backup battery.\nSpeaker C: Hmm.\nSpeaker A: Just to cover those moments when for whatever reason the remote has some experts.\nSpeaker A: Well, people like to be in their basement.\nSpeaker C: Like we're we can't guarantee someone everywhere.\nSpeaker C: So having a factory source is probably.\nSpeaker A: Yeah.\nSpeaker A: It works about the same as a solar powered calculator.\nSpeaker A: And you know how those this don't really require that much light.\nSpeaker A: Obviously a little more light than calculated, but we're not talking about a lot of light doesn't have to be out taking some bath.\nSpeaker A: Right.\nSpeaker A: Right.\nSpeaker A: Okay.\nSpeaker D: What do you think?\nSpeaker B: I'm going to do agree with everything that's been said.\nSpeaker B: Okay.\nSpeaker B: I have to say though that another idea is coming up in my head.\nSpeaker B: If we're really not handling the remote control to a great extent, we could possibly get away from the idea of having a handheld remote control and maybe kind of have a round remote control that kind of looks like a paperweight or something like that.\nSpeaker B: Kind of a sweet little thing that sits on your table or something.\nSpeaker C: Can you say?\nSpeaker B: Why?\nSpeaker A: Why moving away from handheld?\nSpeaker A: Why?\nSpeaker B: Why not?\nSpeaker B: Well, if you don't need to pick it up, it could kind of be a selling point.\nSpeaker C: It's got voice recognition.\nSpeaker C: It can be technically anywhere in your room and still do its job.\nSpeaker A: Do you think people that are people that buy remote, are they always going to want to use the voice recognition?\nSpeaker A: It's just something that they do sometimes.\nSpeaker C: True.\nSpeaker C: And probably I think we're banging on selling it to more than just voice recognition people like we want it to work fundamentally.\nSpeaker C: We have to.\nSpeaker D: Right.\nSpeaker D: But that's done.\nSpeaker D: They might have.\nSpeaker D: Who is it?\nSpeaker D: Apple makes these really pretentious speakers with the subwoofers, like clear and glass and the little pyramidal type of speakers.\nSpeaker D: Why not have a little rounded kind of thing?\nSpeaker D: It could still have the basic buttons on it because we're going for basic functionality primarily as well.\nSpeaker D: It's about, yeah.\nSpeaker D: And maybe a menu button and so forth, you know, use the channel button to scroll through the menu if they want to record programs or whatever, you know.\nSpeaker A: I think you're onto something because we need to escape the traditional shape of the remote.\nSpeaker A: Maybe something that looks nice on a table would be good, even though, and handheld at the same time.\nSpeaker B: I think it's all about following Apple's week.\nSpeaker C: Yeah, I'm thinking of the airport portal, you know, like that little pod looking thing.\nSpeaker D: Yeah, I mean, and I, although we do, as I'm recalling what she mentioned, we need to get away from the surgical white kind of brush to the aluminum thing and get back to it.\nSpeaker D: But you could have a very tasteful wood colored or earth tone kind of.\nSpeaker C: Terracotta bowl or something.\nSpeaker D: Still, I mean, yeah, along those lines.\nSpeaker D: Yeah, I like that.\nSpeaker D: I like that idea a lot.\nSpeaker D: Let's see what we can do as far as that goes. Okay. And the material like the plastics and so forth, we were discussing that being using like a rubber kind of softer feel.\nSpeaker D: You know, I feel the tip on this pen is a bit gives just a bit.\nSpeaker D: Yeah, something where it's a more advanced form of plastic that has some kind of a tactile response to it.\nSpeaker A: Right.\nSpeaker A: Just kind of the squishy feel.\nSpeaker C: Yeah, which is the next big thing. So that's not going to hurt us either.\nSpeaker D: Yeah, let's see if we can do a squishy, non remote control looking remote control.\nSpeaker D: But to be fair, yeah, I mean, you just could just put it literally put it on the night, the coffee table next to the telly and say volume up.\nSpeaker D: I like it. I like that.\nSpeaker D: That's good.\nSpeaker D: And we've sort of discussed costs.\nSpeaker D: I guess that's going to be a thing if we run a bit over budget, it might be okay.\nSpeaker A: Sorry about the lack of information costs. I just haven't been provided information by manufacturers.\nSpeaker A: Well, I have more ideas.\nSpeaker D: We'll have more of an idea later on.\nSpeaker D: Yeah, go from there.\nSpeaker D: So we've revisited the touch screen and more or less ruled that out. I think so we're more or less in agreement that we want to have a simple kind of function.\nSpeaker D: You know, not too complex.\nSpeaker C: Right.\nSpeaker C: Well, when the majority of people are only using the most primary functions on a daily basis, although I'm not saying we should completely rule out major functions, they should be secondary, at least if not functionally then visually like those shouldn't be.\nSpeaker D: Right.\nSpeaker D: Like maybe have menu things.\nSpeaker A: If we're not going to touch screen root, then we can just incorporate maybe something that folds out like what you often see on these kinds of remotes is the most basic functions up here.\nSpeaker A: And then something that slides down into the field, you know, the more complicated things.\nSpeaker C: We want to consider like an iPod screen, which isn't a touch screen, but you're still scrolling through menu options.\nSpeaker B: Then we're hitting our cost issue.\nSpeaker B: Sure.\nSpeaker D: And we've also got the thing of we're going to have a non-remote looking remote.\nSpeaker D: How do we?\nSpeaker D: Yeah.\nSpeaker D: But no, I mean, we could do a slide or a compartment, you know, like if it's say it's a little vaguely avoidable type of thing, it'd be easy to have a compartment in there.\nSpeaker D: Or, you know, a series of, you know, three or four buttons with a menu button and then a side and up and down type of thing like on a DVD player.\nSpeaker D: You see the modern DVD players will just have a menu button on the side and then four buttons around them.\nSpeaker D: And you can just kind of maneuver through the menu like that.\nSpeaker A: So are we talking, we need to figure out what kind of buttons we're going to use.\nSpeaker A: Are we going to use scroll buttons, rubber buttons?\nSpeaker D: Well, it seems like, I don't know, it seems to me that we could just do the stick with the rubber, because we're probably going to be using some kind of rubber for the outside case.\nSpeaker D: Myles will stick with that.\nSpeaker D: Right.\nSpeaker B: I think to a certain extent, we have to stick with kind of a little bit traditional in terms of the buttons and then, and then make our unique future casing and whatnot.\nSpeaker C: And our place.\nSpeaker C: Well, those basic, just like four directions that are that can use as menu or channel and volume or however you want to do it are really versatile and every's already got them to some extent.\nSpeaker C: I'm a moat there.\nSpeaker C: So it's not like we're dealing with everybody re-learning things because it's not so many.\nSpeaker C: It's been one of my new four.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: We've already kind of covered this as well.\nSpeaker D: Seems to be selling.\nSpeaker D: And we're more or less agree that we want to target this youth market.\nSpeaker D: Especially now with 18 to 35 year olds being a large quantity of the population.\nSpeaker C: Particularly in technological fields.\nSpeaker C: So it's exactly where we're headed.\nSpeaker D: Okay.\nSpeaker D: Yeah. All right. Well, more or less covered.\nSpeaker D: But we need to cover, I think, in the final thoughts before we think about doing the product, moving out of the prototype.\nSpeaker B: Well, what are we actually doing?\nSpeaker B: Right.\nSpeaker D: I was just going to step on to.\nSpeaker C: Oh, it wasn't in the way.\nSpeaker D: Oh, it wasn't.\nSpeaker D: Oh, my bad.\nSpeaker D: Sorry.\nSpeaker D: No, no, we're not.\nSpeaker D: Yeah, okay, I was just going to assign tasks in the next.\nSpeaker D: Any other final thoughts before we go ahead and cool.\nSpeaker A: So I would say that we are going to go with different style cases for different people.\nSpeaker A: Or we just going to go with one.\nSpeaker A: It's very, it's very hard thing to predict because you have different cases that might open up the market a little bit.\nSpeaker A: Obviously, but you have just one case and it doesn't go over well.\nSpeaker A: Yeah, it's a situation, but obviously having more cases also costs more.\nSpeaker D: Well, then again, colors wouldn't be so hard to do.\nSpeaker D: You could have a, you know, I am kind of a natural wood color, like a stained wood and, I don't know, all of green or something.\nSpeaker D: That wouldn't be so much of a problem to incorporate into the color of the rubber, I don't think.\nSpeaker B: And again, copying IMAX kind of for iPod Mac apples, color scheme.\nSpeaker C: Yeah, totally.\nSpeaker D: Yeah, I think that's probably good. Okay, so let's work on multiple case colors.\nSpeaker D: But yeah, stick with the same kind of, kind of.\nSpeaker D: Yeah, the same basic non-remote kind of remote design.\nSpeaker D: Cool.\nSpeaker D: All right, so we'll have the next meeting in about half an hour.\nSpeaker D: I want the, I'd like, Nathan, I'd like you to work on just the basic look and feel.\nSpeaker D: What can we accomplish given these parameters that we've sort of sorted out this kind of a non-remote remote.\nSpeaker D: What are broad constraints before we design a prototype?\nSpeaker D: And, Ron, if you can figure out how probably best to lay out this idea of the simple design with the voice recognition built in and also this kind of drop down or on the side kind of menu options simple, somehow work out how we can get this all in the same place.\nSpeaker D: And if you can check product evaluation with some pilots and stuff, I need you guys to work together on making a prototype using prototype building materials.\nSpeaker D: And also specific instructions will be sent to you by your coaches as well.\nSpeaker D: So that's what to start with for now. Is that all right?\nSpeaker D: That's so clear about this.\nSpeaker D: All right.\nSpeaker D: I guess we'll just hit the bricks. Thanks guys.\nSpeaker C: Cool.\nNone: Thank you.\n"}, {"meeting_id": "ES2004d", "summary": "The team mainly planned the remote control project down to the last detail in the final meeting. Industrial Designer presented the prototype as a clam-shell design with a flip-top LCD touch screen and fully closed flat buttons at the bottom. There would be a wide choice of colours and patterns due to customization. Inside the remote control, there would be a standard chip and a kinetic battery. The team was under the budget at present. Once the budget increased, however, the voice recognition would be incorporated. The remote control's fashion style and location function were most highly rated among the team members, but the minimization of repetitive strain injury was not satisfying enough. All members gave their sincere feedback on teamwork and leadership.", "dialogue": "Speaker E: Okay.\nSpeaker E: Minutes. We decided to use a kinetic charger.\nSpeaker E: Standard chip, because it can come in various different sizes, it wasn't going to be a problem factor.\nSpeaker E: We wanted a standby function. The case materials can be soft, rubbery, changeable.\nSpeaker E: Buttons with a combination of LCD and rubber according to design.\nSpeaker E: Bright funky designs, inspired by fruit, keeping with the hip kind of feel.\nSpeaker E: And to try and incorporate voice recognition software into our design until we can find out more about the cost of things like that.\nSpeaker E: And the rubber buttons that we'll use will be anti-RSI.\nSpeaker E: Okay.\nSpeaker E: Prototype presentation.\nSpeaker F: Is that for us?\nSpeaker A: I think that would be you.\nSpeaker A: Okay.\nSpeaker A: Me and William both don't prototype and I think William is going to make a presentation on that.\nSpeaker F: Yeah, shall I show?\nSpeaker F: Do you want to hold it?\nSpeaker F: Do you want to hold it?\nSpeaker F: I'll see you at the presentation.\nSpeaker E: Yes, yes, I'll get it.\nSpeaker F: It looks crazy over here.\nSpeaker F: Thank you.\nSpeaker E: Just have one last thing you can just take it off.\nSpeaker A: Oh, that other hand is...\nSpeaker F: Okay, so this is our look and feel presentation, our final presentation.\nSpeaker F: And we'll first look at the exterior of what we've come up with over there.\nSpeaker F: It's going to have a plastic body with a sort of standard color, either we're thinking something fairly neutral like a white or a light blue or something.\nSpeaker F: This is underneath the rubberized outer casing, which there'd be sort of a wide choice.\nSpeaker F: They would be attached, but we can count with that.\nSpeaker F: The wide choice of colors and sort of patterns.\nSpeaker F: So you've got a lot of customization with it.\nSpeaker F: It's obviously in a sort of clamshell design.\nSpeaker F: The top LCD screen that you can see there would be sort of...\nSpeaker F: A little black and white touch screen.\nSpeaker F: Yeah, it would be sort of inset into the top and the bands at the bottom would say it'll fully close.\nSpeaker F: Flat.\nSpeaker F: And working on the inside, we've already decided on the kinetic batteries, which I actually think about it now, because you could also probably attach to the flipping open and shut as well, so that you could probably get a bit...\nSpeaker E: Sort of get in the top section now, the bottom section, because it's in the top of the box.\nSpeaker F: We decided that the voice recognition system, they did actually say on the email that they were sort of coming in and they were fairly easy to get hold of, so we presumed that they'd also be quite cheap.\nSpeaker F: So you'd have something like, you'd shout out where is the remote, and it'll shout back, I'm here or something.\nSpeaker C: And then...\nSpeaker F: You can think.\nSpeaker F: Yeah.\nSpeaker F: It won't shout out, I'm here or something. It'll just shout out, I'm here.\nSpeaker A: Okay.\nSpeaker F: Or something too similar.\nSpeaker F: Yeah, if it was getting that...\nSpeaker F: Well, that could be something for next time, maybe.\nSpeaker F: You can have a remote that tells you exactly where it is.\nSpeaker F: The standard...\nSpeaker F: There's a standard transmission with the TV using all the standard chips that we talked about.\nSpeaker F: It would have, obviously, because it's split over two different layers.\nSpeaker F: It would need two separate PCBs that would be joined at the hinge through some sort of cabling.\nSpeaker F: And because, obviously, all TVs use this, the same infrared medium would just be using the same thing to transmit the data.\nSpeaker F: The infrared...\nSpeaker F:...sender would be on one of the bottom layers just at the front of the...\nSpeaker F: I don't think we actually put it on...\nSpeaker F: Yeah, someone like that.\nSpeaker F: Infrared could be built on the front side of that.\nSpeaker A: Oh, right.\nSpeaker A: So when it's open here, the signal is on the right side.\nSpeaker F: So when you've actually got it open, it would be facing the TV.\nSpeaker F: And then finally, on the interface, the top screen, as we said, would be an inset black and white LCD touch screen, which...\nSpeaker F: Yeah, would have all of the available functions for the remote on it.\nSpeaker F: Whereas the bottom screen would just contain the standard buttons, like the volume up and down, channel up and down, the hour on and off.\nSpeaker F: And things to that effect.\nSpeaker F: And now, we've also decided on the inside, we could possibly either have some kind of sort of bezeled logo on it or something inset, or maybe an engraving of the logo on the top.\nSpeaker F: But not interfering with the...\nSpeaker E: No, not actually...\nSpeaker F: No, not interfering with the whole look of the product when it's out on the thing.\nSpeaker F: And finally, that's how we put the fashion back into electronics.\nSpeaker F: Thank you very much.\nSpeaker E: That's a company logo.\nSpeaker E: Oh, don't...\nSpeaker E: Okay.\nSpeaker E: I've now got evaluation criteria.\nSpeaker E: Certainly.\nSpeaker A: So, this is the one.\nSpeaker A: Sorry.\nSpeaker A: Okay.\nNone: Thank you.\nSpeaker B: Oops.\nSpeaker B: Sorry.\nSpeaker B: That was delicious.\nNone: Yeah.\nSpeaker A: I think this looks like a chip.\nSpeaker E: It's quite similar to what it was before there.\nSpeaker D: Okay.\nSpeaker D: This stage of the evaluation is really for us as a team to evaluate if we have now got a product specification and prototype design that meets the criteria that we got from our market research.\nSpeaker D: So, this is the first stage of the evaluation.\nSpeaker D: Now, the collection of the criteria as we saw in our previous meeting was based on the user requirements and trends found in the marketing reports and marketing strategy of our company.\nSpeaker D: So, it's what we've discussed in the last meeting.\nSpeaker D: Are we actually meeting those trends and requirements?\nSpeaker D: Now, the findings that we came up with just to recap are here.\nSpeaker D: The criteria that we want in this remote control are a fancy look and feel, technological innovation.\nSpeaker D: It should be easy to use.\nSpeaker D: It should incorporate current fashion trends and those the two main ones that were the spongy texture and the fruit and vegetable strong design colors.\nSpeaker D: The design should minimize our eyes and be easy to locate.\nSpeaker D: And we were still a slightly ambivalent as to whether to use voice recognition there, but that did seem to be the favored strategy, but there was also on the sideline, the thought of maybe having a beeper function.\nSpeaker D: Okay.\nSpeaker D: So, we can come back to that slide if you don't have a note of those.\nSpeaker D: I'll just show you how we're going to evaluate our own feedback to this, to what we have so far.\nSpeaker D: We're going to use a seven point scale where one is true and seven is false.\nSpeaker D: We'll look at each of those criteria that I've just mentioned, I'll call that slide back up.\nSpeaker D: And I will just do a preliminary rating of all those criteria on the white board here.\nSpeaker D: Does that seem clear? Any questions there?\nSpeaker D: So, we're going to look at these.\nSpeaker A: Is it everybody's going to evaluate or just some of them?\nNone: Yes.\nSpeaker D: We're going to come to, we'll discuss each one and we'll come to our consensus rating between one and seven.\nSpeaker D: Okay.\nSpeaker D: One is true, seven is false.\nNone: Right.\nSpeaker D: So, I won't write all of that out again.\nSpeaker D: It will just be criteria one, two, three, four, five, six or A, B, C, D, E, F to confuse it with the number rating.\nSpeaker D: This is where I realize how tiny I actually am.\nSpeaker E: Just write, Molly.\nSpeaker E: Criteria.\nSpeaker D: And rating.\nSpeaker D: Actually, it might be an idea if we each did give our own individual rating, we could take an average at the end.\nSpeaker D: Yeah, so you can.\nSpeaker D: Okay.\nSpeaker D: So, I'll just separate the ratings by obliques and if we go one, two, three, four, we know who's here.\nSpeaker D: Okay.\nSpeaker D: Criteria A, the fancy look and feel.\nSpeaker D: How do we feel about this prototype model relating to fancy, is a fancy look and feel one is true, seven is false.\nSpeaker D: My own rating for that would be a two.\nSpeaker D: One is true and seven is false.\nSpeaker F: Yeah, I'd probably put it.\nSpeaker F: Two, yeah, two or three. No, three, three.\nSpeaker C: Okay.\nSpeaker C: Two.\nSpeaker A: I would say four.\nSpeaker D: Four.\nSpeaker D: Okay.\nSpeaker D: I think those up.\nSpeaker D: We've got six and five, eleven divided by four is what?\nSpeaker D: Two and three quarter.\nSpeaker A: Yeah.\nNone: Three.\nSpeaker D: Two and three quarter.\nSpeaker A: I think.\nSpeaker A: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Two point seven five. There we go.\nSpeaker D: Okay.\nSpeaker D: Criteria B, Criteria on B. Technologically innovative.\nSpeaker D: I would give that A three.\nSpeaker F: I'd give it a one.\nSpeaker D: Not that you're biased.\nSpeaker D: No, not at all.\nSpeaker D: The designer.\nSpeaker C: A two.\nSpeaker A: Oh, sorry.\nSpeaker A: I got it wrong.\nSpeaker A: The first one reading, I'm sorry.\nSpeaker A: Can you just make it two?\nSpeaker A: The average.\nSpeaker D: Yeah, I just thought you were rating to be a two.\nSpeaker A: Is that what you're saying?\nSpeaker A: Yeah, I just got.\nSpeaker D: So I'll work out the average for that again at the end.\nSpeaker D: It's a very slightly altered.\nSpeaker D: Okay.\nSpeaker D: I'm just waiting for you.\nSpeaker D: Two point five for that one.\nSpeaker A: One is a good choice.\nSpeaker D: So you're waiting for this one part?\nSpeaker D: Two.\nSpeaker D: Okay.\nSpeaker D: So that is eight.\nSpeaker D: That brings it down to two.\nNone: Two.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Ease of use.\nSpeaker D: Ease of use.\nSpeaker D: Based on what you've said there, I would say one.\nSpeaker D: Two.\nSpeaker F: Two.\nSpeaker F: Two.\nSpeaker E: I would say a two.\nSpeaker E: A two.\nSpeaker C: Two.\nSpeaker D: I should have said a two to make the arithmetic.\nSpeaker D: We'll just put almost two because we're not going to get into silly decimally places.\nSpeaker D: Okay.\nSpeaker D: We want to really bring it down.\nSpeaker D: Now we're looking at it incorporates current fashion trends.\nSpeaker D: And that's particularly in relation to our market research finding about the spongy texture to the exterior and the fruit and vegetable design colors.\nSpeaker D: Yeah.\nSpeaker D: Right.\nSpeaker D: So I'm just thinking before I give it my rating, you were limited in the use of materials for your prototype here.\nSpeaker D: Yeah.\nSpeaker F: Is this actually going to be the colors that you choose?\nSpeaker F: No, the base color was white or all light blue, but the changeable faces would allow you to get any basically any one of a number of colors.\nSpeaker F: It's full, full customer.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And the spongy feel is no problem with that.\nSpeaker F: No, no, because that's the spongy feel would be in the rubber that you put around it.\nSpeaker F: That otherwise is just sort of hard plastic.\nSpeaker D: In that case, it's got to be a one for me.\nSpeaker F: Yeah.\nSpeaker F: I'll give it a one as well.\nSpeaker C: Okay.\nSpeaker A: One.\nSpeaker D: That part was nice.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Moving on to does the design minimize repetitive strain and juice?\nSpeaker D: I don't think we've really touched on that.\nSpeaker D: We've discussed it, we haven't really come up with anything that we felt could feasibly reduce that.\nSpeaker D: We've talked about pointers, but the very use of a remote control, if you're someone who's zapping, who's sitting like that, we've found so many people did.\nSpeaker D: How do you minimize that in such a small device?\nSpeaker E: Well, the type of button that we're going to use in the bottom here, the material.\nSpeaker E: It's not my society, it's meant to.\nSpeaker E: Maybe because it's like with the size of the guy, it's quite small, but with the amount of stuff we're putting on it isn't that much.\nSpeaker E: So maybe because there's more space, not kind of moving around to accurately the buttons in between, it's quite obvious, just big buttons.\nSpeaker E: Right.\nSpeaker D: I think I'm going to have to be mutual on that and give it a four.\nNone: Yeah.\nSpeaker F: Yeah, I'd go for a five actually.\nSpeaker F: Okay.\nSpeaker C: Four.\nSpeaker C: Four.\nSpeaker C: Four.\nSpeaker C: Okay.\nSpeaker C: Twenty one.\nSpeaker C: Twenty one.\nSpeaker C: So that's four point two five.\nSpeaker B: Yep.\nSpeaker D: And finally, last but not least, easy to locate.\nSpeaker D: So we talked about voice recognition.\nSpeaker D: We talked about a deeper, we really haven't come to any.\nSpeaker F: Yeah, it was a voice recognition.\nSpeaker D: It was the voice recognition.\nSpeaker D: And we're happy with the costs.\nSpeaker D: Yeah, yeah, that's feasible.\nSpeaker D: That sounds good then.\nSpeaker D: I'd go for, we can adjust the volume on that just as we could volume on TV.\nSpeaker D: I think it would be a standard.\nSpeaker F: I think it would probably be a standard.\nSpeaker F: It would be a standard.\nSpeaker F: It would be quite loud.\nSpeaker E: So it would be built into, if you didn't hear it in the room, you were standing in and you realized it wasn't in that room.\nSpeaker D: Okay.\nSpeaker D: Logical.\nSpeaker D: That's the one for me.\nSpeaker D: Yeah, one.\nSpeaker D: One.\nSpeaker C: And one.\nSpeaker F: I do realize that we might be being fairly biased because it is a problem.\nSpeaker D: So how do we feel about this?\nSpeaker D: We've got the highest rating of meeting specifications that is definitely true for two of the six criteria there.\nSpeaker D: That is for incorporates current fashion trends and an easy to locate.\nSpeaker D: The lowest rating we've got, which is really, it's not terribly low, it's closer and neutral, is for the minimization of repetitive strain engineering.\nSpeaker D: Do we feel on the basis of these evaluation findings that we can go ahead and produce this as a prototype and market it, or do we have to make further modification?\nSpeaker E: I don't think so.\nSpeaker E: I think, yeah, I think.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: So I think for most of those rating that it's high enough, at the upper end of the scale for us to go ahead with that, and I really doubt if on the basis of current technology and our current capability, we could actually do much more than that.\nSpeaker D: And then we're having a huge break.\nSpeaker D: I know.\nSpeaker D: Okay.\nSpeaker D: And I hope that's clear to the team.\nSpeaker D: Is there anything you would like to ask me about the findings for a lot?\nSpeaker D: No.\nSpeaker D: Thank you.\nSpeaker D: I'm going to clean it now.\nSpeaker E: Okay. Thank you.\nSpeaker E: Okay.\nSpeaker E: I've got finance here.\nSpeaker E: I'm going to plug this in so you can all see it.\nSpeaker E: That's okay.\nSpeaker E: Oh, yeah.\nSpeaker A: We do want to plug that.\nSpeaker F: Yeah.\nNone: So this.\nSpeaker C: Right.\nSpeaker E: Now I presume.\nSpeaker E: That scream over that.\nSpeaker E: Okay.\nSpeaker E: We just type numbers into this and we come out with the final value.\nSpeaker E: So are we still on for kinetic?\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: So it would have been handy to have this at the beginning.\nSpeaker E: It might have been starting to start choice.\nSpeaker E: Right.\nSpeaker E: What's happening with the electronics?\nSpeaker F: There's regular chip on print and.\nSpeaker F: Oh, no, no, no.\nSpeaker F: The, um.\nSpeaker F: Yeah.\nSpeaker F: It's you, you've put in three for the number of kinetic cells.\nSpeaker F: This should just be one.\nSpeaker F: In the top, it's the number.\nSpeaker B: Oh, right. Okay.\nSpeaker E: Thanks.\nSpeaker F: Yeah.\nSpeaker E: So would it be two?\nSpeaker F: Yeah.\nSpeaker F: Just, no, one reg.\nSpeaker F: One chip.\nSpeaker F: Yeah.\nSpeaker F: One of them and one sample sensor and sample speed.\nSpeaker C: Okay.\nSpeaker F: Yeah.\nSpeaker F: And they're double curve. No.\nSpeaker E: One double curve.\nSpeaker F: Two because it's too low.\nSpeaker D: We don't have a flight.\nSpeaker D: We can fit.\nSpeaker E: So what's a single curve?\nSpeaker E: Yeah, I'd say, I'd say it was.\nSpeaker F: Yeah.\nSpeaker B: I think, I think it's.\nSpeaker F: So just one double.\nSpeaker E: One double curve.\nSpeaker F: And what no, because one.\nSpeaker F: Yeah.\nSpeaker F: One's double curved and then the other one is.\nSpeaker D: Yeah.\nSpeaker D: Plastic.\nSpeaker D: Plastic and rubber.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: That's tickle one and maybe a little one color.\nSpeaker A: No, I think rubber since it's being used just as a casing, we can put 0.5.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Because there are, I think it allows a 0.5.\nSpeaker A: We can use that.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: What does it mean if you put 0.5.\nSpeaker A: Yeah.\nSpeaker D: It means we are not using a lot of rubber, actually.\nSpeaker D: We are using this.\nSpeaker A: We like saying we're using a very low quantity of rubber compared to plastic.\nSpeaker A: See, it says case material.\nSpeaker E: So we're not actually using plastic in a case.\nSpeaker E: No, no, that's, it's as an extra.\nSpeaker E: It's a greeting.\nSpeaker E: It's a greeting.\nSpeaker F: So that shouldn't be actually on there because that's not incorporated in the cost of the remote you get.\nSpeaker F: But it is going to be part of the total question.\nSpeaker D: There's nowhere else we can, we can put that in as there.\nSpeaker D: There's nowhere else we can code with the rubber used in the case.\nSpeaker D: So do we not have to code put in the cost somewhere there?\nNone: Okay.\nSpeaker E: We'll just put it in as, we'll put it in as half.\nSpeaker E: We should just put it in as one because plastic is zero anyway.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker E: There we go.\nSpeaker E: Okay.\nSpeaker E: Special color.\nSpeaker E: Do we need that?\nSpeaker D: You might do.\nSpeaker E: I'm going to go for some of the more exotic ball regimes and such like color.\nSpeaker E: Into face.\nSpeaker F: Push button.\nSpeaker F: Yeah.\nSpeaker F: The push button is one and I'll see display one.\nSpeaker F: One.\nSpeaker F: One button.\nNone: One button.\nNone: One button.\nNone: One button.\nNone: One button.\nSpeaker A: I think we could change the battery also instead of going for kynoting how about going for the standard battery.\nSpeaker D: Is that not me Danny?\nSpeaker D: If you click off that square now, it doesn't need any difference.\nSpeaker D: Was that not?\nSpeaker D: No, it was 7.5, it changed.\nSpeaker D: So it's uncurbed completely and just actually making the rubber case the curve thing is not\nSpeaker A: going to make a difference. Uncode flat.\nSpeaker D: It's not me Danny, it's been a minute.\nSpeaker F: No, well I just surprised this one.\nSpeaker F: No, no, you've got to click off to calculate it again.\nSpeaker E: Okay, there we go.\nSpeaker E: It does slightly.\nSpeaker F: It might be assuming that that is in euros, it could be in dollars.\nSpeaker F: And then it would be fine because the exchange rate would make it a better point.\nSpeaker E: We haven't been dealing with dollars though.\nSpeaker A: I think it's the new one.\nSpeaker E: Okay, so the highest regard is the electronics here.\nSpeaker D: If we turn the interface, if we moved away from our much loved idea of a kinetic battery, just like in stone, the battery would make a huge difference.\nSpeaker F: Yeah, the standard, it would make one difference.\nSpeaker F: The biggest one would be taking away if you took away.\nSpeaker A: What I feel is customers never said anything about the battery.\nSpeaker A: It's internal, nobody looks into the battery.\nSpeaker A: But shape and color, that's something we should take away.\nSpeaker F: If you take away the voice recognition, then you've got it.\nSpeaker A: Where's that special form?\nSpeaker A: Do we see what difference it makes?\nSpeaker D: Where's the voice recognition?\nSpeaker F: It's samples sent to sample speaker.\nSpeaker F: It took away that they'll make it 12.35.\nSpeaker F: Well the kinetic is three.\nSpeaker E: If we change it to the battery, that's minus three.\nSpeaker F: Yeah, but you've minus three plus two.\nSpeaker E: Oh, right.\nSpeaker E: We can do it some other way.\nSpeaker E: It doesn't have to be voice recognition.\nSpeaker E: We could do the voice recognition for business class or something.\nSpeaker E: Like an upgraded version.\nSpeaker E: You could choose to have that.\nSpeaker F: But I mean, I don't see why we have to set up 25 years now because they were saying they were quite willing to pay more for a better product.\nSpeaker D: We're still working to...\nSpeaker D: So should we just change it to head off?\nSpeaker D: Head off.\nSpeaker D: Head off.\nSpeaker D: We're working to head off the specifications.\nSpeaker D: This is what this project team is working for.\nSpeaker D: We can do it in our recommendations with what we've found and the consensus that we've come to as a lot of meetings.\nSpeaker D: We can.\nSpeaker D: But we need to work to that specification to start.\nSpeaker D: Yeah, we can say.\nSpeaker D: The voice recognition sounds wonderful, but our object is so distinctive that that in itself is going to make it easier to locate as an in our first instance.\nSpeaker D: As you say, we can offer the voice recognition initiative to business class customers.\nSpeaker E: Okay, we can.\nSpeaker E: Yeah, I make the price fit and then say, we recommend.\nSpeaker E: We would have had this because it also sets up as a product.\nSpeaker E: Because we don't have a background.\nSpeaker E: Yeah.\nSpeaker E: Because someone can be different associated with your company.\nSpeaker E: Right, okay.\nSpeaker D: So if we take voice recognition, I...\nSpeaker F: Spell the 12.35.\nSpeaker F: Is it 12.50?\nSpeaker E: No, it's 12.25.\nSpeaker E: Are you sure?\nSpeaker D: Sorry.\nSpeaker E: Not meaning to doubt your work.\nSpeaker E: 12.25 times two years.\nSpeaker E: 25.\nSpeaker E: Are they going to be...\nSpeaker D: It's 12.5.\nSpeaker D: I didn't.\nSpeaker D: I'm quite zero, one over euro.\nSpeaker D: Which is less than 10.\nSpeaker A: I don't remember, but it's said 50% is the cost.\nSpeaker A: So half of the price was...\nSpeaker E: The other one was where the price was not there.\nSpeaker A: I don't know.\nSpeaker C: What is the first one?\nSpeaker E: That's two days.\nSpeaker A: 12.5?\nSpeaker C: Oh, so we are under the position.\nSpeaker C: So we're okay.\nSpeaker E: Okay.\nSpeaker E: So, is that...\nSpeaker E: Got us covered for the electronics then.\nSpeaker E: We don't need something else to take that place.\nSpeaker F: Nope.\nSpeaker F: So that was just a bolt on X-ray.\nSpeaker F: You could just take that out.\nSpeaker F: And it would be fine.\nNone: Okay.\nSpeaker A: It seems fine.\nSpeaker A: Yeah, I think it was.\nSpeaker A: X-A-Boys are good mission.\nSpeaker A: Everything is...\nSpeaker E: All right.\nSpeaker E: Done that.\nSpeaker E: Okay.\nSpeaker E: Project evaluation.\nSpeaker E: We've done room for creativity, haven't we?\nSpeaker E: Yeah.\nSpeaker E: Got decisions that we would have made that we weren't.\nSpeaker E: Didn't feel able to make.\nSpeaker E: Appaging we discuss leadership and teamwork?\nSpeaker F: Yes.\nSpeaker E: In the sense that...\nSpeaker E: Did you feel like a team or did you feel like...\nSpeaker F: All right, so it's a team.\nSpeaker D: I think we're talking.\nSpeaker D: There was so much interaction, so much that we needed to...\nSpeaker D: Find out, come together.\nSpeaker D: And I've certainly felt heard, listened to, and that in relevant areas, we've been able to give and take, and adjust our remit.\nSpeaker D: Yeah.\nSpeaker D: We're in SSO.\nSpeaker A: Yeah, I think it is a very constructive feedback by everybody.\nSpeaker A: It's not like people trying to cut each other.\nSpeaker A: It is more of a through-cohesivity move.\nSpeaker A: We came to a very pretty creative design.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: And...\nSpeaker A: Yeah, I think...\nSpeaker A: Sarah, you coordinated the work very well.\nSpeaker A: Thank you.\nSpeaker E: How did you find it?\nSpeaker F: Yeah, no, I thought it went really well.\nSpeaker F: And I, yeah, I feel that everyone was listened to, and all the points that were raised seem to have been sorted out.\nSpeaker F: But they didn't quite make the voice recognition in there, so that's good.\nSpeaker F: Yeah.\nSpeaker F: I think it would have made it into the final product if it was actually.\nSpeaker F: If we'd seen the marketing before the initial specification was put out, I think maybe it would have come out a little different.\nSpeaker E: Yeah.\nSpeaker E: I like the fact that we could say an idea, and it'd be...\nSpeaker E: I suggested that there wasn't the best idea, but no one felt my short time.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: You know, it didn't matter, saying what you thought, because if it wasn't something that was relevant, then it didn't matter, because it was just another idea.\nSpeaker E: Mm.\nSpeaker E: Um...\nSpeaker E: Teamwork, well, you two created that wonderful specimen of a plaido model.\nSpeaker E: Yeah.\nSpeaker A: I think that was the best part of that.\nSpeaker A: Yeah.\nSpeaker E: Uh...\nSpeaker E: I'm still not caught up.\nSpeaker E: That's nothing to do with the teamwork, but...\nSpeaker D: Um...\nSpeaker D: Maybe we should think of branching out into children's toys, maybe mobile phones or something like that.\nSpeaker D: If they go on...\nSpeaker E: Yeah, that was quite fun.\nSpeaker E: Um...\nSpeaker E: Means whiteboard digital pens, etc.\nSpeaker E: What does that mean?\nSpeaker E: Um...\nSpeaker D: I'd be evaluated for materials we had for communicating and sharing information.\nSpeaker D: Yeah.\nSpeaker D: Could it have been better? Was it adequate?\nSpeaker F: Yeah, I think we probably would have used the actual email system more had we actually...\nSpeaker E: Had time to kind of...\nSpeaker F: Had more time, and if we'd been separated more.\nSpeaker E: Yeah, because we could just sort of say, sorry, what did you say about that, or what do you think about that, rather than having to email it?\nSpeaker E: Yeah.\nSpeaker E: But it was nice having it there.\nSpeaker E: Um...\nSpeaker E: Like the whole picture of the thing.\nSpeaker A: Yeah, but in like moving around a little.\nSpeaker A: I think it's good, like...\nSpeaker A: We spend time individually...\nSpeaker A: I never thought of a remote control with a flip top.\nSpeaker E: Yeah, I think it's really boring to other areas.\nSpeaker D: It's bringing things from other areas in.\nSpeaker D: I mean, nothing in me.\nSpeaker D: But it's applying it to a different area.\nSpeaker D: Vegetables.\nSpeaker A: It's mine.\nSpeaker A: Yeah.\nSpeaker A: Yeah, no.\nSpeaker A: Yeah, no.\nSpeaker E: Yeah, no.\nSpeaker E: Well, they have to come from somewhere.\nSpeaker E: Absolutely, yeah.\nSpeaker A: But they...\nSpeaker A: As...\nSpeaker E: Sorry, guys.\nSpeaker E: Um, as we're shown by your presentation with all of the other controls, remote controls, no one's thought about it, particularly.\nSpeaker E: I mean, they're slightly different.\nSpeaker E: So, someone's been shot in a room and said, make ours a bit different from everyone else's, because there's something we should think about.\nSpeaker E: But obviously, no one's put any great deal of thought into it.\nSpeaker F: I don't think the companies are really concerned.\nSpeaker F: They're just, like, we've got a DVD player.\nSpeaker F: Does anyone remember the remote control?\nSpeaker F: Can't you get the one that we used for the last one?\nSpeaker F: Yeah, just jazz it off of it.\nSpeaker F: They don't really think about it, because normally, the remote control isn't the product, which they're trying to ship.\nSpeaker F: They're trying to ship the DVD player, the video player, the TV.\nSpeaker E: But then when everything is really smart...\nSpeaker E: Yeah.\nSpeaker E: And you've just got this big chunk of black thing sitting on your coffee table.\nSpeaker E: It doesn't go.\nSpeaker E: If you could have something that's proper, funky thing, a funky item that's individual.\nSpeaker E: Individual to you.\nSpeaker E: I mean, they could even go into...\nSpeaker E: You go in with an idea, and they design it on Photoshop, and then they just get it printed on the plastic.\nSpeaker E: You pay a lot extra because it's individually being plastic was being made for you.\nSpeaker E: Yeah.\nSpeaker E: But people couldn't have anything that they wanted.\nSpeaker A: Surprising to be just like people give a lot of attention to modifying mobile phones.\nSpeaker A: Like every month you see a new launch or something like that.\nSpeaker A: And new television products coming up.\nSpeaker A: But nobody giving a much idea to this.\nSpeaker D: I think this has been very market research-based.\nSpeaker D: Because just going back to mobile phones, I mean, this is the first change in remote control devices that I can really say is obvious and visible.\nSpeaker D: We see it in mobile phones a lot, and that's where we've heard a lot of ideas from.\nSpeaker D: There are innovations in that that people don't really want.\nSpeaker D: I see people wanting a model of the phone that really helps you with it, and they can't get it anymore.\nSpeaker D: It's innovation for innovation's sake.\nSpeaker D: And I think it's wonderful that our companies do R&D based.\nSpeaker E: Well, it's innovation for money's sake.\nSpeaker D: People have to keep up.\nSpeaker D: The forcing it on to people.\nSpeaker D: Yeah.\nSpeaker D: Things that, you know, they might want to buy the thing they do.\nSpeaker E: And you can't get...\nSpeaker E: You've got your hands at the works fine, but you can't get battery anymore.\nSpeaker E: And that type of phone is the phones that have been done.\nSpeaker D: Oh, there's a cover to fit it on the wire.\nSpeaker D: Yeah.\nSpeaker C: That's nice.\nSpeaker C: And...\nSpeaker E: In closing...\nSpeaker E: Closeable.\nSpeaker E: Um...\nSpeaker E: Our costs are below budget, with recommendations that the budget be increased.\nSpeaker E: But I think...\nSpeaker E: I don't know.\nSpeaker E: What do you think?\nSpeaker E: Do you think that takes into account, um...\nSpeaker E: Overheads.\nSpeaker E: Like, us being well paid for the thing.\nSpeaker E: And the heating for the building.\nSpeaker E: Do you think our budget includes everything, all the costs that are going out?\nSpeaker E: I think it was just the product.\nSpeaker E: I think that was just the physical.\nSpeaker E: So it looks, well, it looks like it's going to be 50% profit.\nSpeaker E: Yeah.\nSpeaker E: Still gone.\nSpeaker E: All of the overheads might have been.\nSpeaker E: So maybe increasing it.\nSpeaker E: You would also have to increase the price today.\nSpeaker E: So don't.\nSpeaker E: Yeah, I think...\nSpeaker E: But you were saying that that's correct.\nSpeaker E: Yeah, I don't.\nSpeaker D: I think in the remake it's very specific.\nSpeaker D: I think we've done what we required to do.\nSpeaker D: And I think like the other project team, or a new project for us to look at those kind of things.\nSpeaker D: I don't think that's something we have to look at.\nSpeaker D: No.\nSpeaker D: And the way we're breathing the cash for.\nSpeaker F: Yeah.\nSpeaker D: It was...\nSpeaker D: We've done very well to get within budget.\nSpeaker D: And we still make such an innovative item.\nSpeaker D: Yeah.\nSpeaker D: So I think people are really going to want to...\nSpeaker E: To shame it won't ever get me.\nSpeaker E: Maybe we're going to steal our ideal balance.\nSpeaker D: It's top secret.\nSpeaker F: It's conspiracy going on here.\nSpeaker E: The project has been evaluated well and truly.\nSpeaker E: Yeah.\nSpeaker E: And we've got the meeting summary in question here.\nSpeaker E: There's a final question.\nSpeaker E: So we've got about 50 minutes to catch up with everything that I felt behind.\nSpeaker E: So I'll be delegating it.\nSpeaker E: In your last project, superbising.\nSpeaker E: So good luck.\nSpeaker A: Celebration, you didn't talk about that.\nSpeaker E: I think it was celebration at the 25th time.\nSpeaker E: That will be dance on the dance.\nSpeaker E: That was...\nSpeaker F: Okay.\nSpeaker F: Was that it then?\nSpeaker E: Yeah.\nSpeaker E: Just the last I think.\nNone: Awesome.\nSpeaker E: I think I've got minutes for two of the meetings and...\nSpeaker E: The last thing...\nSpeaker E: It's a question I've done.\nSpeaker D: Well, I didn't have to rate you on how much you include the meeting this time.\nSpeaker F: Is that not the question asked or the how you thought you'd done in your...\nSpeaker F: Thing?\nSpeaker D: I'm right.\nSpeaker D: But do we not sometimes evaluate the meeting?\nSpeaker D: Yeah, there's this.\nSpeaker F: Yeah.\nSpeaker F: I think...\nSpeaker F: Yeah, question is seven years.\nSpeaker F: So I'm going to go to the 40 meeting.\nSpeaker B: Yeah.\nSpeaker E: I just got to...\nNone: Okay.\n"}, {"meeting_id": "IS1009c", "summary": "After Project Manager recapped the decisions made in the last meeting, such as having a child lock and display clock, User Interface started the presentation regarding how to make the interface more attractive to customers. Then, Industrial Designer gave the presentation on the component design of the remote, which was divided into two parts - the components in the remote and the working design. Afterwards, Marketing reported findings for marketing like the people wanting fewer buttons which the functions were obvious. In the end, Project Manager wrapped up the meeting, concluding what had to be done by each of them for the next meeting.", "dialogue": "Speaker E: Any good here?\nSpeaker E: Okay. Hello everyone.\nSpeaker E: Hi.\nSpeaker E: How are we doing?\nSpeaker E: Yeah, good.\nSpeaker E: First, we're going over the minutes of the last meeting, more or less.\nSpeaker E: In the last meeting, the marketing manager had presented her method of working, meaning gathering suggestions from everyone to see how she best could market this product within the budget that was given.\nSpeaker E: In general, the idea is that it should be something that is not difficult to use.\nSpeaker E: It's also an item that people lose a lot, so we should address that.\nSpeaker E: And of course it should be something that is very simple to use.\nSpeaker E: In addition to that, to make it sell, of course, the marketing manager wishes that it be very attractive, or like she says, put some sizzle into it in one way or another, so that the people are buying it now because in particular with smaller items, that's a very important fact, because if they say, well, I go home and think about it, that won't work.\nSpeaker E: Also mentioned was it should have a very short learning curve, and maybe it could be sold by using a slogan.\nSpeaker E: Our technical manager has then said that he feels it should have a chip that has infrared bits, and it has an interface that controls the chip. Therefore messages will be controlled in the same manner.\nSpeaker E: There should be extra features like lid buttons, maybe a beep.\nSpeaker E: If too many buttons are pressed, a child lock, and maybe a display clock, so that people could see the time, you know, what show they want to watch.\nSpeaker E: Also mentioned was maybe different shapes.\nSpeaker E: So the components of the thing should be button, bulbs, infrared bulbs, battery chips, wires, and maybe some kind of a holder for the item.\nSpeaker E: Pranchina, who is our interface designer, has mentioned that the, of course, should have an on-off button, and also has mentioned an interesting feature that it should have maybe a channel lock, particularly with maybe small children that they couldn't watch a channel that is undesirable.\nSpeaker E: It should be compact, her personal favorite was it should be T-shaped, and maybe having an alarm clock and the material should possibly be of non-allergic nature.\nSpeaker E: The different systems that exist are infrared or radio waves, maybe it should have electricity saving feature, and even possibly a timer so that people can program their favorite program on right from the remote.\nSpeaker E: Are we all in agreement that that's bad?\nSpeaker E: What we discussed last time?\nSpeaker E: Okay, I got pretty much the same.\nSpeaker E: Okay, then we are looking for three presentations, and I don't know whether the order matters much.\nSpeaker E: I don't think so.\nSpeaker E: So whoever wants to...\nSpeaker E: I can stop.\nSpeaker E: Yeah, okay.\nSpeaker B: Am I asleep, please?\nSpeaker E: Okay.\nSpeaker E: Come on, close already.\nSpeaker E: And that's number two, right?\nSpeaker E: Three.\nSpeaker B: Three.\nSpeaker B: But it's been three.\nSpeaker B: Yes.\nSpeaker F: Okay.\nSpeaker B: Now as an interface designer, I would give more emphasis on the interface, how the remote looks like, so that it is sellable, it is attractive to customers.\nSpeaker B: Next, please.\nSpeaker B: Okay, now the function of a remote is to send messages to the television.\nSpeaker B: This messages could be a switch on-off message, or switch to next channel message, or swapping the channels, or switching on to a particular channel.\nSpeaker B: Like you can have the numbers one, two, three, four, up to nine.\nSpeaker E: Nine, what, nine channel switches?\nSpeaker E: Pardon me?\nSpeaker B: Nine channel switches?\nSpeaker B: Yes.\nSpeaker B: Nine numbers.\nSpeaker B: And then you have swapping button, by which using which you can swap the channels.\nSpeaker B: If you don't want to see the third channel, you can swap it to the fourth channel, or vice versa.\nSpeaker B: Then it should have a next button, a next button channel, by which you can keep on scrolling the channels one by one.\nSpeaker B: Next slide, please.\nSpeaker B: Then you should have a button which can be used for increasing or decreasing the volume.\nSpeaker B: Then there should be a button which can give subtitles for a particular program, which is going on a television.\nSpeaker B: For example, if you are watching a French program and you would like to have a subtitles in English, then there should be a channel which can trigger this mechanism in the television, so that the subtitles on the screen.\nSpeaker B: Then there should be some buttons which can control features like the color of the picture, the contrast, sharpness, brightness of the picture.\nSpeaker B: Now there should be a memory switch.\nSpeaker B: There should be a mute button.\nSpeaker B: Suddenly if a viewer gets a telephone call, and if he doesn't want to switch off the TV, but he can reduce the sound.\nSpeaker B: He can bring the volume down and he can watch, while talking he can watch the TV.\nSpeaker B: Now the most important feature I would like to have in my remote would be the speech recognition feature.\nSpeaker B: It's an integrated programmable sample sensor speaker unit.\nSpeaker B: So remote can be designed which can have the voice recognises.\nSpeaker B: You can record your own voice, which can be recognised by a voice recogniser in the television.\nSpeaker B: For example, if you want to see the ninth channel, if you say just say ninth channel, the remote will automatically switch to the ninth channel.\nSpeaker B: So the TV will have some recognises which will recognise the user's voice, and accordingly it will change its functionality.\nSpeaker B: So this is one of the very important features a remote control can have.\nSpeaker B: So this is one of the interface which can be created.\nSpeaker B: A very simple interface which has all the important features.\nSpeaker B: Then these are some of the remotes which are different in shape and colour, but they have many buttons.\nSpeaker B: So sometimes the user finds it very difficult to recognise which button is for what function and all that.\nSpeaker B: So you can design an interface which is very simple and which is user friendly.\nSpeaker B: Even a kid can use that.\nSpeaker B: So can you go on to the next slide?\nSpeaker B: Yeah.\nSpeaker B: So this is one of the interface or one of the remote which has this voice recogniser.\nSpeaker B: And this has multi-purpose use.\nSpeaker B: It can be used for TV, it can be used for cable satellites, it can be used for VCR DVDs and audio.\nSpeaker B: And this has inbuilt voice recogniser.\nSpeaker B: Can you go on to the next slide?\nSpeaker B: Yeah.\nSpeaker B: Now this is an interface for a remote which a child can use.\nSpeaker B: This is user friendly, it's very attractive and children can use it as well as they can play with it.\nSpeaker B: And this comes with different colours, different shapes.\nSpeaker B: And this child interface has minimum buttons and all the important buttons are there in this small compact attractive child interface.\nSpeaker B: Next slide please.\nSpeaker B: Now this is a big oversized remote which cannot be misplaced or it's impossible to display.\nSpeaker B: I'm sorry.\nSpeaker C: You don't know me.\nSpeaker C: I can lose that in a minute.\nSpeaker C: This is a very big container.\nSpeaker B: Let's play with it anyway.\nSpeaker B: So this is a Jambu universal remote control and it's impossible to misplace or lose.\nSpeaker B: This is one such interface which can be created.\nSpeaker B: And the personal preference would be to incorporate speech recognises which will respond to users voice for a particular function.\nSpeaker E: Right.\nSpeaker B: Thank you.\nSpeaker B: That's it.\nSpeaker E: Okay. Thank you very much.\nSpeaker E: Any comments on her presentation?\nSpeaker D: Well, looks like we still have quite a choice of things out there.\nSpeaker D: I'm no suggestion is bad but we're going to have to narrow it down a little more.\nSpeaker D: I don't think that we can get the T-shape is good, the child one is good.\nSpeaker D: It's a too big to misplace, I think it's just funny.\nSpeaker D: I don't think that's going to be our impulse purchase at the checkout counter.\nSpeaker E: No, I think these are her presentations but as far as the decision making we get to that.\nSpeaker E: I'll come back to that later.\nSpeaker E: Okay.\nSpeaker E: But if I just wanted to know whether anybody had anything to add to her presentation.\nSpeaker D: No, I think her presentation was good and she really explored all the options.\nSpeaker E: Who wants to go next?\nNone: Okay.\nSpeaker A: The next one.\nSpeaker A: Oops.\nSpeaker E: The components design.\nSpeaker E: Okay.\nSpeaker A: This time I'm going to concentrate more on the components and the technical side of the remote controller design.\nSpeaker A: Can we go on to the next slide please.\nSpeaker A: I have just brief down a few components which we require for the remote control construction.\nSpeaker A: The first one is case to keep all the components like integrated circuit, battery etc.\nSpeaker A: It can be a plastic one, hard plastic so that it can be strong even if it falls down then it doesn't break.\nSpeaker A: So it should be strong and there are no harmful materials used in that and it should be recyclable.\nSpeaker A: And also using of coloring compounds like if we want to have different colors blue, red, green.\nSpeaker A: So we have to use some coloring component components.\nSpeaker A: And the second important thing is integrated circuit.\nSpeaker A: We can use a highly sophisticated one because it's like the heart of the remote controller.\nSpeaker A: If it is not efficient then everything is going to be less efficient.\nSpeaker A: So we should have a highly sophisticated one and it should be resistant to high as well as low temperatures.\nSpeaker A: Suppose if it is 38 degrees outside, 40 degrees outside it should be able to resist the temperature high temperature and low temperatures.\nSpeaker A: And it should be equipped with timer and alarm facility.\nSpeaker A: And the other component we have in the remote controller is a resistor which is very much important for the electricity flow through the remote controller.\nSpeaker A: And also a capacitor which is a battery capacitor. Can you go to the next slide please?\nSpeaker A: A diode, a transistor, a resonator. These are all the technical electronic components which we have to use in a remote controller.\nSpeaker A: If we need to make a rechargeable battery then we need not go for a high performance battery.\nSpeaker A: Even if it is a low performance battery it can't charge much. It's not a high voltage battery.\nSpeaker A: If it is a rechargeable one then people can use it for a long time.\nSpeaker A: So in that way we can cut the cost but that we have to make the battery as rechargeable one.\nSpeaker A: And we have a circuit board in a remote controller. Can you go to the next slide please?\nSpeaker A: And how it works, how the remote controller works.\nSpeaker A: When you press a button...\nSpeaker E: Go away.\nSpeaker A: When you press a button, when you do that you complete a specific connection.\nSpeaker A: That means when you press a button there will be a small circuit underneath the button and it will send some signals through the wires.\nSpeaker A: And then the chip will send that connection and knows that which button is pressed.\nSpeaker A: Suppose you have pressed channel 1 button, number 1 you have pressed.\nSpeaker A: Then the chip will know that the number 1 button was pressed. It produces a more score line signal specific to that button.\nSpeaker A: Every button, every individual button has its own more score.\nSpeaker A: Suppose the user has pressed button 1 then it will have a circuit will generate a specific more score to that button.\nSpeaker A: And the transistor will amplify the signal and send them to the LED which translates the signal into infrared light.\nSpeaker A: Like you have got a signal by pressing a button that more score has been generated by the integrated circuit.\nSpeaker A: Now that signal that more score has to be amplified by the transistor.\nSpeaker A: That is the use of transistor which we use in the remote controller.\nSpeaker A: It will amplify the signal and it will send it to the LED and which translates the signal into an infrared bits.\nSpeaker A: The sensor in the TV can see the infrared light and seeing the signal, it reacts appropriately.\nSpeaker A: That when it sees the amplified more score signal then it will know which action it has to do.\nSpeaker A: Then it will do the appropriate action.\nSpeaker A: So this is how the remote controller works.\nSpeaker A: Can I go to the next slide please?\nSpeaker A: I have few pictures when you look at the remote controller.\nSpeaker A: This is a normal remote controller.\nSpeaker A: And to the next slide please.\nSpeaker A: And we had a few buttons and all.\nSpeaker A: And if you open the remote controller you have this circuit board and few electronic components.\nSpeaker A: Like you can see a chip there which is having 18 pins and also a capacitor, 3 resistors and also resonator.\nSpeaker A: And the diode transistor.\nSpeaker A: All of the electronic components have all those things like chip and diode transistors.\nSpeaker A: You can see the TA, double 1, 8, 3, 5 labeled as chip.\nSpeaker A: You can also see the green things are resistors and just beside that you can see a transistor.\nSpeaker A: And the cylinder shape that one is a capacitor and also there are resistors.\nSpeaker A: Sorry, there is a diode.\nSpeaker A: Can you go on to the next slide?\nSpeaker A: So this is the circuit board.\nSpeaker A: In one is a circuit board.\nSpeaker A: A building a circuit board is pretty easy and also inexpensive.\nSpeaker A: It costs less than what you print on a paper.\nSpeaker A: Because when you are building some circuits and also wires it is better to go for printing.\nSpeaker A: Because you can build these kind of circuit boards on a bulk and it is just printing.\nSpeaker A: Nothing like you know you do not need to use wires and all.\nSpeaker A: It is not exactly wires we are using.\nSpeaker A: It is just printing something on a board.\nSpeaker A: Those prints will act as wires and circuits.\nSpeaker A: So that green thing is a circuit board.\nSpeaker A: And also you can see there are like access for buttons.\nSpeaker A: Like when you press a button the circuit under the button will be activated.\nSpeaker A: It will get some signals from it and it will send the signals to the integrated circuit.\nSpeaker A: Can you go to the next slide please?\nSpeaker A: So these are the circuits underneath the buttons.\nSpeaker A: Can you see the black round marks?\nSpeaker A: They are the circuits.\nSpeaker A: Next please.\nSpeaker A: And like we have designed before we have seen some few things like instead of buttons we have some scrolls.\nSpeaker A: But a push button requires a simple chip underneath it.\nSpeaker A: But whereas a scroll wheel requires normally a regular chip which is a higher price range.\nSpeaker A: Like for scrolls we have to go for a sophisticated and full complete chip.\nSpeaker A: And as energy source we offer a basic battery.\nSpeaker A: More engineers are hard dynamo.\nSpeaker A: A kinetic provision of energy.\nSpeaker A: Modern watches that you shake casually to provide energy.\nSpeaker A: So with that also we can have in a battery or we can use solar cells.\nSpeaker D: That's interesting.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: The product can be delivered into different cases.\nSpeaker A: Usually the cases uncarved flat that we see usually normal remote controller.\nSpeaker A: Okay.\nSpeaker E: Yeah.\nSpeaker E: You have more pictures.\nSpeaker E: We have five minutes to the end of the meeting.\nSpeaker A: Maybe I have just one more slide I think.\nSpeaker A: Like we can have materials such as plastic, rubber wood, titanium but titanium we can't use.\nSpeaker A: And also for electronics we can use a simple and regular or an advanced chip on the print.\nSpeaker A: Also it includes infrared sander.\nSpeaker A: And for the movie just to develop samples and so to speak sample speaker.\nSpeaker A: Yeah.\nSpeaker A: That's it.\nSpeaker A: That's all for me now.\nSpeaker F: Okay.\nSpeaker F: Thank you.\nSpeaker E: Any particular comments by anybody?\nSpeaker D: On the scroll on the push button.\nSpeaker D: Can you can achieve scrolling by repeatedly pushing a button?\nSpeaker A: No.\nSpeaker A: The scrolling was different.\nSpeaker A: Like you can go for switches like buttons or scrolls which we used to do before 10 years before I think now nobody uses it because you need a sophisticated chip and all.\nSpeaker A: So I think it's better we go for ordinary buttons.\nSpeaker D: We'll just go for push buttons for interest of cost.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: We don't have a lot of time left and we will still have to make a decision at the marketing.\nSpeaker E: Expert has to present her.\nSpeaker E: Okay.\nSpeaker F: You're right in my next slide.\nSpeaker F: Okay.\nSpeaker D: All right.\nSpeaker D: My method is I'm interested in what the competition is doing and want to see how we can make ourselves different from the competition.\nSpeaker D: So I've really been looking at the press and the ads that are out there for other remote controllers.\nSpeaker D: I spend a lot of time on the internet surfing around doing the same thing.\nSpeaker D: And then when I'm out in people's houses or meetings or anything like that.\nSpeaker D: I try to notice what kind of remote controls people have.\nSpeaker D: And if it's convenient in the conversation I ask about it.\nSpeaker D: And I would encourage you all to do the same.\nSpeaker D: And my findings from this is that, you know, small is beautiful.\nSpeaker D: People like something that really fits in their hand.\nSpeaker D: Simple is beautiful.\nSpeaker D: They don't want to have to squint at small print.\nSpeaker D: They want buttons whose functions are obvious and they want as few buttons as possible.\nSpeaker D: And they don't care for the mode thing.\nSpeaker D: They want each button to do something.\nSpeaker D: And eye catching is important.\nSpeaker D: It's got to look cute.\nSpeaker D: It's got to look appealing.\nSpeaker D: Go ahead.\nSpeaker D: I'm trying to finish fast for you.\nSpeaker D: And our preference is as far as I'm concerned, we've got to get to the market before the competition.\nSpeaker D: Ours has to be look really great and it has to come out before the others so that we have a leg up on time to sell it and push it before other people get out there Christmas item.\nSpeaker D: And we should develop one or two features we can really dwell on in our ad campaign.\nSpeaker D: If we try to tell people it has too many great features, the consumer just gets confused and we don't get anywhere.\nSpeaker D: So we've got to narrow our selection down to two things I think that we're going to say are really great about our new product.\nSpeaker D: And I've been looking around at what designs every year different things are popular.\nSpeaker D: And in my research this year I found out that fruit and vegetable shapes are really popular.\nSpeaker D: And people are tired of hard plastic and hard metal.\nSpeaker D: They're more back into soft feel, spongy feeling things, things with maybe a little cloth on them.\nSpeaker D: So those are things we want to look at as far as saleability of the item.\nSpeaker E: Okay, that concludes the presentation of everyone.\nSpeaker E: And what we really have to decide in this meeting is the concept of the remote.\nSpeaker E: And so what do we think on the concept of the remote?\nSpeaker D: You want to try to come back to yours?\nSpeaker B: Yes, I would like to include this feature which is called as voice recognizer.\nSpeaker B: Okay.\nSpeaker D: So speed tracking is most important as far as you're concerned?\nSpeaker B: This could be one feature which could be sellable.\nSpeaker D: Yeah, that could be our star feature. That might be really good.\nSpeaker D: I agree with that.\nSpeaker A: But I'm very much doubtful how far it will work because speech recognizer has its own...\nSpeaker A: Distance problem?\nSpeaker A: Yeah, it's not distance problem.\nSpeaker A: It's recognizing a person's voice.\nSpeaker A: Like maybe different people will be having different voices.\nSpeaker A: Okay, like every time you say something.\nSpeaker A: You have to teach.\nSpeaker B: I get a good recognizing system.\nSpeaker B: No, it's like you're recording a question already.\nSpeaker B: And then you're expecting an answer from...\nSpeaker B: For example, you have a TV system. I'm the user.\nSpeaker B: And my family members are the user.\nSpeaker B: I've already recorded a question like good morning, like around 8 o'clock.\nSpeaker B: I want to see the news in the television.\nSpeaker B: So I'll say this good morning and the TV will switch on.\nSpeaker B: It will recognize my voice.\nSpeaker E: Okay, before we get too far off here, the components of the concept is the energy.\nSpeaker E: What kind of energy do we foresee?\nSpeaker D: I think we all agreed on that.\nSpeaker D: That's going to be most cost effective and the best thing.\nSpeaker E: Okay, then chip on print.\nSpeaker E: Yes.\nSpeaker E: Okay.\nSpeaker E: And the case. And I think we all agree on the case.\nSpeaker E: We want to have something maybe bright, colorful.\nSpeaker E: Yeah.\nSpeaker D: Bright, colorful, trendy design and compact.\nSpeaker E: Do you need design, compact and strong?\nSpeaker E: Yes.\nSpeaker E: User interface concept, interface type supplements.\nSpeaker E: I will be your area.\nSpeaker E: I think.\nSpeaker E: Yeah.\nSpeaker A: Like the switches, like we use buttons for user interface.\nSpeaker A: Push buttons.\nSpeaker E: Push buttons.\nSpeaker E: I guess for, yeah.\nSpeaker E: And I'm not sure what they mean by supplements.\nSpeaker E: Well, I think.\nSpeaker B: There's like different types of features buttons.\nSpeaker C: Yeah, or like, or like her speech recognition, it would also be a supplement.\nSpeaker C: Her speech recognition.\nSpeaker C: Right.\nSpeaker D: Right.\nSpeaker D: So why don't we put down speech recognition, if possible, pending some more research from our industrial engineer on how expensive that is.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker E: Our next meeting will be in 30 minutes.\nSpeaker E: And the ID is to, this is the individual actions to be taken until then.\nSpeaker E: To have the look and feel design.\nSpeaker E: And the UID is supposed to come up with the user interface design.\nSpeaker E: Okay.\nSpeaker E: And the marketing expert with the product evaluation.\nSpeaker E: Okay.\nSpeaker E: In this phase, the two of you, Yana and Francine, have to work together on a prototype using modeling clay.\nSpeaker E: It says here.\nSpeaker E: You will receive specific instructions will be sent to you by your coaches.\nSpeaker E: Okay.\nSpeaker E: If you have any questions, you know, you can always contact me or your coach, I suppose.\nSpeaker E: We're, they're hiding.\nSpeaker E: So, I think we conclude the meeting here and we come back in 30 minutes.\nSpeaker E: Okay.\nSpeaker E: According to our timetable here.\nSpeaker F: Thank you very much.\nNone: Thank you.\nNone: Thank you.\nNone: Thanks.\n"}, {"meeting_id": "IS1004d", "summary": "Project Manager started meeting on the detailed design of the remote control. User Interface and Industrial Designer gave a presentation on the prototype with a banana leaf base station and a hand-sized remote with two scroll wheels, the turbo button, the teletext button, the calling button, the rechargeable-batteries power source, and the speaker regular chip. Next, the group evaluated based on Marketing's list of user requirement criteria. Then, groupmates discussed the cost estimation, including the component production cost. They agreed to use plastic instead of rubber material for the dual chips. Also, they agreed not to use LCD and no button supplements to avoid over budget. Lastly, the group discussed the project process. The group agreed they had creativity, while User Interface suggested that they should find out more target markets. Besides, Project Manager suggested doing a street survey. Also, User Interface suggested that they could simplify the interface and the circuit board.", "dialogue": "Speaker B: Okay, good after this again.\nSpeaker B: So we should have our final meeting about the detailed design of the product, although not control.\nSpeaker B: So there is the agenda for today.\nSpeaker B: I'm just going to go quickly through the minutes of the last meeting.\nSpeaker B: Then we have a presentation of prototype Vue2.\nSpeaker B: Sounds interesting.\nSpeaker B: And I will have a presentation of evaluation criteria by our marketing experts.\nSpeaker B: Then we'll have to go through finance evaluation of the cost of the thing.\nSpeaker B: And hopefully we should fit the target of 12.5 euro.\nSpeaker B: Okay, so let's go.\nSpeaker B: If I go quickly to the minutes of the last meeting.\nSpeaker B: So we went through, we took these following decisions.\nSpeaker B: No LCD, no speech recognition technology.\nSpeaker B: We went through a bad analog in field for the remote control.\nSpeaker B: We went through the use of wheels and buttons.\nSpeaker B: And also the use of a basic station for battery charging and also to code the remote control when it is lost.\nSpeaker B: Okay, good.\nSpeaker B: So guys, let's present this wonderful thing.\nSpeaker D: Okay, so we can go to the slides.\nSpeaker B: Oh, yeah, sorry.\nSpeaker B: Number two, sorry.\nNone: Which is?\nSpeaker D: The final design, final design.\nSpeaker D: Okay, so, Michael, you can go ahead.\nSpeaker A: Yeah, so following our decision to make a yellow, well, to make a banana.\nSpeaker B: Yeah, can you show it to the camera maybe?\nSpeaker A: Okay, so.\nSpeaker A: We actually have a...\nSpeaker D: You can pull it out first.\nSpeaker A: Well, first of all, we made an attractive base station with a banana leaf.\nSpeaker A: Look and feel.\nSpeaker A: And banana sits in there, you know, nicely weighted so that it's not going to tip over.\nSpeaker A: And this is the remote itself.\nSpeaker A: It's kind of...\nSpeaker A: It's ergonomic if it's in the hand rather well.\nSpeaker A: We've got the two scroll wheels here, which, you know, one on the left for the volume and the one on the right for the channel.\nSpeaker A: And underneath we have the turbo button, which is in like a nice trigger position for pressing quite naturally.\nSpeaker B: Why is the use of the turbo button already?\nSpeaker A: This is when you're scrolling through the channels, you can tell it to skip the past channels that you quickly rather...\nSpeaker B: When you stop, it stops.\nSpeaker A: Yeah, well, when you stop scrolling the wheel, it stops.\nSpeaker A: But normally with...\nSpeaker A: It will just stay on each station briefly so you can see it.\nSpeaker D: And we do have one more functionality.\nSpeaker D: If you take the banana as such and you press the turbo button, so it's switch on the TV.\nSpeaker D: The turbo button.\nSpeaker A: So, rather than having an extra button for the on-off switch, we just use the turbo button.\nSpeaker A: Where is this button for?\nSpeaker A: This is the teletext button.\nSpeaker A: Okay.\nSpeaker A: So once you press that, then you get teletext and you can use the channel selector scroll wheel as...\nSpeaker A: Two navigation screws.\nSpeaker C: You want to go to page 700?\nSpeaker A: Yeah.\nSpeaker B: Yeah.\nSpeaker A: How many wheels is he?\nSpeaker A: Well, then you can...\nSpeaker A: You have like a little number selection thing.\nSpeaker A: You press the teletext button to move between the field and then you can just scroll the number back and forth so you have... you scroll to seven and then zero zero and then you can...\nSpeaker A: I don't understand.\nSpeaker C: Can you repeat this?\nSpeaker A: Well, you can press the teletext button.\nSpeaker A: Yeah.\nSpeaker D: Then you can...\nSpeaker D: Then both the scroll buttons, they are for teletext browsing.\nSpeaker D: And you can tell it...\nSpeaker C: Ah, okay, okay.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Once you press the teletext button, then the scroll buttons, they are more for teletext, they are no more for channel or volume.\nSpeaker D: I see, I see.\nSpeaker C: Okay.\nSpeaker A: And this is the infrared port.\nSpeaker A: That's right.\nSpeaker A: Also the top of the banana.\nSpeaker A: Yeah.\nSpeaker A: Excellent.\nSpeaker A: And then we have in the base station, we have the button at the front for calling the banana.\nSpeaker B: Yeah.\nSpeaker B: And the leaves play through all of the antennas.\nSpeaker A: Actually they do.\nSpeaker A: It's...\nSpeaker A: Yeah.\nSpeaker A: That's form and function in the one...\nSpeaker D: So it always means whatever the race goes back, they get reflected and then...\nSpeaker D: You are having a better coverage.\nSpeaker D: It's like antennas.\nSpeaker D: Yeah.\nSpeaker D: So...\nSpeaker A: But yeah, that's just like...\nSpeaker A: That's an attractive base station.\nSpeaker B: Great.\nSpeaker A: So...\nSpeaker B: So...\nSpeaker B: So...\nSpeaker D: So...\nSpeaker D: What else?\nSpeaker D: And for the power source, we are having solar cells and rechargeable batteries in this.\nSpeaker D: And the base station is going to have the input from the power line for...\nSpeaker D: For charging the batteries.\nSpeaker C: Is it really white?\nSpeaker C: Is it light?\nSpeaker C: Or...\nSpeaker C: It is very light.\nSpeaker A: Yeah.\nSpeaker A: It's very light.\nNone: It's about the weight of a banana.\nSpeaker A: Okay.\nSpeaker A: You know, to give you the correct look and feel.\nSpeaker D: The bridge.\nSpeaker D: And we have put these different colors so that people don't mistake them.\nSpeaker D: Mistake it as a banana.\nSpeaker D: Otherwise it's...\nSpeaker C: Yeah, yeah.\nSpeaker C: A child comes in.\nSpeaker A: I understand.\nSpeaker A: I think a child would try to eat it anyway.\nSpeaker A: So maybe we should consider that.\nSpeaker A: Maybe help them safety aspects.\nSpeaker D: Or we didn't think of that yet.\nSpeaker B: So for the power source, apparently you want to use both solar cells.\nSpeaker B: Oh, yes.\nSpeaker B: That's right.\nSpeaker B: I mean...\nSpeaker B: Okay.\nSpeaker A: I don't really know if the solar cells are actually necessary anymore for a recharging base station.\nSpeaker B: Where are you going to place them?\nSpeaker D: It'll be always the top somewhere there.\nSpeaker A: If I was going to place them, I'd put them on the top here since that's like...\nSpeaker A: Do you have enough surface?\nSpeaker A: But yeah, I don't really think it's necessary to have the solar cells.\nSpeaker D: Yeah, because now we have rechargeable batteries.\nSpeaker D: Okay.\nSpeaker B: So that is...\nSpeaker B: What would be the autonomy roughly?\nSpeaker A: The what, sorry?\nSpeaker B: The autonomy.\nNone: Autonomy.\nSpeaker B: What do you mean?\nSpeaker A: How long did the batteries now?\nSpeaker B: How long did it be out of a station?\nSpeaker A: A long time.\nSpeaker D: Yeah.\nSpeaker D: Eight to ten hours.\nSpeaker A: No, no, no.\nSpeaker A: It should be...\nSpeaker D: What?\nSpeaker D: So the time it's not being used.\nSpeaker B: Yeah.\nSpeaker B: So it's...\nSpeaker A: Yeah, but people don't like to put it back in the base station all the time.\nSpeaker A: People want to leave it on the couch, so...\nSpeaker D: No, eight or ten hours of working.\nSpeaker D: If you're just leaving like that, it'll be much longer.\nSpeaker B: Yeah, weeks.\nSpeaker D: Yeah.\nNone: Yeah, that's right.\nSpeaker B: All right.\nSpeaker B: Next slide.\nSpeaker B: Yeah.\nSpeaker D: And we are having the speakers, regular chip for control.\nSpeaker D: Everything was a factor, so that's why we have gone for our regular chip only, not the advanced chip.\nSpeaker D: And that's it.\nNone: Okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: This is look at all the ways.\nSpeaker B: That's right.\nSpeaker B: Okay.\nSpeaker B: Those are very sounds.\nSpeaker B: Very good.\nSpeaker B: Nothing else to add.\nNone: So it's...\nSpeaker E: I like all that.\nSpeaker E: Yeah.\nSpeaker C: I like it.\nSpeaker C: Maybe the thing that convinced me the less is the multifunctional buttons.\nSpeaker C: Looks a bit...\nSpeaker D: You want to have more functional buttons?\nSpeaker C: Looks a bit puzzled.\nSpeaker C: I don't know how to say it.\nSpeaker C: The buttons change their function depending if it's not...\nSpeaker C: It's not...\nSpeaker D: We want to keep it simple so that this button is for teletext, which is usually also the case that usually there is a teletext button and once you press that, the channel buttons, they become the scrolling buttons.\nSpeaker C: And the volume button will become...\nSpeaker D: It's up to you means.\nSpeaker B: Well, in fact, both will be...\nSpeaker B: Could be useful.\nSpeaker D: That means...\nSpeaker D: Let's say this can move the larger digits and this can move the smaller digits.\nSpeaker A: Or can move between positions and the numbers.\nSpeaker C: What about people who want to use digits?\nSpeaker C: But...\nSpeaker C: Real buttons?\nSpeaker C: Wow.\nSpeaker D: Yeah.\nSpeaker D: So there was a constraint that the surface area which we have on this banana on one side because of the shape.\nSpeaker D: So we are targeting a segment which is just very trendy kind of thing that they don't care about the buttons anymore.\nSpeaker C: Okay.\nSpeaker C: And...\nSpeaker C: Have you thought about configuration and all this kind of stuff?\nSpeaker A: Also, all of it.\nSpeaker C: It's all automatic.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Very good.\nSpeaker C: Yeah.\nSpeaker C: We are living in a wonderful world.\nSpeaker B: Yeah.\nSpeaker B: Banana everywhere.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker C: It's automatically comfortable.\nSpeaker C: It's...\nSpeaker B: So we have to go through now variations.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker B: Your slice already.\nSpeaker B: Oh, you're four.\nNone: I...\nSpeaker B: Yeah.\nSpeaker B: So this is one...\nSpeaker B: Which one is this one?\nSpeaker E: Yeah.\nNone: Okay.\nNone: I...\nSpeaker C: I constructed a list of criteria based on the general user's requirements.\nSpeaker C: And its criteria is...\nSpeaker C: Will be evaluated.\nSpeaker C: It's logical criteria.\nSpeaker C: So we must... users must say if it's true or if it's false in a scale ranging from 0 to 7.\nSpeaker D: Why does strange vector of 7?\nSpeaker C: Because it...\nNone: Actually...\nSpeaker C: Sorry.\nSpeaker C: Sorry.\nSpeaker C: Usually I've seen that scale from one.\nSpeaker C: Sorry.\nSpeaker C: It's from 1 to 7.\nSpeaker C: It's from 1 to 7.\nSpeaker C: Sorry.\nSpeaker C: So you can even...\nSpeaker C: It should be an even...\nSpeaker C: Okay.\nSpeaker C:...scale.\nSpeaker C: And 5 is too short and 9 is too long.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Fine.\nSpeaker B: Go ahead.\nSpeaker B: I'm...\nSpeaker B: To have enough granularity.\nSpeaker B: Sorry.\nSpeaker B: It's in order to have enough granularity in the evolution.\nSpeaker C: The variance is...\nSpeaker C: Okay.\nSpeaker C: That's many more.\nSpeaker C: Okay.\nSpeaker C: Great.\nSpeaker C: I'm answering your question.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: I found more useful.\nSpeaker C: I think I could write the criteria on the whiteboard.\nSpeaker C: And we all four could range.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: Could evaluate the...\nSpeaker C: Okay.\nNone: So you can say fancy.\nSpeaker E: Handy.\nSpeaker E: Okay.\nSpeaker D: Okay.\nSpeaker D: So you can say fancy.\nSpeaker C: Handy.\nSpeaker C: Handy.\nSpeaker C: Okay.\nSpeaker C: Let's evaluate if it's fancy.\nNone: Yeah.\nSpeaker D: It's fancy.\nSpeaker D: According to me.\nSpeaker C: Seven.\nSpeaker C: But...\nSpeaker B: Yeah.\nNone: Six.\nSpeaker D: Seven.\nSpeaker C: Seven by me.\nSpeaker C: Six.\nSpeaker C: I would say seven.\nSpeaker D: It's quite fancy.\nSpeaker D: So you can add seven plus six plus seven plus...\nSpeaker B: No way.\nSpeaker A: Yeah.\nSpeaker A: Five.\nSpeaker A: Five.\nSpeaker A: Five.\nSpeaker A: Maybe six.\nSpeaker B: It's...\nNone: Okay.\nSpeaker B: Six point five.\nSpeaker B: Handy.\nSpeaker A: Again, I'd give it a six.\nSpeaker A: I think it's probably more handy than my current remote.\nSpeaker A: Yeah.\nSpeaker A: I'd say scroll wells, but maybe it was the point for not having, you know, the extra buttons when you're real, if you do need them for some reason.\nSpeaker A: But, you know, you can always use your other remote.\nSpeaker B: So seven.\nSpeaker A: Seven.\nSpeaker B: Seven.\nSpeaker B: Six.\nNone: Yeah.\nNone: Six.\nNone: Six.\nNone: Four.\nSpeaker B: Five.\nNone: Fung Channel.\nNone: I'll give five.\nSpeaker D: Four.\nSpeaker D: I would say...\nSpeaker C: Well, it depends when you say functional.\nSpeaker A: Do you mean it does what we wanted to do or does what?\nSpeaker A: It does, you know, every coffee.\nSpeaker A: You know.\nSpeaker A: Every thing.\nSpeaker B: Every thing remote control does you have all the capabilities.\nSpeaker A: Yeah.\nSpeaker C: It's compared to the all...\nSpeaker A: That's right.\nSpeaker D: Criminal control.\nSpeaker D: That's before.\nSpeaker D: The standards.\nSpeaker D: What is available in the market of the show?\nSpeaker D: I have to say four.\nSpeaker C: Actually, I don't know what are the real specifications of a universal...\nSpeaker C: Well, it's not a universal remote.\nSpeaker C: I have not a universal...\nSpeaker E: But it's for all kind of TVs.\nSpeaker B: Yeah.\nSpeaker A: Well, all TVs, but only TVs, I guess.\nSpeaker B: So it's universal, but for TVs.\nNone: Yeah.\nSpeaker B: So, four.\nSpeaker B: Five.\nSpeaker D: Five.\nSpeaker D: Four.\nSpeaker D: Four.\nSpeaker D: Four.\nSpeaker D: So four point two.\nSpeaker D: Just four.\nSpeaker D: Four.\nSpeaker A: Obviously, there's an outliers.\nSpeaker A: So...\nSpeaker B: Okay.\nSpeaker B: Cool.\nSpeaker B: Cool device.\nSpeaker D: There I'll give it seven.\nSpeaker C: It means cool features like new features, actually.\nSpeaker C: That's right.\nNone: Which...\nSpeaker D: While a TV, the most important feature which I felt was the locator, which is a cool feature.\nSpeaker D: And then scroll buttons are again cool features.\nSpeaker D: We don't have LCD for it.\nSpeaker D: But that we decided we don't want to have.\nSpeaker B: Yeah.\nSpeaker C: Seven.\nSpeaker C: I go to type five.\nSpeaker C: I'll say five.\nSpeaker B: Six.\nSpeaker B: Seven.\nSpeaker B: Plus six.\nSpeaker B: Yeah.\nSpeaker B: I say, I six seven.\nSpeaker A: So it's...\nSpeaker A: You hit seven?\nSpeaker A: Yeah.\nSpeaker A: Because it's five, five, seven, seven.\nSpeaker A: Okay.\nSpeaker A: Definitely easy to use.\nSpeaker A: Definitely seven.\nNone: Seven.\nSpeaker B: Seven.\nNone: And you?\nSpeaker B: Five.\nSpeaker B: I'll play here.\nSpeaker B: Seven.\nSpeaker B: I'll play here.\nNone: Seven.\nNone: Okay.\nNone: I have them.\nSpeaker E: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker A: Now here's a 60 million dollar question.\nSpeaker A: Well, twenty five.\nSpeaker A: Twenty five.\nSpeaker B: You're right question.\nSpeaker B: Of course.\nSpeaker B: I would buy the mango butter.\nNone: Thank you.\nNone: What do you...\nSpeaker A: What do you got right?\nSpeaker A: Of course.\nSpeaker C: Of course.\nSpeaker C: The more difficult question for the end.\nSpeaker D: I'll say five.\nSpeaker B: Twenty five euros.\nSpeaker C: I find it quite cheap.\nSpeaker C: I don't know if...\nSpeaker C: It depends if you live in Switzerland or you live in...\nSpeaker B: Yeah.\nSpeaker B: So the target price is for all Europe, only for rich countries.\nSpeaker B: It's more targeting UK or...\nSpeaker C: I don't know.\nSpeaker C: The initial specifications were for the whole Europe or...\nSpeaker B: So this is selling cost, not production cost.\nSpeaker B: Yeah.\nSpeaker C: This is the initial specifications.\nSpeaker B: Yeah.\nSpeaker B: I'm going to say six.\nSpeaker C: It's quite cheap actually.\nSpeaker A: I'd say two.\nSpeaker A: One.\nSpeaker A: I don't want to banana or make...\nSpeaker A: I don't want to banana or make banana or make banana.\nSpeaker D: No, it's very handy actually if you see.\nSpeaker D: It is handy.\nSpeaker A: It's so handy.\nSpeaker A: It's handy but it's terrible.\nSpeaker D: Anybody who comes to your home, he'll at least ask once, what is this?\nSpeaker A: Yeah, but it's not a positive thing.\nSpeaker D: It's a very positive thing.\nSpeaker A: Well, it's handy.\nSpeaker B: It's ergonomic but it's a banana.\nSpeaker B: Don't forget...\nSpeaker B: Ooh, targeting or so.\nSpeaker B: I'm sorry, maybe it's...\nSpeaker A: I would buy this.\nSpeaker B: No, well, yeah.\nSpeaker B: I would be...\nSpeaker B: Yog.\nSpeaker B: Not telling that you are young.\nSpeaker B: Like a teenager, for instance.\nSpeaker A: No, it's I.\nSpeaker A: I would buy.\nSpeaker B: You're crazy teenager and you like fun things.\nSpeaker B: You want to plant this video girlfriend?\nSpeaker B: Yeah, you want to show the beautiful banana you have?\nSpeaker D: Or maybe it does some other kind of thing.\nSpeaker A: Still, I'd say too.\nSpeaker A: I don't think at any stage in my life I would want to banana or my control.\nSpeaker A: Really?\nSpeaker A: Yeah, crazy.\nSpeaker A: I can say maybe there is a market for it.\nSpeaker A: Yeah, yeah, yeah.\nSpeaker B: So you say two.\nSpeaker D: I say five.\nSpeaker B: I say five.\nSpeaker B: You say?\nSpeaker C: I changed the question.\nSpeaker C: So what's your question?\nSpeaker C: I'm not sure.\nSpeaker C: I have loads there.\nSpeaker B: You have to reload?\nSpeaker B: Yeah, I think so.\nSpeaker A: Okay, so it depends if...\nSpeaker C: Yeah, it's two different situations.\nSpeaker C: If you really need a universal remote control or if you would change your remote control for a new one.\nSpeaker A: Yeah, if I had to spend 25 euro, if that was like my limit, maybe I would buy it.\nSpeaker A: Because the other 25 euro remote controls are probably going to look worse than a banana.\nSpeaker D: And they might not be as easy as they are.\nSpeaker A: Yeah, this is going to, you know, hand it to you.\nSpeaker C: So, when I'm...\nSpeaker C: I stick to a knife.\nSpeaker C: Although it still has six.\nSpeaker A: It still has the word of course at the beginning.\nSpeaker A: I don't know.\nSpeaker B: We have six.\nSpeaker A: I'll give it a...\nSpeaker A: I'll give it a four now.\nSpeaker C: So we are...\nSpeaker B: Six?\nSpeaker B: Six.\nSpeaker C: Five.\nSpeaker C: Four.\nSpeaker C: Six.\nSpeaker C: So it's 5.5 more or less.\nSpeaker E: Yeah.\nSpeaker D: Okay.\nSpeaker B: So the last question would I change my remote control for a new banana?\nSpeaker B: Zero?\nSpeaker B: No, we can't.\nSpeaker B: So one.\nSpeaker C: Actually, yeah.\nSpeaker B: Well, if it's the same thing, I will.\nSpeaker B: Yeah.\nSpeaker B: Let's output two.\nSpeaker A: I'd say three.\nSpeaker A: I mean, my remote control at time is pretty terrible.\nSpeaker A: If it was...\nSpeaker A: I changed my remote control of my DVD player for a manna banana, then I would be more inclined to.\nSpeaker A: For the TV.\nSpeaker A: But, because it's really bad.\nSpeaker A: But I'd say a three.\nSpeaker A: I'll still give it five.\nSpeaker D: Five.\nSpeaker C: Yeah.\nSpeaker C: Two, three, five.\nSpeaker C: Yeah, I'll go to three.\nSpeaker C: You're a romantic, really.\nSpeaker E: I would say two.\nSpeaker B: And two.\nSpeaker B: So it's...\nSpeaker B: So it's...\nSpeaker B: We had three point five.\nSpeaker B: Yeah, three point five.\nSpeaker C: Who is that?\nSpeaker C: Liar?\nSpeaker C: Who?\nSpeaker C: You said five?\nSpeaker C: No, I said five.\nSpeaker B: You say five.\nSpeaker B: Ease is the outlier.\nSpeaker B: Okay.\nSpeaker B: Just do some.\nSpeaker C: I don't know if...\nSpeaker C: It's...\nSpeaker A: It's not very promising, but, you know...\nSpeaker C: Well, because there are more...\nSpeaker C: Yeah.\nSpeaker C: We shouldn't sound like that.\nSpeaker B: Well, maybe we should...\nSpeaker C: Because the last two questions are much more important than the rest, actually.\nSpeaker C: Otherwise, we wouldn't...\nSpeaker A: I said some formula you're using that says you have to sound the mouse.\nSpeaker C: No, I didn't prepare anything.\nSpeaker A: Well, just leave it at that then.\nSpeaker B: Yeah.\nSpeaker A: Oops.\nSpeaker B: Maybe we should stick to a J.R. feeling.\nSpeaker B: Yeah.\nSpeaker B: We can have an out of these numbers, which is that...\nSpeaker B: Well, we should go for it.\nSpeaker C: Do you want me to sound?\nSpeaker A: I think it's not...\nSpeaker A: I think it's not...\nSpeaker A: You just lose information if you sound it.\nSpeaker B: Okay.\nSpeaker B: So let's move.\nSpeaker B: Let's move on.\nSpeaker B: Yeah, sure.\nSpeaker B: Okay, sound...\nSpeaker B: No, no, you have to estimate the cost.\nSpeaker B: Okay.\nSpeaker B: So I prepare an exe...\nSpeaker B: an exe...\nSpeaker B:...an exe...\nSpeaker B:...what we are going to calculate the production cost.\nSpeaker B: We should...\nSpeaker B: We should be below 12.5.\nSpeaker B: So I already put some...\nSpeaker B: some numbers here.\nSpeaker B: Okay.\nSpeaker B: We are going to go through...\nSpeaker B: So this is the number of components we need.\nSpeaker B: For this thing.\nSpeaker B: So it appears that there were things that we didn't thought about.\nSpeaker B: And also things that I forget to put like solar cells.\nSpeaker A: Well, we just started against the solar cells.\nSpeaker A: Oh, yes.\nSpeaker A: I knew we stayed out.\nSpeaker B: Okay.\nSpeaker B: So let's go through all the lines.\nSpeaker B: So how did I move?\nSpeaker B: This is something we didn't thought about.\nNone: But...\nSpeaker A: You mean charging it by shaking the banana?\nSpeaker B: Yeah.\nSpeaker D: I think rechargeable batteries will take care of the power thing.\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: To bring the cost.\nSpeaker B: Okay.\nSpeaker B: So we stick to battery one.\nSpeaker B: Yeah.\nSpeaker B: No kinetic.\nSpeaker B: So why don't I don't see the difference between kinetic and hand-in-o?\nSpeaker A: Well, maybe dynamo is like you have to actually...\nSpeaker B: Oh, you have to...\nSpeaker B: Okay.\nSpeaker B: So kinetic is really shaking the banana.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker B: So it's better, nice falling.\nSpeaker B: What's that?\nSpeaker B: Let's go ahead.\nSpeaker B: Okay.\nSpeaker B: So we only have one for battery.\nSpeaker B: And then for electronics.\nSpeaker B: So I didn't put anything for the month.\nSpeaker D: So we have the regular chip on the print.\nSpeaker D: Yeah.\nSpeaker D: Which is one.\nSpeaker D: Yeah.\nSpeaker B: And that's it.\nSpeaker B: Okay.\nSpeaker B: So we hit...\nSpeaker D: And we have sample speaker.\nSpeaker D: Yeah.\nSpeaker D: So one.\nSpeaker D: Yeah.\nSpeaker B: Oh, yeah.\nSpeaker B: The cost is crazy.\nSpeaker A: Well, actually that sample speaker is not...\nSpeaker A: We're not using that.\nSpeaker A: We're just using the...\nSpeaker A: The beep.\nSpeaker A: The very beep, simple beep.\nSpeaker A: That sample thing is like the voice recording and everything.\nSpeaker A: Okay.\nSpeaker A: So I remove it.\nSpeaker A: Yeah.\nSpeaker A: I say that...\nSpeaker A: Yeah.\nSpeaker B: So don't we need...\nSpeaker B: Well, there is no listing for radio-frequency thing.\nSpeaker B: Yeah.\nSpeaker D: So we will put some extras.\nSpeaker D: If there is something...\nSpeaker B: Yeah, maybe.\nSpeaker B: We'll see later.\nSpeaker B: Okay.\nSpeaker B: So for the case, I put single curved.\nSpeaker B: Okay.\nSpeaker D: To reduce the cost.\nSpeaker D: It's okay.\nSpeaker A: Well, wait a second.\nSpeaker A: Because what we think...\nSpeaker A: It's double curved.\nSpeaker A: It's got a...\nSpeaker A: Oh, it's...\nSpeaker A: It's got walled at the direction.\nSpeaker A: So don't worry.\nSpeaker A: Well, yeah.\nSpeaker A: It's monotonic.\nSpeaker A: But...\nSpeaker A: It's got walled at the direction.\nSpeaker D: It's got...\nSpeaker A: But if you hold it...\nSpeaker A: If you hold it that way, that's too curved.\nSpeaker A: One on the side, one on that side.\nSpeaker A: But the opposite side.\nSpeaker C: Actually, where's this section?\nSpeaker A: I mean, this is probably...\nSpeaker A: It's a good one here.\nSpeaker A: This probably actually costs more than three issues.\nSpeaker A: Yeah.\nSpeaker B: So let's put one here in the back end instead of single lock.\nSpeaker B: Okay.\nSpeaker B: Right?\nSpeaker B: Yeah.\nSpeaker B: So we stick to plastic.\nSpeaker A: It costs nothing.\nSpeaker A: That's right.\nSpeaker A: Well, wait a minute.\nSpeaker B: No, didn't we say we wanted to do a rubber?\nSpeaker B: No.\nSpeaker A: Oh, if you drop it.\nSpeaker A: Too expensive.\nSpeaker B: Too expensive.\nSpeaker A: Oh, we're already at 11.\nSpeaker A: We'll come back.\nSpeaker A: We'll come back and see if we can fill it in.\nSpeaker B: Okay.\nSpeaker B: So I'll put a rubber one.\nNone: Okay.\nSpeaker B: So special core.\nSpeaker B: You know?\nSpeaker B: Yeah.\nSpeaker B: For the interface, we have...\nSpeaker D: We don't have any push buttons.\nSpeaker D: No, we have two push buttons.\nSpeaker D: We have three.\nSpeaker D: No, that is a scroll wheel itself.\nSpeaker D: It'll be put in that.\nSpeaker B: No, no.\nSpeaker B: We have...\nSpeaker B: We call and we have three push buttons.\nSpeaker D: Okay.\nNone: Okay.\nSpeaker C: Actually, when you wrote regular tip, you should put two because there is another tip here.\nSpeaker B: No, it's always no cheap.\nSpeaker B: This is just radio frequency thing.\nSpeaker B: Yeah.\nSpeaker B: Yeah, but cheap.\nSpeaker D: No.\nSpeaker D: There's no chip there.\nSpeaker D: It just emits the signal.\nSpeaker D: It's just...\nSpeaker D: And the receiver accepts it and...\nSpeaker D: Yeah.\nSpeaker C: Sure.\nSpeaker C: It does nothing?\nSpeaker D: Actually.\nSpeaker A: No.\nSpeaker A: It's a rechargeer thing.\nSpeaker B: Okay.\nSpeaker B: We didn't think it thought about integrated core wheel push buttons.\nSpeaker A: Well, I actually did think about it myself, but I thought, you know, because you could potentially, you know, you could be pushing it down as you scroll it for a...\nSpeaker A: Instead of a turbo.\nSpeaker A: Yeah.\nSpeaker A: But, you know, the turbo button just adds that extra class.\nSpeaker B: Yeah.\nSpeaker A: You know?\nSpeaker A: So, I mean, if we're over budget, then maybe we could rethink that.\nSpeaker A: Okay.\nSpeaker A: So, no LCD.\nSpeaker B: So, for... we have no buttons, deployments.\nSpeaker B: Yeah.\nSpeaker B: Right?\nSpeaker B: No.\nSpeaker B: Uh...\nSpeaker B: But in fact, couldn't we consider this button as a buttons supplement because...\nSpeaker B: Oh, no.\nSpeaker B: These are for colors and special forms, special colors and special materials.\nSpeaker B: No, we're not.\nSpeaker B: We don't need anything special.\nSpeaker B: Okay.\nSpeaker A: So, we are over budgeted.\nSpeaker A: Yeah.\nSpeaker B: So, first thing which we should take care of is...\nSpeaker D: Make it plastic and forever.\nSpeaker D: Let it be plastic.\nSpeaker A: And then we're basically on budget except for, you know, in 10 cents.\nSpeaker A: And that much money will be required for the base station, which is not there.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: So, maybe, in fact, we have to put two here.\nSpeaker B: Because it costs nothing.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker A: Yeah, well, yeah, the base station is made out of maybe yours.\nSpeaker D: It might be 90 centimetres for the remaining things which the cord and everything is too cool.\nSpeaker B: Exactly.\nSpeaker B: So, we have an margin for that stuff.\nSpeaker B: That's right.\nSpeaker A: Does that include charging circuitry and nothing?\nSpeaker A: Yeah, maybe.\nSpeaker B: Okay.\nSpeaker B: Good.\nSpeaker B: Yeah.\nSpeaker B: What?\nNone: Excellent.\nSpeaker A: So, what do you do with the extra profits?\nSpeaker B: We'll invest in R&D.\nSpeaker B: Okay.\nSpeaker A: Let me extra for it.\nSpeaker B: Yeah.\nSpeaker B: So, well, we're under the cost.\nSpeaker B: So, we can go through two project evaluation.\nSpeaker C: Okay.\nSpeaker C: So, now we have a product which nobody would buy.\nSpeaker C: Would...\nSpeaker C: Yeah, would buy.\nSpeaker B: Sorry.\nSpeaker B: No, we have a product which none of us would buy.\nSpeaker B: The evaluation project.\nSpeaker B: Which is different?\nSpeaker B: None of us would buy.\nSpeaker A: No, it's people in Milan and in Paris.\nSpeaker A: I was like, I'm going to buy it.\nSpeaker A: Massively, yeah.\nSpeaker A: Yeah.\nSpeaker A: We're not in Milan or Paris.\nSpeaker C: Yeah.\nNone: Okay.\nSpeaker C: You have been in Milan a couple of times, so...\nSpeaker C: Actually, there were a lot of batteries.\nSpeaker C: And you said the lowest.\nSpeaker C: The lowest is caution.\nSpeaker C: This is what we would...\nSpeaker B: You can.\nSpeaker B: The tangible battery.\nSpeaker B: Because I'm sick of...\nSpeaker B: Yeah, but that is...\nSpeaker B: Extra battery.\nSpeaker B: Yeah.\nSpeaker A: Okay.\nSpeaker A: So, project versus.\nNone: Wow.\nSpeaker B: In fact, I didn't know really what to say here.\nSpeaker B: If you have any ideas or what we can say.\nSpeaker B: So, I don't understand what they mean by satisfaction on the four example.\nSpeaker B: Should it be more like a status of these meetings?\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker B: Well, in fact, we use a little bit of the whiteboard and the digital pen.\nSpeaker B: Not that much.\nSpeaker A: I think we had a fair bit of creativity.\nSpeaker B: Oh, yeah.\nSpeaker B: It's really creative.\nSpeaker A: And...\nSpeaker A: But I think one thing we missed out of this whole process was a focus group with...\nSpeaker A: The actual people were targeting.\nSpeaker A: We needed some of these kind of young transettes to come in and play with the banana.\nSpeaker A: And, you know, see if they liked it.\nSpeaker A: See if I like it.\nSpeaker B: Yeah, maybe we should go through, yeah, in the evaluation.\nSpeaker A: Because, yeah, the evaluation for us is kind of pointless.\nSpeaker B: So, maybe we should do a kind of evolution in the streets like asking to young people.\nSpeaker B: Well, do you like to have a banana as a remote control?\nSpeaker A: Because it would also be interesting to find out if we have a market in really young children as well.\nSpeaker A: To see, you know, how to market this thing.\nSpeaker A: Because, you know, well, I mean, maybe you don't want to give all your kids their own remote because they'll be changing the channel all the time.\nSpeaker A: But, you know, maybe in some households where there is, you know, TV for each kid, then, you know, banana remote control could be fun for them as well.\nSpeaker A: But, you know, we might have to draw a face on it.\nSpeaker A: So, but I think that's something we need to work on next time is really finding out more about the target market.\nSpeaker B: Yeah.\nSpeaker B: Any new ideas we could investigate next time?\nSpeaker B: I don't know, oranges?\nSpeaker D: Yeah, the cost of the thing can be made more than it might be.\nSpeaker D: Because I think it's just the optimal what we have done for the cost which we are looking forward to.\nSpeaker A: Yeah, I think there are ways we could maybe simplify the interface.\nSpeaker A: Well, just the circuit board that we're using inside, I'm not sure really how complicated our needs are.\nSpeaker A: I mean, all we have is two push buttons and two scroll wheels.\nSpeaker A: Do we really need an integrated circuit to, you know, to process that?\nSpeaker C: So, more general remote control instead of just focusing on TV.\nSpeaker C: The complexity shouldn't be much higher.\nSpeaker C: You said if it goes for DVD, then I...\nSpeaker A: And also for, you know, for mood lighting and stuff that would be rather cool.\nSpeaker A: I think that's actually something that should be in version two, is the ability to control things other than the TV.\nSpeaker A: And not just electronic equipment, but, you know, the whole environment of the room.\nSpeaker B: Okay. Very good.\nSpeaker D: So...\nSpeaker D: More dense?\nSpeaker B: Well done. I think we can go home.\nSpeaker B: Home?\nSpeaker B: Happy that is right.\nSpeaker B: Maybe you can have the production department to make a real prototype that we can use at home to evaluate.\nSpeaker B: Okay, so thanks very much.\nSpeaker E: Okay. Bye.\nSpeaker E: Thank you.\nNone: We didn't get the hope of this time that meeting is five minutes.\nNone: Yeah.\nNone: It was longer than yours.\nNone: Okay.\nNone: No.\nNone: I didn't need that.\nNone: Oops.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "Bmr026", "summary": "The meeting concerned mostly status updates and logistical affairs. The participants learned that the transcription status was good. The progress on the DARPA demo was also good, except the interface did not look as pretty as they had wanted to. The storage disks, which they had been waiting for, had finally arrived, but the person responsible for installing them was going on break. The participants also decided to schedule less meetings, so they have updates to report when they do. The meeting ended with technical discussion about model performance.", "dialogue": "Speaker F: Okay, we're on. So I think this is going to be a pretty short meeting because I have four agenda items.\nSpeaker F: Three of them were requested by Jane who is not going to be at the meeting today. So The first was transcription status. Does anyone besides Jane know what the transcription status is?\nSpeaker G: Sort of. I do it peripherally Well, first of all with IBM, I got a note from Brian yesterday saying that they finally used the tape for the thing that we sent them a week or week and a half ago and that it's gone out to the transcribers and hopefully next week will have the transcription back from that and I've been pan Jane seems to be Moving right along on the transcriptions from the X-E side. She's assigned I think probably Five or six. Yeah, I think we're up to MRR 13 or something. Yeah, so I guess she's hired some new transcribers and which meetings is she transcribing?\nSpeaker G: Well, we've we've run out of 80 years. Okay, because a certain number of them are sort of awaiting to go to In the rest are in processing\nSpeaker C: We're doing it. So in parallel. She has transcribers right now who are risky sitting idle because there's no data back from oh\nSpeaker F: No, no, no, we're not waiting on the process. So they're they're doing the full transcription process. Oh, okay\nSpeaker C: So you're doing the real thing. It's we're doing parallel. Yeah, I need to ask Jane whether it's it would be okay for her Some of her people to Transcribe some of the initial data we got from the smart com data collection, which is these short like five or seven minute sessions And we want it, you know, we need again, we have a similar Logistics setup where we're supposed to send the data to Munich and get it back But to get going we would like some of the data transcribe right away so we can get started. Yeah, that's what I mean And so I wanted to ask Jane if You know, maybe one of the transcribers could could do I mean since these are very short that should really be\nSpeaker D: There's only two channels\nSpeaker C: Yeah, yeah, it's a synthesis thousand have to be transcribed I think so yeah So so it's basically one channel to transcribe and it's one session is only\nSpeaker E: So that should have made many fewer and it's also not a bunch of interruptions\nSpeaker C: Right and some of it is red speech. Yeah, so we could give them the the thing that they're reading and they just make sure it's right and so I guess I was gonna ask her but she's not around Yeah, yeah, yeah, okay with you to you know, yeah\nSpeaker E: Yeah, if we're held up on this other stuff a little bit in order to compensate that's okay because I Still have high hopes that the IBM pipeline will work out for the results\nSpeaker G: Yeah, and also related to the transcription stuff So I've been trying to keep a web page up to date Showing what the current status is of the trends of all the things we've collected and what stage each meeting is in in terms of Whether you mail that out to the list. Yeah, well, that's the thing that I say out just a few people saying can you update these pages?\nSpeaker G: And so that's okay, okay, I'll send it out to the list\nSpeaker F: I haven't done that so I have lots of stuff to add Yeah, just in my own directly try to get to that okay, so Jane also wanted to talk about Participant approval, but I don't really think there's much to talk about. I'm just gonna do it and If anyone objects too much then they can do it instead You are going to I'm gonna send out to the participants With links to web pages which contain the transcripts and allow them to Suggest edits and then believe them out Okay, for the ones that we have\nSpeaker D: But it's just transcripts not the not the audio. No, they'll have access to the audio also\nSpeaker F: I mean that's my intention because transcripts might not be right\nSpeaker G: So you want people to be able to listen to them so the audio that they're gonna have access to will that be the Un compressed version or will you have scripts that like uncompressed the various? Oh, that's a good point\nSpeaker F: That's good point. Yeah, it's it's probably gonna have to be the uncompressed versions because It takes too long to do random access to compression\nSpeaker G: Yeah, I was just wondering because we're Running out of the unbacked up\nSpeaker F: That was the other point. Oh, was that another yet another agenda item so But that is a good point so we'll get to that too DARPA demo status not much to say the backend stuff is working out fine. It's more or less ready to go I've added some stuff that Indexed indexes by the meeting type MRE to you, etc. And also by the user ID So that the front end can then do filtering based on that as well the backend is Going more slowly as I think I said before just because I'm not much of a tickle TK programmer and And Dave Galbart says he's a little too busy. So I think Dawn and I are gonna work on that and And you and I can just talk about it offline But the backend was pretty smooth. So I think we'll have something it may not be as As pretty as we might like but we'll have something I want to when we would reach Dave's saturation point\nSpeaker E: Been volunteering for everything and yeah Okay, finally said he was too busy\nSpeaker F: I guess we reached it. Yeah, he actually he volunteered, but then he's then he retracted it so oh well\nSpeaker B: And also I was just showing Andreas I got I'm an x waves kind of display and I don't know how much more we can do with it with like the Prasadik stuff where we have like stylized pitches and signals and The transcripts on the bottom. So right now it's just an x waves and you have three windows, but I don't know Look pretty nice. I'm sure it's potential for a little something. Yeah, so I\nSpeaker E: Okay, so again the issues for July the issues gonna be what can we fit into a windows machine?\nSpeaker B: Okay, so it might just be slides. Yeah, okay. Well, we'll see\nSpeaker D: I've been putting together transcribe all things for Windows So and I installed it on Dave galots PC and work just fine. So hopefully that because there's some people\nSpeaker C: It would be cool if we could get that to work at SRI because the\nSpeaker F: We have more windows machines transcribe is tickle tk very generic with snack so basically anything\nSpeaker D: Yeah, it's snack to run on yeah, but the problem is the version Transcribe works with the snack version is 1.6 whatever and that's not anymore supported It's not on the web page anymore, but I just wrote an email to to the author to the snack author and he sent me the 1.6 Whatever library and so it breaks it was packaged with transcribe yeah, but then you can't add our patches at then The new version is is totally different and in in terms of the source code You you can't find the tickle files anymore. It's some whatever wrap thing and you can't you can't access that so you have to install First install tickable then install snack and then install the transcribe a thing and then do the patches\nSpeaker C: I wonder if if we should Contribute our changes back to the author so that they maintain those\nSpeaker G: We have yeah, it's just hasn't made it into the release yet. Okay, so did you put the The anti version out on the meeting report page or no, I haven't done that yet\nSpeaker D: No, but I definitely will do that so\nSpeaker E: Can some of the stuff that don't talk me about somehow fit into this? I mean you just have a set of numbers that are associated with that\nSpeaker B: Yeah, it's basically ask you files or binary files whatever representation. It's just three different It's a waveform and just a stylized pitch vector basically so it's I mean we could do it a mat I mean you could do it a number of different places. I'm sure but it would be cool if the transcriber\nSpeaker C: Yeah, interface had like another window for the you know, maybe above the waveform Yeah, yeah, some arbitrary Valid function that yeah, that'd be very cool\nSpeaker G: You know, I'm using that again. It's time synchronous with the waveform. Yeah, more tickle tk\nSpeaker F: Programming so someone who's familiar with tickle tk has to do it, but it wouldn't be hard to do\nSpeaker C: But it would almost be like having another waveform displayed. Yeah, yeah\nSpeaker F: Yeah, yeah, maybe we could look into that. Yeah, but it seems to me that I it doesn't seem like having that real time is that necessary So it seems to me you could do images\nSpeaker B: What do you mean by real time? Do you mean like you will scroll through it instead? Okay? Yeah, yeah, it would be cool to see it it would be cool like to see To hear it and see it and see the pitch contours also. I think you can do all that just statically in PowerPoint Yeah, I mean just record the clip and show an image. I just thought a few minutes lies. I thought you meant like just like\nSpeaker E: View graphs or something you know, yeah, so now we're talking about on the computer right and I think when we were talking about this before we had this little demo meeting we sort of set up a range of different degrees of live-ness that you could have and the more live the better But given the crunch of time we may have to retreat from it to some extent So I think for a lot of reasons I think it'd be very nice to have this transcriber interface be able to show some other Interesting signal along with it. So the big good thing to get in there, but Anyway, just looking for ways that we could actually show what you're doing and to people Because a lot of this stuff particularly for communicator Certainly significant chunk the things that we weaved our arms about originally had to do with prasadx and Nice to show that we can actually get them save them\nSpeaker F: And last that item on the agenda is disc issues yet again So we're doing okay on backed up. We're we're only about 30% on the second disc so We have a little bit of time before that becomes critical, but we are like 95% 98% on the scratch discs for the expanded meetings Yeah, and my original intention was like we would just delete them as we need more space But unfortunately we're in the position where we have to deal with all the meeting data All at once in a lot of different ways. There's a lot of transcribers, too Yeah, there are a lot of transcribers so all of those need to be expanded and then people are doing chunking and I want to do The permission form so I want those to be live so there's a lot of data that has to be around and Jane was going to talk to Dave Johnson about it one of the things I was thinking is we just got these hundred or excuse me 10 Spark blade So they command send blades. Yeah, something like him in the other day came in but they're not set up yet And so it seems to me we could hang Scratch disc on those because they'll be in the machine room They'll be on the fast connection to the rest of the machines And if we just need unbacked up space we could just hang discs off them. Well, is there why not just hang them off of\nSpeaker G: Abbott is there because there's no more room\nSpeaker F: I'm a disc racks on Abbott\nSpeaker E: What are we gonna get well?\nSpeaker C: You can just store the discs somehow well, but the sunblades have spare drive base You can put a new oh you mean you put them inside into boxes\nSpeaker F: Is the sun that these sun blades take commodity hard drives so you can just go out and buy a PC hard drive\nSpeaker E: And stick it in but if Abbott is going to be our disc server It file server it seems like we would want to get a second disc rack or something\nSpeaker C: Plus we're talking about buying a second disc file server\nSpeaker F: Well, I mean there are lots of long-term solutions what I'm looking for is where do we expand the next meeting? Yep\nSpeaker E: Well for the next meeting you might be out of luck with those 10 you know Dave jassen is gone for like 10 days Oh, I didn't know it left already. Well tonight. Oh well\nSpeaker F: I mean he won't set up the how much space to wait for this. I don't know what is going to be the baddest thing\nSpeaker B: Yeah, I have I have an 18 gig drive hanging off on my computer all right, so Your computer's name You had an 18. Yeah, well, it's about I think there's about 12 gig left so with and you have an X drives Install yeah, okay, so I didn't realize it was so critical I mean I'm not doing anything on it right now until I get new meetings to transcribe with that a new transcriptions coming in\nSpeaker G: I really can't do anything. I'm not that I can't do anything. I just gave T-Lo some about 10 gigs the last 10 gigs of space that there was on habit and So but that but which one was that XG\nSpeaker C: XG that's also where we store the The up five Training set I don't think that's on XG on XG is only car but I've also been storing the feature files there and I guess I can Study leading some because now know what the best features are and we won't be using the old ones anymore\nSpeaker B: I have a lot of space then isn't that XH? I have a lot of space and it's not there's very little yeah, not for long But I mean it's not going on outside, but I'm using it\nSpeaker F: Probably only about four gig is on X on your X drive will definitely pay\nSpeaker B: I think it's about four five gig because I love to see four meetings on there three or four meetings\nSpeaker F: Right, so okay, so that will get us through the next couple days We need another gig a quad\nSpeaker E: There should I should just be I should have a button Just press each meeting saying we need more despair Just we skip the rest of the conversation\nSpeaker G: We've collected so far something like 65 meetings and how much does each meeting take about a gig uncompressed\nSpeaker D: It's a little bit more as I usually don't do not uncompressed all of the pzm and the pda things right So if you uncompressed everything is even more\nSpeaker G: I pressed how much are they like half a gig for all of them yeah So we're definitely restoring You know all of those so there's what 30 some gig\nSpeaker E: I was just meeting so far so it's a hundred gig or something Because we have the uncompressed right also well, we haven't uncompressed all the meetings, but oh\nSpeaker F: We'd like to\nSpeaker E: Yeah, well, I mean it's they really are cheap. I mean it's just a question of figuring where they should be hanging them, but but we could you know if you want to get four discs get four discs. I mean it's it's\nSpeaker G: Small these things are just few hundred dollars. Yeah, why send that message out to I guess you and Dave asking for if we could get some disc I sent this out there a day ago, but in Dave didn't respond so I don't I don't know How the whole process works. I mean does he just go out and get them and yeah, it's okay and So I was assuming he was gonna take over that, but he's probably too busy given that he's leaving\nSpeaker E: Yeah, I think you needed direct conversation with him and just say they just asking that you know what what should you do and My answer back was a sure you just want one so I mean I think that what you want to do is plan ahead a little bit and Figure well here's what we've got figuring and doing for the next few months\nSpeaker F: I know what they want the Sissadmins would prefer to have one external drive per machine So they don't want to stack up external drives And then they want everything else in the machine room right so the question is where are you gonna hang it?\nSpeaker G: Mm-hmm. I don't know what the space situation is in the machine room, right?\nSpeaker E: Right, so this is a question. It's pretty hard to solve without I think part of the reason why Dave can't get\nSpeaker G: What the new machines up is because he doesn't have room in the machine\nSpeaker C: What what rearrangement one one thing to and to do when you need to conserve spaces I bet there's still some old like nine gig discs Around and you can probably consolidate them on to larger discs and and you know recover the space. Yeah\nSpeaker E: Yeah, no, I think Dave knows all these things the question and so he always has a lot of plans of things They're just gonna do to make things better in many ways, but I know that runs out of time\nSpeaker F: Generally their first priority has been for backed up disc and so I think what he's been concentrating on is the back the Backup system rather than on new disc so\nSpeaker E: Well, so but this this is a very specific question for me basically we can easily get One to four discs. I mean you just got to get four and we've got the money for it. It's no big deal But the question is where they go and I don't think we can solve that here\nSpeaker C: We can put some this in the in that back room there. Yeah, really\nSpeaker E: I attached to\nSpeaker C: Upcorn to the machine that collects the data So then you you could at least temporarily store stuff there\nSpeaker F: It's not on the net so it's a little what do you mean it's not on the net? It's not It's behind lots of firewalls that don't allow any service\nSpeaker C: Oh because it's because it's in a serial machine. Oh\nSpeaker F: And also on the list is to get it onto the normal X-NAT but Who knows when that will happen that might be it can't be that hard I mean sure No, the problem with that apparently is that they don't currently have a wire Running to that back room that goes anywhere near one of the X-E routers So they actually have to run a wire somewhere. Yeah\nSpeaker E: Yeah, Jen, you know any one of these things is certainly not a big deal if there is a person dedicated to doing it It would happen pretty easily, but it's it's Everybody everybody has a has well all of us have long less of different things we're doing But I mean I think that there's a there's a longer current thing and there's immediate need and I think we need a conversation with Maybe maybe after after T or something you and I can go down and talk to him Just say what you know what should we do right now how long is David gonna be gone?\nSpeaker E: 11 days or something Yeah, basically tomorrow and all of the week after\nSpeaker F: And that's all I have\nSpeaker E: Let's see the only thing I was gonna add was that I Talked briefly to Mari and we had both been busy with the things So we haven't really connected that much since the last meeting we had here But we agree that we would have a telephone meeting the Friday after next and I I wanted to make it After the next one of these meetings so something that we want to do next meeting is is to put together Kind of reasonable list for ourselves of what is it?\nSpeaker E: That we've done I mean just sort of bull ties. I mean, I can I can dream up text But this is basically gonna lead to the annual report so So just that we from tomorrow. Yeah, yeah, so we can this is gonna be in the morning\nSpeaker C: Or because you know at Fridays I have to leave around two\nSpeaker E: Sorry if it could be before that no, no, but I don't need other folks for the meeting Okay, all right all I'm saying is that I'm sorry. I miss understood. Yeah, so what I meant was on the meat this meeting Okay, if I was something I'm making a major thing on the agenda is I want to help in getting together a list of what it is That we've done so I can tell her I think I have a pretty good idea but but And then the next day like late in the day I'll be having that discussion with her\nSpeaker C: One thing I mean we in past meetings we had Also very very slowly talked about the work that It was happening sort of on the on the recognition side but isn't necessarily related to meetings specifically so and I wondered whether we should maybe have a separate meeting between You know whoever's interested in that because I feel that there's plenty of stuff to talk about but it would be sort of Maybe the wrong place to do it in this meeting Well, it's that It's just Gotta be very boring for people who are not you know sort of really interested in the details of the\nSpeaker E: Well, okay, so how many how many people here would not be interested in a meeting about recognition?\nSpeaker G: Jane\nSpeaker C: Well, I know well Jane and where you mean in a separate meeting or talking about it We talked about how much overlap there will be yeah, so you're so so\nSpeaker E: Liz and Jane probably okay, so we're gonna have a guys meeting For all that way real real real men Real men do decoding\nSpeaker C: I Mean it's sort of I mean when when the talk is about data collection stuff sometimes I've\nSpeaker E: You know, I'm bored so it's I can sympathize with them not wanting to to be you know, I could you know this could need a better developed feminine side I The end of the week Yeah, I I've heard some comments like this and it could be I mean\nSpeaker C: And we don't have to do it every week we could do it every other week or you know whatever or whatever we feel like we alternate this meeting every other week\nSpeaker F: Or just alternate the focus that's what I mean. Yeah, so yeah, even weeks have to do that basic on I\nSpeaker C: Personally, I'm not in favor of more meetings\nSpeaker G: Um, right because I am I mean a lot of times lately it seems like we don't really have enough for a full meeting on meeting recorder\nSpeaker F: Right, so we keep going for a full time. Well cuz we get into these other times\nSpeaker G: Yeah, we feel obligated to collect more data So if we could alternate it just focus of the meeting\nSpeaker E: Why don't we just start with that and then if we find you know, we're just not getting enough done There's all these topics not coming up Then we can expand into another meeting, but I think that's a great idea. Uh, so It's chat about it with listen Jane. You can't see what they think that would be good\nSpeaker G: I mean Andreas and I have various talks in the halls and there's lots of things you know details and stuff that I think will be interested in I you know, where do we go from here kind of things and That would be good\nSpeaker E: Yeah, and you're you're attending the the front-end meeting as well as the others so you have you have Probably one of the best you and I guess have them and the main ones who sort of we're at sea the bridge between the two We are doing recognition in both of them\nSpeaker C: So So so we can talk a little bit about that now if there's something that would be for next week I just so the latest result was that um You know, I tested the sort of final version of the PLP configuration on development test data for for this years up five tests and and the recognition performance was Exactly, and I mean exactly up to the you know the first decimal the same as with the Marcapscha front-end for both emails and meals. Yes. Oh, well, it was a little bit of a Overall they were the males. I think we're slightly better and the females were slightly worse but nothing really I mean definitely not significant and then the really nice thing was that if if we combine the two systems We get a one and a half percent\nSpeaker F: So just with Rover\nSpeaker C: With an investor over which is like our new and improved version of Rover which actually uses the whole Invest list from both systems\nSpeaker E: So it's that I mean the only key difference between the two really is the kind of smoothing at the end Which is the other aggressive versus the yeah?\nSpeaker E: sort of trinkets and\nSpeaker C: And and so after I told the my colleagues at us right about that you know Now they definitely want to you know have a next time we have an evolution they want to do you know basically at least the system combination And you know why not sure\nSpeaker F: Clearly gonna add a few more features though\nSpeaker C: What do you mean more features in the sense of front-end features or in the sense of just bells and whistles? No\nSpeaker F: front-end features you know we do PLP and Melcaps drillitz oh yeah, well gee right so we could yeah that's\nSpeaker C: There's one thing I mean you don't want to overdo it because every front end you know if you you know You basically multiply your effort by and where and it's the number of different systems and so so one one compromise would be to only To have the everything up to the point where you generate lattices be basically one system and then After that you risk or your lattices with multiple systems and combine the results and that's a fairly painless I think you think we still get one and a half I think so yeah, maybe a little less because at that point the error rates are lower and so you know Maybe it's only one percent or something, but that would still be worthwhile doing so just You know just want to let you know that's working out very nicely and Then we had some results on digits We so this was really sort of just to get Dave going with his experiments and so But as a result, you know, we were sort of wondering why is the Hub 5 system doing so well on the digits and the reason is basically that there's a whole bunch of threads Each data have five training set including the end together and not all of no it's actually digits only a maybe a Fifth of it the rest is red is red timid data and Edas data and what's your travel?\nSpeaker E: Right, but a fifth is how much?\nNone: Fifth would be maybe Two hours, yeah, so I mean that's actually not that different from the right one training\nSpeaker C: But it definitely helps to have the other red data in there because we're doing You know, they are right as half of what you do if you train only on timid not timid T.I. Digits Which is only what two hours something right and so More red speech data definitely helps and you can leave out all the conversational data with no performance penalty\nSpeaker E: Yeah, that was the interesting thing that was it right because it was apparent if you put in a bunch more data\nSpeaker C: It would be better, but is there even more red speech data around? Oh, yeah, so we're only for the hub five training We're only using a fairly small subset of the microphone database So you could beef that up and probably do even better. I could also put in\nSpeaker F: Focus condition zero from hub four from broadcast news, which is mostly from here at speech\nSpeaker C: It's not exactly red speech, but it's pretty close right now. I mean there's plenty of red speech data I mean what's the journal. Yeah, that's right take one example but So you know that might be useful for the people who trained the Digit recognizes to use something other than T.I. Digits\nSpeaker E: Well, they've been using timet okay, that's the Experimented for a while with a bunch of different databases with French and Spanish and so forth because they're multi-lingual tests and And actually the best results they got were using timet But Using now but yeah, certainly if we If we knew what the structure of what we were doing there was I mean There's still a bunch of messing around with different kinds of noise robustness algorithm So we don't know exactly which combination we're gonna be going with Once we know then the trainable parts of it may be great to run lots of lots of stuff through\nSpeaker C: Well, that was that and then I said guess Chuck and I had some discussions about how to proceed with the tandem system and You want to\nSpeaker G: Yeah, so Andreas brought over the alignments that the SRI system uses and so I'm in the process of converting those alignments into Label files that we can use to train a new net with and so then I'll train the net and\nSpeaker C: And one side effect of that would be that it's that the fun site would change So the MLP would be trained on I think only 46 or 48 48 phones Which is smaller than the Then the phone site that that we've been using so far. Yeah, that that would probably help actually because the fewer dimensions It's the less trouble probably with as far as just the Just you know, we want to try things like delta's on the tandem features and so you have to multiply everything by two or three and so you know Fewer dimensions in the Frontside would be actually helpful just from a logistics\nSpeaker E: Sure, although we I mean it's not that many fewer and run and we take a KLT\nSpeaker C: Exactly, so so that was the other thing and then we wanted to just limited to maybe Something on the same order of dimensions as we use in a standard Frontend so that would mean just doing the top I don't know 10 or 12 or something of the KLT\nSpeaker E: Yeah, and I think the mention and we should again check we should start with Stefan my impression was the one we did that before that had very little He didn't lose very much. Right. I just taking the top whatever. Yeah\nSpeaker C: But then and then something once we have the new MLP trained up One thing I wanted to try just for the front of it was to actually run like a standard hybrid system that is based on you know those features and Retrained MLP and also the\nSpeaker E: You know the dictionary that we use for the up five system and the base starting off with the base of the alignments that you got\nSpeaker C: Exactly from pretty decent system. Yeah, so that would basically give us a More hopefully a better system Because you know compared to what Eric did a while ago where he trained up. I think a system based on broadcast news and then Train retrained it on switchboard. Yeah, and But he I think he didn't he probably didn't use all the training data That was available and his dictionary probably wasn't as tuned to Conversational speech as As ours so that's that's certainly one thing and the dictionary made a huge difference We made some improvements to the dictionaries to the dictionary about two years ago which resulted in a Something like a 4% absolute Air rate reduction and\nSpeaker E: The other thing is dipping deep to history into a resource management days when we were collaborating with us right before It was I think it was a real key Starting point for us that we actually got our alignment when we were working together We got our initial alignments from decipher At the time And later we got away from it because because once we had decent systems going Then it was it was typically better to use our own systems because they were self-consistent But that's certainly to start off when we were trying to recover from our initial 140% error Or But that was a good way to start it. We're not quite that bad with our our switchboard systems But it was certainly on as good as this arise\nSpeaker G: What is the performance of the best switchboard system that we've done?\nSpeaker E: Well the hybrid system we never got better than about 50% Here and it was I think there's just a whole lot of things that no one ever had time for we never did really Fix up the dictionary We always had a list of a half dozen things that we were going to do a lot of them were pretty simple and we never did But that we never did never did any adaptation right we never did any and that number\nSpeaker C: I think was on switchboard one day in our right Where they are right now is in the 20s Yeah, so yeah, we're probably least a factor too often so it would be so it would be good to sort of Re just at least to give us an idea of how well the hybrid system would do\nSpeaker E: Yeah, but I think again, it's yes, the cover is these conversational speech bit because our broadcast new system is actually pretty good\nSpeaker C: Right And the other thing that that will help us to evaluate is to see how well the NLP is trained up right because it's a pretty good Indicator of that so to the vicinity check of the NLP outputs before we go ahead and train up the You know use them as a basis of the tandem system. Yeah, it'll still probably worse\nSpeaker G: I mean it's no because independent should we bother with Using the net before doing Embedded training. I mean should should we even use that or should I just go through that's a good question\nSpeaker C: Yeah, we weren't sure whether it's worth to just use the alignments From the SRI recognizer or whether to actually go through one or more iterations of better track where you're real line\nSpeaker E: You run it keep keep both versions see which one's yeah, I mean I think I agree with I mean basically you would then you proceed with the embedded training Mm-hmm. It's gonna take you a while right training if it's net anyway and while it's training you may as well test the one you have and see\nSpeaker F: I could make arguments either way, but but so I'm well, but in your experience. I mean have you seen\nSpeaker C: Big improvements and on some tasks with embedded training or was it sort of small-ish\nSpeaker E: Improvements well dependent on the task. I mean I think in this one I would sort of expected to be important because we're coming from Alignments that were achieved with an extremely different system, right although\nSpeaker F: I mean we've done it with when we were combining with the Cambridge Recurrent neural net embedded training made it worse Which I've never figured out\nSpeaker C: Right, but I mean I think it's a bug so you started training with outputs from a with alignments that were generated by the Cambridge Yep a system and then Yeah, well that may probably just that was probably because your initial system. I mean your system was worse than Cambridge's and you Was it I don't think it was they were comparable really they were very close. That's weird Excuse me. That's weird. No, I mean it's weird that it's weird that it got worse\nSpeaker E: No Now we've we've seen I mean with the numbers OGI numbers task We've seen a number of times people doing embedded training. Oh actually it's not that weird because we have seen\nSpeaker C: We have seen cases were acoustic retraining the acoustic models after some other change made matters worse rather than yeah\nSpeaker E: But I would I would suspect that something that that had a very different Feature set for instance. I mean they were using pretty different similar feature sets to us Yeah, I would expect that something had a different feature set would would benefit from What about hidden unit size? Oh, wait a minute and the other thing Sorry, the other thing is that what was in common to the Cambridge system in our system is they both were training posteriors right, so I mean that's another pretty big difference and Went back at least you remember soft targets or\nSpeaker C: Sorry, I'm sorry. I missed what what's the key for here. Oh that\nSpeaker E: Both the Cambridge system and our system were we're training posters and if we're we're coming from alignments coming from the RIS system. It's a likelihood based system So so that's another difference. I mean there's different different front end different different training criterion I would think that in that in embedded Embedded training would have at least a good shot of improving it some more, but we don't know\nSpeaker G: Okay, yeah, I was wondering, you know what size net Anybody have any\nSpeaker E: How much training daddy?\nSpeaker G: Well, it's gonna start off with the small train set and how many hours is that so I'm not sure\nSpeaker C: I think that has about well, it would be gender dependent training, right? So so I think it's that's about 30 hours 30 hours per gender\nSpeaker F: Not sure what this will mean the small training I think so I'll\nSpeaker C: Excuse me. It's definitely less than 100. You know, it's more like Like 30 40 hours something like that wrong number\nSpeaker G: I called to tell us that yeah\nSpeaker E: so Too big right so At least a couple thousand units. I mean it's it's the thing I'll think about a little more But the toss up between 2000 to 4000 He definitely wouldn't want the 8000 it's it's more than a thousand. It's too small Well, let me think about it, but I think that that at some point there's diminishing returns I mean it doesn't actually get worse typically, but but there is diminishing returns and you doubling remember\nSpeaker C: You'll have a smaller output layer, so there's gonna be fewer parameters. Not like that 50\nSpeaker F: From the input unit\nSpeaker E: Okay, yeah, yeah, I will have a very tiny effect\nSpeaker F: Or because you use the context when does and so the input to hidden is much. Oh, I see. I see. Yeah, of course. Yeah\nSpeaker E: So it should be way way less than 10% of difference There's\nSpeaker G: How big how big? I think the net that we did use already was 8000 in units\nSpeaker E: That's the one air trained up right and that was trained up on like 140 hours of Such under dependent or independent or so that would be like trained on 60 or 70 hours So yeah, definitely not the 1000 2000 I mean the 4000 will be better and 2000 will be faster and almost good\nSpeaker G: So maybe I'll start off with 2000. Yeah, see\nSpeaker E: Okay, yeah 30 hours is like 10,000 seconds So that's like 11 11 million frames and 2000 in unit net is I Guess about 7800,000 parameters That's probably that's probably fine. I mean a 4000 is well within the range that you could benefit from but the 2000 be faster\nSpeaker C: I actually have to go all right\nSpeaker E: Uncle Bernie's rule is 10 to 1 Bernie Woodrose rule of Here's Uncle Bernie. Yeah\nNone: You Just waiting for you to leave Anything else?\nSpeaker E: Nothing to talk about we only talked for\nSpeaker F: Well, we started late transcript L177 3015 47 548 8 7 0 7 0 8 9 0 8 7 7 7 286 5 4 7 6 1 8 2 7 9 4 6 7 9 8 4 4 2 6 4 9 2 0 8 9 2 6 8 2 1 7 7 7 3 0 1 7 4 4 1 8 7 9 2 4 8 9 0 6 5 6 9 7 1 3 3 7 6\nSpeaker E: transcript L-178 0 4 9 9 6 4 3 3 0 7 5 2 8 1 6 7 8 9 7 1 0 4 7 9 7 1 9 5 3 1 1 0 1 8 8 3613 4 7 6 9 9 4 259 3301 4057 38 498 6 7 0 9 5 4 009 3 7 4 9 7\nSpeaker D: transcript L-179 6 9 1 2 3 3 3 3 6 7 216 11 0 11 8 9 0 9 6 0 6 7 9 8 3 4 6 403 2 2 6 4 6 9 652 5 2 7 6 406 282 904 0025 5 8 0 5 5 372 5 302 6712 4 245\nSpeaker B: transcript L-98 90 60 5 4 46 18 4829 4 3 2 4 357 8 7 96 444 3025 6 152 3434 16 33 93 5 4 0 7 7 11 0 8 4 901 474 98 9 6 0 8 125 368 661\nSpeaker G: transcript L-139 0 9 5 3 4 5 5 971 9 7 9 5 12 4 9 3 8 5 4 2 8 7 4 6 1 5 3 136 8 9 98 8 859 9 118 7 6 251 6065 0 8 6 1 6 8 0 219 709 982 9 4 9 477 40\u043d\u0438\u0447\n"}, {"meeting_id": "ES2013b", "summary": "The meeting mainly focused on the functional design of the remote control. After Project Manager reviewed the last meeting briefly, Marketing made a presentation about re-envisioning the menu functionality, current technology, and interaction with other tools of the remote control by discarding unnecessary buttons and adding the speech recognition to the remote control according to market research. User Interface argued that they should discard the existing standard, and only keep the basic and user-friendly functions like channel selection, volume and enter key. The group then discussed the source of power whether to have a charger or solar power. Although solar energy was fancy and environmentally friendly, the cost was much higher than using a charger. The group faced some difficulty when deciding the target group whether for the young end or for the lower end of the age range who have more money than sense. In the next part, the group reached a consensus in terms of the keys of the remote control, which is power, channel, volume, enter, number.", "dialogue": "None: Okay.\nSpeaker D: Welcome to this second meeting.\nSpeaker D: It's now quarter after 12 and we're given 40 minutes for this meeting.\nSpeaker D: This is a meeting on functional design.\nSpeaker D: And I want to welcome you all and thank you all for doing some research in between.\nSpeaker D: I did took minutes from the first meeting and I'll show them to you in a moment.\nSpeaker D: I know each of you have a presentation and in thinking about the 40 minutes, I thought it would take only like three minutes for the previous minutes.\nSpeaker D: Each of you having about seven minutes or maybe a little less for your presentations and a little discussion because I happen to have been told there was some new project requirements and we have to make some decision on what functions it will have.\nSpeaker D: Okay.\nSpeaker D: Does everybody agree with this?\nSpeaker D: Yes.\nSpeaker D: Okay.\nSpeaker D: And after the meeting, there'll be things to be done and as you can see, it says we get to get lunch and then some more individual work and then putting minutes away and individual actions.\nSpeaker D: Now, for the minutes of the first meeting.\nNone: Good.\nSpeaker D: That one.\nSpeaker D: As you can see, it was this earlier today.\nSpeaker D: Kate, Steph, Sarah and myself in our full capacities were present.\nSpeaker D: I opened the meeting.\nSpeaker D: The product was developed and reviewed and we talked about the financial end of it and some implications, the four million sales target and new ideas of not too many buttons, bright colors and some of the influence of the Japanese and we closed early so you could then proceed with your research and getting reports together for this meeting.\nSpeaker D: Anybody have any questions on those minutes?\nSpeaker D: Are they complete?\nSpeaker D: Did they discuss everything that we covered last time?\nSpeaker D: Kate?\nSpeaker C: Did I miss something?\nSpeaker C: We talked about the individual roles that we tried as well.\nNone: Yeah.\nSpeaker D: I'm afraid I incorporated that when I said who was present.\nSpeaker D: Okay.\nSpeaker D: Yes.\nSpeaker D: We did and we did a little bit of thought team building of making the pictures but I didn't think those were appropriate to the minutes necessarily.\nSpeaker D: So as a group, I think they are accepting minutes.\nNone: Okay.\nSpeaker A: Is that what we were supposed to say?\nSpeaker B: Yeah.\nSpeaker D: Yeah, I did.\nSpeaker D: Good.\nSpeaker D: Then we'll move to the three presentations.\nSpeaker D: Okay.\nSpeaker D: Okay.\nNone: Okay.\nSpeaker D: We need to move this.\nSpeaker D: Who wants to go first?\nSpeaker D: As far as it goes.\nSpeaker B: Not really meant to touch these.\nSpeaker B: Okay.\nSpeaker B: It doesn't have any on this one.\nNone: Excellent.\nSpeaker C: Thank you.\nSpeaker D: She said we didn't need to screw it in.\nSpeaker A: Okay.\nSpeaker D: It's doing it.\nSpeaker A: There we are.\nSpeaker A: All right.\nSpeaker A: Thank you very much.\nSpeaker A: One of the biggest issues I thought about from last meeting was the fact that we need to sell 4 million of these remote controls.\nSpeaker A: I think that this was an opportunity to really take real reaction in the direction of similar of handheld tools that have been used and are used by many of us and to kind of bring the remote control into the same realm as an accessible, useful electronic device as opposed to something that is lost in the couch and for having it.\nSpeaker A: My main goal here is to re-envision remote control in this context and to think about menu functionality and current technology and the fact that it could be interactive with other tools.\nSpeaker A: Some of the research in the market has shown that people really are not happy with remote controls as they are now.\nSpeaker A: That means we do need to make some decisions about what keys or buttons on the remote control to perhaps keep and what ones to discard.\nSpeaker A: If we devote some energy into this, I think the recent productions of real reaction, the IGO, everywhere power and the high definition DVD players, although it makes immediate sense to have our remote control interact with these, I think we can also use this as a platform to make it interact with other tools.\nSpeaker A: In fact, I think the high definition DVD players and all of this will come along and will only benefit from the positive feedback from our well-designed tool.\nSpeaker A: Again, most users really dislike the current look and feel of remote controls.\nSpeaker A: 50% of all these numbers, most important, is 50% of users say they only use 10% of the buttons and 80% of users.\nSpeaker A: If we think about this, there are a lot of television, DVD, stereo, remote control users out there, 80% and spend more money on a remote control that looks fancy.\nSpeaker C: Can I ask for these figures to come up with this?\nSpeaker A: There was market research and there were 100 people from the room, so 80 out of 100 said they would spend more money.\nSpeaker D: Now, in between, as the project manager, they sent me an email from the powers that be.\nSpeaker D: That teletext is outdated and the internet is coming in as important, but that they want this remote control to only be for TV with incorporating the corporate image, color and slogan.\nSpeaker A: I think we can really focus on this remote and, again, bring the real reaction brandy in and get some positive marketing for other tools, even if we directly don't advertise for the I-Go everywhere.\nSpeaker A: An interesting element was the, would you pay more for speech recognition question?\nSpeaker A: So these market research questionnaires looked into your concern about technology and specifically wanted to find out if you're not speech recognition.\nSpeaker A: Okay, now the early adopters, those of us who grew up with technology and lucky for us, have the cash to pay for it, the young age group without the mortgages and responsibilities, 91% of them would pay more for speech recognition and a remote control.\nSpeaker A: Very interesting, I leave this up to the group to decide if we want to use this in the designers, but 91% is a large enough target market to target it.\nSpeaker A: Well, I think, especially in terms of growth, I think this would be a very smart group to target, I mean, three quarters of the next age group, 25 to 35 are interested in with the technologies improving if we can get to the same.\nSpeaker D: In real numbers, this is the 91% and the 76% translate to in excess of the 4 million or\nSpeaker A: 8 million, yes. But would you pay more and it doesn't work and is it approachable and did I know that it was a very good question?\nSpeaker A: I don't know if speech recognition should be included, but I think it's an interesting, I think that maybe shows more about being open to technology.\nSpeaker B: Definitely.\nSpeaker B: I've got more research on how much more it would be and any existing examples and what reactions to that would be.\nSpeaker C: Yeah.\nSpeaker C: And this, how well how well target growths would be, the typical price of these things,\nSpeaker A: that can best be a little, but if we're even two, to fill this thing, the total year is 50. Is that a little?\nSpeaker A: Exactly.\nSpeaker A: So, I think that's a little bit of a competition, sussing out what other people are doing and what's in the pipeline is very, very important because there was a question about to want an LCD screen that wasn't responded to, but some of the larger remotes do have screens where you can navigate, you know, so it turns into something perhaps you all have seen and it's attempting to manage this super entertainment system with something that looks like a small train.\nSpeaker A: And I think, you know...\nSpeaker C: What are the easy little age groups you must be one of the small population?\nSpeaker C: No, no, no.\nSpeaker A: You're exactly right there.\nSpeaker A: But I think the key is to get the early adopters, people with familiar technology and they'll be...\nSpeaker D: Well, we're not looking at whether the earlier adopters on that screen, that's looking at age groups.\nSpeaker A: Yes, and I'm making the leap that people who are familiar, younger people are more familiar with technology.\nSpeaker A: Or comfortable.\nSpeaker A: So, okay, you had the other power...\nSpeaker A: I think the most important thing is an attractive streamline remote control and to be extraordinarily reductionist, power, channel volume, and everything else is...\nSpeaker A:...up to the designers.\nSpeaker A: And this is also supported by the market research.\nSpeaker D: Okay.\nNone: Thank you.\nSpeaker A: That's my contribution.\nSpeaker D: All right.\nSpeaker D: And we'll turn to the next presentation.\nSpeaker D: I think she said we don't need to screw it in, just stick it in.\nSpeaker D: And then press what?\nSpeaker D: FN and F8.\nSpeaker D: And then we'll be patient.\nSpeaker B: Yeah, here we go.\nSpeaker C: Ta-da.\nSpeaker C: Okay, just do each one of those.\nSpeaker C: There we are.\nSpeaker C: Yay!\nSpeaker B: Right, technical functions design.\nSpeaker B: Well, I think first off, basically I do agree with what Sarah has defined as your personal preferences and I think we need a more streamlined volume with no extraneous functions.\nSpeaker B: So my method was to look at the existing amounts and what functions they have.\nSpeaker B: And what we all need to discuss is whether we want those functions pretty much the same as what existing amounts have if we can build on this with the speech recognition.\nSpeaker B: That's not something I thought about at all, but it's also something we can discuss.\nSpeaker B: Okay.\nSpeaker B: And I presume we can miss out functions related to a video or DVD remote control if this is only going to be, you know, satellite cable TV remote control.\nSpeaker B: So these are two models of existing remote controls.\nSpeaker B: The one on the left seems to be a fairly standard universal remote control.\nSpeaker B: It has fast forward, stop play, all related to movies.\nSpeaker B: It also has, seems to have channel up and channel down, which is more of what you'd expect from either like a sky or cable remote control where you've got hundreds of channels instead of a nearly terrestrial one.\nSpeaker B: But I think we should be looking more along the lines of the one on the right, which has also a play, stop and pause, or a thing.\nSpeaker B: I don't think we need them at all.\nSpeaker B: I think we just need channel selection, volume up, volume down, and I think an inter function where you can access, it's not like teletext but along the same line, to access things on the screen.\nSpeaker B: Not related to internet one that you mentioned because that would be far outside our budget and what we want this to do.\nSpeaker D: And exceed the requirements to expecting others.\nSpeaker B: Really exceed the requirements because the requirements really I just want to be able to change channels and functions, which is more a text on the screen thing than the actual buttons.\nSpeaker B: I was thinking something, some smooth, sleek little remote control where big user friendly buttons and a menu that you can access.\nSpeaker B: But then I do think we need to discuss the speech recognition possibility.\nSpeaker D: Okay.\nSpeaker D: Anything else you want to add?\nSpeaker B: Okay.\nSpeaker B: So we could go back to the pictures of the, what I thought, the pictures of the remote controls and possibly discuss what we think about them.\nSpeaker B: Or, but I may just hear what Kate has to say first.\nSpeaker D: Yeah.\nSpeaker D: That's your word Kate has to say.\nSpeaker A: Maybe afterwards we could do a whiteboard that you're the one on the right.\nSpeaker D: I think the white, the one on the right is as well as less cluttered.\nSpeaker B: Definitely less quiet.\nSpeaker B: I mean, but still it's.\nSpeaker B: It's there but it's assumed something else I was going to say.\nSpeaker B: The style of these is terrifying.\nSpeaker B: Yeah.\nSpeaker B: I really think we need to not only possibly the materials like the type of plastic used but everything including size and shape of buttons, positioning of buttons, the actual shape of the handheld device, colors, ergonomics.\nSpeaker B: Yeah.\nSpeaker B: Everything to do with this has to be revolutionized.\nSpeaker B: But.\nSpeaker B: So that's that for now.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: So, I want to bring us down to earth again.\nSpeaker C: So, let's go back to what the basic function of remote control is.\nSpeaker C: It's sending a message.\nSpeaker C: And the basic components we've got to build in for our 12 heroes 50 are an energy source, the user interface, which will incorporate an integrated circuit that actually composes the message based on what the button's use it presses, we turn into a message.\nSpeaker C: And then we need a sending mechanism to send it to the receiver.\nSpeaker C: Now, I would have hoped, I think that's my only slide actually, I would have hoped to do you a pretty powerpoint slide of my first cut design but unfortunately the technology defeated me so if you'll bear with me, I'll do it on the whiteboard.\nSpeaker C: So, we want an energy source.\nSpeaker C: So, we've got to think about what that might be.\nSpeaker C: And then we have the user interface.\nSpeaker C: On the main components in the RV.\nSpeaker C: The chip that actually has the intelligence of the machine that translates button presses into a message, which then transfers to some sending mechanism, which encodes it and sends the message to the receiver.\nSpeaker C: So, those are the basic things that we've got to get in for our 12 heroes 50.\nSpeaker D: Thank you.\nNone: Okay.\nSpeaker D: Right.\nSpeaker D: But those things, it's also we can get those components.\nSpeaker D: The block, that rectangle for the user interface is where the user comes in of what's it look like, what do the buttons look like, what does it feel like, that's where the user interface is really coming into its own.\nSpeaker D: The technical end is what's actually going to be in there.\nSpeaker D: But also it has to be easy enough to change or repair if something goes wrong.\nSpeaker D: For example, the battery energy source or what if the chip for whatever reason breaks down after a certain amount of time, do you just replace it?\nSpeaker D: Is there any, because it may be in the same area with several other user interfaces like for DVDs, movies, whatever, does it have to have a way of being segregated from the others in a different frequency or something?\nSpeaker C: Well, I may wrong here, but I've been thinking of those devices being a cheap, less produced device with a Microsoft phone, million of them.\nSpeaker C: That's almost one in every 10th house or whatever it is.\nSpeaker C: And I haven't thought of it as being a repair or anything.\nSpeaker C: You just check it out.\nSpeaker C: And that's why I'm a bit concerned.\nSpeaker C: I like the idea of speech recognition.\nSpeaker C: That's a good idea.\nSpeaker C: But I'm not convinced we've been put it into this box for the price that we're going to need to hit.\nSpeaker A: Do we have some idea of how much it would cost to create a device that has these basic elements?\nSpeaker A: Because then...\nSpeaker B: Isn't that your job?\nSpeaker A: No, the chip composer marketing.\nSpeaker A: Oh.\nSpeaker A: Oh, no, the chip composer sending.\nSpeaker A: I don't believe I know.\nSpeaker C: I'm hoping that my personal coach is going to give me some advice.\nSpeaker C: I'll ask him.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Hmm.\nSpeaker B: It does seem as if we're just to do something really simple and mass-produced, which is pretty much the same as these existing models, just maybe...\nSpeaker B: Inspired?\nSpeaker B: A little bit more inspired.\nSpeaker B: But basically just the same.\nSpeaker B: But what suddenly came into my head is they always take two AA batteries, which is really not very efficient at all.\nSpeaker B: It could be possible to have a rechargeable internal battery, like an MPC player does.\nSpeaker B: You could just...\nSpeaker B: Or a mobile phone or whatever.\nSpeaker B: You could just plug into a parasauce with a couple of hours and then it would be charged for ages and ages and ages.\nSpeaker B: And you don't have to worry about replacing AA batteries.\nSpeaker B: Well, that has a...\nSpeaker A: No, I don't.\nSpeaker A: Which is, if every time you're done using a remote, you put it on a charger, then it has a...\nSpeaker A: You'd never need the batteries.\nSpeaker A: Yeah, but it also has a place.\nSpeaker B: It's not stuck down the back of the sofa.\nSpeaker B: But then again, I don't know if this is within our christ-ringing.\nSpeaker B: That's really good.\nSpeaker C: Well, I think that's a very interesting idea.\nSpeaker C: But I'm not a very good industrial designer and I don't know much about what is this cost.\nSpeaker C: I'll do some research for the next meeting.\nSpeaker C: Okay.\nSpeaker A: It's better than 90 of that solar probably.\nSpeaker A: Yeah.\nSpeaker C: And solar may not be so good when you're watching TV in the night.\nSpeaker C: Yeah.\nSpeaker C: Well, it is just so...\nSpeaker D: It would have to store up the energy.\nSpeaker D: Yeah, I guess.\nSpeaker D: And then use it.\nSpeaker D: Solar can do that.\nSpeaker C: It might be looking quite heavy then.\nSpeaker D: Yeah.\nSpeaker D: We are not pretty heavy and cost too much.\nSpeaker B: But just a rechargeable internal battery would probably...\nSpeaker B: I mean, it might cost more at first to develop and to install.\nSpeaker B: But for a long-term use, there'll be so much more convenient and economic than fiddling around trying to replace chunky AA batteries and not having any in your kitchen drawer.\nSpeaker C: So you think you might make that a selling point if it was something that you probably could afford to develop?\nSpeaker A: Definitely, because I'm thinking in terms of the loss and breakage of remote temperature that is supposed to be not really having a home or a nest.\nSpeaker A: So it's a little dock.\nSpeaker A: Yeah.\nSpeaker A: So if you can dock it, you know, you could argue that.\nSpeaker A: And the dock could look very fancy.\nSpeaker A: Exactly.\nSpeaker D: That could be your inspiration.\nSpeaker B: I don't know if you're decent.\nSpeaker B: Right.\nSpeaker B: Within our crisis.\nSpeaker A: You're talking about another component.\nSpeaker A: I think it's a big one.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Are we agreed as to what our target group is, though?\nSpeaker D: Pretty much so that we'd be looking for the younger end.\nSpeaker A: Exactly.\nSpeaker A: But I think you may see the questions about are there enough for these people out there?\nSpeaker A: Or are they making purchasing decisions?\nSpeaker A: Yeah.\nSpeaker C: Because I agree that there are people with, how can I put it more money than cents?\nSpeaker C: You are the buyer.\nSpeaker C: I usually put more money than brains.\nSpeaker A: Oh, well, I'll make a note of that.\nSpeaker A: Good comment.\nSpeaker C: But what I was going to say was although they may be buying personal music devices, are they necessarily buying TV remote controls because they probably live in a household that has a TV or phone.\nSpeaker C: That is low end of the range.\nSpeaker A: It might be good to know.\nSpeaker A: Who is actually buying television.\nSpeaker A: And are we in the region where people have money and television?\nSpeaker C: I think we've got a big heel to climb it.\nSpeaker C: I mean, we've got this way people who go on remote control because it paid you the telly that they should buy our product instead.\nSpeaker C: Yeah.\nSpeaker D: Or do we make sure that it goes through the manufacturer?\nSpeaker D: We sell to the manufacturer as the remote that goes with it.\nSpeaker B: Yeah.\nSpeaker D: That's probably what would be a more efficient way of doing it.\nSpeaker A: Yeah.\nSpeaker A: And some of our TV players incidentally have them because we have federal issues with them in the department.\nSpeaker A: But more of them need to go through them.\nSpeaker D: Okay.\nSpeaker D: Right.\nSpeaker B: I'm not going to get any resolutions by the end of the day.\nSpeaker D: Probably not.\nSpeaker D: The other thing that we're supposed to do is make decision on our functions.\nSpeaker D: Our functions we've so far decided I think that power channel volume make it attractive.\nSpeaker D: It has to have an enter key and of course the number keys.\nSpeaker D: It has to have big user friendly buttons.\nSpeaker D: And definitely we want to be inspired that the current.\nSpeaker D: Styles are just playing awful.\nSpeaker A: Do we need.\nSpeaker D: Is that agreeable to everyone?\nSpeaker A: Actually.\nSpeaker A: The enter key.\nSpeaker A: I have a chart here that I didn't include.\nSpeaker A: Yeah.\nSpeaker D: Throw some light on that.\nSpeaker D: Actually, we're, you know, we are almost a half hour out of the 40 minutes.\nSpeaker D: So we have to get close to finishing.\nSpeaker A: My only comment is I think maybe we could somehow include an enter key in like a power key.\nSpeaker A: That was my.\nSpeaker A: Okay.\nSpeaker A: Enter power as a.\nSpeaker D: So simultaneous key.\nSpeaker A: Exactly.\nSpeaker A: Again.\nSpeaker A: Is that okay with you?\nSpeaker A: Sorry.\nSpeaker B: I was.\nSpeaker C: I was.\nSpeaker C: I was.\nSpeaker A: Well, if you're pressing enter the thing that I already.\nSpeaker A: And so maybe when you press power.\nSpeaker A: Enter.\nSpeaker A: Well, I was thinking maybe you to turn it off.\nSpeaker C: You'd have to press power twice.\nSpeaker C: And maybe power follows something like a channel.\nSpeaker A: Yeah, or something that has a turning dial where.\nSpeaker A: Who's got an iPod then.\nSpeaker A: Yeah, I don't.\nSpeaker A: I wish.\nSpeaker A: Anyway, that was the only comment about some of the decisions people made.\nSpeaker B: What's most important.\nSpeaker B: Definitely channel volume power.\nSpeaker B: Yeah.\nSpeaker B: Things like screen settings and audio settings.\nSpeaker B: I would generally.\nSpeaker B: Yeah, you have a.\nSpeaker B: You know, a little flat with a little control panel on the actual box itself.\nSpeaker B: I wouldn't know how to do it using a remote control.\nSpeaker B: So.\nSpeaker B: I would say that they are definitely less relevant.\nSpeaker A: Okay.\nSpeaker A: Thanks for looking at that.\nSpeaker D: It's easy to use is the other thing that we want to be able to do.\nSpeaker D: We don't want to be smart.\nSpeaker D: Yeah, don't make it too hard for the granny.\nSpeaker D: I just joined that set last week.\nSpeaker D: First grandchild around.\nSpeaker D: So are we agreed then of those things?\nSpeaker D: And let's go back to agenda.\nSpeaker D: And hook me up.\nSpeaker D: I thought it'd be fun.\nSpeaker D: We probably won't go the first time.\nSpeaker D: I'll probably be in the mess.\nSpeaker D: It lost it off here.\nSpeaker D: No, it was up there, but I couldn't see it down here.\nSpeaker D: This time it should be both.\nSpeaker D: You have to keep doing the FN and F8 to five minutes to finish.\nSpeaker D: Thank you a lot for telling me.\nSpeaker D: Right.\nSpeaker D: We're ready to close that will be completed questionnaire.\nSpeaker D: Then we'll have some time for individual work to continue our research.\nSpeaker D: I'll put some minutes of this meeting together.\nSpeaker D: Your individual assignments are for Kate to do the components for you, Steph, to do the user interface.\nSpeaker D: And for you, the trend watching.\nSpeaker D: And each of us will get help from our coach.\nSpeaker D: We're agreed to get ourselves together and then have lunch.\nSpeaker D: That's the end of this meeting.\nSpeaker D: I hope that's good enough for her to tell her that's the end.\nSpeaker B: We didn't come to any sort of decision on the functions.\nSpeaker D: I think I went over the functions and wrote them down and marked them.\nSpeaker D: That's what I went over and nobody was objecting to them.\nSpeaker A: And you mentioned I was just going to say could you really like the new project requirements?\nSpeaker A: It has to be for a TV.\nSpeaker D: The teletext is outdated.\nSpeaker D: The internet is important.\nSpeaker D: It's only to be for a TV and it must include the corporate image, color and slogan, which I think is more in the user range with Steph.\nSpeaker B: So what actually are these?\nSpeaker B: Is that the yellow and black?\nSpeaker B: It doesn't tell me.\nSpeaker B: I'll just use it on my website.\nSpeaker C: Okay.\nNone: Thank you.\n"}, {"meeting_id": "IS1005b", "summary": "This meeting included three presentations from marketing, user interface and industrial design and discussion about core function of remote control. At first, marketing introduced the conclusion of experiment and questionnaire. Users thought remote controls were ugly and only used a few buttons. Since young users might prefer high technology features like speech recognition. After that, the user interface introduced how electric circuits, keyboard and transmitter reacted together. And group members discussed solutions of wave interference. Then industrial designer talked about working design and how LCD display or speech recognition might influence the design of circuits. And whether adding these functions would make the price out of range was discussed. After discussing new requirements for remote control from management, the group members started with core and basic functions, such as necessary buttons. Finally, they mentioned something about speech recognition and speech commands and their influence on battery life.", "dialogue": "None: get started.\nSpeaker C: So, nice to see you again.\nSpeaker C: So, for this meeting I will take the notes and do the minutes.\nSpeaker C: So we will see our free presentations.\nSpeaker C: We will start with the manager expert.\nSpeaker C: We will talk about user requirements, what user needs and what is desired for this device.\nSpeaker A: Okay, can I have the laptop over here?\nSpeaker C: Oh, I don't think so.\nSpeaker C: Okay, I have to get up.\nSpeaker C: I think it should stay in this meeting.\nSpeaker C: Okay, okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker A: So, basically I am going to present some findings of a study we conducted into what users want in this remote control.\nSpeaker A: So, first of all, what we did is we conducted an experiment with 100 test subjects.\nSpeaker A: We put them in our usability laboratory and got them to play with remote controls and also after they have done that to complete a questionnaire to tell us what they like and what they don't like in remote controls.\nSpeaker A: So, basically the major things we found out was that basically users don't like the look and feel of most remote controls that are currently on the market.\nSpeaker A: They, you know, 75% of the people we did the experiments on found that remote controls that they used in the past were completely ugly.\nSpeaker A: They didn't match the operating behaviour of the user.\nSpeaker A: That is, you know, the way users use remote controls when they are watching TV, that the layout of the remote controls didn't match the way they used it.\nSpeaker A: And thirdly they say that half of the users that we tested said that they only used 10% of the buttons that are on remote controls.\nSpeaker A: So, we also collected some users statistics based on how these test subjects were using their remote control.\nSpeaker A: And from this we basically came up with the figure that the channel buttons, the channel selection buttons are the most, by far, the most used buttons on the remote control.\nSpeaker A: And you can see they used 168 times per hour on average while the user is watching TV.\nSpeaker A: The closest button that was used, the button that was used that was closest to the channel button was the teletext button, which was used 14 times per hour, followed by the volume button, which was 4 times per hour.\nSpeaker A: All the other buttons such as audio and picture selection configuration buttons and things were used less than or equal to 1 times per hour.\nSpeaker A: We also asked users which buttons had the most importance to them, which buttons they felt were the most important buttons on the remote control.\nSpeaker A: And basically they said the channel volume and power buttons had the highest relevance to users.\nSpeaker A: Note the only power was very infrequently used.\nSpeaker A: It only had a usage frequency of about 1 times per hour, but users ranked it as having a very, very high relevance.\nSpeaker A: And the audio and picture settings had a very, well, the users thought that the audio and picture settings weren't very important to them, and they used them very infrequently as well.\nSpeaker A: So we asked users what frustrates them the most about current remote controls.\nSpeaker A: And 50% of the users said that what frustrates them is losing their remote control, somewhere in the room and not being able to find it.\nSpeaker A: They also said that it takes a lot of time to learn a new remote control, especially when there's many buttons and it's an unintuitive interface.\nSpeaker A: And then thirdly, some users commented on the fact that the way that you have to hold and press buttons on a remote control are bad and cause repetitive strain injury.\nSpeaker A: We also asked some users about some specific features that they'd like to see in the remote control, in particular, do they want an LCD display?\nSpeaker A: And secondly, do they think speech recognition is a useful feature to have on our remote control?\nSpeaker A: Basically, our findings are that amongst younger age groups, the answer is overwhelmingly yes, they want these features, they want these high technology features.\nSpeaker A: For instance, 91% of people aged between 15 and 25 said yes, they want these features, whereas the trend was as users became older and older, they were less likely to want these features in a remote control.\nSpeaker A: So I guess it depends on where we're focusing out our market.\nSpeaker A: And as our company motto is putting fashion in electronics, I think we're focusing on the younger target demographic.\nSpeaker A: And so maybe we should think about adding these high technology features into our remote control.\nSpeaker A: That's my presentation, thank you.\nSpeaker C: Thank you.\nSpeaker C: So maybe now we can talk about user interface and about technical function of this device.\nSpeaker C: So Peter, can you talk about something about that?\nSpeaker D: Okay, yeah, but you think I'm using this.\nSpeaker B: I'm using it.\nSpeaker D: Sorry, I'm sorry.\nSpeaker B: Okay.\nSpeaker B: If I could go there.\nSpeaker B: This, okay.\nSpeaker D: You're scaring me with LCD man.\nSpeaker D: And speech recognition in real-o-tunate will be very expensive.\nSpeaker A: Yeah, it's true, but you know, the features that users want.\nSpeaker A: Yeah.\nSpeaker A: And our production cost of 1250 euros per unit is fairly high, I think.\nSpeaker A: So I think we can afford to add these sort of features into our remote.\nSpeaker D: At least we have a couple of months to work on it, so it will be cheaper by the way.\nSpeaker D: It's true.\nSpeaker B: There's D-H, but...\nSpeaker B: Okay.\nSpeaker B: Oh.\nSpeaker C: Oh, I hope I'm...\nSpeaker A: That's the wrong one, I think.\nSpeaker B: Yeah, it's still more...\nSpeaker C: Presentation tree?\nSpeaker C: Because you should have put yes.\nSpeaker C: Oh, yes.\nSpeaker C: Oh, yes.\nSpeaker C: Okay.\nSpeaker B: Okay, so here is my presentation about technical function design.\nSpeaker B: I will talk about different components of system and how they react together.\nSpeaker B: Okay.\nSpeaker B: First, what is the remote control?\nSpeaker B: Simply device, as you know, for sending some commands by some waves to another device to tell different commands with this device.\nSpeaker B: And the main idea is we don't use any cable and we can react simpler with the device.\nSpeaker B: It has different blocks, different blocks.\nSpeaker B: First, this remote control should have some electric circuits, making interface with keyboards and reading the keyboard commands and then make these keyboard commands, interpret these keyboard commands, and then there should be an electronic circuit making electronic signals according to these commands.\nSpeaker B: And finally, there is a transmitter which is a cord or a diode making waves to transmit through the air and this wave will be received by the other device like a television or whatever to realize their command.\nSpeaker B: Okay.\nSpeaker B: About what I found about these different blocks are usually there are two different metals for designing a remote control.\nSpeaker B: They are based on infrared waves or radio waves.\nSpeaker B: There are two different solutions.\nSpeaker B: This can be the differences between different kind of waves, infrared or radio waves.\nSpeaker B: And also, as I understood and I think it was a part of the presentation, people prefer to have remote control with less bottom.\nSpeaker B: So for the electronic part working and interfacing with buttons, we should try to design a remote control, having some simpler buttons or some rolling buttons just to search between different options and showing something on TV and putting less stuff on the remote control.\nSpeaker B: And personal preferences are certainly a remote control with working with radio waves is preferred because you can take it in any direction and you don't need to tune it any way.\nSpeaker B: And again, using bigger buttons and less number of buttons are also preferred.\nSpeaker B: As I see. Okay. That was my presentation.\nSpeaker A: I have a question.\nSpeaker A: Do you think radio waves will interfere with other appliances in the home?\nSpeaker B: I don't think so because we can make this wave in a specific frequency.\nSpeaker B: So they can be in a range which is not interfering with other devices inside the home.\nSpeaker C: Can we use any frequency? We have the right to use any frequency?\nSpeaker B: No, but as I know there is a range for this stuff, for designing this circuit.\nSpeaker B: We can tune our transmitter to work in this range.\nSpeaker B: For this range, we don't need to ask any permission.\nSpeaker C: Okay. And what happens with radio waves when two neighbors have the same remote control, for example?\nSpeaker C: Do they have the same frequency?\nSpeaker B: For this, I don't know the solution, but one solution can be something like putting a password or something inside the wave.\nSpeaker B: So only your team can understand identification.\nSpeaker D: But since it's exactly my feel so, it's kind of hand shaking when you start to communicate with your TV.\nSpeaker D: It's like hand shaking protocol with your remote.\nSpeaker D: So if two devices are trying to communicate with the TV set, then the one which has more energy in the wave is chosen.\nSpeaker D: This can be a problem sometimes, but most of the time it works okay.\nSpeaker B: The task force may simply or identification code may simply solve this problem, a specific remote control.\nSpeaker D: Yeah, but we don't have to think about this because I think as a function designer that we will use the already made circuits, which we probably buy its words to buy, and they have these problems solved.\nSpeaker D: So we don't have to think about these.\nSpeaker C: So maybe you can talk about the function.\nSpeaker D: Yes, I have only a couple of things because I had struggled a bit with the software that I'm supposed to use in this company.\nSpeaker D: I was used to use Linux before, but I tried to break through this.\nNone: Yes.\nSpeaker D: Okay.\nSpeaker D: So I will speak about working design.\nSpeaker D: That's the first slide.\nSpeaker D: Okay, so presently I'm looking what is possible to use, what circuits to use and stuff like that because I didn't work with these circuits so far.\nSpeaker D: So I have to look what is available on the market for the communication for the IR circuits and so on.\nSpeaker D: So I'm currently looking what is available on the web and I wanted to ask you maybe afterwards after our discussion if we have some contacts in some company, which can report on what is going on there.\nSpeaker D: So I would be glad if you can tell me about that.\nSpeaker D: Okay, findings, that's the point that I'm working on currently, but so far I was looking what are the circuit radio wave radio frequency circuits are available now, but the prices are rather high.\nSpeaker D: So I know that the user interface people and these speak about radio frequency waves because you can make the TV do what you want, even if you are in the bathroom or so on.\nSpeaker D: But when you're not close to the TV you probably want to need to change the program and so on.\nSpeaker D: We will discuss it later maybe.\nSpeaker A: Are we planning to take off the shelf and OEM component for the radio wave circuit or are we planning to construct their own circuit board?\nSpeaker D: So we just buy a circuit board.\nSpeaker D: So I agree on using any kind of LCD, less buttons good for me as a designer of the circuit.\nSpeaker D: But the speech recognition we have to compare whether the price and what does it offer.\nSpeaker C: Yeah, I'm surprised it would be out of French or it would be maybe visible.\nSpeaker D: I was not thinking too much about the price, but if we use the LCD, even the radio frequency communication with the TV set and the backlight and related things like the photodiodes and stuff, it should be okay.\nSpeaker D: If we decide to use the speech recognition then we probably could struggle, but we'll see afterwards.\nSpeaker D: It depends on if the M.E. or U.I.D. wants to have it like in metal or in plastics these things.\nSpeaker D: It depends on you not because the electric devices, prices not so big in comparison to the overall shape and stuff like that.\nSpeaker D: We will discuss it afterwards.\nSpeaker D: This is nothing, this is just my note on what to use.\nSpeaker D: My personal preference is yes, I would like also preferably to use R.W. circuit, but from the point of view of the design and price, I would stick to the IRs, that's my opinion.\nSpeaker D: I mean if infrared circuit not the radio frequency.\nSpeaker C: Why? Because it's simpler.\nSpeaker D: Because the range where you can use it is fair, it's okay I think and the price is fairly cheap.\nSpeaker D: It's a price matter.\nSpeaker D: Just a price, otherwise I don't care what I put there because it's the chip which I buy or which we buy.\nSpeaker A: How much more expensive are we talking three times more expensive?\nSpeaker A: Three times more expensive?\nSpeaker D: Three to five.\nSpeaker D: Not ten times, but it depends.\nSpeaker A: That's a lot.\nSpeaker A: I think it's probably not worth spending the extra money because all the other remote controls on the market have infrared.\nSpeaker A: So people don't expect anything other than infrared.\nSpeaker D: So it's not worth spending the extra money.\nSpeaker D: They can't see the TV.\nSpeaker C: On the other side we want to have something new.\nSpeaker C: We want to have something new.\nSpeaker C: I think we should still think about it.\nSpeaker A: Based on my usability studies I feel that users prioritize the look and the feel and the trendiness above the difference between infrared or radio waves.\nSpeaker A: So I think we're better off spending money.\nSpeaker D: In the usability.\nSpeaker C: I have to inform you I receive an email from the management board today and they have new requirements for the remote control.\nSpeaker C: First, they said that something about teletext.\nSpeaker C: Apparently it becomes, according to them it becomes out of date, outdated.\nSpeaker C: Because of the internet popularity and everybody else internet at home and actually it's not useful to have teletext.\nSpeaker C: So I think we can avoid teletext.\nSpeaker C: The second thing is they suggest that we should use the remote control only for TV, not for DVD and other devices.\nSpeaker C: Because it makes it too complex and because we have no much time for this project we should stay on the specific TV remote control.\nSpeaker C: The third one is about the image of the company.\nSpeaker C: So we should keep the product should be recognized ever.\nSpeaker C: That means we should use the color of our company and maybe put somewhere the slogan of the company which is we put fashion in electronics.\nSpeaker C: So when people see the remote control they should say oh it's from a reaction and they should recognize the company.\nSpeaker C: So now we should take the decision what function we will have on this remote control.\nSpeaker C: So are we going to use LCD speech recognition?\nSpeaker A: Should we start with just the core, the basic functions that we need and then we can move on to the more advanced features?\nSpeaker D: The available things are LCD buttons and everything radio frequency depends and the recognition depends on you guys.\nSpeaker C: But first maybe what are the usual functions of standard control?\nSpeaker A: The obvious one is changing channels.\nSpeaker C: I think we should stick on very useful functions because we want less buttons.\nSpeaker C: So changing channel of course, volume, setting.\nSpeaker D: Just one note to the channel changing will we use only two buttons or numbered buttons?\nSpeaker D: I think it would be a button.\nSpeaker C: Another side we have more and more channels and if you want to pass through all the channels to get the channel you want.\nSpeaker D: Like 10 plus, 5 plus, 1 plus, 1 plus, or using the names and the keyboard.\nSpeaker C: Maybe you could have K buttons like documentary channel and movies.\nSpeaker C: If you want to see a movie you click on the movie button so you have all the movie channels.\nSpeaker D: It requires the use of LCD to tell you all the same feedback.\nSpeaker A: We could maybe incorporate the idea of the slider for the volume with the channel.\nSpeaker A: We have channel and volume.\nSpeaker D: So we are still speaking about common devices or we are inviting.\nSpeaker D: I think so.\nSpeaker A: Basically cool functions we want and then more advanced ones.\nSpeaker C: What about the settings of the TV? Because we don't choose very often but we need it anyway.\nSpeaker B: We don't have any, we don't have visual design, remote control and we don't have any access to the TV design or we can change some design.\nSpeaker B: One solution for this changing channels is to see a summary of all channels, some preview of all channels.\nSpeaker B: On the screen.\nSpeaker B: Not on the screen.\nSpeaker B: On the screen.\nSpeaker D: On the TV screen.\nSpeaker A: I don't think we are the ones.\nSpeaker A: This remote control we are developing is a generic control for all TVs.\nSpeaker D: I think it would be better to stick to the remote control and not bother to TV to print these things.\nSpeaker D: It will be still more expensive but for the LCD and this stuff is no problem in the Pride.\nSpeaker C: So what are we doing with the settings? Because if we want to do settings we need buttons for that and we want less buttons.\nSpeaker C: Maybe with the LCD we can do something with less buttons.\nSpeaker A: But then you don't want to make the LCD display too complicated.\nSpeaker A: At the same time.\nSpeaker A: I mean we can always have these less often used functions hidden somewhere under a cover or a back of under a slide.\nSpeaker D: We could have for example two buttons like simple mode and advanced mode.\nSpeaker D: Or like children and grandfathers modes and the TV manager mode.\nSpeaker A: We have five minutes left.\nSpeaker A: So I think the settings we agreed that they are required.\nSpeaker C: We should hide them somewhere in the menus of the LCD or in the back of the remote control.\nSpeaker A: What else?\nSpeaker A: I mean a power button obviously.\nSpeaker D: Yes. If I was thinking do we need a power button at all?\nSpeaker D: Should we do it like sleep mode after five minutes or not using it?\nSpeaker C: Because generally it's a kind of setting I think.\nSpeaker C: It should fit into settings.\nSpeaker C: Because it's not a very current useful function.\nSpeaker A: Well I think when I say power button they mean to turn the TV on and off.\nSpeaker C: No I think it's after five minutes or a timer.\nSpeaker A: But if you're watching TV for two hours you don't want your TV to turn off.\nSpeaker D: You don't want your TV to turn off.\nSpeaker D: You don't want your TV to turn off.\nSpeaker A: So you should kill this button.\nSpeaker A: I mean based on our usability studies again people said that the power button was a very relevant button.\nSpeaker A: It was nine out of ten.\nSpeaker D: What we could probably do is to keep the LCD and all the buttons and stuff.\nSpeaker D: We could make it like an opening stuff that if you open it you just turn on the TV and if you close it it will.\nSpeaker D: Turn off the TV if you like.\nSpeaker A: Okay.\nSpeaker A: I think we need to concentrate on the major usage of the control which is you sit down, you turn on your TV, you change channels, you change the volume, you turn the TV off.\nSpeaker A: And all the other functionality is not used very often.\nSpeaker B: They can be hired somewhere by a cover or something like this.\nSpeaker B: Like covering.\nSpeaker D: Yeah, like mobile phones.\nSpeaker D: Since we have the LCD we don't need too much buttons.\nSpeaker D: Okay, just the decision on the power button should we make it a button or something which would be for what?\nSpeaker D: I think a button.\nSpeaker A: I think it should be a button.\nSpeaker D: It's a button.\nSpeaker D: Okay, one nice big button.\nSpeaker D: Old-fashioned button to satisfy the grandmothers.\nSpeaker A: Okay.\nSpeaker C: So any other suggestions or functions?\nSpeaker A: What about things like the clock and timers?\nSpeaker D: Do we still have the time?\nSpeaker D: I just wonder.\nSpeaker C: Yeah, we have still one or two minutes.\nSpeaker D: What is the question?\nSpeaker A: You know, some features on their control to display a time or to display.\nSpeaker D: You know, TV or something like that.\nSpeaker D: But since we want to control all the televisions, it would probably be worse to set the timing on the remote.\nSpeaker D: Because if the TV turns on itself, if we add the time, we should be there.\nSpeaker C: Maybe a bigger display or something like that.\nSpeaker C: I mean, our users want to have the time on the remote.\nSpeaker A: It's a question.\nSpeaker C: Yeah, it's a trade-off.\nSpeaker C: In my opinion, we should have only useful things.\nSpeaker C: Because apparently the simplest is better.\nSpeaker D: Did you ever use the timing of turning on the TV?\nSpeaker D: And based on the...\nSpeaker A: Yeah, okay, very occasionally.\nSpeaker A: But I do use the display of the time quite often.\nSpeaker D: Yeah, but it can be on the display in the corner all the time on the remote.\nSpeaker A: Yeah, that's okay.\nSpeaker A: That's true.\nSpeaker B: Okay, so no time.\nSpeaker B: Do we need to include anything about the speech recognition and speech commands?\nSpeaker B: I think it will not take lots of place.\nSpeaker B: We just need a microphone and a software.\nSpeaker B: And it will be a little bit new and interesting.\nSpeaker B: So people may be attracted to buy this stuff.\nSpeaker B: And it's not very difficult to put a software inside the electronic device and put a microphone.\nSpeaker B: It doesn't take that much place and also that much.\nSpeaker B: It doesn't cost that much.\nSpeaker D: We should be careful about the battery life then.\nSpeaker C: So maybe we can think more about that and discuss that maybe last time.\nSpeaker C: Next time.\nSpeaker C: And now the meeting room is busy.\nSpeaker C: Somebody booked a meeting room just at 1pm.\nSpeaker C: So we should live.\nSpeaker C: So no, we are going for a small lunch.\nSpeaker C: It's founded by the company.\nSpeaker C: And after we have 13 minutes to do individual works.\nSpeaker C: And I will do the minutes.\nSpeaker C: And you are going to work on your individual works.\nSpeaker C: And you will receive as usual specific instructions and so on.\nSpeaker A: Okay, thank you.\nSpeaker A: Cool.\nSpeaker B: Thank you.\nNone: Thank you.\n"}, {"meeting_id": "Bmr024", "summary": "The participants discussed how meetings would be transcribed, what kind of information to include in their corpus as well as how to structure it, issues with storing data, and their model. They were particularly concerned with how IBM could assist with transcribing meetings and how they would manage large amounts of data if they include more information in their corpus, given that they were running low on storage. They decided that they could store the data on tapes for backup, and that they would wait and see how IBM transcribes their meetings. As for the modeling, PhD I reported several results and a few members of the team decided to further discuss progress in a smaller meeting later on.", "dialogue": "Speaker E: So we're on.\nSpeaker E: And somewhere is my agenda.\nSpeaker E: I think the most important thing is Morgan wanted to talk about the ARPA demo.\nSpeaker J: Well, so here's the thing.\nSpeaker J: Why don't we start off with, again, I'll get it.\nSpeaker J: I think we want to start off with the agenda.\nSpeaker J: And then given that, listen, Andreas, we're going to be 10, 15 minutes late.\nSpeaker J: We can try to figure out what we can do most effectively without them here.\nSpeaker J: So one thing is, yeah, talk about demo.\nSpeaker E: IBM transcription status.\nSpeaker J: IBM transcription.\nSpeaker J: What else?\nSpeaker J: SmartCom.\nSpeaker J: SmartCom.\nSpeaker E: What's SmartCom?\nSpeaker E: We want to talk about, we want to add the data to the meeting recorder corpus.\nSpeaker F: The data which we are collecting here.\nSpeaker J: What are we collecting here?\nSpeaker E: So why don't we have that on the agenda and we'll get to it and talk about it.\nSpeaker J: Yeah, right.\nSpeaker J: Right.\nSpeaker E: Reorganization status.\nSpeaker J: Reorganization status.\nSpeaker J: Files and directories.\nSpeaker E: Absent, which is the multi-processor Unix Linux.\nSpeaker E: I think it was, Andreas wanted to talk about segmentation and recognition.\nSpeaker E: And update on SRI recognition experiments.\nSpeaker E: And then if there's time I want to talk about digits, but it looked like we were pretty full, so I can wait till next week.\nSpeaker J: Right.\nSpeaker J: Well, let's see.\nSpeaker J: I think the segmentation recognition we want to maybe focus on when Andreas is here, since that was particularly his.\nSpeaker J: And also the small, smartCom also.\nSpeaker J: Andreas.\nSpeaker J: Absent, I think also he has sort of been involved in a lot of things.\nSpeaker J: At least he'll probably be interested.\nSpeaker J: So I mean, I think they'll be interested in all this, but probably if we had to pick something that we would talk on for 10 minutes or so while they're coming here, I guess it would be your organization status.\nSpeaker E: Yeah, I mean, I think Chuck was the one who added up the agenda item.\nSpeaker E: I don't really have anything to say of them, and we still haven't done it.\nSpeaker K: Well, I mean, just basically that maybe I said, maybe we said this before, just that we met and we talked about it and we sort of have a plan for getting things organized.\nSpeaker B: And I think crucial for that is that he's not wanting to do it until right before the next level zero back up,\nSpeaker K: so that there won't be any change. Right.\nSpeaker K: That was basically it.\nSpeaker E: Although Dave basically said that if we want to do it, just tell him and he'll do it level zero then.\nSpeaker K: So maybe we should just go ahead and get everything ready.\nSpeaker E: So I think we do need to talk a little bit about, well, we don't need to do it during this meeting.\nSpeaker E: We have a little more to discuss, but we're basically ready to do it.\nSpeaker E: And I have some web pages on more of the background.\nSpeaker E: So naming conventions and things like that, I've been trying to keep actually up to date.\nSpeaker E: And I've been sharing them with you, UW folks also.\nSpeaker E: Sharing them with the UW folks.\nSpeaker J: Okay. Well, maybe, since I was pretty sure when maybe we should talk about the IBM transcription status, I'm feeling there's an address later.\nSpeaker E: Okay. So we did another version of the beeps where we separated each beeps with spoken digit.\nSpeaker E: Chuck came up here and recorded some, himself speaking some digits.\nSpeaker E: And so it just goes beep, one beep, and then the phrase, and then beep, two beep, and then the phrase.\nSpeaker E: And that seems pretty good. I think they'll have an easier time keeping track of where they are in the file.\nSpeaker G: Maybe we have to on the automatic segmentation.\nSpeaker E: And we did it with the automatic segmentation. And I don't think we didn't look at it in detail.\nSpeaker E: We just sent it to IBM. We sort of spot checked it.\nSpeaker K: I listened to probably five or ten minutes of it from the beginning.\nSpeaker E: Oh, really? Okay. And I sort of spot checked here and there, and it sounded pretty good.\nSpeaker E: So I think it will work. And we'll just have to see what we get back from them.\nSpeaker K: The main thing will be if we can align what they give us with what we sent them.\nSpeaker K: I mean, that's the crucial part. Right? And I think we'll be able to do that with this new beep format.\nSpeaker E: Yep. Well, I think it's also there much less likely to have errors.\nSpeaker E: So the problem with last time is that there were errors in the transcripts where they put beeps where there weren't any.\nSpeaker E: And they put in extraneous beeps.\nSpeaker E: Yeah. And with the numbers there, it's much less likely.\nSpeaker K: Yeah, one interesting note is a problem. I don't know if this was just because of how I play it back.\nSpeaker K: I say, uh, SND play in in the file. Every once in a while, like a beep sounds like it's cut into two beats.\nSpeaker K: Yeah, I don't know if that's a hard effect of playback.\nSpeaker K: I don't think it's probably in that original file. I recommend that too.\nSpeaker E: That's interesting. I didn't hear that. Yeah.\nSpeaker K: But with this new format that hopefully they're not hearing that.\nSpeaker K: And if they are, it shouldn't throw them.\nSpeaker E: So well, maybe Vera was to it again. Make sure. But I mean, certainly the software shouldn't do that.\nSpeaker K: Yeah, that's what I thought. It's probably just, you know, somehow the audio device gets pickups in 30 seconds.\nSpeaker G: So they have one number and they know that there's only one beat next one.\nSpeaker E: Yeah. Yeah. The only part that might be confusing is when Chuck is reading digits.\nSpeaker E: Right.\nSpeaker E: Seven, four, eight, beep, seven, beep, eight, three, two.\nSpeaker E: Yes. Because we don't cut those out.\nSpeaker E: In order to cut them out, we'd have to listen to it. And we wanted to avoid doing that.\nSpeaker E: So they are transcribing the digits.\nSpeaker E: Although we can tell them or it.\nSpeaker E: We could tell them if you hear someone reading a digit string, just say bracket digit bracket and don't bother actually computing the did writing down the digits.\nNone: That'd be what I'm having the transcript.\nNone: It's here to visit.\nSpeaker E: Yeah. And then I wanted to talk about, but as I said, we may not have time.\nSpeaker E: What we should do about digits. We have a whole pile of digits that haven't been transcribed.\nSpeaker J: Let's talk about it because that's that's something that I know Andreas is less interested in.\nSpeaker E: Okay. Do we have anything else to say about transcription about IBM stuff?\nSpeaker K: Brian, I sent, I sent Brian a message about the meeting and I haven't heard back yet.\nSpeaker K: So I hope he got it and hopefully he's maybe he's gone. I don't know.\nSpeaker K: He didn't even reply to my message. So I should probably ping him just to make sure that he got it.\nSpeaker E: Right.\nSpeaker E: So we have a whole bunch of digits. I pretty want to move on to digits.\nSpeaker J: Actually, I want to relate more related thing in transcription. So that's the IBM stuff. We got that sorted out.\nSpeaker J: How are we doing on the on the rest of it?\nSpeaker B: I'm doing well. I hire two extra people already expected to hire two more.\nSpeaker B: I've prepared a set of five, which I'm calling set to, which are now being edited by my head transcribed in terms of spelling here as well.\nSpeaker B: She's also checking through and monitoring the transcription of another transcribed.\nSpeaker B: She's going through out of these kinds of checks.\nSpeaker B: And I've moved on now to one of the calling sets. I sort of thought if I do it and set through supply, then I can have like sort of a terrible processing through the current.\nSpeaker B: And I knew indicated to me that we have a goal now for the dark but demo of 20 hours.\nSpeaker B: So I'm going to go up to 20 hours. Be sure that everything that's processed and released.\nSpeaker B: And that's what level is. Package of 20 hours right now.\nSpeaker B: And once that's done, we'll have to do that.\nSpeaker J: Yeah. So 20 hours, but I guess the other thing is that that's kind of 20 hours ASAP because the longer before the demo, we actually have 20 hours. The more time it'll be for people to actually do cool things with it.\nSpeaker B: I don't think it's possible.\nSpeaker B: Thank you for accuracy.\nSpeaker J: Yeah.\nSpeaker J: Yeah. I mean, I guess the difference if the IBM stuff works out, the difference in the job would be that they primarily would be checking through things that were already done by someone else.\nSpeaker E: Incorrected. Correcting. We'll expect that they'll have to move some time bins and do some corrections.\nSpeaker B: And I've also discovered so with a new transcriber.\nSpeaker B: So let me say that like.\nSpeaker B: So at present, people have been doing these transcriptions at the channel all the time.\nSpeaker B: And that sort of is useful.\nSpeaker B: And once in a while they'll have to refer to the other channels to create something.\nSpeaker B: Well, I realized that we're using the pre-segmented version.\nSpeaker B: And the pre-segmented version is extremely useful.\nSpeaker B: And wouldn't it be useful also to have the visual representation of those segments?\nSpeaker B: And so I trained the new one, the newest one, to use the visual from the channel that is going to be transcribed at any given time.\nSpeaker B: And that's just amazing, we helpful. Because what happens then is you scan across the signal.\nSpeaker B: And once in a while you'll find a blip that didn't show up in the pre-segmentation.\nSpeaker B: Oh, right. I see what you mean.\nSpeaker B: And it's something like a back channeler.\nSpeaker B: Sometimes it seems to be similar to the ones that are being picked up.\nSpeaker B: And they're rare events, but you can really go through a meeting very quickly.\nSpeaker B: You just scroll from screen to screen looking for blips.\nSpeaker B: I think that we're going to end up with better coverage of the back channels, but at the same time we're benefiting tremendously for the pre-segmentation.\nSpeaker B: Because there are huge places where it's just absolutely no activity at all.\nSpeaker K: And the audio quality is...\nSpeaker K: So they can scroll through that pretty quick.\nSpeaker K: Yeah.\nNone: That's great.\nSpeaker B: So I think that that's going to also see the efficiency of this process.\nSpeaker J: Okay. Yeah.\nSpeaker J: So, yeah, so let's talk about the digits in the interior.\nSpeaker E: So we have a whole bunch of digits that we've read and we have the forms and so on.\nSpeaker E: But only a small number of that...well, not a small number.\nSpeaker E: Only a subset of that has been transcribed.\nSpeaker E: And so we need to decide what we want to do.\nSpeaker E: And Liz and Andreas, actually, they're not here, but they did say at one point that they thought they could do a pretty good job of just doing a forced alignment.\nSpeaker E: And again, I don't think we'll be able to do with that alone because sometimes people correct themselves and things like that.\nSpeaker E: But so I was just wondering what people thought about how automated can we make the process of finding where the people read the digits, doing a forced alignment and doing the timing.\nSpeaker J: Well, forced alignment would be one thing. What about just actually doing recognition?\nSpeaker E: Well, we know what they read because we have the forms.\nSpeaker E: No, they make mistakes.\nSpeaker E: Right. But the point is that we want to get a set of clean digits.\nSpeaker K: You're talking about it's a pre-processing step, right, Morgan?\nSpeaker J: Is that what you're hearing?\nSpeaker J: I'm not quite sure what I'm talking about.\nSpeaker J: I mean, we're talking about digits now.\nSpeaker J: And so there's a bunch of stuff that hasn't been marked yet.\nSpeaker J: And it's the one option.\nSpeaker J: I was just asking you to start a curiosity.\nSpeaker J: If with the SRI recognized that we're getting 1% word error, would we do better?\nSpeaker J: So if you do a forced alignment, but the transcription you have is wrong because they actually made mistakes.\nSpeaker J: But that's pretty uncommon.\nSpeaker J: It's much less common than 1%.\nSpeaker E: If we could really get 1% on, well, I guess if we segmented it, we could get 1% on digits.\nSpeaker J: Yeah. So that's just my question.\nSpeaker J: I'm not saying it should be one way or the other, but it's...\nSpeaker E: Well, there are a couple different ways of doing it.\nSpeaker E: We could use the tools I've already developed and transcribe it.\nSpeaker E: Hire some people or use the transcribers to do it.\nSpeaker E: We could let IBM transcribe it.\nSpeaker E: They're doing it anyway, and unless we tell them different, they're going to transcribe it.\nSpeaker E: Or we could try some automated methods.\nSpeaker E: Well, my tendency right now is, well, if IBM comes back with this meeting and the transcript is good, just let them do it.\nSpeaker J: Yeah, you raised a point kind of...\nSpeaker J: You've missed it, but maybe it is a serious problem.\nSpeaker J: What will they do when they go?\nSpeaker J: Here beep, 7, beep, 7, 3, 5, 2...\nSpeaker J: I mean, you think though...\nSpeaker M: It's pretty distinct. The beeps are pre-ordered from mine.\nNone: I'll let me proceed by reading transcripts.\nSpeaker E: So also...\nNone: Yes.\nSpeaker E: I mean, it will be in the midst of a digit string.\nSpeaker E: Sure, there might be a place where it's beep, 7, beep, 8, beep, 8, beep.\nSpeaker E: But they're going to have macros for inserting the beep marks.\nSpeaker E: And so I don't think it will be a problem. We'll have to see it, but I don't think it's going to be a problem.\nSpeaker J: Okay. Well, I don't know. I think that if they are, in fact, going to transcribe these things, certainly any process that we'd have to correct them or whatever needs to be much less elaborate for digits.\nSpeaker J: Right.\nSpeaker J: For other stuff, so why not?\nSpeaker J: That was it?\nSpeaker E: That was it. Just what do we do with digits? We have so many of them.\nSpeaker E: That would be nice to actually do something with them.\nSpeaker F: Yeah.\nSpeaker F: Your mic is a little lower.\nSpeaker J: And Berkeley, yeah.\nSpeaker J: So...\nSpeaker J: You have to go early, right?\nSpeaker H: Well, I think about...\nSpeaker H: Speak 40.\nSpeaker J: All right. So let's make sure we do the ones that...\nSpeaker J: So there are some...\nSpeaker J: In...\nSpeaker J: In Adams's agenda list, he had something from you about segmentation in the last record.\nSpeaker H: Well, yeah. So this is just...\nSpeaker H: Partly to form everybody and help us to get input.\nSpeaker H: So we had a discussion...\nSpeaker H: And I had a discussion actually about how to proceed with...\nSpeaker H: With Don's work and...\nSpeaker H: And one of the obvious things that occurred to us was that...\nSpeaker H: We were, since we now have feelings, segmentation, and it works, you know, amazingly well.\nSpeaker H: We should actually basically re-evaluate the recognition...\nSpeaker H: And results using, you know, without cheating on segmentation.\nSpeaker G: And so...\nSpeaker G: And how do we find the transcripts for those?\nSpeaker G: So the...\nSpeaker G: Yeah. The reference is for the segment.\nSpeaker G: That's not that...\nSpeaker C: Yeah, again.\nSpeaker H: And once.\nSpeaker H: It's a very sophisticated scoring program.\nSpeaker H: That you can give a...\nSpeaker H: A time...\nSpeaker H: You know, you basically just give two time mark sequences of words.\nSpeaker H: And it computes...\nSpeaker K: You know, it does all the work for you.\nSpeaker H: So we just...\nSpeaker H: And we use that actually in half-five to do a scoring.\nSpeaker H: So what we've been using so far was sort of a simplified version of the scoring.\nSpeaker H: And we can handle the type of problem we have here.\nSpeaker G: So basically you give some time constraints for references and for the hypothesis.\nSpeaker I: Yeah, maybe the start of your speech in the end of it or something like that.\nSpeaker H: It does time constraint where you line up.\nSpeaker H: Okay.\nSpeaker H: So that should be possible.\nSpeaker H: I mean, that should be...\nSpeaker H: So that was the one thing and the other was that...\nSpeaker H: Of course, you got a problem. Oh, that Taylor wanted to use the recognize or linements to train up his speech detector.\nSpeaker H: So that you could use...\nSpeaker H: You know, there would be so much hand labeling that to generate training data.\nSpeaker G: Yeah, I'm just in progress of doing that.\nSpeaker G: And I think you're in the positive way that...\nSpeaker K: I'll give you a lot more data, too, I want it.\nSpeaker G: So it's basically, I think, eight meetings or something which I'm using.\nSpeaker G: Before it was 20 minutes of one meeting, so...\nSpeaker G: It should be a little bit more.\nSpeaker H: The alignment's already perfect.\nSpeaker H: Yeah, but it's probably still better than all this extra data.\nSpeaker G: We'll see that.\nSpeaker I: Actually, I had a question about that.\nSpeaker I: If you find that you can lower the false alarms that you get where there's no speech, that would be useful for us to know.\nSpeaker I: There were the false alarms.\nSpeaker I: Right now you get false speech regions when it's just like a breath or something like that.\nSpeaker I: And I'm interested to know that if you retrain, do those actually go down or not?\nSpeaker G: Yeah, I can make it comparison of the old system to the new...\nSpeaker I: Yeah, just to see if by doing nothing in the modeling of just having that training data, what happens?\nSpeaker J: Yeah, another one that we had on the Adams agenda that definitely involved you with something about SmartCom.\nSpeaker E: Right, so Rob Porzel and the...\nSpeaker E: Porzel and the SmartCom group are collecting some dialogues.\nSpeaker E: Basically, they have one person sitting in here looking at a picture and a wizard sitting in another room somewhere.\nSpeaker E: And they're doing a travel task.\nSpeaker E: And it involves starting. I believe starting with a...\nSpeaker E: It's always the wizard, but it starts where the wizard is pretending to be a computer and it goes through a speech generation system.\nSpeaker G: Yeah, actually it's changed to a synthesis system.\nSpeaker E: A synthesis system.\nSpeaker E: And then it goes to a real wizard and they're evaluating that.\nSpeaker E: And they wanted to use this equipment.\nSpeaker E: And so what the question came up is, well, here's some more data.\nSpeaker E: Should this be part of the corpus or not?\nSpeaker E: And my attitude was yes, because there might be people who are using this corpus for acoustics as opposed to just for language.\nSpeaker E: Or also for dialogue of various sorts.\nSpeaker E: So it's not a meeting, right, because it's two people and they're not face to face.\nSpeaker J: So I just wanted to understand, because I haven't quite followed this process.\nSpeaker J: So it's wizard in the usual sense that the person who is asking questions doesn't know that it's a machine.\nSpeaker J: At the beginning.\nSpeaker H: I don't know who came up with the flip, but I think it's a little out of the end.\nSpeaker H: We simulate a computer breakdown and go through the session.\nSpeaker H: So after that, the person's told that they're now talking to a...\nSpeaker F: To a human operator.\nSpeaker E: Of course they don't know that it's the same person both times.\nSpeaker H: Computer and computer data are the same.\nSpeaker J: You might want to try collecting it the other way around sometimes saying that the computer isn't up yet.\nSpeaker J: So then separate out whether it's the beginning or end.\nSpeaker E: I have to go now. You can talk to the computer.\nSpeaker K: So if you tell them that the computer part is running on a Windows machine, a breakdown thing kind of makes sense.\nSpeaker L: A birthday try fail.\nSpeaker I: So do they actually save the far field data?\nSpeaker E: That's the question. So they were saying they were not going to.\nSpeaker E: Well, let's silly if we're going to try to do it for a corpus.\nSpeaker E: There might be people who are interested in acoustics.\nSpeaker G: We were not saying we're not doing it.\nSpeaker H: We just wanted to do it.\nSpeaker E: I see no reason not to do all of them.\nSpeaker E: That if we have someone who is doing acoustic studies, it's nice to have the same for every recording.\nSpeaker J: So what is the purpose of this recording?\nSpeaker J: The acoustic and language model.\nSpeaker J: Training data for smart time.\nSpeaker I: We can have him vary the microphones.\nSpeaker E: For their usage, they don't need anything.\nSpeaker F: But I'm going to try about the legal aspect of that.\nSpeaker F: Is there some contract with smart come or something about the data?\nSpeaker F: Is that our data?\nSpeaker J: We've never signed anything that said that we couldn't use it.\nSpeaker F: That's the question.\nSpeaker J: That's not a problem.\nSpeaker J: It seems to me that if we're doing it anyway and we're doing it for the purposes that we have, and we have these distant mics, we should save it all as long as we've got this space.\nSpeaker J: This is pretty cheap.\nSpeaker J: We save it because it's potentially useful.\nSpeaker J: Now what do we do with it?\nSpeaker J: Anybody who's training something up could choose to include this or not.\nSpeaker J: I would not say it was part of the meetings corpus.\nSpeaker J: But it's some other data we have.\nSpeaker J: If somebody doing an experiment wants to train up including that, then they can.\nSpeaker E: I guess it begs the question of what is the meeting corpus.\nSpeaker E: If at UW, they start recording two person, how many conversations is that part of the meeting corpus?\nSpeaker J: I think the idea of two or more people conversing with one another is key.\nSpeaker E: Well this has two or more people conversing with each other.\nSpeaker I: We just give it a name.\nSpeaker I: That was my intention.\nSpeaker I: Later on some people will consider it a meeting.\nSpeaker E: That was my intention.\nSpeaker E: Part of the reason that I wanted to bring this up is do we want to handle it as a special case?\nSpeaker E: Or do we want to fold it in?\nSpeaker E: Would give everyone who's involved as their own user ID, give it session IDs, and all the tools that handle meeting record or handle it, or do we want a special case?\nSpeaker E: If we're going to special case it, who's going to do that?\nSpeaker H: It's the next answer to the handler with the same infrastructure since we don't want to do the meetings necessarily.\nSpeaker H: But as far as distributing it, we shouldn't label it as part of this meeting corpus.\nSpeaker E: I don't see why not.\nSpeaker E: It's just a different topic.\nSpeaker B: It's a scenario based.\nSpeaker J: It's human computer interface.\nSpeaker J: It's really pretty different.\nSpeaker J: But I have no problem with somebody folding it in for some experiment they're going to do.\nSpeaker J: But I don't think it doesn't match anything that we've described about meetings.\nSpeaker J: Whereas everything that we talked about them doing at UW and so forth really does.\nSpeaker E: So what does that mean for how we're going to organize things?\nSpeaker J: Again, as I think Andres was saying, if you want to use the same tools and the same conventions, there's no problem with that.\nSpeaker J: It's just that it's a different directory.\nSpeaker J: It's called something different.\nSpeaker J: It is different.\nSpeaker J: You can't just fold it in as if it's, I mean, digits are different too, right?\nSpeaker E: Yeah, but those are folded in and you just mark the transcripts differently.\nSpeaker E: So one option is you fold it in and just simply in the file, you mark somewhere that this is this type of interaction rather than another type of interaction.\nSpeaker J: Well, I wouldn't call reading digits meetings, right?\nSpeaker J: I mean, we have to do it.\nSpeaker E: But I put it under the same directory tree.\nSpeaker E: It's in user doctors, speech data, and more.\nSpeaker I: You just have to have it called like other stuff and other.\nSpeaker I: Well, I don't know.\nSpeaker I: I mean, I don't care what the directory is.\nSpeaker J: I mean, that's just.\nSpeaker E: My preference is to have a single procedure so that I don't have to think too much about things and just have a marking.\nSpeaker E: If we do it any other way, that means that we need a separate procedure.\nSpeaker E: If you're someone who has to do that.\nSpeaker J: And so whatever procedure you want that's convenient for you, all I'm saying is that there's no way that we're going to tell people that reading digits is meetings.\nSpeaker J: Right.\nSpeaker J: And similarly, we're not going to tell them that someone talking to a computer to get travel information is meetings.\nSpeaker J: Those aren't meetings.\nSpeaker J: But if it makes it easier for you to put fold them in the same procedures and have them on the same directory tree and act yourself out.\nSpeaker K: There's a couple other questions that I have too.\nSpeaker K: And one of them is what about consent issues?\nSpeaker K: And the other one is what about transcription?\nSpeaker K: Transcription is not a Munich.\nSpeaker K: Okay, so we don't have to worry about transcription.\nSpeaker E: So we will have to worry about format.\nSpeaker H: So that's an argument to keep it separate because it's going to follow the smart contrast description.\nSpeaker H: Oh, okay.\nSpeaker H: Okay, well, I didn't realize that.\nSpeaker E: That's a good point.\nSpeaker J: But I'm sure no one would have a problem with our folding it in for some acoustic modeling or something.\nSpeaker J: Do we have American-born folk reading German, German place names and so forth?\nSpeaker J: Yep, yep, yep.\nSpeaker E: Great.\nSpeaker E: They even have a reading bus.\nSpeaker F: That sounds good, right?\nSpeaker F: You can do that if you want.\nSpeaker J: I don't know if you want that.\nSpeaker J: So...\nSpeaker J: High-dollberg.\nSpeaker E: Disk might eventually be an issue.\nSpeaker E: So we might...\nSpeaker E: Yeah, I'd be pretty good.\nSpeaker E: We might need to get some more disk pretty soon.\nSpeaker E: We're about half way through our disk right now.\nSpeaker E: Are we only half?\nSpeaker E: I thought we were more than that.\nSpeaker E: We're probably a little more than that because we're using up some space that we shouldn't be on.\nSpeaker E: So once everything gets converted over to the disks we're supposed to be using, we'll be probably 75%.\nSpeaker K: Well, when I was looking for space for Tilo, I found one disk that had...\nSpeaker K: I think it was nine gigs and another one had 17.\nSpeaker K: And everything else was sort of committed.\nSpeaker E: Were those backed up or non-backed up?\nSpeaker E: Those were non-backed.\nSpeaker E: Right, so that's different.\nSpeaker E: So you're talking about backed up?\nSpeaker E: I'm much more concerned about the backed up than non-backed up.\nSpeaker E: I haven't looked to see how much of that is.\nSpeaker E: It's cheap.\nSpeaker E: I mean, if we need to, we can buy a disk, hang it off a workstation.\nSpeaker E: If it's not backed up, the citizens don't care too much.\nSpeaker J: Yeah, so I mean, pretty much anytime we need a disk we can get it at the right thing we're...\nSpeaker H: I'm sure we're saying this, but you can just...\nSpeaker H: You know, since the backed up so every night you can recycle the backed up.\nSpeaker E: Yeah, but that's risky.\nSpeaker A: Yeah, you really should say that.\nSpeaker A: I didn't say that.\nSpeaker E: Yeah, that's right.\nSpeaker J: Detail out.\nSpeaker J: We can't allow Dave to listen to these recordings.\nSpeaker J: Yeah, and there's been this conversation going on about getting another file server.\nSpeaker J: And we do that, we'll take the opportunity and get another big raft of disk, I guess.\nSpeaker E: Well, it's really the backup issue rather than the file server issue.\nSpeaker H: You can use our old file server for this to have data that is very fairly accessed.\nSpeaker H: And then have the fast new file server for data that is heavily...\nSpeaker E: My understanding is the issue isn't really the file server.\nSpeaker E: We could always put more disks on...\nSpeaker E: It's the backup system, which is near saturation, apparently.\nSpeaker K: So, I think the file server could become an issue as we get a whole bunch more new compute machines and we've got, you know, 50 machines trying to access data off of that.\nSpeaker E: But we're all right for now because the network's so slow.\nSpeaker H: I think we've tried more, and someone said this is not reliable.\nSpeaker H: We were going to do it back to the...\nSpeaker H: What about putting the stuff on me?\nSpeaker E: Yeah, that was me.\nSpeaker E: I was the one who said it was not reliable. They were out.\nSpeaker E: Yeah, the...\nSpeaker E: But they're out just from sitting on the shelf?\nSpeaker E: Yep, absolutely.\nSpeaker E: No.\nSpeaker E: Reading right don't hurt them too much unless you scratch them.\nSpeaker E: But the right ones and the read rights don't last.\nSpeaker E: So you don't want to put your un reproducible data on them.\nSpeaker K: We're out after what amount of time you're at two.\nSpeaker J: You're at two?\nSpeaker H: Wow.\nSpeaker H: But if that venue would...\nSpeaker H: Thank you, here much more, I'm planning about...\nNone: Yeah.\nSpeaker H: They're awesome.\nNone: I mean, yeah.\nSpeaker E: I don't know many people who do it on CD.\nSpeaker E: I mean, they're with the most...\nSpeaker J: All the LDC distributions are on CD, right?\nSpeaker J: They're on CD, but they're not...\nSpeaker E: That's not the only source. They have them on disk.\nSpeaker E: And they burn new ones every once in a while.\nSpeaker E: But if you go...\nSpeaker I: Or we have like 30, you know, from 10 years old?\nSpeaker I: No.\nSpeaker A: Yeah, 10 years old, 91.\nSpeaker A: And they're still all fine.\nSpeaker I: They're under where they press.\nSpeaker I: I both. I've burned them and they're still okay.\nSpeaker E: I mean, you...\nSpeaker E: The last ones last for...\nSpeaker E: Well, not forever. They've been finding even those degrade.\nSpeaker E: Oh, see.\nSpeaker E: But the burned ones...\nSpeaker E: I mean, when I say two or three years what I'm saying is that I have had disks which are gone in a year.\nSpeaker E: On the average, it'll probably be three or four years.\nSpeaker E: But you don't want to have your only copy on a media that fails.\nSpeaker E: And they do.\nSpeaker E: If you haven't professionally pressed, you know, they're good for decades.\nSpeaker H: So how about putting them on that plus, like, on that or some other media?\nSpeaker E: I think we can already put them on tape.\nSpeaker E: And the tape is very reliable.\nSpeaker E: So the only issue is then if we need access to them.\nSpeaker E: So that's fine if we don't need access to them.\nSpeaker H: Well, if they last say they actually last like five years.\nSpeaker H: And occasionally you might need to regrade one.\nSpeaker H: And then you get your tape on the other way or two.\nSpeaker H: And you just put them on.\nSpeaker A: So just archive it on the tape and then play them CDs one.\nSpeaker E: Oh, so you're just saying put them on CDs for normal access.\nSpeaker E: Yeah.\nSpeaker E: I mean, you can do that, but that's pretty annoying because the CDs are so slow.\nSpeaker K: Yeah.\nSpeaker K: It would be nice as a system that re-burned the CDs every year.\nSpeaker K: Every time it would.\nSpeaker K: The CDs are an obstacle.\nSpeaker I: Just before it goes bad.\nSpeaker E: The CD is an alternative to tape.\nSpeaker E: XC already has a perfectly good tape system and it's more reliable.\nSpeaker E: So for archiving, we'll just use tape.\nSpeaker H: I would think you're saying this.\nSpeaker H: If you have the data, if the meeting data is put on this exactly once.\nSpeaker H: And it's back that once and the backup system should never have to work with it.\nSpeaker E: Regardless, well, first of all, there was a problem with the archive in that I was every once in a while doing a Chimaud on all the directories, a recursive Chimaud and Chone because they weren't getting set correctly every once in a while.\nSpeaker E: And I was just doing a minus R star, not realizing that that caused it to be re-backed up.\nSpeaker E: But normally you're correct, but even without that, the backup system is becoming saturated.\nSpeaker H: But the backup system is smart enough to figure out that something hasn't changed.\nSpeaker E: Sure, but we still have enough changed that the nightly backups are starting to take too long.\nSpeaker J: I think at least the ones that you put it on.\nSpeaker J: It has nothing to do with the meeting.\nSpeaker E: It's just the general XC backup system is becoming saturated.\nSpeaker H: So what if we buy, what do they call this?\nSpeaker E: Why don't you have this conversation with Dave Johnson rather than with me?\nSpeaker H: Maybe something that we can do without involved in Dave and putting more work on that.\nSpeaker H: How about we buy one of these high density tape drives.\nSpeaker H: And we put the data actually on non-backed up disks.\nSpeaker H: And we do our own backup once and for all.\nSpeaker E: And then we don't have to embark on that.\nSpeaker E: Actually, you know, we could do that just with the tape, with the current tape.\nSpeaker H: I don't know what these tapes, at some point, I don't know what kind of tape I'm going to do.\nSpeaker E: I don't know, but it's an automatic robot, so it's very convenient.\nSpeaker E: You just run a program to restore them.\nSpeaker E: The one that we have.\nSpeaker J: Yeah.\nSpeaker J: Am I going to fit with their backups?\nSpeaker J: No, we have something that isn't used by the backup gang.\nSpeaker J: Don't we have something downstairs?\nSpeaker J: Kind of tapes drive.\nSpeaker E: But Andreas' point is a good one.\nSpeaker E: And we don't have to do anything ourselves to do that.\nSpeaker E: They're already right now on tape.\nSpeaker E: So your point is, and I think it's a good one, that we could just get more disk and put it there.\nSpeaker E: That's not a bad idea.\nSpeaker J: Yeah, that's basically what I was going to say is that disk is so cheap.\nSpeaker J: It's essentially close to free.\nSpeaker J: So one thing that costs is the backup issue into first order.\nSpeaker J: And we can take care of that by putting it on non-backed up drives and just backing it up once onto this thing.\nSpeaker J: I think that's a good idea.\nSpeaker J: Yeah, good.\nSpeaker I: So who's going to do these backups?\nSpeaker I: The people that collect it?\nSpeaker E: Well, I'll talk to Dave and see what the best way of doing that is.\nSpeaker E: It's a little utility that will manually burn a tape for you.\nSpeaker E: And that's probably the right way to do it.\nSpeaker K: Yeah, and we should probably make that part of the procedure for reporting the meetings.\nSpeaker E: Well, we're going to automate that.\nSpeaker E: My intention is to do a script that will do everything.\nSpeaker E: We're going to have to take in the drive.\nSpeaker E: No, it's all tape or a lot.\nSpeaker E: So you just sit down at your computer and you type a command.\nSpeaker H: Yeah, but then you're effectively using the resources when the backup system was at a different time.\nSpeaker I: But not at the same time.\nSpeaker K: But you would be anyway.\nSpeaker K: Right?\nSpeaker K: No, he's saying get a whole different drive.\nSpeaker E: But there's no reason to do that.\nSpeaker E: Well, we already have it there and it's.\nSpeaker H: And asking, can I use your tape for about, you will say, well, that's going to screw up our backup.\nSpeaker H: No, he won't.\nSpeaker E: He'll say, if that means that it's not going to be backed up standardly, great.\nSpeaker J: Dave has promoted this in the past.\nSpeaker J: Yeah, it's definitely no problem.\nSpeaker J: Yeah.\nSpeaker I: What about if the times overlap with the normal backup time?\nSpeaker E: It's just a utility which queues up.\nSpeaker E: Okay.\nSpeaker E: It just queues it up and when it's available, it will copy it.\nSpeaker E: And then you can tell it to then remove it from the disk or you can do it a few days later or whatever you want to do.\nSpeaker E: After you confirm that it's really backed up.\nSpeaker E: NW.\nSpeaker E: You're saying NW, okay?\nSpeaker E: NW archive, that's what it is.\nSpeaker E: Okay.\nSpeaker C: And if you did that during the day, it would never make it to the library.\nSpeaker C: Right.\nSpeaker H: Well, you have to put the data on the non-life disk to the game, right?\nSpeaker H: Right.\nSpeaker B: So that otherwise you don't have to...\nSpeaker B: You can have a non-backed disk NWRI.\nSpeaker E: Right.\nSpeaker E: And then it never...\nSpeaker E: Right.\nSpeaker E: Which I'm sure would make it.\nSpeaker E: The assessment's very happy.\nSpeaker E: So I think that's a good idea.\nSpeaker E: That's what we should do.\nSpeaker E: So that means we'll probably want to convert all those files, file systems to non-backed up media.\nSpeaker K: That sounds good.\nSpeaker K: Yeah.\nSpeaker J: Another thing on the agenda that says SRI recognition experience.\nSpeaker J: What was that?\nSpeaker J: Oh.\nSpeaker J: That wasn't me.\nSpeaker E: Well, we have lots of them.\nSpeaker H: I didn't know you talked to you having the updates.\nSpeaker K: I'm successfully increasing the error rate.\nSpeaker K: That's good.\nSpeaker K: That's the hair value approach.\nSpeaker I: Yeah.\nSpeaker K: So I mean, I'm just playing with the number of Gaussian's that we use in the recognizer.\nSpeaker H: You have to set...\nSpeaker H: You have to type people that you're trying the tenant features.\nSpeaker H: Yes, I'm using tenant features.\nSpeaker K: Oh, you are?\nSpeaker H: Cool.\nSpeaker H: And I still take my work to the PLP features.\nSpeaker J: Yeah, I got confused by the results because the meaning before you said, okay, we got it now and the way there...\nSpeaker J: That was on mail.\nSpeaker J: That was before I tried it on the females.\nSpeaker J: Oh.\nSpeaker J: It's the women of the problem.\nSpeaker J: Okay.\nSpeaker I: Well, let's just say that men are simple.\nSpeaker I: Hello.\nSpeaker I: Hello.\nSpeaker I: I had...\nSpeaker C: It was a quick response.\nSpeaker C: I'm well-rehearsed.\nSpeaker H: We had reached the point where the mail portion of the development set, one of the development sets, I think, that the mail error rate with the XCPRP features was pretty much identical with the SRI, which are the XCPRP.\nSpeaker H: So then I thought, oh, great.\nSpeaker H: I'll just let's make sure I think it works on the females and the error rate.\nSpeaker H: Yeah, there was a fee percent difference.\nSpeaker I: Is there less training data?\nSpeaker I: I mean, we don't...\nSpeaker I: This is more training data.\nSpeaker I: This is on just digits?\nSpeaker I: No, it's in a five.\nSpeaker H: Oh, okay.\nSpeaker H: I'll probably be able to...\nSpeaker H: I'll be tested as a moment, sort of.\nSpeaker H: So then...\nSpeaker H: Oh, and plus the vocal track length organization didn't actually make things worse.\nSpeaker H: So it's happening to me seriously.\nSpeaker H: So...\nSpeaker H: Okay.\nSpeaker H: So...\nSpeaker J: But see, now between the males and the females, there's certainly a much bigger difference in the scaling range than there is, say, just within the males.\nSpeaker J: And what you're using before was scaling factors that were just from the SRI front end.\nSpeaker J: And that worked fine.\nSpeaker J: Yeah.\nSpeaker J: But now you're looking over a larger range and it may not be so fine.\nSpeaker H: Well, so the one thing that I then tried was to put in the low pass filter, which we have in the...\nSpeaker H: So most...\nSpeaker H: Most have five systems actually band-limit at about 3700 hertz, although, you know, normally the channel goes to 4000, right?\nSpeaker H: So...\nSpeaker H: And that actually helped a little bit.\nSpeaker H: And it didn't hurt on the males, you know.\nSpeaker H: So... and I now try the...\nSpeaker H: Oh, and suddenly also the vocal track length organization only in the tests on the test data.\nSpeaker H: So you can do a vocal track length organization on the test data only or ongoing training in the test.\nSpeaker H: And you expect it to help a little bit if you do it only on the test and more if you do it on both training and...\nSpeaker H: Yes.\nSpeaker H: So it now helps you do it only on the test and I'm trying to retrain another set of models where it's both in the training and the test.\nSpeaker H: And then we will probably have hopefully a better one.\nSpeaker H: But it looks like there will still be some difference, maybe between one and two percent of the females.\nSpeaker H: And so, you know, I'm often doing suggestions.\nSpeaker H: And it is true that the...\nSpeaker H: You know, we are using the... but it can't be just the VTL because...\nSpeaker H: No, no, I'm not.\nSpeaker H: It's much worse, you know.\nSpeaker H: You know, the females are concerned with the reverse and the...\nSpeaker H: So that must be something else.\nSpeaker I: Well, what's the standard...\nSpeaker I: I thought the performance was actually a little better on females than males.\nSpeaker I: That's what I thought too.\nSpeaker H: That overall, yes, but on this particular development test, they're actually a little worse.\nSpeaker H: But that's beside the point we're looking at the discrepancy between the SRI system and the SRI system when training with each feature.\nSpeaker I: Right, I'm just wondering if that...\nSpeaker I: If you have any indication of your standard features, you know, if that's also different in the same direction or not.\nSpeaker J: Let me ask a more basic...\nSpeaker J: I mean, is this a iterative bound-multchtraining or is it a terribly training?\nSpeaker H: It's a lot of training.\nSpeaker J: And how do you determine when to stop iterating?\nSpeaker H: Well, actually, we just basically do a fixed number of iterations.\nSpeaker H: In this case, four.\nSpeaker H: Which we used to do only three and then we found out we can squeeze.\nSpeaker H: And it was basically...\nSpeaker H: We're keeping down the same side.\nSpeaker H: But you're right, it might be that one more iteration would help, but it's...\nSpeaker J: Or maybe you're doing one too many.\nSpeaker H: No, but this bound-multchtraining shouldn't be over there.\nSpeaker J: Well, there can be sure.\nSpeaker J: Well, you can try each one on a cross-order.\nSpeaker J: Some years ago, Bill Burndyla thing where he was looking at that and he showed that he could get it.\nSpeaker J: But...\nSpeaker H: Well, that's the easy one.\nSpeaker H: Because you've got all these immediate models.\nSpeaker J: In each case, how do you determine the usual fudge factors, the language scaling, acoustic scaling?\nSpeaker H: I'm actually re-optimizing them.\nSpeaker H: Although that hasn't shown to you.\nSpeaker J: Okay, and the question he was asking at one point about perning...\nSpeaker J: Remember that one?\nSpeaker J: It looked like the probability was getting out of PLP versus milk upstream.\nSpeaker J: They looked pretty different.\nSpeaker K: Yeah, the likelihood were lower.\nSpeaker J: And so there's the question.\nSpeaker H: For the PLP.\nSpeaker H: Did you see this in the SRI system?\nSpeaker H: Did this look into the log files?\nSpeaker H: Well, the likelihoods are...\nSpeaker H: You can't recognize them.\nSpeaker H: Because for every set of models, you compute a new normalization.\nSpeaker H: And so these log probabilities, they aren't directly...\nSpeaker H:...you have the normalization cost of the average model in the trial.\nSpeaker J: But still, there's a question.\nSpeaker J: If you have some thresholds somewhere in terms of beam search or something?\nSpeaker J: Well, yeah, that's what I was wondering.\nSpeaker K: I mean, if you have one threshold that works well because the range of your likelihoods is in this area.\nSpeaker H: Very conservatively.\nSpeaker H: I mean, as we saw with the meeting that we could probably tighten the pruning without really...\nSpeaker H: So we basically have a very open view.\nSpeaker J: But you're only talking about a percent or two, right?\nSpeaker J: Here, we're saying that with this G, there's this difference here.\nSpeaker J: And see, because there could be lots of things, right?\nSpeaker J: But let's suppose just for a second that we've sort of taken out a lot of the major differences between the two.\nSpeaker J: I mean, we're already sort of using the male scale and using the same style filter integration.\nSpeaker J: And we're making sure that the low and high...\nSpeaker H: So for the PLP features, we used the triangular filter shapes and for the NDSRI front, we used the triple-modal ones.\nSpeaker E: And what's the top frequency of each?\nSpeaker H: Well, now it's the same. It's 30, 700...\nSpeaker E: Yeah, one's triangular, one's trapezoidal.\nSpeaker J: So...\nSpeaker J: No, no, but...\nSpeaker J: Before, with straight PLP, it's trapezoidal, also, but then we had a slight difference in the scale.\nSpeaker H: So it's currently that the FICAD program doesn't allow me to change the filter shape independently of this.\nSpeaker H: Yeah.\nSpeaker H: I did the experiment on the NDSRI front, where I tried the...\nSpeaker H:...with the standards to be the as trapezoidal purpose.\nSpeaker H: You can actually continuously vary between the two.\nSpeaker H: And so I...\nSpeaker H: That's what I'm trying to...\nSpeaker H: Trying to do what's...\nSpeaker H: And it did slightly worse, but it's really a small difference.\nSpeaker J: So...\nSpeaker J: A couple of tensor representations.\nSpeaker E: Okay, so it's not just losing some frequency range.\nSpeaker J: Yeah. Right, so the other thing that...\nSpeaker J: So we've always viewed it anyway, is the major difference between the two is actually in the smoothing.\nSpeaker J: That the PLP and the reason PLP has been advantageous in slightly noisy situations is because PLP does the smoothing at the end by an aggressive model.\nSpeaker J: And Melcafstrom does it by just computing the lower test row coefficients.\nSpeaker H: Okay, so one thing I haven't done yet is to actually do all of this with a much larger with our full train set.\nSpeaker H: So right now we're using a...\nSpeaker H:...40E.\nSpeaker H: It's a train set that's about...\nSpeaker H:...my effective force model that we want to use when we train the fault system.\nSpeaker H: So some of these moving issues are overfitting for that matter.\nSpeaker H: And the long route should be much less of a factor if you go full-hog.\nSpeaker H: Good to be, yeah.\nSpeaker H: So the strategy is to first treat things with fast train around on the smoothing set.\nSpeaker H: And then when you've narrowed down, you've tried on a larger train set.\nSpeaker H: So we haven't done that yet.\nSpeaker J: Now the other great related question though is, is what's the boot models for these things?\nSpeaker H: The boot models are trained from scratch.\nSpeaker H: So we compute...\nSpeaker H: So we start with a...\nSpeaker H:...alignment that we computed with sort of the best system we have.\nSpeaker H: And then we train from scratch.\nSpeaker H: We do...you know... we collect the...\nSpeaker H:...observations from those linings under each of the future sets that we train.\nSpeaker H: And then from there we do...\nSpeaker H: There's a lot of actually the way it works.\nSpeaker H: The first train, the phonetically tight mixture model.\nSpeaker H: You do a total of...\nSpeaker H: First you do a context independence PTN model.\nSpeaker H: Then you switch to context...\nSpeaker H:...and then you do two iterations of that.\nSpeaker H: Then you do two iterations of the context dependent, but actually the packet mixtures.\nSpeaker H: And then from that you go to a state custom model.\nSpeaker H: And you do four iterations of that.\nSpeaker H: So there's a lot of iterations overall between your original boot models and the final models.\nSpeaker H: I don't think that we have never seen big differences.\nSpeaker H: Once I thought, oh, I can...\nSpeaker H: Now I have these much better models.\nSpeaker H: I'll regenerate my initial alignments and I'll get much better models at the end.\nSpeaker H: That's what's right.\nSpeaker H: Right.\nSpeaker J: Well, making things better, yeah, but this...\nSpeaker J:...for making things worse, that is possible.\nSpeaker J: Another possible partial cause is if the boot models use a different feature set.\nSpeaker H: But there are some boot models in fact.\nSpeaker H: You're not booting from initial models, you're booting from initial alignments.\nSpeaker J: What you got from a different feature set.\nSpeaker H: That's correct.\nSpeaker J: So those features look at the data differently actually.\nSpeaker J: I mean, you know, they will find boundaries a little differently.\nSpeaker J: You know, all that sort of thing is actually slightly different.\nSpeaker J: I'd expect it to be a minor effect.\nSpeaker H: So for a long time we had used boot alignments that had been trained with a same front end.\nSpeaker H: But with acoustic models that were like 15% worse than what we used now.\nSpeaker H: And with a different dictionary, with a considerably different dictionary, which was much less detailed and much less suited.\nSpeaker H: And so then we switched to new boot alignments, which now had the benefit of all these improvements that we've made over two years in the system.\nSpeaker H: Right.\nSpeaker H: And the result at the end was no different.\nSpeaker H: What I'm saying is the exact nature of these boot alignments is probably not a big factor in the quality of the final model.\nSpeaker J: Yeah, maybe not. But it's still see it as, I mean, there's a history of this too.\nSpeaker J: But I don't want to go into it.\nSpeaker J: But I think it could be the things that the data is being viewed in a certain way.\nSpeaker J: That a beginning is here rather than there and so forth because the actual signal processing is doing a slightly different.\nSpeaker J: But it's probably not it.\nSpeaker H: I should really reserve any conclusions until we've met it on the live trains and until we've seen results with the BTO training.\nSpeaker J: Yeah, at some point you also might want to take the same thing and try it on some broadcast news data or something else that actually has some noisy, noisy components.\nSpeaker J: So we can see if any conclusions would come to a halt across different data.\nSpeaker J: So something quick about absent.\nSpeaker E: Just what we were talking about before, which is that I hoarded a blast library to absent and then got it working with fast forward and got a speed up roughly proportional to the number of processors times the clock cycle.\nSpeaker E: So that's pretty good.\nSpeaker E: I mean, the process of doing it for quick net, but there's something going wrong and it's about half the speed that I was estimating it should be.\nSpeaker E: And I'm not sure why, but I'll keep working on it. But what it means is it's likely that for net training and forward passes will absent will be a good machine, especially if we get a few more processors and upgrade the processors.\nSpeaker E: There are five now. It can hold eight.\nSpeaker E: Yeah, we'll just go by and it's also 550 megahertz and you can get a gigahertz.\nSpeaker H: Can you fix the processors with different?\nSpeaker E: I don't think so. I think we'd have to do all.\nSpeaker J: Yeah, thank you for the box.\nSpeaker H: We can press.\nSpeaker E: We'd have to get a almost certainly have to get a netfinity server.\nSpeaker E: They're pretty, pretty specialized.\nSpeaker J: Okay. Is this going back?\nSpeaker J: Yeah. How are you doing?\nSpeaker J: All right.\nSpeaker J: All right. See you.\nSpeaker J: All right. So they're having T out there.\nSpeaker J: So I guess the other thing that we were going to talk about is demo.\nSpeaker J: And so these are the demos for the July meeting and July what?\nSpeaker J: Early July, late July? Oh, I think it's July 15.\nSpeaker J: 16, 13. Yeah. So we talked about getting something together for that.\nSpeaker J: But maybe we'll just put that off for now given that.\nSpeaker J: But I think we should have a sub meeting.\nSpeaker J: I think probably Adam and Chuck and me should talk about.\nSpeaker J: Should get together and talk about that sometimes.\nSpeaker J: Over at Captain Chinatumar.\nSpeaker J: Yeah, something like that. Maybe we'll involve Daniel and some of us as well.\nSpeaker J: Okay. The T is going.\nSpeaker J: So I see just we do a unison.\nSpeaker J: A unison digit. Yeah.\nSpeaker E: Which is going to be a little hard for a couple people because we have different digit forms.\nSpeaker E: I found a couple of old ones.\nSpeaker J: Interesting. So have you done digits before?\nSpeaker J: No. I haven't done it.\nSpeaker E: Okay. So the idea is just to read each line with a short pause between lines.\nSpeaker E: Not between. And since we're in a hurry, we were just going to read everyone all at once.\nSpeaker E: So if you sort of plug your ears and read.\nSpeaker E: Okay. So first read the transcript number and then start reading the digits.\nSpeaker E: Okay. One, two, three.\nSpeaker M: L1154 Bazinai Stars.\nSpeaker L: 0 is 333 Codos,\nSpeaker C: m \uac19as, j Eins,\nSpeaker D: 510-610-165-4-6367-277-5-8659-6978-248-617145-2258-31469-566\nSpeaker J: Okay, we're done\nSpeaker C: And\n"}, {"meeting_id": "Bmr018", "summary": "The group discussed the preparation of a data sample for IBM, the manual adjustment of time bins by transcribers, recognition results for a test set of digits data, and forced alignments. Participants also talked about Eurospeech 2001 submissions, and exchanged comments on the proceedings of the recently attended Human Language Technologies conference (HLT'01). Preliminary recognition results were presented for a subset of digits data. Efforts to deal with cross-talk and improve forced alignments for non-digits data were also discussed.", "dialogue": "Speaker F: We're recording.\nSpeaker F: All right, no crash.\nSpeaker D: I recrashed it.\nSpeaker D: Yeah.\nSpeaker B: It never crashes on me.\nSpeaker D: I think it's actually, it depends on if the temp files are there or not.\nSpeaker D: At least that's my current working hypothesis that I think what happens is it tries to clear the temp files and if they're too big, it crashes.\nSpeaker E: When the power went out the other day and I restarted it, it crashed the first time.\nSpeaker E: So there would be no temp files?\nSpeaker D: No, it doesn't clear those necessarily.\nSpeaker D: They're called temp files, but they're not actually in the temp directory.\nSpeaker D: They're in the scratch.\nSpeaker D: They're not backed up.\nSpeaker D: They're raised either on PowerFailure.\nSpeaker B: But that's usually the meeting that I recorded and it doesn't crash on me.\nSpeaker E: Well, this wasn't, actually this wasn't before your meeting.\nSpeaker E: This was Tuesday afternoon when Robert just wanted to do a little recording and the power had gone out earlier in the day.\nSpeaker F: I don't know when would be a good excuse for it, but I just can't wait to be giving a talk and use the example from last week with everybody doing the digits it wants.\nSpeaker F: I'd love to play somebody there.\nSpeaker F: It was quick.\nSpeaker E: It was, it was really efficient.\nSpeaker E: Talk about a good noise shield, you know.\nSpeaker E: If you wanted to keep people from listening in, you could like have that playing outside the room, nobody could listen in.\nSpeaker B: Well, I had this idea we could make our whole meeting faster that way.\nSpeaker F: Yeah, everybody give the reports and they're doing it exactly the same time.\nSpeaker E: And then we'll go back later to the individual channel.\nSpeaker F: Actually, isn't that what we have been doing?\nSpeaker E: It's just sounds practically hum, we overlapped.\nSpeaker D: What are we doing?\nSpeaker D: I've been gone all week. I didn't send out a reminder for an agenda.\nSpeaker D: Do we have anything to talk about?\nSpeaker E: Should we just re-titch it?\nSpeaker E: I wouldn't mind hearing how the conference was.\nSpeaker E: What conference?\nSpeaker B: I wish about, aren't the UW folks coming this weekend?\nSpeaker B: Yep.\nSpeaker B: Next weekend.\nSpeaker B: Next weekend.\nSpeaker B: That is right.\nSpeaker B: Sorry, not the days coming up.\nSpeaker B: A week from Saturday.\nSpeaker B: Yeah.\nSpeaker B: That's when they're coming.\nSpeaker B: That's correct.\nSpeaker B: So, are we, do we have like an agenda or anything that we should have?\nSpeaker F: No, but that would be a good idea.\nSpeaker F: Okay.\nSpeaker A: Why don't we...\nSpeaker A: I want to deal with that I can be available after like 10, 30 or something.\nSpeaker A: I don't know how early you wanted to.\nSpeaker F: They're not even going to be here to 11 or so.\nSpeaker D: That's good.\nSpeaker B: They're flying up that day.\nSpeaker B: On Sunday?\nSpeaker B: Saturday.\nSpeaker F: Saturday.\nSpeaker F: Well, Saturday.\nSpeaker D: Yeah.\nSpeaker D: Your speeches do on Friday and then I'm going down to San Jose Friday night.\nSpeaker D: So, you know, if we start nice and late Saturday, that's a good thing.\nSpeaker F: Yeah, I mean, they're flying up from...\nSpeaker F: Seattle.\nSpeaker F: Down from...\nSpeaker F: They're flying from somewhere to somewhere.\nSpeaker F: Yeah.\nSpeaker F: And they'll end up here.\nSpeaker F: And also, Brent Kingsbury is actually flying from these coasts on that morning.\nSpeaker F: So, I will be...\nSpeaker F: I mean, he's taking a very, very flight and we do have the time difference running the right way.\nSpeaker F: But I still think there's no way we'd start before 11 and it might end up early being 12.\nSpeaker F: So, when we get closer, we'll find people's playing schedules and let everybody know.\nSpeaker D: So...\nSpeaker D: Yeah, maybe an agenda or at least some things to talk about would be a good idea.\nSpeaker F: Well, we can start gathering those ideas but then we should firm it up by next Thursday's meeting.\nSpeaker C: Well, we have time to prepare something that we...\nSpeaker C: In the format we were planning for the IBM transcribers by them.\nSpeaker D: Oh, yeah, absolutely.\nSpeaker D: So, have you heard back from Brian about that?\nSpeaker E: Chuck?\nSpeaker E: Yes.\nSpeaker E: I'm sorry, I should have forwarded that along.\nSpeaker E: Oh, I think I mentioned the last meeting.\nSpeaker E: He said that he talked to them and it was fine with the beeps.\nSpeaker E: They would be...\nSpeaker E: That's easy for them to do.\nSpeaker D: Okay.\nSpeaker D: So, I hope TTLO isn't here.\nSpeaker D: But I have the program to insert the beeps.\nSpeaker D: What I don't have is something to parse the output of the channelized transcripts to find out where to put the beeps.\nSpeaker D: But that should be really easy to do.\nSpeaker D: So, do we have a meeting that that's been done with?\nSpeaker D: That we've tightened it up to the point where we can actually give it to IBM and have them try it out.\nSpeaker C: He generated a channel-wise pre-signated version of a meeting but it was robustness rather than EDU.\nSpeaker C: So, I guess, depends on whether we are willing to...\nSpeaker E: Well, for this experiment, I think we can use pretty much anything.\nSpeaker E: Okay.\nSpeaker D: Well, we had talked about maybe doing EDU as a good choice though.\nSpeaker E: Well, whatever we have...\nSpeaker E: Whatever we talked about that is being the next ones we wanted to transcribe.\nSpeaker E: Right.\nSpeaker E: And then we're sending him a sample one to...\nSpeaker D: Yeah, maybe it doesn't matter.\nSpeaker C: I'll make that available.\nSpeaker D: Okay. And has it been corrected?\nSpeaker C: Oh, well, wait.\nSpeaker D: And check because that was one of the processes we were trying to do.\nSpeaker E: Right, so we need to run...\nSpeaker E: That's right.\nSpeaker E: TTLO's thing on it and then we go in and adjust the boundaries.\nSpeaker C: That's right.\nSpeaker C: And we haven't done that.\nSpeaker C: And I think I can set someone on that tomorrow.\nSpeaker E: Okay. And we probably don't have to do necessarily a whole meeting for that if we just want to send them a sample to try.\nSpeaker C: Maybe a good number of minutes.\nSpeaker E: I don't know. Maybe we could figure out how long it'll take to do...\nSpeaker D: I don't know. It seems to me we probably should go ahead and do a whole meeting because we'll have to transcribe the whole meeting anyway sometime.\nSpeaker F: Yes, except that if there was a choice between having 15 minutes that was fully the way you wanted and having a whole meeting that didn't get at what you wanted for them...\nSpeaker F: So, I mean, I guess we have to do it again anyway.\nSpeaker D: But...\nSpeaker E: Yeah.\nSpeaker E: Yeah, I guess the only thing I'm not sure about is how quickly can the transcribers scan over and fix the boundaries?\nSpeaker E: And I mean, is it pretty easy?\nSpeaker D: I think it's going to be one or two times real time at...\nSpeaker D: Well, excuse me, two or more times real time, right?\nSpeaker D: Because they have to at least listen to it.\nSpeaker F: Can we pipeline it so that the transcriber gets done with the quarter of the meeting and then you run it through this other stuff?\nSpeaker D: Well, the other stuff is IBM.\nSpeaker D: I'm just thinking that from a data...\nSpeaker D: Keeping track of the data point of view, it may be best to send them whole meetings at a time and not try to send them bits and pieces.\nSpeaker F: Oh, that's right. So the first thing is the automatic thing.\nNone: Right.\nSpeaker F: And then it's the transcribers tightening stuff up and then it's IBM.\nSpeaker F: Okay, so you might as well run the automatic thing over the entire meeting.\nSpeaker F: And then you would give IBM whatever it was fixed.\nSpeaker C: And then fix it over the entire meeting, too.\nSpeaker F: Well, yeah, but it starts from beginning to the end, right?\nSpeaker F: So if they were only halfway through, then that's what you'd give IBM.\nSpeaker F: Okay.\nSpeaker E: As of what point?\nSpeaker E: I mean, I guess the question I'm mind is do we wait for the transcribers to adjust the marks for the whole meeting before we give anything to IBM?\nSpeaker E: Or do we go ahead and send them a sample?\nSpeaker F: Well, if they were going sequentially through it, why wouldn't we give them...\nSpeaker F: I mean, are we trying to get something done by the time Brian comes?\nSpeaker E: Well, that was the question.\nSpeaker F: So if we were, then it seems like giving them something, whatever they had got that I agree.\nSpeaker D: Well, I don't think... I mean, they typically work for what, four hours, something like that?\nSpeaker C: I get them.\nSpeaker D: I think they should be able to get through a whole meeting in one sitting.\nSpeaker D: I would think, unless it's a lot harder than we think it is, which it could be, certainly.\nSpeaker C: It's got like four speakers, then...\nSpeaker C: I guess...\nSpeaker E: We're just doing the individual channels, right?\nSpeaker C: Individual channels, yeah.\nSpeaker E: So it's going to be depending on the number of people in the meeting.\nSpeaker E: Well...\nSpeaker C: I guess there is an issue of, you know, if the segmenter thought there was no speech on a particular stretch, on a particular channel, and there really was, then if it didn't show up in a mix signal to verify, then it might be overlooked.\nSpeaker C: So, I mean, the question is, should transcribe, listen to the entire thing or can it be based on the mix signal?\nSpeaker C: And I, as far as I'm concerned, it's fine to base it on the mix signal at this point.\nSpeaker D: That's what it seemed to me to, and that if they need to, just like in the other cases, they can listen to the individual if they need to, but they don't have to for most of it.\nSpeaker D: That's good.\nSpeaker E: I don't see how that will work though.\nSpeaker F: So you're talking about tightening up the time boundary?\nSpeaker D: Yeah. So they have the normal channel trans interface, where they have each individual speaker has their own line.\nSpeaker D: Yeah.\nSpeaker D: But you're listening to the mix signal, and you're tightening the boundaries, correcting the boundaries.\nSpeaker D: You shouldn't have to tighten them too much, because Delos program does that.\nSpeaker B: Except for...\nSpeaker B: It doesn't do well in short things.\nSpeaker B: Right, so you'll have to...\nSpeaker B: Yeah, I think that will miss most of the really short things.\nSpeaker B: Like that.\nSpeaker B: But those would be...\nSpeaker B: Ah, ha!\nSpeaker B: Yeah, you have to say, ah, ha more slowly.\nSpeaker B: Sorry.\nSpeaker B: Oh, I'm actually serious.\nSpeaker B: So it will miss stuff like that.\nSpeaker D: Well, so that's something that the transgarbers will have to do.\nSpeaker C: Presumably, most of those, they should be able to hear from the mix signal unless they're embedded in the heavy overlap section.\nSpeaker C: That's what I'm concerned about.\nSpeaker C: I'm concerned about that part.\nSpeaker C: Yeah, I am too.\nSpeaker C: And I think it's a little...\nSpeaker E: Can we... couldn't we just have, um...\nSpeaker E: I don't know, maybe this just doesn't fit with the software, but I guess if I didn't know anything about transgarber and I was gonna make something to let them adjust boundaries, I would just show them one channel at a time.\nSpeaker E: Oh, I think so.\nSpeaker D: But then they have to do...\nSpeaker D: And then for this meeting they would have to do seven times real time.\nSpeaker D: Yeah.\nSpeaker D: And it would probably be more than that.\nSpeaker D: Right, because they'd have to at least listen to each channel all the way through.\nSpeaker E: But it's very quick, right?\nSpeaker E: I mean, you scan.\nSpeaker E: I mean, if you have a display of the waveform, you're talking about visually.\nSpeaker D: I just don't think...\nSpeaker C: The other problem is the breath, because you also see the breaths on the waveform.\nSpeaker C: I've looked at the... I tried to do that with a single channel.\nSpeaker C: And you do see all sorts of other stuff besides just the voice.\nSpeaker D: And I think that they're going much more on acoustics than they are on visuals.\nSpeaker D: Well, that, that I'm not sure.\nSpeaker C: The digital task that you had your interface, I know for a fact that one of those...\nSpeaker C: She could really well...\nSpeaker C: Yeah, that's actually true.\nSpeaker C: What number was that you're on?\nSpeaker D: Yeah, you're absolutely right.\nSpeaker D: I mean, I found the same thing that when I was scanning through the waveform, I could see when someone started to read digits just by the shapes.\nSpeaker D: Yeah, she could tell which one was on.\nSpeaker F: Maybe.\nSpeaker F: So I don't...\nSpeaker F: But I'm now entirely confused about what they do.\nSpeaker F: So they're looking at a mixed signal or looking...\nSpeaker F: What are they looking at visually?\nSpeaker C: Well, they have a choice.\nSpeaker C: They could choose any signal to look at.\nSpeaker C: I've tried looking, but usually they look at the mixed.\nSpeaker C: But I've tried looking at the single signal and in order to judge when it was speech and when it wasn't.\nSpeaker C: But the problem is then you have breaths which show up on the signal.\nSpeaker F: But the procedure that you're imagining, I mean, people vary from this, is that they have the mixed signal waveform in front of them.\nSpeaker F: Yes.\nSpeaker F: And they have multiple...\nSpeaker F: Well, let's see, there isn't...\nSpeaker F: We don't have transcription yet.\nSpeaker F: So, but there's markers.\nSpeaker F: Right. That have been happenautomatically.\nSpeaker F: No show up on the mixed signal.\nSpeaker C: Oh, they show up on the separate ribbons.\nSpeaker C: Right, the separate ribbons.\nSpeaker C: So that was separate ribbons for each channel.\nSpeaker C: And it'll be because it's being segmented as channel at a time with Tilo's new procedure, then you don't have correspondence of the times across the bins, across the ribbons.\nSpeaker F: And is there a line moving across the waveform as it goes?\nSpeaker F: Yes.\nSpeaker F: Okay, so the way you're imagining is they kind of play it and they see how this happened then, and if it's about right, they just sort of let it slide.\nSpeaker F: Right.\nSpeaker F: And if it...\nSpeaker F: There's a question on something they stop and maybe look at the individual waveform.\nSpeaker D: Right. Well, they wouldn't look at it at this point.\nSpeaker D: They would just listen.\nSpeaker F: They might look at it, right?\nSpeaker D: Well, the problem is that the interface doesn't really allow you to switch visuals.\nSpeaker D: Not really.\nSpeaker D: The problem is that the tickle-tk interface with the visuals, it's very slow to load waveforms.\nSpeaker D: That's it.\nSpeaker D: And so when I tried, that was the first thing I tried when I just started it, right?\nSpeaker C: You can switch quickly between the audio, but you just can't get the visual display to show quickly.\nSpeaker C: So you have to...\nSpeaker C: It takes, I don't know, three, four minutes to...\nSpeaker C: Well, it takes a long enough off.\nSpeaker C: Yes, very slow.\nSpeaker C: It takes a long enough off.\nSpeaker C: Because that's to reload the...\nSpeaker C: I don't know exactly what it's doing.\nSpeaker C: Thank you.\nSpeaker C: It takes a long enough that it's just not a practical alternative.\nSpeaker D: Well, it does some sort of shape pre-computation so that it can then scroll it quickly.\nSpeaker D: Yeah.\nSpeaker D: But then you can't change the resolution or scroll quickly.\nSpeaker D: So...\nSpeaker C: Now you could set up multiple windows, each one with a different signal showing, and then look between the windows, maybe that's...\nSpeaker D: I mean, we could do different interfaces, right?\nSpeaker D: I mean, so we could use like X-Waves instead of transcriber.\nSpeaker D: And it loads faster, certainly.\nSpeaker A: What if you were to pre-load all the channels from initially?\nSpeaker A: Well, that's what I tried originally.\nSpeaker D: So I actually, before Dave Galbart did this, I didn't interface, which showed each waveform and a ribbon for each waveform.\nSpeaker D: The problem with it is even with just three waveforms, it was just painfully slow to scroll.\nSpeaker D: So you just scroll screen and it would go, go, Curr Chunk.\nSpeaker D: And so it just was not doable with the current interface.\nSpeaker C: You know, I am thinking if we have a meeting with only four speakers, and you could fire up a transcriber interface for, you know, in different windows, multiple ones, one for each channel, and it's sort of a hack, but I mean, it would be one way of seeing the visual.\nSpeaker D: I think that if we decide that we need, that they need to see the visuals, we need to change the interface so that they can do that.\nSpeaker B: So that's actually why I thought of loading the chopped up waveforms.\nSpeaker B: I mean, you know, that that would make it faster.\nSpeaker B: What's the problem is if if anything's cut off,\nSpeaker E: you can't expand it from the chopped up. Right, but if you...\nSpeaker D: And wouldn't that be the same as the mix signal?\nSpeaker B: No, I mean the individual channels that were chopped up, that it'd be nice to be able to go back and forth between those short segments.\nSpeaker B: Because you don't really need like nine tenths of the time you're throwing most of them out.\nSpeaker B: But what you need are that particular channel, that particular location, and might be nice, because we save those out already to be able to do that.\nSpeaker B: But it won't work for IBM, of course.\nSpeaker B: It only works here because they're not saving out the individual channels.\nSpeaker C: Well, I do think that this will be a doable procedure.\nSpeaker C: Okay.\nSpeaker C: And have me starting with the mix, and then when they get into overlaps, just have them systematically check all the channels to be sure that there isn't something hidden from audio view.\nSpeaker D: Yeah, hopefully, I mean, the mix signal, the overlaps, are pretty audible, because it is volume equalized.\nSpeaker D: So I think they should be able to hear.\nSpeaker D: The only problem is counting how many, and if they're really correct or not, I don't know.\nSpeaker B: I don't know that you can locate them very well for the mix signal.\nSpeaker D: Right, but once you know that they happen, you can at least listen to their close talking.\nSpeaker D: But right now, to do the slummitation,\nSpeaker F: the switching is going to be switching of the audio. Right.\nSpeaker F: So, did they use any of the areas to do these?\nSpeaker D: Did Dave do that change, where you can actually just click rather than having to go up to the menu to listen to the individual channels?\nSpeaker D: I had suggested it before I just don't know whether you did it or not.\nSpeaker C: I'm not sure what, click on the ribbon, and you can get the switch audio.\nSpeaker C: Yeah.\nSpeaker C: Not last I tried, but in many cases.\nSpeaker C: We should get them to do that,\nSpeaker D: because I think that would be much, much faster than going to the menu. There's a reason I disagree,\nSpeaker C: and that is that it's very good to have a dissociation between the visual and the audio. There are times when I want to hear the mix signal, but I want to transcribe on the single channel.\nSpeaker D: So, maybe just button stand at the bottom.\nSpeaker D: Maybe ask for it.\nSpeaker D: Just something so that it's not in the menu option, so that you can do it much faster.\nSpeaker C: Well, I think that might be a personal style thing.\nSpeaker C: I find it really convenient the way it's set up right now.\nSpeaker D: Well, it just seems to me that if you want to quickly, well, was that chain known, was that chuck known, was that morgan right now?\nSpeaker D: You have to go up to the menu, and each time go up to the menu, select it, listen to that channel, then click below, and then go back to the menu, select the next one, then click below.\nSpeaker D: So you can definitely streamline that with the interface.\nSpeaker C: You know what I mean?\nSpeaker C: In the ideal world.\nSpeaker C: What?\nSpeaker C: No, I agree. That'd be nice.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker F: So, put that down with that.\nSpeaker F: Forget it. Is anybody working any your speech submission related to this?\nSpeaker D: I would like to try to do something on digits, but I just don't know if we have time.\nSpeaker D: I mean, it's due next Friday.\nSpeaker D: So we have to do the experiments and write the paper.\nSpeaker D: So I'm going to try, but we'll just have to see.\nSpeaker D: So actually, I want to get together with both Andreas and Stefan with their respective systems.\nSpeaker F: Yeah.\nSpeaker F: Yeah, that's where we had one conversation about what did it mean for one of those speakers to be pathological.\nSpeaker F: Right. And I haven't had a chance to sit down and listen.\nSpeaker A: I was going to do that this afternoon.\nSpeaker A: But there must be something around here.\nSpeaker D: Well, Mayor Gennari, we're having a debate about that.\nSpeaker D: Whereas I think it's probably something pathological.\nSpeaker D: And actually, Stefan's results, I think, confirm that he did the Aurora system, also got very lousy average error, like 15 or 15 to 20% average.\nSpeaker D: But then he ran it just on the lapel and got about 5 or 6% word error.\nSpeaker D: So that means to me that somewhere in the other recordings there are some pathological cases.\nSpeaker D: But that may not be true.\nSpeaker D: Maybe just some of the segments they're just doing a lousy job on.\nSpeaker D: So I'll listen to it and find out since you actually split it up by segment.\nSpeaker D: So I can actually listen to it.\nSpeaker E: Did you run the Andreas?\nSpeaker E: Are I recognized or on the digit?\nSpeaker D: Oh, I thought he had sent that around to everyone.\nSpeaker A: Did you just send that to me?\nSpeaker A: No.\nSpeaker A: Since I considered those preliminary.\nSpeaker A: It was primodal.\nSpeaker A: It was a tri-modal.\nSpeaker A: Oh, it was a tri-modal.\nSpeaker F: So there was zero a little bit and a lot.\nSpeaker A: One bump at zero, around zero, which were the native speakers.\nSpeaker A: Zero percent error.\nSpeaker A: And there was another bump at 15 or something.\nSpeaker A: This is error you're talking about?\nSpeaker A: Yeah.\nSpeaker A: Those were the non-natives.\nSpeaker A: There was another distinct bump at like 100.\nSpeaker A: Wow.\nSpeaker A: Which must have been some problem.\nSpeaker A: What is pathological?\nSpeaker D: Just something really wrong with a bug is what I mean.\nSpeaker A: So that it's like...\nSpeaker A: There was this one meeting I forget which one it was where like six out of the 8 channels were all like 100%.\nSpeaker D: Which probably means like there was a recording interface crashed or there was a short, you know, one was jiggling with a chord or I extracted it incorrectly.\nSpeaker D: It was labeled it was transcribed incorrectly.\nSpeaker D: Something really bad happened.\nSpeaker D: I just haven't listened to it yet to find out what it was.\nSpeaker A: It was like, I excluded the pathological ones.\nSpeaker A: Well, I don't.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: I don't know.\nSpeaker A: The gravel wasn't too heavy either.\nSpeaker E: And it didn't matter whether it was the lapel or whether it was the...\nSpeaker A: I haven't split it up that way.\nSpeaker B: But there's no overlapping in legit reading experience.\nSpeaker F: No, but there's a little difference.\nSpeaker F: And we haven't looked at it for digits.\nSpeaker F: Yes, I was curious about that.\nSpeaker F: Because what I was seeing when I looked at those things, I was almost going to call Quadromodal because there was a whole lot of cases where it was 0%.\nSpeaker F: They just playing got her go right yeah, and then there and then there was another bunch that were a couple percent\nSpeaker A: I just Instagram that yeah, it was a nice Normal was zero was the most of them but then there were there others was decaying from there. Yeah, yeah\nSpeaker F: I see I see\nSpeaker D: Yeah, some of our non-natives are pretty non-nated So\nSpeaker C: Yeah, did you have something in the report about about for forced alignment?\nSpeaker A: Well, yeah, so I've been struggling with the forced alignment So this scheme that I drew on the board last time where we try to Allow reject models for the speech from other speakers most of the time it doesn't work very well so And the I haven't done I mean the only way to check this right now is for me to actually load these into x waves and you know plus the linings and Lay them and see where that and it looks and so I looked at all of the Utterances from you Chuck on that one conversation I don't know which you probably know which one I mean it's where you were on the lapel and Morgan was sitting next to you and you can hear everything Morgan says But and some of what you I mean you also hear quite a bit of a cross-talk so I Actually went through all of those there were I think 55 segments in the next wave and sort of did a crude check and More often than not it gets it wrong so there's either the beginning mostly the beginning word Where you You know talk talks somewhere into the segment But the first Word what he says often I but it's very reduced I that's just aligned Beginning of someone else's speech That's I'm still tickling with it might well be that we can't get\nSpeaker F: Last maybe we do this\nSpeaker B: cancellation right but I mean that was our plan but it's clear from Dan that this is not something you can do in a short amount of time Oh the shorter amount of time You know we it's been a long time writing up the HLT paper and we wanted to use that kind of analysis but the HLT paper has You know, it's a very crude measure of overlap. It's not really something you could scientifically say is overlap. It's just whether or not the High correlation segments that were all synchronized whether there was some overlap somewhere and You know pointed out some differences so we thought well if we can do something quick and dirty because Dan said the Cross cancellation it's not straightforward if it were straightforward then we would try it but so sort of good to hear that it was not straightforward thinking If we can get decent forced alignments then at least we can do sort of an overall report of what happens with actual overlap in time but\nSpeaker E: I didn't think that what we said it wasn't straightforward\nSpeaker B: Well, I thought he just saying I have to look over a longer time window. I need to but there are some issues of this timing Yeah, yeah, yeah\nSpeaker E: Right, so you just have to look over a longer time when you're trying to align the things you can't you can't just look well\nSpeaker D: Are you talking about the fact that the recording software doesn't do time synchronous? Is that what you're referring to? That seems to me you can do that over the entire file and get a very accurate\nSpeaker A: I don't think that was the issue. Yeah, I didn't think so either You have to have you first have to have a pretty good speech detection of the individual channels\nSpeaker B: And it's dynamic so I guess it was more dynamic than some simple models Would be able to so so there are some things available and I don't know too much about this area Where if people aren't moving around much then you could apply them and it should work pretty well if you took care of this Recording time difference right which should be pretty straightforward. It least is well defined. Yeah But then if you add the dynamic Aspect of adapting distances than it wasn't I Guess it just wasn't something that he could do quickly in time for us to be able to do something by two weeks from now so So I don't know what we can do if anything that's sort of worth\nSpeaker E: You know a year old speech paper at this point. Well Andreas. How well did it work on the non lapel stuff?\nSpeaker A: Yeah, so it's a check-through. It's very tedious to check this We would really need ideally a transcriber to time mark You know the at least the beginning ends of continuous speech And you know that with the time marks you can do an automatic comparison of your\nSpeaker E: Because really the the least in terms of how we were going to use this in our system was to get an ideal an idea For each channel about the start and end boundaries. We don't really care about like intermediate word boundaries\nSpeaker A: So that's how I've been looking at it. Yeah, that the individual words are like yeah, but you don't want\nSpeaker E: Right exactly so that's why I was wondering if it I mean maybe if it doesn't work for lapel stuff we can just not use that\nSpeaker A: I have I have just haven't had the time to do the same procedure on one of the So I would need a I would need a channel that has the speaker who's Who has a lot of overlap that's you know is a not lapel mic and We're preferably also there's someone sitting next to them who talks a lot so So meeting with me and\nSpeaker E: We can you know what maybe the best way to find that would be to look through these Because you can see the seat numbers and then you can see what type of mic they were using and so we just look for you know Somebody sitting next to Adam\nSpeaker B: That one of the meeting we can tell from the data that they have Yeah, there's a way to tell it might not be a single person who's always overlapping that person but any number of people and If you align the two Hypothesis files across the channels, you know just word alignment you'd be able to find that so so I guess that's sort of a last There there's sort of a few things we could do one is just do like non lapels if we can get good enough Alignments another one was to try to get somehow align T-lows energy segmentations with What we have but then you have the problem of not knowing where the words are because these meetings were Done before that segmentation, but maybe there's something what what is be done? Why do you need the\nSpeaker E: The forced alignment for the HLT. I mean for the euro speech paper well\nSpeaker B: I guess I wanted to just do something not on Recognition experiments because that's the way too early but to be able to report You know actual numbers like if we if we had Hand transcribe good alignments or hand checked alignments then we could do this paper It's not that we need it to be automatic But without knowing where the real words are so it was to get it was to get more data and better\nSpeaker E: So to squeeze the boundaries and know what an overlap really it's really an overlap. Yeah, or if it's just a\nSpeaker B: A segment correlated with an overlap and I guess that's the difference to me between like a real paper and a sort of promissory paper so if we It might be possible to take T-lows output and like if you have Like right now these meetings are all forgot the digital camera again every meeting You know they're time in line. So these are two different channels and somebody's talking here and somebody else is talking here Just that word if T-lows Can tell us that they're boundaries here. We should be able to figure that out because the only thing transcribed in this channel is this word but You know if there are things two words Yeah, if you have two and they're at the edges like here and here and there's a feature then it doesn't really help you so T-lows won't put down two separate mark T-lows will but it would but we don't know exactly where the words are because the Transcriber gave us two words in this time been and we don't really know\nSpeaker C: What's emerging problem if you had if you had a script which would I thought about this I mean if you have any ideas I discussed it with T-low The I mean I in principle I could imagine writing a script which would Approximate it to some degree but well maybe this problem of slippage\nSpeaker B: Maybe that will get enough of the cases to use for I mean that that would be really helpful\nSpeaker D: Yeah, it's because it seemed like most of the cases are in fact the single word Swords or at least a single phrase\nSpeaker C: In most of the pants I wouldn't make that generalization because sometimes people will say and then I and there's long pause and Finish the sentence and and sometimes it looks coherent and the I mean it's it's not a simple problem But it's really and then it's coupled with the problem that Sometimes you know with with a fricative you might get the beginning of the word cut off And so it's coupled with the problem that T-lows isn't perfect either. I mean, right?\nSpeaker C: It's like you have emerging problem plus so merging plus this problem of Not you if the speech on speech were perfect to be with the detector that would already be an improvement But that's impossible. Yes, there's too much to ask and so it and I mean I think I think that there always there would have to be some hand tweaking but it's possible that a script could be written to merge those two types of things I've discussed it with T-l In terms of not him doing it, but we we discussed some of the parameters of that and how hard it would be to In principle to write something that would do that\nSpeaker B: And I guess in the future it won't be as much of an issue if Transcribers are using the Titan boundaries to start with then we have a good idea of where the force alignment is constrained to\nSpeaker C: It's just you know matter. I know the revolution. We had the revolution of improved interface One month too late, but it's like you know, it's wonderful to have the revolution So it's just a matter of you know for now on we'll be able to have things Chanalyzed to begin with right and we'll just have to see how hard that is\nSpeaker D: Yeah, that's so so whether the corrections take too much time I was just thinking about the fact that if T-lows miss these short segments that might be quite time consuming for them to insert them\nSpeaker B: But he also can adjust this minimum time duration constraint and then what you get is Spurious noise is mostly, but that might be okay. It might be easier to delete something that's wrong\nSpeaker D: Then to insert something that's missing. What do you think?\nSpeaker F: If you can feel confident that what that yeah, but there's actually something\nSpeaker D: Yeah, because then you just deleted and you don't have to pick a time\nSpeaker C: It's a really good question and I really find it a pain the neck to delete things because you have to get the mouse up They're on the on the text line and otherwise you're just using it to get down I mean it depends on how long there's so many extra things that would make it one of them harder than the other Visitors that's not a simple question, but you know in principle like you know if one of them is easier than to buy\nSpeaker D: Is it toward whichever ones easier? I guess the semantics aren't clear when you delete a segment right because you would say\nSpeaker B: You would have to determine what the surroundings were you could just say it's a noise though and right you know a post processor Will just all you have to do is really a noise?\nSpeaker B: Well, just say it's just put X you know like not speech or something. I think it's easier to add than to delete\nSpeaker C: Yeah, because you have to maneuver around on that on both windows and\nSpeaker D: To add or to delete Okay, anyway, so I guess that maybe that's an interface issue that might be addressable But I think it's the semantics that are that are questionable to me that you delete something So let's say someone is talking to here and then you have a little segment here Well is that part of the speech is a part of the non-speech?\nSpeaker D: I mean what do you embed it in?\nSpeaker B: There's something nice though about keeping this is probably another discussion keeping the stuff that Teelos detector detected as possible Speech and just marking it as not speech then deleting it. Oh, I see so then they could just like but that's what you meant by just put an X there Reject model or whatever and you're an interesting idea with the automatic system\nSpeaker D: So all they so that all they would have to do is put like an X there or some so blank for Blank for silence S for speech X for whatever something else\nSpeaker B: That's actually a better way to do it because the the force alignment will probably be more consistent\nSpeaker C: Well, I mean if it's a complication which is that that you can have speech and noise And you know in the same channel the same speaker so now sometimes you get a microphone pop and I mean there are these fuzzy Hybrid cases and then the problem of the boundaries that have to be shifted around Simple simple\nSpeaker B: Anyway, quick question though at a high level to people think Let's just say that we're moving to this new era of like using the Pre-segmented you know non-synchronous Conversations it does it make sense to try to take what we have now which are the ones that you know We have recognition on which are synchronous and not time tightened and try to Get something out of those for sort of Purposes of illustrating the structure and the nature of the meetings or is it better to just you know forget that?\nSpeaker D: Well, I think we'll have to eventually and my hope was that we would be able to use the force alignment to get it But if we can't\nSpeaker B: But if we can't then maybe we just have to but is it worth if we can't and we can fake it even if we're we report You know we're wrong 20% of time or kind of well\nSpeaker D: I'm thinking are you talking about for a paper are you talking about for the corpus?\nSpeaker B: That's a good question actually because for the corpus it would be nice if everything were Because we'd have to completely redo those meetings and we have like ten of them now we wouldn't have to redo them\nSpeaker D: We would just have to edit them\nSpeaker C: Well now also I still haven't been forced to line but I think that when Brian comes this will be an interesting aspect to ask him as well when\nSpeaker D: When Brian I thought you said Ryan and it's like it's right, okay?\nSpeaker B: No, that's a good point though because for feature extraction like for Prodigy or something I mean the meetings we have now it's a good chunk of data. Yep. We need to get a decent. Okay\nSpeaker C: We should let's try it and that's what that's right ever since the the February meeting that I transcribed from last year First alignment has been on the table right on table right later And so I'm hopeful that that's possible I know that there are complications in the overlap sections and with lapel mics\nSpeaker B: But I mean we might be able at the very worst we can get transcribers to correct the cases where I mean You sort of have a good estimate where these places are because the recognition so poor\nSpeaker E: Right and so you know we were never gonna just go with these as the final alignments\nSpeaker B: We're always gonna run and pass some way to push these first chunk of meetings into a state where we get good alignments\nSpeaker A: I'm probably gonna spend another day or so trying to improve things by By using Acoustic adaptation The right now I'm using the other death that Models for the first alignments and it's possible that you get that's gonna be better results if you manage to Adapt the Ford models to the speaker at the reject model to all the other speech\nSpeaker E: Could you could you at the same time Adapt the reject model to the speech from all the other channels That's what you're saying oh not just the speech from that of the other people from that channel But the speech from the actual other channels\nSpeaker D: I don't think so I don't think that would work right because you a lot of it's dominated by channel\nSpeaker B: Properties, but what you do want to do is take the even if it's Clujie take the segments the Synchronous segments the ones from the HLT paper where only that speaker was talking and use those for adaptation because if you If you use everything then you get all the cross-talk in the adaptation It's just sort of blurring and that we know I mean we have that and it's about roughly two thirds I mean very roughly averaged it's not completely negligible like a third of it is bad for adaptation Cool, I thought it was higher than that. It's really it depends a lot this just sort of an overall\nSpeaker F: Well, I know it we're not turning into your speech a redo of the HLT paper\nSpeaker D: I don't want to do that. Yeah, I'm doing that for a V.S\nSpeaker B: But I think we're more against one very Yeah, really I think Morgan's talk went very well\nSpeaker D: It was you know, it was really it well presented especially the battery meter popping up that was hilarious right when you were talking about that\nSpeaker F: You know that was the battery meter saying that it was full nature. Yeah\nSpeaker D: He was on to the bullet points about talking about The you know the little handheld and trying to get lower power and so on and Microsoft pops up a little window saying your batteries are now fully charged Yeah I'm thinking about scripting that for my talking about a little script in there to say your batteries are low right when I'm saying\nSpeaker F: Yeah, no, I mean in your case when you were joking about it, but I mean in your case The fact that you're talking about similar things at a couple conferences. It's not These are conferences that have really different emphases whereas HLT and and your speech are too close Yeah, pretty similar so I I can't see really\nSpeaker B: Just putting in the same thing. No, I don't think that paper is really the HLT paper is really more of an introduction to the project paper and yeah\nSpeaker D: Yeah, yeah, we want some results. We can get them well. Yeah, it's probably wouldn't\nSpeaker F: Or some or some I mean I would see your speech if we have some your speech papers These will be paper submissions. These will be things that are particular things Aspects have a detailed look yet rather than you know overall tempted a global paper about it\nSpeaker C: I did go through one of these meetings I had one transgarbers go through and tighten up the bins on one of the MSA meetings and then I went through afterwards and double checked it so that one is really very Very accurate. I mentioned the length I'm trying to remember the number off-hand so one of the NSA's I sent email before the conference before last week That might have been the one I'm sure that that was accurate have been through it\nSpeaker B: That might actually be useful, but they're all native speakers\nSpeaker D: Yeah, so it's gonna say the problem with those are\nSpeaker B: Extremely hard to follow like word-wise. I've got the transgar I mean I have no idea what they're talking about\nSpeaker C: I corrected a friend number the words. I'm sure\nSpeaker B: There's tough for language model probably But but that might be useful just for okay Andreas is leaving leaving the building\nSpeaker D: See you Oh I guess it's all right for you to talk a little without the mic I know you adjusting the mic a lot did it not fit you well\nSpeaker C: Why what I know is when you turned your head it would it would tilt maybe it wasn't just tightened it never\nSpeaker E: Yeah, this thing that you have actually if if you have a larger head that mic's got to go farther away Which means the the balance is gonna make it want to tip down anyway?\nSpeaker D: Yeah, okay, yeah, I'm just thinking you know we've been talking about changing the mics Yeah, for a while and if these aren't Acoustically they seem really good, but if they're not comfortable we have the same problems we have with these stupid things\nSpeaker C: I think it's come this is the first time I've worn this I find it very comfortable I find it very comfortable too\nSpeaker D: But it looks like Andreas was having problems. I think Morgan was saying it well, but I had it on this morning And it was fine. Oh, you did wear it this morning. Yeah, okay, it's off\nSpeaker E: I yeah, I don't want it on this I just want to Say what I think is a problem with this if you are wearing this Over your ears and you've got it all the way out here Then the balance is gonna want to pull it this right whereas if somebody with a smaller head has it back here It's more balanced. Yeah, then it then it falls back this way. So what what it's supposed to do is the back strap is supposed to be under your crown\nSpeaker D: And so that should be should be It's right against your head there, which is what it's supposed to be that balances it so it doesn't slide out This is supposed to be right right below And so it's supposed to be right under that so it's really supposed to go more like this Yes, exactly, but then isn't that going to that I guess you can that tilts right and lots and lots of different ways\nSpeaker B: So I'm not saying anything about that head's small head size\nSpeaker C: Would be an advantage If he was wearing it over there instead of under his ear, I think probably it was work on compressing the heads\nSpeaker D: It probably just wasn't tied enough to the back of his head I mean so the directions do talk about bending the hair side pants way off the back, which is not really what we want\nSpeaker C: That's good\nSpeaker D: We did that We at Boeing I used I was doing augmented reality so they had head mounts on and we had a little jury rig I don't know the welders how much and we had just a bag with a bunch of marble sentences\nSpeaker F: Well, maybe this could be helpful just for evening the conversation between people people those who talk a lot have to wear heavier weights Anyway So I was gonna say oh yeah, I was gonna say I had these conversations with NIST folks also So they they have their their plan for a room with mics in the middle of the table and Close-mounted mics and they're talking about close-mounted and the pals And raise and the ray and cameras and yeah multiple multiple video cameras covering covering everybody every place in the room The the mics in the middle the head mounted mics the lapel mics the array with Well, there's some discussion of the nine they might go down to 57 because There was some pressure from a couple people to meeting for them to use a keem our head I forget what keem our stands for but what it is is it's dummy head Oh, that's right. Yep, and and so what they're actually doing is they're really there's really two recording systems So they may not be precisely synchronous, but there but there's two two recording systems one with I think 24 channels No one with 64 channels and 64 channel one is for the array, but they've got some empty channels there and anyway They like they're saying they may give up a couple or something before for the keem our head if they go go with that is a good idea\nSpeaker D: Yeah Jonathan viscous did say that They have lots of software for doing calibration for skew and offset between channels and that they found that's just not a big deal\nSpeaker F: Yeah Yeah, not too worried about that was yeah, but they're still\nSpeaker D: Planning to do like fake scenario based they have to do right there. They're legal issues want to lab them to do otherwise But it sounded like they were pretty well thought out there. They're gonna be real meetings It's just that they're with with people who would not be meeting otherwise did they give a talk on this or was Inform us we just had some discussions various discussions. Yeah, I also sat and chatted with several of the nests books\nSpeaker E: They seem like a good group. What was the the paper by Lori the Mel that you mentioned?\nSpeaker F: Yeah, we should just have a heavy read it, but I mean Well got these little proceedings, but basically It was about Going to a new task where you have insufficient data and using using data from something else and adapting and how well that works So in effect it was pretty related to what was nandras did right except that this was not with meeting stuff. It was with I could think they didn't they start off with broadcast news\nSpeaker D: They're broadcast news was their acoustic models and then all the other tasks were much simpler Yeah, so they were command and control and that sort of thing. Yeah digits was one of them. Yep, and\nSpeaker E: What was there? Yeah, Red Bull's what was their conclusion it works. Yeah\nSpeaker D: Yeah, yeah, yeah, that was one of the ones that I liked that it not only works in some cases It was better which I thought was pretty interesting, but that's because they in control for parameters so You know broadcast news nets were not nets\nSpeaker E: Did they ever try going complex going the other direction from simpler tasks to more complicated tasks not in that paper\nSpeaker F: That'd be hard\nSpeaker D: Yeah, well one of the big problems with that is is often the simpler task isn't fully doesn't have all the phones in it and that makes it very hard But I've done the same thing. I've been using broadcast news nets for digits. Yeah, like for the speech proxy thing that I did that's what I did Yeah, sure it works\nSpeaker F: Yeah, yeah, and they have I mean they have better Adaptation than we had in that yeah that system so they You mean they have some Yeah, we should probably what actually what we should do I Anything about this were probably the five is should pick out a paper to that that You know got our interest and we should Go around the room at one of the Tuesday lunch meetings. Yep, so you know what you're talking about reference. Yeah\nSpeaker B: Well the summarization stuff was interesting. I mean, I don't know anything about that field but for this proposal and meeting summarization I mean sort of a far cry because they weren't working with meeting type data but Get sort of an overview of the different approaches. Do you remember who those groups were that were a lot of different Last day, but I mean there's that's a huge field and probably the groups there may not be representative of the field. I don't know exactly That everyone submits was whether folks from bbn presenting yet there was let's see Smider bn ibn Maryland\nSpeaker F: It was\nSpeaker B: The order one the sentence ordering one was that barzile and these guys? I'm just so bad anyway I it's in the programmer should have read it to remind myself, but that's sort of useful and I think like when Mari and Katrin and Jeff are here be good to figure out some kinds of things that we can start doing maybe just on the transcripts because we already have\nSpeaker D: Yeah, we do have word transcripts\nSpeaker C: Well, I like the idea that Adam out of Maybe generating minutes based on some of these things that we have because it would be easy to do that Just right and it has to be though someone from this group because of the technical nature of the thing someone who actually does take notes\nSpeaker D: I think there's all these right down the wrong things\nSpeaker B: You know how do you evaluate whether the summary is good or not and that's what's what's interesting to me is that there's different ways to do it\nSpeaker E: Yeah, was SRA one of the groups talking about some organization now\nSpeaker D: As I said, I like the Microsoft talk on scaling issues and words since this impaguation. That was interesting\nSpeaker F: Yeah, that was an interesting discussion\nSpeaker D: It it it it was the only one it was the only one that had any sort of real disagreement\nSpeaker F: Well, I didn't have as much disagreement as I would have liked but I didn't want to I didn't want to get into it because It was the application is when I didn't know anything about so it just would have been you know He getting up to be argumentative but but I mean the missing thing so so what they were saying So only thing is you know all you need is more data sort of but I mean it that's that's missing it I mean it was a nice study They were doing this it wasn't word sense to some regulation. Well, it sort of was it was a word\nSpeaker D: It was it was a very simple case of two versus two versus two and there there there that you could do better with more data\nSpeaker F: I mean that's really and so what they did was they had these different kinds of learning machines and they had different Mots of data and so they did like you know eight different methods that everybody you know argues about about oh my my kind of learning machine is better than your kind of learning machine and They were started off with a million words that they used which was evidently a number that a lot of people doing that Particularly kind of task had been using so they went up being Microsoft and went up to a billion And then they had this log scale showing that you know and and then they went up to a billion They that's a big company. I didn't mean is anything negative\nSpeaker D: But it's in the bigger the comfort and more words they use the reason they can do that is that they assumed that Text that they get off the web like from Wall Street Journal is correct and edited So that's what they use this training data is just saying if it's in this corpus. It's correct Okay, but I mean yes\nSpeaker F: Of course, there was the kind of effect that you know what would expect that that you got better and better performance with one more data But the their real point was that the the different learning machines were sort of all over the place and and by by going up Significantly in data you get a much bigger effect than by switching learning machines and further more which learning machine was on top kind of Depended on where you were in this picture. So this was my concern about the recognizer in Aurora that\nSpeaker E: That the differences we're seeing in the front end. Yeah, our relevant our irrelevant once you get a real recognizer at the back end\nSpeaker F: Yeah, you know, yeah, could well be so so I mean that was that was kind of you know, it's a good point But the problem I had with it was that the implications out of this was that The kind of choices you make about learning machines were therefore irrelevant which is not as far as I know in in tasks I'm more familiar with it is not at all true What is is true is the different learning machines have different properties and You want to know what those properties are and someone else sort of implied that well We you know all the study of learning machines. We still don't know what those properties are We don't know them perfectly But we know that some kinds use more memory and some other kinds use more computation and some are are\nSpeaker E: Limited kind of discrimination, but are just easy to use and others are it doesn't their conclusion just sort of you could have guessed that before they even started because If you assume that these learning things get better and better and better than As you approach there's a point where you can't get any better right you get everything right?\nSpeaker D: No, but they're all spread. They weren't all if they weren't converging\nSpeaker E: They were all still the bread they have to as they all get better. They have to write right sure better\nSpeaker D: They hadn't even come close to that point all the tasks were still improving when they hit a billion\nSpeaker E: Yeah, but they're all going the same way right so you have to get closer. What they didn't get closer\nSpeaker D: Oh, they did they just switched position\nSpeaker F: Well, that's getting close. I mean Yeah, the spread was still pretty wide. That's true, but but I Think it would be earned to intuition that this would be the case But to really see it and have the intuition is quite different. I mean, I think somebody So you was talking about earlier that the effect of having a lot more data is quite different in switchboard It depends on broadcast news\nSpeaker B: Yeah, it depends a lot on whether you know a disinviguation is exactly the case where more data is better right? Yeah, yeah, you can assume similar distributions But if you wanted to do disinviguation on a different type of Test data than your training data then that extra data wouldn't generalize so\nSpeaker D: But I think one of their they had a couple points I Think one of them was that well maybe simpler algorithms and more data is better less memory faster operation simpler Right because they're simplest most brain dead algorithm did pretty darn well When you got gave it a lot more data and then also They were saying well you have access to a lot more data Why are you sticking with a million words? I mean their point was that this million word corpse that everyone uses is apparently 10 or 15 years old And everyone is still using it so yeah\nSpeaker F: But anyway, I think it's just we could talk about this stuff It's it's not really the conclusion they came to so much as the conclusion that some of the Commenters in the crowd right came up with that you know this therefore is further evidence that you know more data is really All you should care about and that I thought was just kind of going to a lighter way and the the one one person gave it got up and made a brief defense But it was a different kind of grounds it was that that The reason people were not using so much data before was not because they were stupid or didn't realize data was important But in fact they didn't have it available But the other point to make again is that Machine learning still does matter but it matters more in some situations than in others and also there's there's not just mattering or not mattering but there's mattering in different ways I mean you might be in some situation where you care how much memory you're using right are you care?\nSpeaker F: You know what recall time is are you care you know and are you only have a million words?\nSpeaker D: Yeah, for you some new task or\nSpeaker B: Another language. Yeah, I mean you see there's papers on portability and write Prototyping and blah blah blah and there's people saying oh just add more data and there's cost and there's like two different\nSpeaker F: Relatives cost. Yeah, it's just like cost, you know, so so these I mean the in the speech side the thing It always occurs to me is that if you if you One person has a system that requires 10,000 hours to train on and the other only requires a hundred and they both do about the same Because the hundred hour one was smarter That's that's gonna be better. Yeah, because people I mean there isn't gonna be just one system that people train on and then that's it for For all of time. I mean people are gonna be doing other different things and so these things matters matter Yeah, so that's one of the\nSpeaker C: Providence slides this up and it's like this is this people kept saying can I see that? Slime and then they make a comment one person said Well, one person said you know before you dismiss 45 years of research. Well, you know the same thing has happened in\nSpeaker B: Computational and risks right you look at the ACL papers coming out and now there's sort of a turn backwards Okay, we've learned statistic. You know, we're basically getting what we expect out of some statistical\nSpeaker D: Methods and you know the there's arguments on both sides. I think the matters is the thing that that was misleading Is that all all of them are based on all the others right just you can say you said focus or something? Yeah, I mean So and I was saying the same thing happened with speech recognition right for a long time people were hand quote coding linguistic rules And then they discovered machine learning worked better and now they're throwing more and more data and worrying and then you have Surrying less and less about the exact details of the algorithms Except when they have a year speech paper Anyway, anyway, so we read tickets he is starting are we gonna do one at a time or should we read them all again? Oh once again Let's do it all once we have to say we let's try that again Okay, maybe we won't laugh so remember to read the transcript number so that everyone knows that what is and Ready three two one L22 transcript L-27465453738 07 6 9 4\nNone: 8 6 7 7 6 7 8 8 9 6 9\nSpeaker E: 9 6 7\nSpeaker F: 6 8\nSpeaker G: 8\nSpeaker D: 8\nNone: 8 8 8 8 8 8 8\n"}, {"meeting_id": "ES2009a", "summary": "The meeting was about a preliminary idea of new remote control, covering the price, the functions, the appearance and the name. After a brief self-introduction, Project Manager assigned the task. One of the most important issues of the meeting was about the price. Project Manager supposed the product should be sold at 25 euros with a one hundred percent profit. As it would be a multifunctional remote control, the members were confident that it would stand alone. Moving on to the issue of the appearance, the group analyzed the problems of the existing remotes and briefly talked about the user interface as well as came up with a name of the product.", "dialogue": "None: Okay.\nSpeaker E: How we placed in terms of...\nSpeaker E: You're okay.\nSpeaker B: I guess I should probably try to see up straight.\nSpeaker C: Like that?\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker B: I think mine's falling off.\nSpeaker D: I'm just going to start this PowerPoint real quick.\nSpeaker A: Yeah.\nSpeaker A: Very official.\nSpeaker A: Yeah.\nSpeaker E: I kind of like this.\nSpeaker A: I'm kind of getting into it.\nSpeaker A: Right.\nSpeaker A: So, just to kick off the meeting basically.\nSpeaker A: So we're working for a real reaction.\nSpeaker A: And this is...\nSpeaker A: Just got an agenda to set out what we're going to try to accomplish in this particular first meeting.\nSpeaker A: We're going to just do a quick opening and we can hopefully all get acquainted with one another.\nSpeaker A: Then we're going to talk a little bit about tool training.\nSpeaker A: Essentially that means getting used to the only thing that we haven't tried out yet, the whiteboard.\nSpeaker A: We've got a general plan for the project, how we're going to go about accomplishing this and then just a bit of discussion to close up.\nSpeaker A: IRL, I guess, you know, gamers lying in real life.\nSpeaker A: So, yeah.\nSpeaker A: Basically, I want to...\nSpeaker A: I'm just going to...\nSpeaker A: Of course we can discuss that.\nSpeaker A: I'm thinking about proposing that since we've got this weird blend of ourselves and our roles that we just don't ask, don't tell.\nSpeaker A: So, if you say something about marketing, right, sort of.\nSpeaker A: I'm just going to believe you can hook up from there.\nSpeaker A: Exactly.\nSpeaker A: I mean, if you guys...\nSpeaker A: At the same time, if you logically, if something doesn't look like a comic, we're going to sell a remote control that's the size of this paper book.\nSpeaker A: You say, well, that doesn't seem like such a good idea because of X.\nSpeaker A: Obviously, go with it and we'll discuss it.\nSpeaker A: But I'm not going to ask, do you know that?\nSpeaker A: It seems like...\nSpeaker A: Prove it.\nSpeaker A: Yeah.\nSpeaker A: Exactly.\nSpeaker A: Because we're sort of role-playing.\nSpeaker A: You're going to tap into your own knowledge as well.\nSpeaker A: And that's the same for when we do introductions.\nSpeaker A: I mean, and you talk about your background.\nSpeaker A: You know, have fun.\nSpeaker A: Maybe you went to...\nSpeaker A: You know, maybe you're like me and you went to UCSB, but you want to say you went to Harvard or something like that.\nSpeaker A: Why not?\nSpeaker A: You know, you can...\nSpeaker A: I guess we can have a little bit of a phone with it.\nSpeaker A: So, are you guys okay with that?\nSpeaker A: Does that seem logical?\nSpeaker C: Oh, yeah.\nSpeaker C: It's fine.\nSpeaker D: It works for me.\nSpeaker C: It works for me.\nNone: It works for me.\nSpeaker A: Yeah.\nSpeaker A: So, I guess we're totally...\nSpeaker A: We're making a remote control, which is thrilling.\nSpeaker A: But the idea is that we can make something...\nSpeaker A: Based on the whole corporate motto, I don't know if you guys had time to check the...\nSpeaker A: In your life.\nSpeaker A: I don't know if you guys would check the corporate website.\nSpeaker A: We've got to make something as fashionable as possible.\nSpeaker A: That's kind of the corporate strategies.\nSpeaker A: We're going to try to take ordinary stuff that nobody really thinks about and try to make it nice.\nSpeaker A: You know, like John Lewis, nicer, you know, if you go to Devanems or something.\nSpeaker A: So, basically, we are reinventing the wheel, but we want to try to do it in a user-friendly, slick, sleek kind of way.\nSpeaker A: What we're going to go about doing that is basically, at first, we're going to start on the basics.\nSpeaker A: And that's where I'm going to need you guys to use your interface designers and the other designer that I can't remember the idea, the idea, the industrial designer.\nSpeaker A: Hey, I'm not getting into it.\nSpeaker A: To guide me and guide us on this project, because you're going to be...\nSpeaker A: You guys are the bottom.\nSpeaker A: You know, you can't do that.\nSpeaker A: You can't have X and Y at the same time.\nSpeaker A: And then we'll work up from what is necessary to more like what would be good.\nSpeaker A: You know, like...\nSpeaker A: I think you guys probably got the same email as I did, but the idea of, yes, a coffee pot needs to be able to hold coffee.\nSpeaker A: But it's also better if it's not like really cheap glass, so that if you touch it, you hurt your hand or something like that.\nSpeaker A: And so we'll work up from there and then we'll meet on and talk about it. And then finally, we'll incorporate as kind of the last stage.\nSpeaker A: You know, what you guys build or tell us what's possible and then you tell us what we can hope for and what...\nSpeaker A: And what way to go, take the basics and make it nicer.\nSpeaker A: And then obviously, the UID and the ID, you know, you can keep on, you know, cutting edge of how to get about maximizing what is possible to make it easier.\nSpeaker A: To try to sync it all up. So that's a detailed design. So it's a three stage kind of thing.\nSpeaker A: Right, so for now, just for the whiteboard, basically, just to get used to it, I haven't tried it yet either.\nSpeaker A: I'm just going to start and carry like five remotes around.\nSpeaker A: And just write down, I'm just going to write down one of the names of my desert discs.\nSpeaker A: You know, if you were trapped on a desert island and you can only bring five CVs along with you, name one of them that you could.\nSpeaker A: Not all five. If you want to write all five, go for it. But name one of them that you could.\nSpeaker A: Oh, we skipped introductions. Nice. I'm an excellent project manager. I'm Marty.\nSpeaker A: I went to uni at UC Santa Barbara and I'm here working on a PhD in psychology.\nSpeaker A: Yeah, so...\nSpeaker D: I'm Sarah. I went to Michigan and I'm here in cultural studies and I'm a marketing manager or something, marketing expert.\nSpeaker A: Don't play yourself down. That's me.\nSpeaker B: I'm Ron. I once upon a time studied in Victoria and I am the user interface designer.\nSpeaker C: I'm Nathan. I'm from California and I'm here doing a master's screen in social anthropology.\nSpeaker C: What did you go to uni at UC?\nSpeaker A: No, I'm really cool. My little brother goes there.\nSpeaker A: Right, so desert island discs.\nSpeaker D: So we have to wait for you to write it down or you're going to tell us?\nSpeaker A: Well, I'll just get a couple of them down. See, I'm a big music fan. I don't know if you guys are.\nSpeaker A: I'm assuming everybody likes music because it's some lesser, greater extent.\nSpeaker A: But there's some other options. If you're a TV slut like I am, like small, terrible television show, but I happen to love it.\nSpeaker D: It's rubbish, but I love it. I went to high school with Tom Willing, actually.\nSpeaker A: The main guy. Wow. Is he a wanker? Yeah.\nSpeaker D: Very much so. I love a soccer player, but a total bastard.\nSpeaker A: He looks really tall. He's got to be like six seconds. He's a big guy.\nSpeaker A: Okay, so I really like Jeff Buckley. You guys heard Jeff Buckley?\nSpeaker A: That's cool because not very many people have.\nSpeaker A: Oh, well, I might as well throw a British person in there.\nSpeaker A: You can't go wrong with radio head.\nSpeaker D: Can you call?\nSpeaker A: Okay, so it really works just like a panel and it makes noises, I think.\nSpeaker A: It's kind of weird. Anyway.\nSpeaker A: Yeah, you like pressing your...\nSpeaker A: You'll see.\nSpeaker A: Alright, so whoever wants to get up next, you can write down some telly that you watch or whatever you want.\nSpeaker D: I guess I'll go next next.\nSpeaker D: Right on.\nSpeaker D: Okay, we lose all my mics plugged in here. Okay.\nSpeaker D: This is basically just pen practice, huh?\nSpeaker D: Okay. Oh, you're much taller than me, so I'm going to write down here.\nSpeaker D: Right now I'm listening to a lot of...\nSpeaker D: Somebody never heard of Chris Bathgate, local Michigan, folk singer.\nSpeaker D: Nice.\nSpeaker D: Really lame.\nSpeaker D: And what else did I bring with me?\nSpeaker D: Probably classical, the total leaky.\nSpeaker D: Okay, yeah, yeah.\nSpeaker D: I think...\nSpeaker D: And my family guy DVDs.\nSpeaker A: Oh, yeah.\nSpeaker A: We don't need to write that one down.\nSpeaker A: Oh, family guy.\nSpeaker A: Do you watch the new season?\nSpeaker D: No.\nSpeaker D: Are you getting it online or is it on the sky?\nSpeaker D: I think I'm going to start downloading it, yeah.\nSpeaker D: Yeah.\nSpeaker B: That'd be nice.\nSpeaker B: Alright.\nSpeaker B: I think I'm just going to put down one...\nSpeaker B: One CD.\nSpeaker B: Anybody?\nSpeaker B: I'm afraid not.\nSpeaker B: Afrobeat orchestra.\nSpeaker B: Very cool.\nSpeaker B: Afrobeat orchestra.\nSpeaker B: Yeah.\nSpeaker B: Very cool.\nSpeaker B: They like 15 members from Brooklyn.\nSpeaker B: And I'm hoping to go to the concert in Belgium.\nSpeaker B: Wow.\nSpeaker B: Exciting.\nSpeaker B: Yeah.\nSpeaker B: Supposed to be across the center.\nSpeaker B: Thing I love about Enra.\nSpeaker D: I didn't even read those.\nSpeaker D: I should admit that.\nSpeaker A: That's what a PowerPoint presentation is for.\nSpeaker A: They're designed specifically to ignore.\nSpeaker D: It's brilliant.\nSpeaker D: It's the five by five.\nSpeaker D: I can't read that much.\nSpeaker A: Ah, yes.\nSpeaker A: Yes, okay.\nSpeaker A: I see that.\nSpeaker A: Vomit.\nSpeaker A: Yes.\nSpeaker E: Streetteats.\nSpeaker D: That's so brilliant.\nSpeaker D: Yeah.\nSpeaker D: It's so horrible.\nSpeaker A: I've seen more urine in this city than ever before.\nSpeaker A: Seriously.\nSpeaker B: But I just came from Glasgow and I'm happy to say that there's more vomit.\nSpeaker B: There's the same quantity, approximately.\nSpeaker B: It's so mean.\nSpeaker A: It really is.\nSpeaker A: Does...\nSpeaker B: Yeah.\nSpeaker B: Ready?\nSpeaker B: Alright.\nSpeaker B: Yep.\nSpeaker E: Yeah.\nSpeaker B: Mingen.\nSpeaker B: Nice.\nSpeaker B: Yeah.\nSpeaker A: I'm going local.\nSpeaker A: Yeah.\nSpeaker A: I'm going local.\nSpeaker A: I have to be here for three years.\nSpeaker D: I'm going to look at the terminology, right?\nSpeaker D: I've already got more than I can keep track of.\nSpeaker D: I'm going to go home next week and everyone's like, oh my God.\nSpeaker D: You're trying to get into one of those.\nSpeaker D: Have you been home yet?\nSpeaker A: No.\nSpeaker A: They'll be like, say something British and you're like, oh, shut up family.\nSpeaker D: Oh, it should be interesting.\nSpeaker D: Wait until I tell them I'm not coming back.\nSpeaker D: I don't know that one.\nSpeaker A: Are you going to stay here?\nSpeaker D: Probably.\nSpeaker D: Or at least get a work visa for a while and then decide.\nSpeaker D: Better religion?\nSpeaker C: Yeah.\nSpeaker C: Some music I grew up listening to.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker D: And...\nSpeaker D: Yeah, I can think of so many other ones.\nSpeaker D: Well, yeah, that's why.\nSpeaker C: That's how it works.\nSpeaker C: Yeah.\nSpeaker C: Something I miss about my hometown.\nSpeaker A: I miss coffee.\nSpeaker A: Doritos.\nSpeaker A: Doritos.\nSpeaker C: I cost less than...\nSpeaker C: Oh, yeah.\nSpeaker C: Two bucks.\nSpeaker C: Anything that are like...\nSpeaker C: Where are you from in California, by the way?\nSpeaker C: I grew up in San Diego, but...\nSpeaker A: Did you really?\nSpeaker C: What part?\nSpeaker A: La Jolla, P.V.\nSpeaker A: Yeah, I'm from San Diego as well.\nSpeaker A: Beachy.\nSpeaker A: Yeah, really?\nSpeaker C: I last lived in San Francisco having lived in Cal.\nSpeaker C: Well, I have lived in Southern California since I was 18.\nSpeaker A: Going to North Cal, I'm sorry.\nSpeaker A: You just can't get a bit of bread over what's available in San Diego.\nSpeaker C: It's different because in San Diego, the tortillas are cooked on the grill and then throw the California de steamed.\nSpeaker D: It must make all the difference.\nSpeaker A: Yeah, it really does.\nSpeaker A: Well, it's...\nSpeaker A: There's other things too.\nSpeaker A: You just can't place it.\nSpeaker A: Like, when I went to school in Santa Barbara, which is Central California, the Mexican food is okay.\nSpeaker A: It's just not good.\nSpeaker A: Like...\nSpeaker A: And yeah, it's like two bucks.\nSpeaker A: Like, literally two bucks for this massive...\nSpeaker A: Right.\nSpeaker A: Yeah.\nSpeaker A: Good call on that.\nSpeaker A: Yeah.\nSpeaker A: Where are you from in San Diego?\nSpeaker A: Just literally just metropolitan San Diego.\nSpeaker A: I live like five minutes from the zoo.\nSpeaker A: Okay.\nSpeaker A: So, North Park, actually, if you want to get real...\nSpeaker C: Yeah, my grandparents lived on 32nd.\nNone: Yep.\nSpeaker C: Clust, you know, where a clardoline coffee shop is.\nSpeaker A: Yes.\nSpeaker A: On university.\nSpeaker A: Yeah.\nSpeaker A: It's actually like literally half a mile from my house.\nSpeaker A: Cool.\nSpeaker A: Yeah, pretty cool.\nSpeaker A: Small world.\nSpeaker A: Yeah.\nSpeaker A: As we were discussing before, especially when we're all from the same general region.\nSpeaker A: Right.\nSpeaker A: So, okay.\nSpeaker A: Success on the whiteboard.\nSpeaker A: There you go.\nSpeaker A: You can harness the awesome power a little bit of interactions.\nSpeaker A: We talked about some of our CDs and things we like about the city.\nSpeaker A: I think we'll...\nSpeaker A: Right.\nSpeaker A: So, moving on to not fun stuff.\nSpeaker A: Project finance.\nSpeaker A: Basically, what we're trying to do is sell this remote for 25 euros.\nSpeaker A: This is what the finance department has told me, the CFO.\nSpeaker A: But, I don't know.\nSpeaker A: I'm not sold on this.\nSpeaker A: It's pretty dear.\nSpeaker A: I mean, that's like, you know, 40 bucks for remote.\nSpeaker A: It would have to pretty much like do my laundry for me.\nSpeaker A: So, we can maybe work on that later.\nSpeaker A: But, we're going to make a lot on it.\nSpeaker A: We have profit names to make 50 million euros on it internationally.\nSpeaker A: So, one of the things that I was going to mention to you guys, the designers, is that we probably need a...\nSpeaker A: It needs to be a universal remote control probably.\nSpeaker A: Okay.\nSpeaker A: So, something that could do NTSC as well as PAL as well as various other formats.\nSpeaker A: Like if it's going to control DVDs.\nSpeaker A: But, you know, I'll leave that to you guys.\nSpeaker A: But that's something that is going to be an international sold thing.\nSpeaker A: But, we want to try to make it for 1250.\nSpeaker A: So, we want to try to make 100% profit on it if we can't.\nSpeaker A: So, just to close up, I'm not sure how much time I've used the next time, right?\nSpeaker A: Project manager, sort of.\nSpeaker A: We'll meet in another half an hour or so.\nSpeaker A: And, I'd like the industrial designer to think about what needs to be done.\nSpeaker A: Like what the basic function of it.\nSpeaker A: You ID...well, yeah, your assignments are up there.\nSpeaker A: And you'll also get assignments from in your email as well.\nSpeaker A: More specific on what to do.\nSpeaker A: Basically, so, I mean, do you tell us what the user's going to want?\nSpeaker A: So, actually, in a way, you guys maybe in our next meeting chat a bit about what the user's going to want and what the user can have.\nSpeaker A: You know, like...\nSpeaker A: You'd go and shape that.\nSpeaker A: Yeah, well, it is when we'll discuss the trade-offs in between.\nSpeaker A: And so, yeah, specific instructions will be sent in your email.\nSpeaker A: But, I think that that is more or less a good place to start for now.\nSpeaker A: And as more things come up, we'll have meetings and you'll get emails and so forth.\nSpeaker A: Any questions before we get started?\nSpeaker B: I assume that we're building a standalone remote control.\nSpeaker B: We can't kind of build it into other products.\nSpeaker B: You mean, for instance, like a mobile phone.\nSpeaker C: Hmm, that's interesting.\nSpeaker C: Yeah.\nSpeaker D: I don't think there's any rules about it yet.\nSpeaker C: Maybe you're a personal coach.\nSpeaker B: Or, you know, can we sell a remote control phone for 25 pounds, you know?\nSpeaker A: Well, have a think about it.\nSpeaker A: I mean, okay.\nSpeaker A: I'm certainly open.\nSpeaker A: It seems like, yeah, it seems like it's certainly doable, isn't it?\nSpeaker A: Or if we can't have a full mobile phone, maybe a remote that has some other kind of useful function.\nSpeaker A: The clap.\nSpeaker A: No, good idea.\nSpeaker A: Good idea.\nSpeaker C: Let's see.\nSpeaker C: Every mode with changeable faces, like face-to-face.\nSpeaker A: I like that.\nSpeaker A: I like that.\nSpeaker A: Hot.\nSpeaker A: Yeah.\nSpeaker A: That's true.\nSpeaker A: I guess we probably have some time.\nSpeaker A: Maybe we should brainstorm a bit like what we want to do.\nSpeaker A: I'll go back to, um, I don't really have any.\nSpeaker A: Let me bring up something about our basic goals here, what we want to accomplish.\nSpeaker A: Project announcements.\nSpeaker E: Hmm.\nSpeaker E: Yeah.\nSpeaker A: Not so much.\nSpeaker A: All right, we'll find that we're on our own.\nSpeaker B: Now, are we also discussing kind of our initial ideas at all?\nSpeaker B: Yeah, yeah, let's do it.\nSpeaker A: Let's do it.\nSpeaker B: Does anybody have any initial ideas?\nSpeaker A: I'm going to go ahead and take notes on this too, because...\nSpeaker D: Good idea.\nSpeaker D: Start two minutes.\nSpeaker A: Yeah, I mean, I don't remember.\nSpeaker A: So, initial ideas.\nSpeaker D: Well, it's pretty much given it's going to be universal.\nSpeaker D: Yeah.\nSpeaker D: We decided that already.\nSpeaker D: And it may be functioning for other things.\nSpeaker D: As soon as you said that, I was thinking like all the other things you get in remote too, like your microwave or your front door, like, to have everything on one thing.\nSpeaker D: But then, never been a fan of those huge remotes that have like a million buttons you can't do.\nSpeaker B: Smaller is better.\nSpeaker B: Simple.\nSpeaker B: I'm thinking...\nSpeaker B: Specific.\nSpeaker B: I'm thinking kind of PDA, uh, design, so touchscreen design, rather than...\nSpeaker B: Okay.\nSpeaker B:...but also that you can kind of flip around all sorts of different things.\nSpeaker A: Yeah, that slick, isn't it?\nSpeaker A: I mean, like stylist.\nSpeaker A: Yeah, yeah.\nSpeaker A: Sure.\nSpeaker A: So, we got five minutes more to chat about that.\nSpeaker A: That's perfect.\nSpeaker A: So, we've got this kind of an idea of a trade-off between, um, size and functionality.\nSpeaker A: Right.\nSpeaker D: And we also...\nSpeaker D:...month multifunctional, but at the same time, if you get it to do too much, you're not going to be able to tell them apart.\nSpeaker C: It's going to be too complicated.\nSpeaker C: Yeah.\nSpeaker C: It's a crowded button, some things.\nSpeaker A: I'm also going to note for future reference, this idea of, um, so you, like, remember, I can, like, a touchscreen type of remote?\nSpeaker B: Mm-hmm.\nSpeaker B: Possibly.\nSpeaker A: I don't think one of this.\nSpeaker A: You know, it should be an option.\nSpeaker C: It needs one outstanding feature to set it apart from all the different relates.\nSpeaker A: Definitely.\nSpeaker A: Yeah, all the other universal remotes.\nSpeaker A: Um, I don't know if there's such thing out there.\nSpeaker A: I guess we'll do some research on it or one of us can do some research on it about whether or not there are, um, multi-format.\nSpeaker A: Like, um, you know, PAL, NTSC, Region 1.\nSpeaker B: Right.\nSpeaker B: I'm pretty sure there is.\nSpeaker B: I mean, I have a friend who has a PDA.\nSpeaker B: Okay.\nSpeaker B: That he just points at his tel- any television he wants, and it'll figure out the specifications of it and we'll control it.\nSpeaker B: Interesting.\nSpeaker A: Awesome.\nSpeaker B: Okay.\nSpeaker B: I assume that that can be done with going around the world.\nNone: Okay.\nSpeaker A: Um, all right.\nSpeaker A: So I'm liking that idea, this idea of a touchscreen remote with multi-format features.\nSpeaker A: Um, right.\nSpeaker A: Um, let's see.\nSpeaker C: I think making it out of a nice material would be very important because so many of those remotes that you see, these universal remotes look so cheap and low quality.\nSpeaker D: Yeah, it'd be nice and slick would be important.\nSpeaker D: And I don't know, like, there's such a problem with losing them.\nSpeaker D: That adding this whole, like, PDA pen business is only one more thing to lose.\nSpeaker D: So we're going to be careful of what, like, just something like, keep in mind, we start actually dealing with this stuff, but that would be really cool.\nSpeaker A: Let's see.\nSpeaker B: Um, I like the idea of the multi-plate.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker D: What do they call those faceplate things?\nSpeaker D: They're named as... Are they?\nSpeaker C: Yeah.\nSpeaker C: I like...\nSpeaker B: We'll have to come up with a fuzzy one as well.\nSpeaker B: Lumbercraven.\nSpeaker B: Lumbercraven.\nSpeaker C: I think it wouldn't be such a bad idea to have a locator device.\nSpeaker C: Maybe a simple button that you have on your television to help you find your remote.\nSpeaker A: But if we're bundling it, unless we're selling their tele with the remote.\nSpeaker A: Right.\nSpeaker A: Um...\nSpeaker B: Well, if we bundle it as a phone, then you can always call it.\nSpeaker B: True.\nSpeaker B: If you're not doing that, then we can have something that just kind of rings from either...\nSpeaker B: Well, there used to be those whistling devices, but that's a little bit annoying.\nSpeaker A: Could we not do something where like just a little... Like literally just a very small kind of thing that comes with the remote?\nSpeaker A: That you can place somewhere else that you press and it makes the remote page.\nSpeaker A: Kind of like how on a lot of cordless regular phones you have a page button that goes, right?\nSpeaker A: Could we do something like that?\nSpeaker B: That's cool.\nSpeaker B: I think so.\nSpeaker B: I think we could design that.\nSpeaker D: Yeah.\nSpeaker D: Good.\nNone: Um...\nSpeaker A: Yeah, I think this material quality as well, like I guess we can think about what kind of...\nSpeaker A: You know, Apple has been really successful with this surgical white kind of business or this sleek kind of...\nSpeaker D: Yeah, and that titanium, the new silver sleek ones, the last couple of years.\nSpeaker D: Yeah.\nSpeaker D: It's pretty much so.\nSpeaker D: Curves.\nSpeaker A: Yeah, we do have the minimum amount.\nSpeaker A: I mean, we were talking finances, I don't know, selling a 40-pound remote, or a 40-dollar remote, 25-year-old remote would be pretty...\nSpeaker A: You know, it's pretty expensive.\nSpeaker A: So maybe we might want to trade off some of the features for a lower price.\nSpeaker A: Without getting in about a whole, like, you know, go down to bargain store remote, you know, bargain store universal remote that's black and, you know, massive, some kind of... I don't know, a balance there in somewhere.\nSpeaker A: Definitely.\nSpeaker A: But, um, have a think about what we can do, have a think about what we want to do, how we're going to sell it.\nSpeaker D: And our users in mind, like these...\nSpeaker D: Grammies are not going to be into this whole new...\nSpeaker D: Let's design the modes, the buttons.\nSpeaker D: So I'll have to be careful of exactly who we're marketing this to, and who we're going to be able to get it out of it.\nSpeaker D: It's true.\nSpeaker C: We're talking 25 pounds or 25 euros.\nNone: 25 euros.\nSpeaker B: It's a slight difference, I guess.\nNone: Yeah.\nNone: They're all weaker than... they're all stronger than the dollar.\nNone: Although, computer parts.\nSpeaker B: Oh, if you can upgrade your computer by the time you're in a computer, you can't get it.\nSpeaker A: They're all weaker than... they're all stronger than the dollar.\nSpeaker A: Although, computer parts.\nSpeaker A: Oh, if you can upgrade your computer by the states.\nSpeaker A: Like, do you guys know Fries?\nSpeaker A: Huge computer electronics store?\nSpeaker A: They serve... or they say, yes, they will sell things overseas.\nSpeaker A: So you can buy stuff in America and have it shipped over for like 23 pounds, like that.\nSpeaker A: I'm going to put these initial ideas that we've got in the project documents.\nSpeaker A: So if you guys need a reminder about what we've talked about on the different trade-offs that we've got and the other ideas, you can consult them at your leisure.\nSpeaker A: And right, so thanks for that.\nSpeaker A: Let's just head back to work on what we were talking about.\nSpeaker A: Getting into...\nSpeaker D: You have an hour?\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker A: Perfect.\nSpeaker A: Cool.\nSpeaker A: Thank you.\nNone: Thanks for watching.\n"}, {"meeting_id": "ES2008d", "summary": "This meeting was about the prototype presentation and evaluation. Firstly, the group presented the prototype with agreed features including a latex case in fruity colours that's curved, push buttons with an on-screen menu button as well as eight extra buttons. Next, the group had a discussion about the prototype evaluation. They gave a high scale of the score on creativity, ease of use, integration of fruits as well as the simplicity. Also, they gave an average score on technology innovation and ease of learning. Then, the group agreed to set the target price at 12.5 Euros. Lastly, the group had a discussion about the project process, including management and leadership as well as group mates' personal learning.", "dialogue": "None: Okay.\nSpeaker D: Okay.\nSpeaker E: Welcome to our detailed design meeting.\nSpeaker E: I'm pretty excited.\nSpeaker E: Let's start.\nSpeaker E: It's approximately 1540 or something like that.\nNone: Okay.\nSpeaker E: The agenda we're going to do an opening and then I'll talk about the minutes from the last meeting, what we discussed.\nSpeaker E: Then we'll have the prototype presentation and look at the evaluation criteria.\nSpeaker E: We'll look at the finances and finally do production evaluation and close.\nSpeaker E: So starting off with the last one.\nSpeaker E: I don't have it here.\nSpeaker E: But we talked about energy.\nSpeaker E: We're going to use a kinetic battery.\nSpeaker E: We want to use a simple chip because we're not going to need a shuffle.\nSpeaker E: I'm going to need a scroll.\nSpeaker E: We're choosing a latex case in fruity colors.\nSpeaker E: It's curved.\nSpeaker E: And we're using push buttons with a supplement of an on-screen menu.\nSpeaker E: It sounded like we had set like eight or nine buttons, including five preset channels.\nSpeaker E: Okay.\nSpeaker E: Let's do the look and feel design presentation first.\nNone: One.\nSpeaker A: Right.\nSpeaker C: Well, we made three different prototypes.\nSpeaker C: And I guess we'll start with this one.\nSpeaker C: We have our colors not fixed, but this is the general shape.\nSpeaker C: You hold it sort of either like this in your left hand or you switch it over and it's easily adaptable to either hand.\nSpeaker C: You can push the buttons with your thumb like a mobile phone or you can push them with your index finger of your other hand or even then there's a whole variety.\nSpeaker C: You can hold it like this and press it with your same index finger.\nSpeaker C: We have the on off button at the tip.\nSpeaker C: Very visible.\nSpeaker C: Very big.\nSpeaker C: We have our up and down buttons, which are also going to be our channel selectors.\nSpeaker C: And we have our little menu button here.\nSpeaker C: If you push, if you're just pushing these normally, they're the menu buttons.\nSpeaker C: If the volume buttons rather, if you press select once, they become channel changing buttons.\nSpeaker C: If you press select three times, the menu with the other features and possibly also with your TV channel choices shows up and you have your five presets down here.\nSpeaker C: People want to grab a hold of that and see how it feels in your hand.\nSpeaker C: That's our number one prototype.\nSpeaker C: You want to present the potato or shall I fill with the varsity?\nSpeaker E: You're just acting a little lightning bolt in it.\nSpeaker E: Very cute.\nSpeaker E: I can do about one of the wrong bits.\nSpeaker A: This one's known as the potato.\nSpeaker A: It's not presented.\nSpeaker A: It's an ergonomic shape.\nSpeaker A: It fits in your hands.\nSpeaker A: It's designed to be used to be the left hand or the right hand.\nSpeaker A: I've gone here just before buttons on this one.\nSpeaker A: There are two blue buttons here after adjusting the volume.\nSpeaker A: This is called volume up and volume down on the other side here.\nSpeaker A: The back one is the artboard, and check the channels.\nSpeaker A: Channel up and channel down.\nSpeaker A: The moves between the bigger channels that you've selected.\nSpeaker A: This middle button here is the top-down screen menu.\nSpeaker A: When you're working in the on-screen menu, you use the other four buttons to navigate the middle button.\nSpeaker A: That's basically it.\nSpeaker E: On off?\nSpeaker A: That would be one of your channels.\nSpeaker A: Channel zero would be...\nSpeaker A: It turns off.\nSpeaker B: Could the middle button of the on-screen menu function as a power button?\nSpeaker A: That's not really...\nSpeaker A: It's not really hard to do.\nSpeaker C: If you press and held it, maybe.\nSpeaker C: You're going to call it down.\nSpeaker B: Yeah.\nSpeaker C: That would be on.\nSpeaker C: On-off.\nSpeaker C: That's the possibility of...\nSpeaker C: And then finally, we have the Martian or the pair either way.\nSpeaker C: It's a bit different.\nSpeaker C: Just a little bit more of a creative feel.\nSpeaker C: You have the on-off toggle stem on the top.\nSpeaker C: We have the five preset Z.\nSpeaker C: And then you have all the sides to make it a little bit more three-dimensional.\nSpeaker C: You have your channel changing volume changing buttons and your menu button right here in the middle.\nSpeaker C: That's for your consideration as well.\nSpeaker C: Plus it's an interesting talking point to have standing up.\nSpeaker C: We figured it could stand up like this on your table if I made the bottom flat.\nSpeaker C: Sorry, what's the other one in the middle?\nSpeaker C: The menu select button.\nSpeaker E: Very interesting.\nSpeaker E: That was my favorite.\nSpeaker A: I'm not sorry, I'm sorry.\nSpeaker C: We were thinking that normally we'd go for fruity colors but maybe we're also thinking that you're sort of middle aged man.\nSpeaker C: For an example, customer might not want a fruity colored remote.\nSpeaker C: So maybe we'd have one version that's a bit toned down, maybe with less contrast on it.\nSpeaker C: There's something still a little bright to make it hard to lose but...\nSpeaker E: That was one thing that we brought up over email.\nSpeaker E: I don't know if you picked up your email but the feature that we considered for not getting lost.\nSpeaker C: Right.\nSpeaker C: We were talking about that a little bit when we got that email.\nSpeaker C: And we think that each of these are so distinctive that it's not just like another piece of technology around your house.\nSpeaker C: It's going to be somewhere that it can be seen.\nSpeaker C: So we're not thinking that it's going to be as critical to not cover in a couch.\nSpeaker E: Would it be very difficult to just have an external device that like, I don't know, you taped to your TV that when you press it, a little like beep goes off?\nSpeaker E: Do you think that would be conceptually possible?\nSpeaker A: I think.\nSpeaker A: I think that's the most most. It's probably under the city and then not because of the uncertainty that's in the city.\nSpeaker A: That's true.\nSpeaker A: I wonder if it's more just the gimmick than anything else.\nSpeaker A: I don't know how many times the service would be doing that to go with the device like that.\nSpeaker C: There might be something that you can do in the shortboard in the chip to make it make a noise or something, but it would take a lot more development than we have this afternoon.\nSpeaker E: Okay.\nSpeaker E: That's a fair evaluation.\nSpeaker E: Getting lost.\nSpeaker E: We've decided not to worry about that for now.\nSpeaker E: Okay.\nSpeaker E: Well, the designs are very bright.\nSpeaker E: So you're right.\nSpeaker C: They're going to stick out.\nSpeaker C: So do people have a preference as far as feel and functionality?\nSpeaker B: I feel like this is similar to sort of what already exists.\nSpeaker B: So we're trying to think of something sort of like new and fun, even though this is like what I'm initially drawn to, just because it's like comfortable and like not different.\nSpeaker B: I sort of like this one.\nSpeaker B: Like I don't know why it just it's like small, but still sort of like cute looking.\nSpeaker B: I don't know.\nSpeaker B: But I also like the side buttons on that one.\nSpeaker B: Like I think that's kind of me, but I don't know how much any of this has to do with the fashionable sort of cool looking thing that we also need to focus on.\nSpeaker E: Could we maybe have like an extra button on the top for on off?\nSpeaker E: So then we wouldn't have to do it.\nSpeaker C: Yeah, that's good.\nSpeaker C: Here.\nSpeaker C: Stick it on.\nSpeaker E: Sure.\nSpeaker E: Um, why don't we do a product evaluation using your criteria?\nSpeaker B: Well, to be one, like I think we're supposed to have one that we do it for.\nSpeaker B: Oh, okay.\nSpeaker B: Okay.\nSpeaker B: That was, I was a little vague on what exactly I'm supposed to do, but let me, I have to like write something on the whiteboard.\nSpeaker E: So do you need this or just right on the right?\nSpeaker B: No, I actually don't have like a PowerPointy thing.\nSpeaker B: Okay.\nSpeaker B: Because I think it would be redundant.\nSpeaker B: I don't know.\nSpeaker E: It's kind of like a joystick kind of thing.\nSpeaker E: Hey.\nNone: Okay.\nSpeaker E: Maybe a little smaller than that.\nSpeaker E: No, I kind of like it.\nSpeaker E: That's hard to mess.\nSpeaker E: Oh, it does.\nSpeaker A: It's kind of like.\nSpeaker E: Okay.\nSpeaker E: There could have a stem like that because I do kind of like the stem.\nSpeaker E: Yeah.\nSpeaker E: It almost helps you keep a grip too.\nSpeaker E: Because it goes in between fingers.\nSpeaker E: Just think.\nSpeaker C: I like this one.\nSpeaker C: Okay.\nSpeaker C: That would be the most important thing that I'm missing.\nSpeaker C: I like the idea of the, I mean that's really gotten the simplicity of the buttons down.\nSpeaker E: That one.\nSpeaker E: I am worried about like using a menu.\nSpeaker E: Um, in that, like within menus there are sub menus.\nSpeaker E: And so how do you get back to the main menu?\nSpeaker E: Well, on the iPad, you just, every time you want to go back you hit the menu button.\nSpeaker E: That has a menu button separate from a select button.\nSpeaker E: Whereas if this one's both the menu and the select button.\nSpeaker A: This is, this is the up and down button.\nSpeaker A: So you use the scrolling up and down for the list of choices.\nSpeaker A: Okay.\nSpeaker A: So you just, you just divide the button.\nNone: Okay.\nSpeaker C: Could these be used for going sub menus or maybe, yeah, maybe it could be one of those if you press down and hold for two seconds.\nSpeaker C: Then it brings you back one level or something.\nNone: Okay.\nSpeaker E: Well, as long as we have that in mind as we're designing it still.\nSpeaker B: Okay.\nSpeaker B: So which one are we sort of roughly looking at to address whether or not it meets our, um, necessities, but yellowy one is that potato?\nSpeaker B: Are we leaning towards the pertaining?\nSpeaker B: So, okay.\nSpeaker B: Well, we can obviously change it after we go through these different ones.\nSpeaker B: So basically what we need to do is some of the things that we've talked about before.\nSpeaker B: We need to make sure that that remote actually does conform to the things that we said it was going to.\nSpeaker B: So what we sort of want to do is we each need to separately rank each of the following things and then I'll tabulate an average just to make sure that it doesn't meet that.\nSpeaker B: So we'll just go through them one at a time and we'll just go around and each of you can tell me on a scale of one to seven with one being really extremely true and seven being not true at all or false if the remote that we've created conforms to the following criteria.\nSpeaker B: So we can do this one first.\nSpeaker B: First, we want to know if it meets the fancy looking feel objective.\nSpeaker B: So like in my opinion, for now at least the yellow one is probably somewhere in the middle.\nSpeaker B: So I'm going to say it's like a three.\nSpeaker B: That's just my opinion.\nSpeaker B: What is each of you?\nSpeaker E: I kind of think it's unique enough that I'd give it a one or two.\nSpeaker B: Okay.\nSpeaker B: Well, give it the number.\nSpeaker B: Sorry.\nSpeaker E: I will give it a one.\nSpeaker C: I don't know if it's creative.\nSpeaker C: I don't know if fancy is the word I would use.\nSpeaker C: I don't know if any of them are fancy.\nSpeaker C: I'd say two because it's unique.\nSpeaker B: And two.\nSpeaker B: Awesome.\nSpeaker B: All right.\nSpeaker B: And same sort of scale for functionality.\nSpeaker B: Is it functional?\nSpeaker B: I think it's extremely functional.\nSpeaker B: I'm going to give it a one.\nSpeaker B: One.\nSpeaker B: One.\nSpeaker A: I think it's functional.\nSpeaker A: It's also pretty basic.\nSpeaker E: It's under the answer.\nSpeaker E: Okay.\nSpeaker E: Functional.\nSpeaker E: I think it'll get everything done.\nSpeaker E: I think it might be a little confusing at first.\nSpeaker E: Well, there's some other points I will address that.\nSpeaker E: Okay.\nSpeaker E: Then I'm going to give it a two.\nSpeaker B: Awesome.\nSpeaker B: Okay.\nSpeaker B: We want to know next if it's technologically innovative.\nSpeaker B: Did you give a functional?\nSpeaker B: Yeah.\nSpeaker B: She said this one.\nSpeaker B: Is it technologically innovative?\nSpeaker B: Not really.\nSpeaker B: I mean, not so much because we don't have the LCD screen.\nSpeaker B: We don't have fancy chip.\nSpeaker B: Other than what it looks like.\nSpeaker B: I don't know if it's really a battery.\nSpeaker B: It's a kinetic battery is a big one.\nSpeaker B: So.\nSpeaker C: How many people noticed that though?\nSpeaker B: But we know it's there.\nSpeaker E: They'll be like, Hey, I have never changed the battery.\nSpeaker B: And if it's made of like latex, that whole idea.\nSpeaker B: That's pretty cool.\nSpeaker B: I'll give it a three.\nSpeaker B: Because it could have picked a lot of features.\nSpeaker B: I would have made it really.\nSpeaker C: Yeah.\nSpeaker C: I would say that it's.\nNone: Yeah.\nSpeaker C: Like fancy versus creative.\nSpeaker C: It's different.\nSpeaker C: It does that.\nSpeaker C: You put innovative.\nSpeaker C: I don't know.\nSpeaker C: I'll give it a three.\nSpeaker B: Why?\nNone: I don't know.\nSpeaker A: It's certainly not the case.\nSpeaker A: It's not the main thing.\nSpeaker A: It's just.\nSpeaker A: It's just.\nSpeaker A: It's just.\nSpeaker A: It's just.\nSpeaker A: So.\nSpeaker E: I think I'm going to go with the four as well.\nSpeaker E: Okay.\nSpeaker B: Next is it easy to use just so you know, easy to learn will be separate.\nSpeaker B: Is it all?\nSpeaker B: Okay.\nSpeaker B: Overlap them.\nSpeaker B: I think it's really easy to use.\nSpeaker B: I'll give it.\nSpeaker B: Two.\nSpeaker B: I'll give it a one.\nSpeaker C: Pretty hard to mess up.\nSpeaker B: I'll let's say two.\nSpeaker B: All right.\nSpeaker B: We next want to see if it has a spongy quality.\nSpeaker B: And indeed it's made of latex or rubber.\nSpeaker B: It's spongy all the way.\nSpeaker B: Give it a one.\nSpeaker A: Yeah.\nSpeaker A: One of the bounces when you drop it.\nSpeaker C: Oh, then you could.\nSpeaker C: It'd be harder to break.\nSpeaker E: It'd be less impact.\nSpeaker A: What do you give it?\nSpeaker B: All right.\nSpeaker B: And the next is does it integrate some notion of fruits and vegetables?\nSpeaker B: Well, is it going to be yellow?\nSpeaker C: It might be.\nSpeaker E: That's our corporate color, isn't it?\nSpeaker E: That's right.\nSpeaker E: Yeah.\nSpeaker E: Corporate color.\nSpeaker E: We didn't keep that.\nSpeaker E: We want to keep it yellow.\nSpeaker E: Well, if we.\nSpeaker E: I know it would make it a little less a little more confusing, but if we had all the buttons in black and the design and the outside in yellow.\nSpeaker E: That would be our corporate one.\nSpeaker E: And then we could also have alternative colors.\nSpeaker E: One a more conservative one.\nSpeaker E: One is more fruity.\nSpeaker B: Yeah, it's like a silvery kind of white or something.\nSpeaker B: Yeah.\nSpeaker E: Can we have like an R R inscribed on the bottom or something?\nSpeaker E: Sure.\nNone: All right.\nSpeaker B: So I think it was.\nSpeaker C: It was inspired by the potato.\nSpeaker C: So I think it's kind of mango-y too.\nSpeaker E: Yeah.\nSpeaker B: I gave it a little one.\nSpeaker B: The mango kind of put me over.\nSpeaker C: That's a much worse.\nSpeaker B: I know.\nSpeaker B: What's everyone's numbers?\nSpeaker A: One.\nSpeaker B: All right.\nSpeaker B: And does the design match the appropriate behavior?\nSpeaker B: Remember earlier we discussed that people don't use a lot of buttons that they use the channel flipping in the volume the most.\nSpeaker B: I think we really took that into account a lot.\nSpeaker B: So I'm an evil one.\nSpeaker C: Yeah.\nSpeaker C: Me too.\nSpeaker B: Do you say one?\nSpeaker B: Yes.\nSpeaker D: Yes.\nSpeaker B: Okay.\nSpeaker B: Also we talked earlier about RSI and wanting to prevent any sort of like carpal tunnel-y kind of thing.\nSpeaker B: Do we think that the latex sort of grip appropriately takes that into account?\nSpeaker B: I think I'll give it to you because I almost feel like no matter what you do something is going to happen.\nSpeaker E: And if it's repetitive movement, it is going to be only four buttons that you're constantly pushing.\nSpeaker E: But I'm worth the rest.\nSpeaker E: I like how it fits in the hand though.\nSpeaker E: I'd go with it too.\nSpeaker C: I'll say two as well because older people that aren't used to testing, but if I might bend it a bit.\nSpeaker C: I'll add that first.\nSpeaker A: All right.\nSpeaker B: Awesome.\nSpeaker B: And the ease of learning it.\nSpeaker B: I know you were saying that you were a little bit nervous about that.\nSpeaker B: I don't know.\nSpeaker B: It sort of reminds me of the iPod.\nSpeaker B: I just got my excellent read the instruction book and I'm doing okay.\nSpeaker B: And I'm not good at learning technology.\nSpeaker B: So I'll give it a two.\nSpeaker C: The menu system and the fact that multiple buttons are used for different things might be a bit confusing.\nSpeaker C: But I think if it's one of those things that it might take you five minutes to figure out, but you'll have it afterwards.\nSpeaker C: So I think I'd give it a two I guess.\nSpeaker A: I think it's probably a little harder than most of the way.\nSpeaker A: Because you have to use the menu system and you have to tell it what the type of travel is on.\nSpeaker A: Oh, good point.\nSpeaker A: Once you've learned how to use it, I think it'll be a four.\nSpeaker A: Okay.\nSpeaker E: I think I'd give it a four too.\nSpeaker E: It's a pretty high learning curve.\nSpeaker E: It'll be easy once you've done it, but.\nSpeaker B: All right.\nSpeaker B: Also earlier, we had something about losing it.\nSpeaker B: And so now we're not addressing that at all.\nSpeaker E: So we, I think we can kind of say we addressed it with color.\nSpeaker E: Okay.\nSpeaker B: So in terms of not losing it, do you think that on a scale of one to seven, how easy or hard is it to lose?\nSpeaker B: I'm going to, I'm going to give it a four because I think that you can still, if it's in between somewhere where you can't see it, you're kind of not going to find it, but anywhere else it's going to stand out.\nSpeaker C: I said, give it a three.\nSpeaker C: It depends how tidy you are normally.\nSpeaker E: You give it a four.\nSpeaker A: Okay.\nSpeaker A: Small too.\nSpeaker B: All right.\nSpeaker B: We also said simplicity.\nSpeaker B: How, how does it address just being simple?\nSpeaker C: Simple to use or simple in design.\nSpeaker B: I think overall, because we had said before our two main things were simplicity and fashion.\nSpeaker B: So those are the next two things we're going to look at.\nSpeaker B: Separate from fancy like that sort of thing.\nSpeaker B: Um, it like wants to be simple, but it's not like totally totally simple.\nSpeaker B: So I'm going to give it a two.\nSpeaker C: I want to give it a three, I guess.\nSpeaker A: Three.\nNone: All right.\nSpeaker B: And fashionable.\nSpeaker E: It's totally fashionable.\nSpeaker B: It's hot.\nSpeaker B: I mean, it's a mango.\nSpeaker B: I mean, how fashionable can you make a road?\nSpeaker B: I think it's bringing technology and fashion together really really well.\nSpeaker C: I don't know.\nSpeaker C: I don't think it's quite as fashionable as my robot or boat.\nSpeaker C: I do like the little more.\nSpeaker B: Yeah, the toggle on off switch.\nSpeaker B: It's really appealing.\nSpeaker B: Number of two.\nSpeaker B: One.\nSpeaker B: And does it meet our like demographic need sort of for international appeal.\nSpeaker B: That whole thing.\nSpeaker B: Just that it would serve our audience.\nSpeaker B: I don't see why not.\nSpeaker E: I think as long if we offer in a in at least three different color arrangements.\nSpeaker E: Yeah.\nSpeaker E: So I'll give it a two.\nSpeaker C: Yeah.\nSpeaker B: All right.\nSpeaker B: Does anyone have any other features that they think are important that we didn't talk about?\nSpeaker E: Well, we didn't we didn't address the fact that it does need to have a corporate logo.\nSpeaker E: So let's make sure we keep that in mind that we have that one of our color concepts is corporate and has an R.\nSpeaker E: R on it.\nSpeaker E: Well, I think all of them are.\nSpeaker B: And so we're going to do that.\nSpeaker B: So we'll address it.\nSpeaker B: Fine.\nNone: Okay.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: Now we're going to look at finances.\nSpeaker E: I have an Excel sheet that we're actually going to calculate the cost.\nSpeaker E: So let me add this first.\nSpeaker E: Okay.\nSpeaker D: I know.\nSpeaker E: Okay.\nSpeaker E: Hand dynamo.\nSpeaker E: We're using kinetic matter.\nSpeaker E: Yeah.\nSpeaker E: And we're having one per.\nSpeaker E: One.\nSpeaker E: Okay.\nSpeaker E: So we have a little bit of electronics.\nSpeaker E: Single.\nSpeaker E: Simple.\nSpeaker D: And simple, rather.\nSpeaker D: Okay.\nSpeaker D: Case.\nSpeaker E: Double curved.\nSpeaker E: Yeah.\nSpeaker C: It's very curvy.\nSpeaker C: So, okay.\nSpeaker C: Yeah, I didn't get a picture of those.\nSpeaker C: So I don't really know.\nSpeaker C: Our case material supplements.\nSpeaker B: Oops.\nSpeaker C: We just get one.\nSpeaker C: We need plastic and.\nSpeaker C: No, we the plastic is the plastic for the inside is.\nSpeaker C: Assume.\nSpeaker C: The wood.\nSpeaker B: Oh, I guess it was rubber rather than rubber.\nSpeaker D: It was rubber and special color.\nSpeaker B: Yeah.\nSpeaker B: More than one special color.\nSpeaker E: Yeah, we're using.\nSpeaker E: We're going to need at least two special colors.\nSpeaker E: I don't know what the basic color is.\nSpeaker B: Yeah.\nSpeaker B: Well, but we know that we're having at least three colors.\nSpeaker C: Well, are we talking about on each color combination or we know like we'll have yellow and black?\nSpeaker C: Is that two special colors?\nSpeaker E: I don't know.\nSpeaker C: Or that I think white and black or.\nSpeaker C: I guess it's three.\nSpeaker C: Three three three.\nSpeaker C: All right.\nSpeaker E: Maybe the R.R. will be in color as well.\nSpeaker E: So, yeah.\nSpeaker E: Interface.\nSpeaker E: We're doing push buttons.\nSpeaker E: Yeah.\nSpeaker E: And how many buttons do we have?\nSpeaker E: Sex.\nSpeaker E: I don't know.\nSpeaker C: Five.\nSpeaker C: With the power.\nSpeaker E: Anything else?\nSpeaker C: No.\nSpeaker C: Oh, we'll do.\nNone: Do we want.\nSpeaker C: Are the buttons in special colors?\nSpeaker C: I didn't get information.\nSpeaker E: Oh, wait.\nSpeaker E: Oh, so, so the case material will just have one color, right?\nSpeaker E: Well, does it?\nSpeaker C: We'll be in special colors.\nSpeaker C: But if we're making multiple varieties of this is where I'm getting confused.\nSpeaker E: We're saying per unit.\nSpeaker C: Yeah.\nSpeaker E: Okay.\nSpeaker E: So each unit will only have one color.\nSpeaker C: All right.\nSpeaker E: And each.\nSpeaker E: But the cases could have up to three.\nSpeaker E: I mean, the buttons could be up to three colors.\nSpeaker E: That's how it's designed there.\nSpeaker E: I like it like that.\nSpeaker E: Okay.\nSpeaker E: Special form.\nSpeaker E: They're all kind of just push button.\nSpeaker E: No, I think they're fine.\nSpeaker C: Special material.\nSpeaker C: We want them rubber as well.\nSpeaker C: Probably.\nSpeaker E: Oh, do I have to do it per button?\nSpeaker E: Do I?\nSpeaker C: No, I don't think so.\nSpeaker C: I think they're all going to be rubbers.\nSpeaker C: That's what it matters.\nSpeaker C: Yeah.\nSpeaker B: Yeah, because for the whole case material, it's only one.\nSpeaker B: Yeah.\nSpeaker B: I mean, it's true to make it rubber.\nSpeaker B: Okay.\nSpeaker B: Oh, wait.\nSpeaker E: So maybe.\nSpeaker E: Thirteen point seven.\nSpeaker E: Uh oh.\nSpeaker E: Yeah.\nSpeaker E: What can we reduce?\nSpeaker C: Okay, let's have our buttons all be one color.\nSpeaker B: I kind of like the button.\nSpeaker E: I would do.\nSpeaker E: It's only going to bring us down to thirteen point three anyway.\nSpeaker E: All right.\nSpeaker E: Okay.\nSpeaker E: Are we sure this is double curved?\nSpeaker E: Maybe it's single curved.\nSpeaker E: We have to know what idea.\nSpeaker E: We have no idea.\nSpeaker C: I don't know.\nSpeaker C: I didn't get any pictures.\nSpeaker D: It's single curved.\nSpeaker D: Why?\nSpeaker E: Well, it's not the.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: It's a kinetic battery that's kind of expensive, but we have a simple chip.\nSpeaker E: Single curve.\nSpeaker E: Case material is rubber and it's a special color.\nSpeaker E: Yeah.\nSpeaker E: But that's important.\nSpeaker E: That is important.\nSpeaker C: I don't know.\nSpeaker C: I don't know.\nSpeaker C: I don't know.\nSpeaker C: I don't know.\nSpeaker C: I don't know.\nSpeaker C: I don't know.\nSpeaker E: I don't know.\nSpeaker E: It just happened.\nSpeaker E: Now it's twelve.\nSpeaker E: Twelve.\nSpeaker E: Twelve.\nSpeaker A: Five.\nSpeaker A: Hey.\nSpeaker A: So we're okay.\nSpeaker E: I think.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: We're all set.\nSpeaker E: I saved that to our.\nSpeaker E: Our.\nNone: Our.\nSpeaker E: Our.\nSpeaker E: Our.\nSpeaker E: Big, shared folders.\nSpeaker E: So you know.\nSpeaker E: Okay.\nSpeaker E: Back to agenda.\nSpeaker E: Are the costs under twelve fifty euro?\nSpeaker E: Yes.\nSpeaker E: Yeah, they are.\nSpeaker E: Let's move on to the project evaluation.\nSpeaker E: Project process.\nSpeaker E: Satisfaction with, for example, the room for creativity, leadership, team work, the means, and you know, idea is found.\nSpeaker E: So I guess what we're going to talk about here is just evaluating how we created this project, the information we got, and used how we used it.\nSpeaker E: If we were able to, you know, use our creativity with the information.\nSpeaker E: How well I guess I let it.\nSpeaker E: How well we worked together as a team.\nSpeaker E: The digital pen, the whiteboard.\nSpeaker E: Okay.\nSpeaker C: I felt very creative.\nSpeaker C: I enjoyed making the prototype.\nSpeaker A: I think it was being successful in the two almost only put a gun in the house for the second semester.\nSpeaker A: Yeah.\nSpeaker A: And we'd come up with a thing about the new gifts that are the gifts.\nSpeaker C: I think we could have done even better with a little bit more information, like what's a single curve case, how many colors, what do colors count, things.\nSpeaker C: But given what we had, I think we did really well.\nSpeaker C: I think we worked together pretty well.\nSpeaker B: Yeah.\nSpeaker B: I mean, if I had more market research, other than the fruits and vegetables, maybe we could have taken that into account.\nSpeaker C: But the fruits and vegetables stipulated by creativity.\nSpeaker C: I know.\nSpeaker B: I really did.\nSpeaker B: The whole mango idea was great.\nSpeaker B: I mean, the pen was a little distracting for me personally.\nSpeaker B: It's heaviness.\nSpeaker B: I've just been so conscious of turning the pages.\nSpeaker B: I don't know.\nSpeaker B: That was a bit of a distraction.\nSpeaker B: But with the last one, I chose not to do a powerpoint because I didn't think it made sense to.\nSpeaker B: So I liked that I had the option to do that and just to take my own notes and that whole sort of thing.\nSpeaker E: I think we all made very significant contributions.\nSpeaker E: I don't think anybody dominated it, which I thought was really good.\nSpeaker E: Like each of us was able to, like, if you had your information and I tried to facilitate it without taking over.\nSpeaker E: Yeah.\nSpeaker D: A little finished bodos.\nSpeaker B: That's kind of a lot one.\nSpeaker C: Can we work at this as the back go remote?\nSpeaker C: Should we have that somewhere on the packaging?\nSpeaker D: I have a little more powerpoint.\nSpeaker B: I'm trying to think of a good pun that I could add there.\nSpeaker B: I know.\nSpeaker E: It's like a little jingle.\nSpeaker B: I like the R.R. that's going to be etched in.\nSpeaker C: Yes.\nSpeaker C: Hopefully not with just my thinking out at some point.\nSpeaker C: But kind of a useful little gadget.\nSpeaker C: Oh, thanks to Ian for the design of that one.\nSpeaker E: Okay.\nSpeaker E: What do we have?\nSpeaker C: Our ideas found.\nSpeaker C: People should really base their boat's, our troops, our vegetables.\nSpeaker E: Or at least be creative enough to think of toggle switches.\nSpeaker C: I, et cetera.\nSpeaker C: I am really into the idea of comedic batteries now.\nSpeaker C: I'm so excited about them.\nSpeaker C: Can I even know they existed?\nSpeaker C: I knew you could get watches that had them.\nSpeaker C: Like, really quite expensive watches.\nSpeaker C: That just would never need a battery because you're always moving your wrist.\nSpeaker C: But in other things, I think it would be really good.\nSpeaker E: And it, yeah.\nSpeaker E: Okay, closing at the cost of the budget.\nSpeaker E: With the project evaluated.\nSpeaker E: And then just find a questionnaire meeting summary.\nSpeaker E: So this is a great product, kids.\nSpeaker E: I think we've created something really interesting and then we have a market for it.\nSpeaker E: Especially if we can produce it at 12.3, which we hope.\nSpeaker E: Yeah.\nSpeaker E: Make sure in your questionnaire to put down criticisms of both the process and the final results.\nSpeaker E: Yeah, really?\nSpeaker E: I do like the Martian remote.\nSpeaker E: If we could choose more than one, that would be my second choice.\nSpeaker B: Oh, that would definitely be my second choice.\nSpeaker E: Although the toggle.\nSpeaker E: I'm afraid I would break it.\nSpeaker E: I would break it.\nSpeaker C: I think I would break it.\nSpeaker C: It started because I wanted to have it as a stem.\nSpeaker C: It started as a pair, but it started looking weird.\nSpeaker C: We're like, what if I could penguin?\nSpeaker D: I like that it stands up.\nSpeaker C: Well, maybe I should market it to some remote control company.\nSpeaker C: So our art.\nSpeaker D: A little thing.\nSpeaker E: Okay.\nSpeaker E: Congratulations.\nSpeaker B: Nothing will come up until after our meeting.\nSpeaker B: It's supposed to be over at five.\nSpeaker E: I got more master classes.\nSpeaker E: That's really funny that you got spamming your work.\nSpeaker E: Check it out.\nSpeaker E: You need to keep in mind your knowledge management.\nSpeaker B: I want to make sure you do.\nSpeaker B: It's very work relevant because we did.\nSpeaker E: I'm very sad that I didn't get any links to the corporate website.\nSpeaker C: You can see it through the corporate website.\nSpeaker B: It's really deep.\nSpeaker B: It takes a little while to get excited to live.\nSpeaker B: That thing is pretty cool.\nSpeaker B: Yes, I love Excel.\nSpeaker B: It's one of my favorite programs.\nSpeaker C: I'm not sure.\nSpeaker C: My inspiration from last time is the interior of remote control being taken apart bit by bit.\nSpeaker B: I learned a lot.\nSpeaker C: I could probably take a part of remote control now if I needed to.\nSpeaker C: I thought it was a little tricky having to hand around this thing.\nSpeaker B: It's a technology out today.\nSpeaker B: It's really interesting.\nSpeaker E: I thought it was a little tricky having to hand around this thing.\nSpeaker E: Yes.\nSpeaker E: It's really cool.\nSpeaker E: I don't see how we could have.\nSpeaker E: At least for me.\nSpeaker C: If I had pictures of the different parts of the case, I would have probably drawn them up.\nSpeaker B: I would have been able to put them up.\nSpeaker C: I would have to have a PowerPoint that we can all look at.\nSpeaker E: You can do all of those things on the PowerPoint.\nSpeaker B: These might have made us more willing to take notes than to write them up here.\nSpeaker C: I would have to have them up separately in the whiteboards in this room.\nSpeaker C: Because we were all going to be working in different places, if we were all going to stay here all the time, then having the notes up on the whiteboard would have been all right.\nSpeaker E: Everyone needs their own specific notes.\nSpeaker E: When you were creating your prototypes here, did you work together\nSpeaker A: or do you do separate projects? I would have to have them up and forth.\nSpeaker C: It was a bit of both.\nSpeaker C: We just started out by each taking a color of clay and just fooling around with it.\nSpeaker C: I came out with that rhombus shape and he came out with a potato-y, mango-y shape.\nSpeaker C: Then I just went from there.\nNone: I felt well-managed.\nSpeaker C: I think we did well in the first festival, keeping our meetings designed.\nSpeaker A: We did a good job of making some decisions with men.\nSpeaker C: Sorry.\nSpeaker C: That happened to me all the time though.\nSpeaker A: I know what happened sometimes is that men have meetings and people said that the way without anything else would really be decided.\nSpeaker A: I think that women have been on the ground and decided to decide someplace now that men are looking to set up the new way to go on.\nSpeaker C: I was quite worried at the end of the last meeting that we weren't getting.\nSpeaker C: We didn't really have enough information to make from decisions, but we were able to do it regardless.\nSpeaker C: I'm not usually a very decisive person, so it helped to have people say this thing is a little bit different than five minutes.\nSpeaker E: I found that we could have used another five or ten minutes.\nSpeaker E: Especially last time.\nSpeaker E: I think the last time we had a lot of information, but at the same time, not quite enough.\nSpeaker E: You know what I mean?\nSpeaker E: We couldn't answer every single question.\nSpeaker C: I felt the first two meetings that I was coming in with no information.\nSpeaker C: That sort of made me really...\nSpeaker C: I don't know.\nSpeaker C: Throwing together a PowerPoint of no information.\nSpeaker C: Last time it was the opposite.\nSpeaker C: I had so much information and so much to talk about.\nSpeaker E: It was interesting what came out later.\nSpeaker E: As I was doing the breakdown on the agenda that they gave me, that more points came out from your presentation even.\nNone: I'm a little...\nSpeaker E: I am a little disappointed that we didn't do something for losing the remote because that was kind of a big point.\nSpeaker C: That was something like...\nSpeaker C: In order to talk about that, we would have had to have actual knowledge about circuit boards.\nSpeaker B: Even when we were creating from the Excel file, there wasn't an option to select to somehow have it included.\nSpeaker B: There was no way we could be like, yeah, I have them included.\nSpeaker B: There's no way for us to have written down that it was really bare.\nSpeaker C: I think we were just getting over enthusiastic with the task.\nSpeaker C: Well, it's interesting really got into it.\nSpeaker C: I mean, I got into it as the day went on.\nSpeaker C: I got really like, oh, I'm designing remote control.\nSpeaker E: I don't know if that's just me, but yeah, I suppose.\nSpeaker E: It was interesting though that they put that 50% of people say they have frustration with losing their remote and yet they didn't provide us with information too.\nSpeaker E: Yeah, I don't want to be able to discuss that.\nSpeaker E: I know that people have things that can put on their keys, that you press it and it'll be.\nSpeaker E: Yeah, but I'm sure that would require some kind of technology that I just don't know what it is.\nSpeaker C: I think there are some of those infomercial remote that have things so you won't lose them.\nSpeaker C: I don't know.\nSpeaker C: I mean, we were talking about it.\nSpeaker C: In my household, at least there's only about two places that the remote is ever.\nSpeaker C: There's only one TB and there's only like two chairs.\nSpeaker B: That's insane. You're never going to lose your keys.\nSpeaker B: I always do anyway.\nSpeaker B: You'll lose them in your pocket.\nSpeaker B: Like you just won't forget that we're going to put it down somewhere that like made sense and then not remember there's always ways to lose things.\nSpeaker E: Yeah, it depends on how organized you are personally.\nSpeaker C: Yeah, or like, I guess what the sense of the house is too.\nSpeaker C: But I am notorious for losing my keys.\nSpeaker C: I guess I've just never lost the remote.\nSpeaker C: I put my keys in the refrigerator the other time.\nSpeaker C: I find them.\nSpeaker C: So I was putting groceries away.\nSpeaker C: That's why.\nSpeaker C: You're taking stuff off a bag and putting them in the refrigerator.\nSpeaker C: Back to my room.\nSpeaker E: Can't get in.\nSpeaker E: Look all around the kitchen.\nSpeaker C: Definitely in the vegetable drawer.\nSpeaker B: So I always do the living and my coat and using a different coat.\nSpeaker B: Yep.\nSpeaker D: I guess that's good.\nSpeaker C: We don't really carry room out around that much.\nSpeaker C: Yeah.\nSpeaker B: Can't always get into the other room.\nSpeaker C: Maybe it was our little robot when we could have had him have a robot alien pair or whatever he is.\nSpeaker C: Have a little voice.\nSpeaker C: I am located.\nSpeaker C: A GPS.\nSpeaker C: It's an internal GPS.\nSpeaker D: And then...\nSpeaker C: There you go.\nSpeaker C: Although if it's sitting still for a moment.\nSpeaker E: We should make one that walks by itself.\nSpeaker E: It really could get up and walk away.\nSpeaker E: Or some crazy electromagnet that you push it and it'll zoom to the TV and stick there.\nSpeaker C: Or just a wheel.\nSpeaker C: Just if you have to ever vote for your robot.\nSpeaker E: Well, but if you could attach them to the TV.\nSpeaker E: Yeah.\nSpeaker C: All kinds of possibilities.\nNone: Okay.\nSpeaker E: Sorry, I'm just trying to update my minutes.\nSpeaker E: I decided to, you know how I sent you the email saying that PowerPoint minutes will be complimentary.\nSpeaker E: Yeah.\nSpeaker E: Rather than like repeating.\nSpeaker E: I'm just going to make them full minutes so that it'll include all of the agenda and all that.\nSpeaker E: Well, because it seems a little more useful.\nSpeaker C: So you had the most typing and organizing to do.\nSpeaker E: But I didn't have like information to swap.\nSpeaker E: Yeah.\nSpeaker C: How much of your PowerPoint was already done for you?\nSpeaker C: Every meeting.\nSpeaker E: Most of it.\nSpeaker E: I added slides.\nSpeaker E: I added a couple slides each time.\nSpeaker E: But that's better.\nSpeaker C: I didn't even think about adding slides because I would just get slides with titles on them and build them all in.\nSpeaker C: Didn't even think about adding more.\nSpeaker E: Well, the thing was it would provide you an agenda with like several points, but it wouldn't have a slide for each point.\nSpeaker E: And that's the only way I remember that I need to go over that point.\nSpeaker E: I know personally when I do PowerPoint, that's what I do.\nSpeaker E: So I added ones, even if it was just like the title of it, like the three presentations.\nSpeaker E: Okay.\nSpeaker B: You have to have a slogan on it or can it just be like the slogan on it?\nSpeaker B: No, different.\nSpeaker B: We could go around the other side.\nSpeaker C: No, I don't think we need to.\nSpeaker C: Yellow and black one.\nSpeaker C: Yellow and blue.\nSpeaker C: Let me go to the web page.\nSpeaker E: I was just kind of going by the web page because they didn't give me any clear yellow gray.\nSpeaker B: Gray.\nSpeaker B: Gray.\nSpeaker B: It's better than black.\nNone: Doesn't look so bumble lady.\nSpeaker C: I really understand the reading.\nSpeaker C: Finish meeting now.\nSpeaker E: Wasn't it interesting that I thought it was interesting that our marketing expert did not agree with the marketing choices.\nSpeaker E: Yeah.\nSpeaker B: Well, because I just think you're really doing like a really big market evaluation.\nSpeaker B: You wouldn't just have like one set of sources kind of and they weren't so not backed up.\nSpeaker B: It would just be a sentence being like we did a survey.\nSpeaker D: Yeah, I don't know.\nSpeaker C: People are stupid.\nSpeaker C: I guess it's sort of like gray, isn't it?\nSpeaker C: Yellow and gray, but then the slogan is in blue.\nSpeaker B: Well, we got some gray and we got some blue buttons.\nSpeaker B: We're good.\nSpeaker C: Yeah.\nSpeaker C: Well, they don't have the right color clay anyway.\nSpeaker E: Maybe like, I don't know.\nSpeaker C: I think I always be something.\nSpeaker B: Well, we're not sadly meant to actually be producing this.\nSpeaker B: Oh, if they ever come out with potato potato bag.\nSpeaker B: I'm going to clean the intellectual property.\nNone: Yeah.\nSpeaker C: I can't believe the whole day.\nSpeaker C: I feel like it's been that long.\nSpeaker C: It's like a...\nSpeaker C: I don't know.\nNone: I'm actually there.\nNone: Yeah.\nNone: Yeah.\n"}, {"meeting_id": "ES2005c", "summary": "Unlike the last meeting, the team further discussed the details about the features they wanted on their new remote controls this time. First, they unanimously agreed that their design shall be the combination of beauty and fashion. Second, after looking at all the other designs existing in the market now, they drew a conclusion that the new remotes would be standardly shaped and children-friendly. Lastly, most of the time was devoted to the discussion about the details of the desired features: rechargeable battery, blue main body and yellow slogan, fruits based shapes and sponge latex rubber material.", "dialogue": "Speaker D: Every time we do something we get home.\nSpeaker F: Alright, I'm attached, I'm here.\nSpeaker D: Here?\nSpeaker D: Here.\nNone: Excuse me.\nSpeaker F: We are, the meeting will have about the same format as the last time, so I'm switching over up just left and left first two screens out.\nSpeaker F: I have mailed to you the minutes of the last meeting, just to save time.\nSpeaker F: Okay, cool.\nSpeaker F: And, is there any questions you have for the right from last meeting, particularly bothering you?\nSpeaker B: No, I don't think so.\nSpeaker F: Okay, cool.\nSpeaker F: Then we shall start with a presentation from that.\nSpeaker A: Yeah.\nSpeaker A: I've made a joke again.\nSpeaker A: In this meeting, I'm going to discuss about the trend watching.\nSpeaker A: How this trend is going to affect our market potential and how important is this.\nSpeaker A: So we have to look on this.\nSpeaker A: First of all, methodology.\nSpeaker A: The methodology to find out the trend was done in a way.\nSpeaker A: We have done a not only recent remote control market survey, but we also considered the latest fashion trends of the market, because we think that this is also a factor which will affect our sales and profit.\nSpeaker A: So, what are our findings?\nSpeaker A: In our findings, we have seen that when we did our remote control market survey, we found that people do have preference for fancy remote controls, which look and feel very good.\nSpeaker A: Rather than having a functional look and feel good.\nSpeaker A: So, this clearly indicates their preference for the design, their outlook of the remote controls.\nSpeaker A: So, we should consider this factor as the most important factor, because this factor is twice as important the second factor, which is further twice as important as the third factor.\nSpeaker A: So, this factor becomes the most important factor in our mark, means in designing our remote controls.\nSpeaker B: The last one is the most important one.\nSpeaker A: No, the first one is the outlook of the mobile.\nSpeaker A: It should have a fancy outlook.\nSpeaker A: It is a fancy design.\nSpeaker A: Rather than just having a functional look and feel good, it should have a fancy look and feel good.\nSpeaker A: The second most important aspect is that remote control should be a technologically innovative.\nSpeaker A: We must have some technological advancement in the remote controls rather than just putting it as it is as the other remote controls are.\nSpeaker A: So, it should be technologically innovative, like glow in the dark or speech recognition, something like that.\nSpeaker A: So, that indicates our technological advancement.\nSpeaker A: And the third most important aspect in this take into consideration is that it should be easy to use.\nSpeaker A: Like, it shouldn't be too much complicated.\nSpeaker A: There shouldn't be too many buttons on this remote control.\nSpeaker A: It shouldn't be too complicated like this way.\nSpeaker A: And customers should be provided with manuals that is easy to understand in their local language, something, so that they could know how to use these remote controls.\nSpeaker A: When we did fashion, our recent fashion update shows that...\nSpeaker B: Sorry.\nSpeaker B: We're in vegetables.\nSpeaker B: How do you know how we're going to incorporate that?\nSpeaker A: Yeah.\nSpeaker A: Yeah, we have to, because you can see how people have related their clothes, shoes and furniture, everything with fruits and vegetables.\nSpeaker A: So, the world is now changing its strength to our organic, becoming more and more organic.\nSpeaker A: We're going to make a big sponge running.\nSpeaker E: Yeah.\nSpeaker A: That's very good.\nSpeaker A: So, something like that we should do.\nSpeaker A: Okay.\nSpeaker A: And people, the feel of the material is expected to be spongy rather than just having a plastic look, hard look.\nSpeaker A: Well, that's good.\nSpeaker A: That's only kind of...\nSpeaker A: Yeah.\nSpeaker A: So, that they could play with it while handling it.\nSpeaker A: Okay.\nSpeaker A: So, that should also be taken into consideration.\nSpeaker A: So, these are my views.\nSpeaker B: Okay.\nSpeaker B: The spongey, not real spongey.\nSpeaker B: You can think like rubber would be good.\nSpeaker B: Does it really want to be?\nSpeaker A: The rubber which is good for health and which is quite disposable.\nSpeaker A: Okay.\nSpeaker A: That we can take into account.\nSpeaker A: Yeah.\nSpeaker A: It shouldn't be have any harm to the environment also because our company is very well adapted for taking all these concerns into consideration.\nSpeaker A: So, we don't want to have any harm to the society.\nSpeaker A: Okay.\nSpeaker A: So...\nSpeaker A: Cool.\nNone: That's all.\nNone: Pretty much.\nNone: There you go.\nSpeaker A: Just why I think of when I think of...\nSpeaker F: I remember control.\nSpeaker F: Yeah.\nSpeaker F: Next slide.\nSpeaker F: Okay.\nSpeaker A: Okay.\nSpeaker B: Are there any factors that weren't important in the survey that they said we don't want to?\nSpeaker A: We didn't find out any such word.\nSpeaker A: Yes, there could be, but we couldn't find out any.\nSpeaker A: Cool.\nSpeaker B: What is it?\nSpeaker B: Functionality.\nSpeaker F: Yeah.\nNone: Okay.\nNone: Yeah.\nSpeaker A: You're so enthusiastic.\nNone: No.\nNone: You're sick.\nSpeaker D: No.\nSpeaker C: No.\nSpeaker B: No.\nSpeaker E: Is that undo?\nSpeaker E: Yes.\nSpeaker E: Yeah.\nSpeaker E: Is that sifting?\nSpeaker E: Yeah.\nSpeaker B: Okay.\nSpeaker B: And then F5, right?\nSpeaker F: Yeah.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: So, the interface concept.\nSpeaker B: Yeah.\nSpeaker B: The interface specification.\nSpeaker B: What people, how they interact with it, basically, I think.\nSpeaker B: So, the method we looked at existing designs, what's good about them, what's bad about them, and looked at their flaws, we're going to look at their flaws and everything.\nSpeaker B: And what the survey told us and what we think would be good.\nSpeaker B: So, a bit of imagination.\nSpeaker B: Okay.\nSpeaker B: The findings.\nSpeaker B: I've got some pictures to show you.\nSpeaker B: That's the rocket.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So, most remote controls use graphical interface where you have got buttons and you point it rather than having the output as a stream of text or something.\nSpeaker B: And we're also found that there's inconsistent layout which makes it confusing.\nSpeaker B: So, I think there are remote control.\nSpeaker B: There is some inconsistency already in existing between remote controls, but I think the standard kind of shape and play and those kind of buttons like the top right for on and off or something.\nSpeaker B: I think people find that important because it's easy to use.\nSpeaker B: And we've got some pictures of some new remote controls to show you.\nSpeaker B: Do I press escape F5?\nSpeaker F: And I'll just escape.\nSpeaker B: Escape, okay.\nSpeaker B: Oh, I still haven't got the glasses on.\nSpeaker B: Yeah, okay.\nSpeaker B: So, these are some of the pictures of existing ones.\nSpeaker B: I'll just talk you through them.\nSpeaker B: This one is voice recognition.\nSpeaker B: And that's the kind of idea we're going for.\nSpeaker B: There's an LCD thing which we thought could get a bit confusing and a bit expensive as well for us.\nSpeaker B: This one has a kind of scroll like a mouse.\nSpeaker B: Middle button.\nSpeaker B: Yeah.\nSpeaker B: But I'm not actually sure how you'd use that.\nSpeaker F: It's kind of like scroll.\nSpeaker F: If I'm thinking of the right one, I've got the same thing in front of my monitor.\nSpeaker F: You scroll it and then when you reach the sort of menu item that you're acquiring, press.\nSpeaker B: It's like the LCD one, is it?\nSpeaker B: But the one below that has got a little scroll function on the side, but I presume that the functions must come up on the TV screen.\nSpeaker B: Yeah.\nSpeaker B: I think that's what that is.\nSpeaker B: So, these are just a few ideas.\nSpeaker B: Again, that's just quite boring shape, grey.\nSpeaker B: It looks quite space-agey, but too many buttons, I think.\nSpeaker B: And that one's the right thing.\nSpeaker B: Looks like something else.\nSpeaker B: Yeah, it does look kind of dangerous.\nSpeaker B: This one I thought was really cool. It's got the program and ability function that we talked about.\nSpeaker B: You can put it in there. It's a few kids and it's quite an organic shape and the little circle around there is pretty cool.\nSpeaker B: And that's really easy to use, bright.\nSpeaker B: So, I like this one a lot for our design.\nSpeaker B: I think something like that would be good.\nSpeaker B: Of course, yeah.\nSpeaker F: I mean, the one thing I think about it, but these ones is these secured areas.\nSpeaker F: I've seen a lot of them with the cover of missing.\nSpeaker F: Right, yeah.\nSpeaker B: So, like, have a head and shoulders.\nSpeaker B: Yeah, that's true. Yeah.\nSpeaker B: So, maybe that could be built into one of the things when it comes up on the TV or something.\nSpeaker B: And this one, the oversized one.\nSpeaker B: I don't know about you, but I think it's a bit too gimmicky.\nSpeaker B: I don't think that will sound very well.\nSpeaker F: I mean, is that not sort of just the blind or something, is it?\nSpeaker B: I guess so. I don't know.\nSpeaker E: I think that's a bit...\nSpeaker E: Blind or watch TV.\nSpeaker B: Yeah, exactly.\nSpeaker B: Yeah.\nSpeaker B: And this one is just pointing out...\nSpeaker B: I like some of these things, the raised symbols and everything, but it's pointing out that this one, the volume, it's kind of pressing down, but it would actually go up because of the shape.\nSpeaker B: So, that's a bit confusing.\nSpeaker B: But the buttons on this, I think, are... it's just showing you how you can have different...\nSpeaker B: different buttons. I don't have to be all the same.\nSpeaker B: So, that's quite cool.\nSpeaker F: People tend to recognize certain shapes to do certain things.\nSpeaker B: Yeah, exactly.\nSpeaker B: F5.\nSpeaker B: So, there is some of the findings.\nSpeaker B: So, we need to combine those ones.\nSpeaker B: And I've just got an email from our technical department.\nSpeaker B: Saying that they have broken through with some new speech recognition software that you can program in.\nSpeaker B: Yeah, it is. Just entirely, very handy.\nSpeaker B: So, I think maybe incorporating that in our design would be good.\nSpeaker B: You program it, like you say, record, and then play, and then record the same machine and stuff like that.\nSpeaker B: So, that's... and it's much...\nSpeaker B: Yeah.\nSpeaker B: So, that's quite cool.\nSpeaker B: Personal preferences, just an imagination.\nSpeaker B: The raised symbols, I thought, were good.\nSpeaker B: The LCD.\nSpeaker B: It does look smart, but I think maybe for our budget, do you think that would be a bit too expensive to have the...\nSpeaker B: The LCD.\nSpeaker E: Yeah.\nSpeaker E: And the other stuff, I think.\nSpeaker B: And the speech recognition.\nSpeaker B: Because I think we're definitely going for the speech recognition.\nSpeaker A: But in our market survey, we have seen that people are willing to pay more.\nSpeaker A: But they want the quality.\nSpeaker A: They want fancy look.\nSpeaker A: They want some new design, something new.\nSpeaker A: A lot budget.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: So, even if we increase our costs a little bit within some limits, and we give something new, technological advancement, as well as new design with fancy outlook, I think we will meet the requirements and we will be able to have a good sales in the market.\nSpeaker B: I'm not sure if it's a 25-year raised...\nSpeaker B: 12 euros 50 manufacturing costs, the thing that might...\nSpeaker F: I can't see.\nSpeaker F: Although, I mean, three...\nSpeaker B: The LCD.\nSpeaker F: Sure, they have got...\nSpeaker F: I mean, they are going crazy with the LCD technology now, so that you've got your LCD TVs and everything.\nSpeaker F: So maybe the small one.\nSpeaker F: But I mean, like, I have the black and white, I guess.\nSpeaker F: It just doesn't look funky enough.\nSpeaker F: But, I mean, like, even mobile phones or whatever, they have color LCD screens.\nSpeaker F: Yeah.\nSpeaker F: I mean, I wouldn't know about the cost, so...\nSpeaker C: Uh-huh.\nSpeaker E: And, if I was not withstanding, is it too complicated?\nSpeaker E: Is it going to be too much overload?\nSpeaker B: Yeah, that's the thing, because...\nSpeaker A: It will be easy, because there will be on LCD screens, there will be different, different icons.\nSpeaker A: They can just click, click, okay, whatever they want.\nSpeaker A: Yes.\nSpeaker E: When you use the remote control, you never look at it, right?\nSpeaker E: Yeah.\nSpeaker C: You're looking at TV.\nSpeaker E: Yeah, that's true, yeah.\nSpeaker E: And it just seems kind of like a...\nSpeaker E: And one of the survey findings was that they were...\nSpeaker B: And one of the findings was that they wanted it easy to use.\nSpeaker B: So, I think, I'm not sure about the LCD.\nSpeaker B: It's a great and good idea, but for our budget and for...\nSpeaker B: The thing we're trying to go for, easy to use, it's not the thing we should go for, I think.\nSpeaker B: Child friends, I thought this was good.\nSpeaker B: As you pointed out, the bit...\nSpeaker B: It often goes missing, especially with children.\nSpeaker B: But it's a good shape, and the organic is kind of...\nSpeaker B: We could make a vegetable-y kind of round shape, I think.\nSpeaker B: So, which vegetable?\nSpeaker B: Yeah, I know, carrot.\nSpeaker F: So, since we're going for the company colors, I think your lemon wasn't that important.\nSpeaker F: Yeah.\nSpeaker B: If it doesn't work, you know.\nSpeaker B: But we don't want it to be lemon.\nSpeaker B: Yeah.\nSpeaker B: The child friend, the easy to use, it seems crazy.\nSpeaker B: She's like the different shapes of the buttons and stuff.\nSpeaker B: I think that's a good idea to take care of.\nSpeaker B: I like the colorful buttons as well.\nSpeaker B: Yeah.\nSpeaker B: And the mouse one, I thought it was a good idea because people use mouse.\nSpeaker B: My mouse is now with this scrolling thing.\nSpeaker F: I mean, we are marketing to sort of...\nSpeaker F:...35 to 35, so most people will have common content.\nSpeaker F: Yeah.\nSpeaker B: So, they'd be able to use that.\nSpeaker B: As I said, I think I'd presume it would come up on the screen.\nSpeaker B: Yeah.\nSpeaker B: So, there you go.\nSpeaker F: That means you get to bump that bit to the TV makers.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker B: So, that's the user interface.\nSpeaker B: It's low.\nNone: Okay.\nSpeaker B: Take this out now, then.\nSpeaker B: Okay.\nNone: Okay.\nSpeaker E: There are a lot of options that we're going to have to choose from among.\nSpeaker E: Yeah.\nSpeaker E: And I'll give you the technical considerations for those.\nSpeaker E: I don't even use the whiteboard just because we haven't used it.\nSpeaker D: Yeah.\nNone: I was just thinking to self-tune back.\nSpeaker E: So, who I'm going to do this is we're going to take a look at some old and old controls.\nSpeaker E: See how they work.\nSpeaker E: We used the final kind of...\nSpeaker E:...and I'm going to use the whiteboard.\nSpeaker E: The final kind of essential pieces of it.\nSpeaker E: And then we'll throw in our new innovations and keep it all within budget.\nSpeaker E: So, looking inside a very simple amount of control.\nSpeaker E: This is what they sent me.\nSpeaker E: Okay.\nSpeaker E: Here's the competition, I suppose.\nSpeaker E: You open it up.\nSpeaker E: There's a circuit board inside.\nSpeaker E: And there's a chip, a processor, the TA11835, which receives input from the buttons.\nSpeaker F: This is just a standard off the shelf kind of a chip.\nSpeaker E: Right.\nSpeaker E: It's very, very cheap.\nSpeaker E: Like this would cost nothing.\nSpeaker E: Yeah.\nSpeaker E: So, that takes the signals from the buttons and translates it into a sequence of pulses that it sends to the amplifier, which is made of some transistors and amplifiers, op amps.\nSpeaker E: And then that gets sent to the LED light, which I can kind of see is a little red light bulb at the end.\nSpeaker E: And that sends out the infrared light signal to the television.\nSpeaker E: So, this is kind of the bare essentials that we need to have in our remote control.\nSpeaker E: That's what it defines.\nSpeaker F: So, can we make them to pretty much any size we like or is there a minimum?\nSpeaker E: No, I mean, this is very old one, so now with a new technology, this is going to be a minimally small.\nSpeaker F: Almost a key thing.\nSpeaker E: Right.\nSpeaker E: So, this is what we need to have for a certain.\nSpeaker E: So, as I said, we've got the outer casing, the tray off the side.\nSpeaker E: You know, it's going to be the board we have to use this, the same setup.\nSpeaker E: And then we have the processor, we'll probably use the more advanced processor that I had, amplifier and transmitter all standard.\nSpeaker E: So, for the casing, this email I got from our manufacturing team, you know, we have a bunch of options from wood, titanium, rubber, plastic, whatnot, latex, double curve, so lots of choices, what do you think?\nSpeaker F: Well, I mean, latex has a kind of spongy feel to it, doesn't it?\nSpeaker E: It's very elastic for sure.\nSpeaker F: Yeah.\nSpeaker F: And that would also give a kind of durability.\nSpeaker F: And it's also relatively cheap to cast.\nSpeaker F: So, I'm going to base a sort of plastic, initial plastic with a latex kind of sheath.\nSpeaker E: So, I hear a plastic latex.\nSpeaker B: I like the rubber, the stress balls, I think, you know, that could be a bit of a gimmick, like it's good to hold.\nSpeaker E: So, something would give to it?\nSpeaker E: Yeah.\nSpeaker B: Okay.\nSpeaker B: And that might be quite durable and easy to chuck around.\nSpeaker B: The color is yellow, right?\nSpeaker B: Yeah.\nSpeaker B: Or at least incorporated.\nSpeaker B: Yellow incorporated, yeah.\nSpeaker F: Okay.\nSpeaker F: I mean, if we're going to, I don't know what other standard silver kind of other parts.\nSpeaker B: Yeah, the buttons, like, because there's going to be the cover, the rubber or the plastic casing, and then the buttons in probably two different colors.\nSpeaker B: Or if we're having buttons, actually.\nSpeaker E: So, yellow, and body.\nSpeaker E: And then what color for the buttons?\nSpeaker F: I quite like the multi-colored buttons.\nSpeaker F: Yeah.\nSpeaker E: So, multi-colored buttons.\nSpeaker B: You do have ones like playing, could be green or on and off as reds and stuff like that.\nSpeaker B: Yeah.\nSpeaker B: Or, yeah.\nSpeaker B: Makes it easy to use.\nSpeaker F: It makes it easier to use.\nSpeaker F: It doesn't look too childish.\nSpeaker B: Yeah.\nSpeaker B: That's true.\nSpeaker B: Because that blue one did look quite challenging.\nSpeaker F: Yeah.\nSpeaker F: Although, I mean, this also comes to shape as well.\nSpeaker F: I mean, if we are going to make it a double, I mean, double curved sounds good to me.\nSpeaker F: If we're talking about some ergonomic and easy to use, that would be comfy.\nSpeaker F: Yeah.\nSpeaker E: Okay.\nSpeaker E: So, the shape, what we want to go.\nSpeaker E: How exactly?\nSpeaker B: Like an hourglass kind of figures that were you thinking of?\nSpeaker F: Or just like it.\nSpeaker F: That'd be, that's sort of comfortable to hold, easy to hold, so you don't drop it.\nSpeaker D: What?\nSpeaker F: Yeah.\nSpeaker B: We could make novelty remote controls.\nSpeaker E: Well, yeah.\nSpeaker E: We could have a big banana shape remote control.\nSpeaker E: Because it's yellow.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And a lemon might be a little bit hard to print.\nSpeaker B: But then how would you print it?\nSpeaker B: Yeah.\nSpeaker E: How would you point it?\nSpeaker E: It doesn't matter which I need to print.\nSpeaker E: We could have little LEDs in each end.\nSpeaker E: Maybe we could have some ink.\nSpeaker F: Yeah.\nSpeaker A: Yeah.\nSpeaker A: I appreciate this idea because then this will help us in our advertisement also and we can relate it with fruits and vegetables, the people's choices that whatever data shows that.\nSpeaker A: So this.\nSpeaker B: Would you, I'm sure a rather banana idea.\nSpeaker F: So, it's fun.\nSpeaker F: Yeah.\nSpeaker F: Rubber banana.\nSpeaker F: I mean, yeah.\nSpeaker B: Okay.\nSpeaker F: That does seem again childish for me.\nSpeaker B: Yeah.\nSpeaker B: I think maybe just draw on the kind of fruit and vegetables shape.\nSpeaker B: What else did you say about fashion?\nSpeaker A: The fashion shows that fruits and vegetables.\nSpeaker A: Like people now.\nSpeaker A: And sponge.\nSpeaker A: And a file.\nSpeaker E: Yeah.\nSpeaker E: So it has a stem perhaps.\nSpeaker E: Maybe.\nSpeaker E: Yeah.\nSpeaker B: It'd be some.\nSpeaker B: Like what's that?\nSpeaker B: I don't even know the name of it.\nSpeaker B: So I'm kind of, you know, it's like, it looks like a little snowman kind of thing.\nSpeaker B: I don't know the name of that.\nSpeaker E: So it looks like this kind of.\nSpeaker B: Yeah.\nSpeaker B: That's what I was thinking.\nSpeaker B: Like a gourd.\nSpeaker B: Yeah.\nSpeaker B: Maybe that's like that.\nSpeaker B: Like a squawk.\nSpeaker B: Because that you can hold it in like the bottom bit and.\nSpeaker E: And it has a clear top and bottom.\nSpeaker E: Yeah.\nSpeaker E: So you know, transmits from the center.\nSpeaker F: Yeah.\nSpeaker F: Why the hell not?\nSpeaker F: I don't know.\nSpeaker B: What do you think?\nSpeaker F: Well, I guess it's kind of, you don't necessarily have to have it sort of clearly identified as a fruit.\nSpeaker F: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Then only we can relate it.\nSpeaker A: Yeah.\nSpeaker A: Something.\nSpeaker E: Yeah.\nSpeaker E: Exactly.\nSpeaker E: So double curved.\nSpeaker E: Single curved.\nSpeaker E: Yeah.\nSpeaker A: Or we can do something we can design two, three shapes and we can have a public survey.\nSpeaker A: Let the public choose what they want.\nSpeaker F: That's a good way.\nSpeaker E: That's a good idea.\nSpeaker E: Okay.\nSpeaker E: I guess since you're the marketing guy.\nSpeaker A: Yeah.\nSpeaker A: Sure.\nSpeaker E: Okay.\nSpeaker B: And buttons.\nSpeaker B: Did we say different shapes or buttons?\nSpeaker F: I mean, for the specific functions, you know, up and down, play stock.\nSpeaker F: They've got, I mean, they've got standard sort of intuitive things that are always used.\nSpeaker F: And.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker F: Yeah.\nSpeaker E: That's cool.\nSpeaker E: I like it.\nSpeaker B: With a scroll wheel or no?\nSpeaker B: Yeah.\nSpeaker B: What about the scroll wheel?\nSpeaker F: Speech recognition.\nSpeaker F: Speech recognition, I think.\nSpeaker F: So we need a microphone.\nSpeaker F: I can put the microphone.\nNone: Yeah.\nSpeaker E: That's the microphone.\nSpeaker F: Yeah.\nSpeaker F: Worship.\nSpeaker F: I mean, are we sure that scroll wheel does give ease of use?\nSpeaker B: Yeah, I'm not sure.\nSpeaker B: I mean, those ideas I saw were just for inspiration, I think.\nSpeaker B: I do not do this for you.\nSpeaker B: Yeah, I can.\nSpeaker B: No, I'm not sure.\nSpeaker B: Okay.\nSpeaker E: Well, we can do some user tests with scroll wheel.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: And I think if this new software for the same recognition as any key point,\nSpeaker E: keep it in the microphone. So should the microphone be just anywhere on it?\nSpeaker F: I would put it sort of substantially.\nSpeaker F: So it's.\nSpeaker F: Yeah.\nSpeaker F: So it can be sort of held and really, really going to need to hold it if it's going to be a voice recognition.\nSpeaker E: Not what we can.\nSpeaker D: Oops.\nSpeaker B: So let's not use the whiteboard anymore.\nNone: Oops.\nNone: Oops.\nNone: Easy.\nSpeaker F: Oops.\nSpeaker E: Sorry.\nSpeaker B: And so what else was that?\nSpeaker B: The, what about the glow in the dark thing, the strip around it?\nSpeaker B: We just can leave that?\nSpeaker B: I still like it.\nSpeaker B: I like it.\nSpeaker F: But that's me.\nSpeaker B: Because we've got the technological innovation with the speech recognition.\nSpeaker B: Yeah.\nSpeaker F: Some of it is just going a bit too far.\nSpeaker F: I mean, we are pushing it probably because of the fun and the shape.\nSpeaker F: Yeah.\nSpeaker F: And don't want to sort of.\nSpeaker B: Especially with yellow.\nSpeaker B: I don't know.\nSpeaker F: Because I mean, like, I mean, how good is the speech recognition thing?\nSpeaker F: Do we want to go for buttons at all?\nSpeaker F: Do we want to just have a device that maybe sits and pretends it's a fruit?\nSpeaker B: You put it in the fruit bowl.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: You know, and then you just work from it.\nSpeaker F: I mean, everybody's got a fruit bowl in front of the tail.\nSpeaker F: Mm-hmm.\nSpeaker E: It could even encourage healthier habits for television watchers.\nSpeaker E: Yeah.\nSpeaker E: Make them think of fruit.\nSpeaker E: Yeah.\nSpeaker E: Yes.\nSpeaker F: Make sure they don't eat their mouth.\nSpeaker F: I mean, yeah.\nSpeaker F: Yeah.\nSpeaker F: Do we need to?\nSpeaker F: I mean, I think of a fruit that could sit.\nSpeaker F: So then the pen lay on its own, like an apple.\nSpeaker F: It's just apples.\nSpeaker F: So yellow apples.\nSpeaker F: I quite like the shape.\nSpeaker B: I quite like the design of that.\nSpeaker B: Because that could sit on its own.\nSpeaker B: Yeah.\nSpeaker B: It's quite, got quite a sturdy base.\nSpeaker B: Groovy.\nSpeaker B: And as we say, we don't want to be too ridiculous with the fruit things.\nSpeaker B: Yeah.\nSpeaker E: Yeah.\nSpeaker E: About the speech thing, it doesn't have to be handheld.\nSpeaker E: Of course, it can set a distance.\nSpeaker F: Yeah.\nSpeaker F: So, I mean, like, you could actually...\nSpeaker F: Yeah.\nSpeaker A: We can do one thing.\nSpeaker A: We can just have a remote control and casing some different different shapes, different fruit shapes in such a way that any casing could be fit into this mobile general piece.\nSpeaker A: So whatever people want, like, if somebody wanted it in banana shape, we will put that casing onto that mobile phone.\nSpeaker A: Absolutely.\nSpeaker A: Okay.\nSpeaker A: It will look...\nSpeaker A: And yeah.\nSpeaker A: Yeah.\nSpeaker A: He said about disposable things.\nSpeaker A: Sorry?\nSpeaker B: Yeah.\nSpeaker A: Like, if this is a mobile phone, we will design casing in such a way like half of...\nSpeaker A: We need not to have a full cover.\nSpeaker A: Yeah.\nSpeaker A: We will just have a half of cover, okay?\nSpeaker A: Like, if somebody wants it in banana shape, we will fit banana shape casing onto that.\nSpeaker A: So it will give a banana shape look.\nSpeaker A: If somebody wanted it in apple shape, we will design that.\nSpeaker A: We will put apple shape casing on that.\nSpeaker A: It will give apple shape look.\nSpeaker A: So in that way, you can have any...\nSpeaker A: Means whatever you want.\nSpeaker A: Okay.\nSpeaker B: Without...\nSpeaker B: Yeah.\nSpeaker B: We need the buttons in the same place.\nSpeaker A: Yeah.\nSpeaker A: Buttons will be on the upper side.\nSpeaker A: Yeah.\nSpeaker A: Buttons will be on the upper side.\nSpeaker A: Yeah.\nSpeaker A: Buttons will be on the upper side.\nSpeaker A: Lower side, we will just put the casing.\nSpeaker A: So half of that will be looked...\nSpeaker A: Oh, half afraid.\nSpeaker A: Yeah.\nSpeaker A: Not the upper side.\nSpeaker A: So from lower, you can...\nSpeaker A: Means while you are holding from this side.\nSpeaker A: You can have banana look or apple look, whatever.\nSpeaker A: So in that way, we need not to have different different shape, mobiles, everything.\nSpeaker A: We will just design cravings through shape.\nSpeaker A: Okay.\nSpeaker F: I think if you are going to have a fascia, then you want to have it so that it does go for the buttons.\nSpeaker F: Because if you think about it, if they are wanting it because they want to look at it, if they are using it, and what they want to look at is facing away from them, it doesn't really...\nSpeaker F: You know, because that would be in the palm of their hand.\nSpeaker F: Yeah.\nSpeaker F: Unless you have sort of...\nSpeaker F: You have the buttons option on one side.\nSpeaker F: You get the fascia on the other side with the microphone so that you can place it face down.\nSpeaker F: And you have got the fascia.\nSpeaker F: And you can just talk at them.\nNone: Mm-hmm.\nSpeaker F: That'd be...\nSpeaker F: Okay.\nSpeaker F: So...\nSpeaker F: So you have narrowed it down to half a dozen options?\nSpeaker F: Yeah.\nSpeaker E: So, I guess we decided on the material, right?\nSpeaker E: So that's spongy, latex, rubber, everything.\nSpeaker E: Yeah.\nSpeaker E: And the color as we got down.\nSpeaker E: Yeah.\nSpeaker E: Your shape, maybe we'll just make it kind of mix and match type of shape or...\nSpeaker B: Well, because...\nSpeaker B: Well, I'm not sure if we should go so far on the whole fruit thing because I think we should maybe just take the inspiration from the fruit.\nSpeaker B: And because...\nSpeaker B: Okay.\nSpeaker B: We stick with what we've got later.\nSpeaker B: Yeah, I think we're trying to get to 25-35 year olds want it quite trendy as well.\nSpeaker B: They said they want it something that looks fancy.\nSpeaker B: And I think maybe fruit could be a bit of too much of a gimmick but something ergonomically shaped and organic, like good to hold based on fruits and natural things like that.\nSpeaker B: Okay.\nSpeaker B: Because already we're going a bit gaudy with the yellow, you know?\nSpeaker B: Mm-hmm.\nSpeaker B: I mean, we could make it a nice pale yellow.\nSpeaker F: Well, it's kind of got to be our company, so...\nSpeaker C: Okay.\nSpeaker F: So, again, I mean, like we could have, I mean, we could quite easily have the main body be a difference.\nSpeaker F: Yeah.\nSpeaker B: Maybe we could have that pale yellow and then an outside bit bright yellow with, you said, the logo.\nSpeaker B: The slogan.\nSpeaker F: The slogan.\nSpeaker F: Yeah.\nSpeaker F: Because...\nSpeaker F: I mean, even if... I mean, not necessarily that the whole body has to be of the company color.\nSpeaker F: So, you know, blue and yellow tend to go through well together.\nSpeaker F: So you have main body blue with the yellow, local and slogan running up one side of it.\nNone: Yeah.\nSpeaker C: Uh-huh.\nSpeaker F: Sure.\nSpeaker E: Okay.\nSpeaker E: As for the energy source, you know, almost every commercial is just batteries, but we don't have to be limited by that.\nSpeaker E: We can use a hand dynamo.\nSpeaker F: Uh-huh.\nSpeaker F: I think it's basically the more you move it, it's got everything inside that just kind of powers it.\nSpeaker E: So those watches that you see.\nSpeaker D: Yeah.\nSpeaker E: Oh, it's a dynamo?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Like with those watches that you...\nSpeaker B: Yeah.\nSpeaker B: So it's fine too.\nSpeaker E: It's not working.\nSpeaker E: I just keep holding the reaction anyways.\nSpeaker E: Yeah.\nSpeaker A: But, but do you think that it will be a good idea to use dynamo?\nSpeaker A: Uh-huh.\nSpeaker A: These type of cells because then people have to like if this hell is out.\nSpeaker A: That's a leap of navigation too.\nSpeaker A: Yeah, too.\nSpeaker F: Especially if they want to use it specifically as voice activated.\nSpeaker A: Yeah.\nSpeaker B: Because most of the people...\nSpeaker B: And they have to pick it up and then activate it and then yeah.\nSpeaker F: Okay.\nSpeaker F: That's true.\nSpeaker E: Right.\nSpeaker E: Um, while we have an option, there's solar power.\nSpeaker A: Uh, solar power will also not be a good idea because then they have to keep their mobiles outside.\nSpeaker A: And the day's when there is no solar sunlight.\nSpeaker A: I'm with Raj on that.\nSpeaker A: I think.\nSpeaker A: Yeah.\nSpeaker A: I've got north facing houses.\nSpeaker A: Yeah.\nSpeaker A: I think we should rechargeable.\nSpeaker A: Uh-huh.\nSpeaker A: Battery will be a good idea.\nSpeaker A: They can...\nSpeaker F: I mean, how about that idea that I thought, um, just on the basis of like reading them with batteries and that kind of bother.\nSpeaker B: And we're very environmentally friendly.\nSpeaker F: Yeah.\nSpeaker F: And we have a rechargeable stand so that not only it almost stands but, um, for using it as recharging it but also for using it as sound recognition.\nSpeaker B: Like a hand, like one of those portable phones.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Exactly.\nSpeaker E: So, uh, a rechargeable battery.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Um, the user face the buttons.\nSpeaker E: I guess we talked about this very much.\nSpeaker F: On print puts, sorry.\nSpeaker F: Nevermind.\nSpeaker E: Yeah.\nSpeaker E: The electronics, um, basically the more features we add.\nSpeaker E: Um, so the more features we add the answer chips we need to buy and put in, which adds to the cost.\nSpeaker E: Um, but, uh, I think we can keep it all in the budget.\nSpeaker E: So, yeah.\nSpeaker E: So the speech thing you said are, are technology research and development department came up with some breakthrough.\nSpeaker F: So, and if we're just having buttons on the speech then we're getting our cheapest option of.\nSpeaker F: Chiping.\nSpeaker E: Right.\nSpeaker F: That's good.\nSpeaker F: Okay.\nSpeaker F: We're kind of, um, I've just deleted that whole thing.\nSpeaker F: Um, we're kind of running at time.\nSpeaker F: So if you could.\nNone: Um, Was that you?\nSpeaker F: That was your bit covered up.\nSpeaker F: Yeah, that was just accidentally deleted what I was supposed to see next.\nSpeaker E: Um, yeah.\nSpeaker B: So, yeah, I'm sure I think.\nSpeaker B: Yeah, mine seems to have turned off.\nSpeaker A: I just touched the pad.\nSpeaker F: It does.\nSpeaker F: That's the bad.\nSpeaker B: No.\nSpeaker F: No.\nSpeaker F: It's actually shut down.\nSpeaker F: It's on, but there's no.\nSpeaker E: Okay.\nSpeaker F: Um, now what we have our next meeting in half an hour.\nSpeaker F: What I would like you guys to do is work on giving me a model in clay.\nSpeaker E: Oh, I guess they do it.\nSpeaker E: Yeah.\nSpeaker E: Cool.\nSpeaker F: Yeah.\nSpeaker F: So, um, you know, I mean, luckily we chose a nice simple shape.\nSpeaker F: Um, and further instructions will be sent by your personal coaches.\nSpeaker E: Okay.\nSpeaker E: That's great.\nSpeaker E: Save everything to the shared documents.\nSpeaker F: Oh, yeah, I hope I can recover this because of that.\nSpeaker F: We deleted it.\nSpeaker F: Which doesn't really help me much.\nSpeaker B: I think I saved mine already.\nNone: Yeah.\nSpeaker F: Can you say if I send that last one again, please, Raj?\nSpeaker F: I still can't find it on there.\nSpeaker A: Uh, it was under a different name.\nSpeaker A: I will show you.\nSpeaker A: In shared documents.\nSpeaker F: Okay.\nSpeaker A: Uh, but the components.\nSpeaker A: Oh, you didn't get that.\nSpeaker A: No.\nSpeaker A: Okay.\nSpeaker A: I will put it in shared documents.\nSpeaker F: Um, yeah.\nSpeaker F: Project project project.\nSpeaker A: Project documents.\nSpeaker A: Sorry.\nSpeaker A: I put it in the shared documents.\nSpeaker A: All right.\nSpeaker F: That's actually that it goes there automatically.\nSpeaker F: If you put it in project documents, project documents is on the, um, desktop.\nSpeaker A: Right.\nSpeaker A: But I can't open that because it asks for some username or password.\nSpeaker A: Okay.\nSpeaker E: I'll show you.\nSpeaker F: All right.\nSpeaker F: I think, um, hold on.\nSpeaker F: Yeah.\nSpeaker F: I don't know if you, it just can't just came up on my, um, on my agenda.\nSpeaker F: Um, presumably this clay somewhere.\nSpeaker F: Um,\nSpeaker E: Yeah. That's good.\nSpeaker F: Whoops.\nSpeaker F: Light, light, please light.\nSpeaker F: Right.\nSpeaker F: There you go.\nSpeaker F: Yeah.\nSpeaker F: Thank you.\nSpeaker F: Yeah.\nSpeaker F: Quite.\nSpeaker F: We're using this sort of basic chipset.\nSpeaker F: So.\nSpeaker F: Sorry.\nSpeaker A: Yeah.\nSpeaker A: You have to keep your pen separate.\nSpeaker A: I used your pen.\nSpeaker F: Oh, sorry.\nSpeaker F: Okay.\nSpeaker F: Still didn't manage to get that.\nSpeaker F: And although that's so we have the chargeable.\nSpeaker F: And.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker D: Okay.\nSpeaker F: See.\nSpeaker C: Cool.\nSpeaker C: Fine.\nSpeaker F: I knew that actually the sort of engineered it so we ended up making a difficult shape.\nSpeaker F: It's for cruelty.\nSpeaker F: Yeah.\nSpeaker B: Starth rate.\nSpeaker F: I mean, like literally make it sort of buttons and everything.\nSpeaker B: No, we can do buttons.\nSpeaker A: Um, we are going to discuss something.\nSpeaker F: Uh, no, I think that's a discussion over and less anybody's got questions.\nSpeaker F: Confusions because I'm confused.\nSpeaker E: Mm hmm.\nSpeaker E: Let's get.\nSpeaker F: Um, we'll probably get a question in a minute.\nSpeaker F: It's, uh.\nSpeaker A: Sorry.\nSpeaker A: Thank you.\nSpeaker A: Yeah.\nSpeaker F: We're going to a warning finish meeting.\nSpeaker F: I wrote it up.\nSpeaker F: I'm going to be going to a document.\nSpeaker F: My project.\nNone: There we go.\n"}, {"meeting_id": "IS1008d", "summary": "Project Manager initiated the final meeting on the topic of detailed design. Industrial Designer and User Interface each gave a presentation on the prototype and discussed the evaluation criteria, especially that of microphone and buttons. Next, the group did calculations on the budget as well as the profit goal before Marketing showed the challenges of this project in terms of budget control and reaching the profit target. The group then discussed using the internet for promotion and ordering and drop-shipping the product to the customer's residence. Finally, the subject moved to product evaluation by putting the prototype on the market and analyzing the feedback. The group discussed whether they should spare time for the redesign process before releasing the product to the market. Then Industrial Designer put up a strategy to do effective user research by choosing sports seasons.", "dialogue": "Speaker A: Welcome back everybody.\nSpeaker A: So this meeting agenda will be the detailed design meeting and opening and PMs are secretary of the meet minutes prototype presentation from Christie and Agni.\nSpeaker C: Agni, yes.\nSpeaker A: And evolution criteria, the finance it's from my side from the management and production evolution then closing.\nSpeaker A: So we have 40 minutes to discuss and finalize and close the product and project and to move further.\nSpeaker A: So let's talk about maybe first prototype.\nSpeaker D: I've done a presentation but it pretty much covers work that we've both done.\nSpeaker D: If I'm missing anything, Christie.\nSpeaker D: So shall I go to correct me?\nSpeaker D: So you did a PowerPoint presentation.\nSpeaker C: Let's go to MI.\nSpeaker D: It's not the biggest PowerPoint presentation in the world.\nSpeaker A: In 2003 or?\nSpeaker D: Free.\nSpeaker D: I think it's the last one.\nSpeaker B: Technical.\nSpeaker B: I would say.\nSpeaker D: No, then that one.\nSpeaker D: Technical design.\nSpeaker D: It is named appropriately.\nSpeaker D: You just couldn't see the name.\nSpeaker D: Okay.\nSpeaker D: Can I have a minute?\nSpeaker D: All right.\nSpeaker D: So from when we were discussing specifying the case in the last meeting, we decided that we wanted an ergonomic shape.\nSpeaker D: The material that we chose was wood and that the color would be customizable because you can stain the wood, whatever color.\nSpeaker D: So in terms of function, you have to be able to turn the TV on and off, volume and channel control, menu control, voice recognition control.\nSpeaker D: And we've incorporated the LCD screen on the panel as part of the design.\nSpeaker D: If we figure out it's too expensive, well, then you just take it off.\nSpeaker D: So to unveil our lovely product.\nSpeaker D: This is our remote control with the flip panel, as you can see.\nSpeaker D: So if you lift up the panel, you can see the lovely yellow LCD display.\nSpeaker D: This is actually hard to do.\nSpeaker D: The yellow button you have is the on off button, so it's really big, hard to miss.\nSpeaker D: You have the red triangles or the toggles for changing the volume.\nSpeaker D: So up, volume up, down, volume down.\nSpeaker D: The green are the channel changing.\nSpeaker D: And it's one of those very light, very touchable displays.\nSpeaker D: And then you have the numeric pad in the dark blue at the bottom.\nSpeaker D: And on the right hand side, you have the access to the menu on the TV.\nSpeaker D: And on the left hand side, you have the ability to turn off the voice recognition.\nSpeaker D: So this is pretty much what we had on the whiteboard the last time.\nSpeaker D: And additional feature on the back is you can have your own customized backing.\nSpeaker D: And I suppose you could do the same thing on the flip case on the front, so that you can really make this a highly, highly customizable remote control.\nSpeaker C: We haven't specified where the speaker or the microphone will be placed.\nSpeaker C: That depends on the design of the circuit board inside and what room is left.\nSpeaker A: I think the microphone is on the top.\nSpeaker A: Yes, okay.\nSpeaker A: Under the flip.\nSpeaker A: So that will be the safe.\nSpeaker A: So it's not on the chip because you need to have microphone out.\nSpeaker C: No, it depends on the design of the circuit board.\nSpeaker C: Yes.\nSpeaker D: But it shouldn't be under the flip either because you can have the remote control closed, but you still might want to activate it by voice.\nSpeaker A: Yeah, but my opinion, I think it's about under the flip because whenever you want to the talk, okay, so then you can speak, then you can close it.\nSpeaker A: Because if you put it on the flip, okay, then technical, I don't think it's feasible because most of the time you speak, then it will be recognized.\nSpeaker D: If you've already got the remote control in your hand, I need to open the flip to use the voice.\nSpeaker D: Why use the voice?\nSpeaker D: Why not just use your hand?\nSpeaker D: I mean, the whole point of the voice is that if the remote control is sitting there and I'm too lazy to reach over and pick it up, I can just use my voice.\nSpeaker C: Maybe I've got my hand in the popcorn bowl and I'm holding my cup of Coca-Cola in the other hand.\nSpeaker C: Yeah.\nSpeaker C: You don't want to let go of either one.\nSpeaker D: No, it doesn't have to be on the flip.\nSpeaker D: It can be on the sides somewhere.\nSpeaker A: The sides may be good.\nSpeaker D: That's good, I mean, I can pass this around if anyone wants to.\nSpeaker C: Yes, the sides may be a good idea.\nSpeaker C: Very fast around with the napkin.\nSpeaker B: You can easily put a microphone on the side that would have no problem, would have not been damaged or anything and it'd be accessible all the time to a voice.\nSpeaker C: Yeah, exactly.\nSpeaker C: The sides may be a good idea.\nSpeaker B: It's a compliment to the artist.\nSpeaker C: It's, you need to work on the weight a little bit.\nSpeaker C: Yes.\nSpeaker C: And maybe the shape of the buttons.\nSpeaker C: I'm satisfied.\nSpeaker D: The little egg shapes aren't the most eaten.\nSpeaker D: We're glad your satisfied.\nSpeaker A: It looks more heavy, but I think when it comes to the reality, maybe it's the less weight.\nSpeaker D: Yeah.\nSpeaker D: I mean, this is plasticine.\nSpeaker D: There's only so much you can do.\nSpeaker D: We could have possibly made it a lot thinner as well.\nSpeaker D: And part of the thing is a lot of people say that they don't like something that's too light because they don't feel like they have enough control over it.\nSpeaker D: I mean, maybe this is excessively heavy, but I think it needs to have some weight.\nSpeaker D: It needs to feel like you're still holding something.\nSpeaker D: So that's pretty much it for our presentation, actually.\nSpeaker A: That's your prototype, Maudet.\nSpeaker A: Yes.\nSpeaker A: Okay, that's good.\nSpeaker A: Thank you very much.\nSpeaker A: So any comments or?\nSpeaker B: Well, the prototype is very well within the design and idea that we've talked about on the previous meetings.\nSpeaker B: Now it goes into this next phase as a financial marketing.\nSpeaker A: Yes, that's right.\nSpeaker A: I'll come back to the...\nSpeaker A: So, evolution criteria, I think that will be good.\nSpeaker A: So when it's come to the finance, I have some calculations which I made as for the budget.\nSpeaker A: So here you can look like the energy source and handy dynamo and kinetic and solar cells.\nSpeaker A: It's optional, some are optional and advanced chip on print.\nSpeaker A: That's what we were talking about that.\nSpeaker A: So then we have sample sensor and sample speaker.\nSpeaker A: Then we have the wood material, then special color and push button.\nSpeaker A: So it's... actually our budget was 12.5 euro, but it's coming to 9.95 euro.\nSpeaker A: So we are under below the budget.\nSpeaker A: Okay, so still we are saving some money.\nSpeaker A: I think it's a good figure.\nSpeaker C: Yes, great.\nSpeaker C: I'm surprised.\nSpeaker C: Congratulations.\nSpeaker C: Thank you.\nSpeaker B: Oh, we haven't covered the money yet, so we're going to go with the rest of the print.\nSpeaker C: You've got a lot of money to market it.\nSpeaker A: So maybe for some money we can utilize for our marketing for the sales.\nNone: Okay.\nSpeaker B: Well, it just depends.\nSpeaker B: If we're going to add on this provisionary cost analysis, we do not have an LCD display.\nSpeaker B: LCD display is going to be very expensive.\nSpeaker D: No, we do, but it's not filled in.\nSpeaker D: It's not necessarily.\nSpeaker B: We don't have the price up there.\nSpeaker B: Oh, yeah.\nSpeaker B: All right.\nSpeaker B: Yes.\nSpeaker B: So if we add approximately 2 to 3 euro per remote, now we're up around about 12, 12 and a half as to what the company had initially requested.\nSpeaker C: That means we can put the...\nSpeaker B: Display in.\nSpeaker B: But as far as production, I'm putting up a question because we're talking about profit also.\nSpeaker B: And in mine, you'll see the problem with our survey, the possibility of how many units can be sold or percentages of the market, et cetera, et cetera, because that has to be taken into consideration.\nSpeaker B: This is just production cost.\nSpeaker B: It is not advertising cost.\nSpeaker B: It's not transportation cost.\nSpeaker B: Yes.\nSpeaker A: So still we have 2.5 minutes.\nSpeaker B: And that will inflate quite a bit on the cost of the unit for the company.\nSpeaker B: So to come up with what the company wants is a 50 million pound profit, we're going to have to go a long way.\nSpeaker A: Yes.\nSpeaker A: This we are talking about one unit.\nSpeaker A: Yes.\nSpeaker A: So then it's going to the quantity and the cost will come down slightly.\nSpeaker C: Although customization, because this is being done on an order basis, might be the quantity you want, the circuit board, you're right, would be in producing quantity, but the cost of the case would be fixed.\nSpeaker C: You got some pretty cheap labor that can do this case for one euro.\nSpeaker C: That's the cost of material and it's really outstanding.\nSpeaker A: But anyhow, still we are under control.\nSpeaker A: So what I will do is I will try to negotiate with the vendors to get the production cost less so that we can save some money to put it into our marketing or the promotions, whatever.\nSpeaker A: So that I will look after.\nSpeaker A: I will speak to the management and how to get some more cost down.\nSpeaker B: If we can go to my display and we'll come back to yours just to give everybody an idea of the market.\nSpeaker B: So now I'm going to scare everybody out of this project.\nSpeaker B: If I'm still here, press the button.\nSpeaker A: You're in full?\nSpeaker B: Yep.\nSpeaker B: The forward gives me a try and watch.\nSpeaker C: It's the same one you did before.\nSpeaker B: No.\nSpeaker C: Okay.\nSpeaker B: It shouldn't be if it's not the right one.\nSpeaker B: No.\nSpeaker B: That's the same one.\nSpeaker B: You have to go back and find another one.\nSpeaker B: Whatever name it popped up under.\nSpeaker B: So functional.\nSpeaker B: Try function on might not be it either, but we'll see.\nSpeaker D: It looks like it.\nSpeaker B: Yep.\nSpeaker B: That's it.\nSpeaker B: So we'll go screen by screen.\nSpeaker B: Since we need to have some type of idea on the market, we had an independent study that says that this market has an availability to absorb 8 million units per year.\nSpeaker B: Our internal company evaluation puts it between 8 to 9 million, which is approximately the same as the independent study.\nSpeaker B: So if we continue, we look at the findings.\nSpeaker B: Next screen.\nSpeaker B: Which means that if we have a target of 2 million, the company has to take 25% of the market in the first year, which is actually a tremendous amount.\nSpeaker B: No kidding.\nSpeaker C: I admit they already expected some.\nSpeaker B: So if we put an inflated price of 50 year old at a production cost, they cannot exceed 25 year old.\nSpeaker B: Okay.\nSpeaker B: We're already in that price.\nSpeaker B: Yes.\nSpeaker B: Okay.\nSpeaker B: With transport promotion labor, because we haven't included the promotion in the cost, transport for the material to the stores or whatever, however, we're going to break this down between our retailers.\nSpeaker B: 25% of the market to get to 2 million units, we have to have a profit of 25 year old per unit to get to the 50 million year old profit.\nSpeaker B: Yep.\nSpeaker B: Okay.\nSpeaker B: So obviously, I just did a rundown of the evaluation of the form, the fancy stylishness of the unit, the ease of use, speech recognition, cost, we've gone through these.\nSpeaker B: Now the company must evaluate the feasibility of being able to take enough of the market to justify a production.\nSpeaker B: Or we project this over two years by being that the market changes very, very quickly.\nSpeaker B: Maybe there's no more interest in buying this thing in 18 months from now.\nSpeaker B: So now we have to come up with a decision.\nSpeaker B: Can the company sell 2 million units?\nSpeaker B: Can it sell it for 50 year olds?\nSpeaker C: Could I go to findings?\nSpeaker C: I would like to explore the possibility of using alternative delivery in sales channel, which would be to use the internet for promotion and ordering and then to drop ship the product to the customers, residents.\nSpeaker C: That way you have no storage, you do have transportation, you still have the labor cost, but you don't have the transport to the point of sale.\nSpeaker C: The point of sale is online.\nSpeaker D: You can do a shipping center somewhere or strategically place shipping centers.\nSpeaker D: Right.\nSpeaker C: To minimize distance costs.\nSpeaker C: You should sell through Amazon, don't you?\nSpeaker C: Or eBay, yeah.\nSpeaker D: Yeah.\nNone: Yeah.\nSpeaker C: That's an idea.\nSpeaker C: That's right.\nSpeaker A: Going with the un...\nSpeaker A: To improve more profit and...\nSpeaker B: Upscale technology.\nSpeaker C: Ah, we're selling a unique product.\nSpeaker D: It actually makes more sense if we're going to make it so highly customizable because on the web people can look at the different options they have, see maybe what other people have done, what the range of possibilities are.\nSpeaker A: There's some companies that come that way.\nSpeaker D: Where you can't...\nSpeaker D: Unless you're a highly imaginative person, you may not really know what it is you want.\nSpeaker D: Whereas on the web, if you have a bunch of pictures, it can trigger ideas and you can\nSpeaker C: even have a movie that you can rotate the object and look at the... The only thing that you're missing really is the weight.\nSpeaker B: The weight, the feel of the product but...\nSpeaker C: We're getting used to that.\nSpeaker C: It's not quite like trying on a shoe but people are getting used to buying things online that they can't touch before buying.\nSpeaker B: Several that have gone through with the watches too.\nSpeaker B: You can customize the watch.\nSpeaker B: You can see how it is at the end of the production.\nSpeaker B: You can change it.\nSpeaker B: There's a lot of online that is doing this now.\nSpeaker B: You're rotating and you look behind and look this way.\nSpeaker B: It's possible to do this.\nSpeaker B: Maybe there's a possibility of selling more than 2 million units in one year which could feasibility lower the price of the unit.\nSpeaker A: You can.\nSpeaker A: Great.\nSpeaker A: I don't think that's not possible.\nSpeaker A: It's...\nSpeaker A: Okay, then let's wait for the production.\nSpeaker A: Then you can evaluate the product.\nSpeaker A: So how it looks like technically and how it looks like the real.\nSpeaker D: What turnaround time do we have?\nSpeaker D: Because I mean production evaluation can be very, very quick or very, very long.\nSpeaker A: It's very quick.\nSpeaker A: Of course.\nSpeaker A: It will come back in two weeks.\nSpeaker A: Okay.\nSpeaker A: It will work in two weeks.\nSpeaker A: Works for me.\nSpeaker A: For evolution.\nNone: Okay.\nSpeaker C: Prototypes you mean?\nSpeaker C: Yes.\nSpeaker A: The prototype.\nSpeaker C: Prototype product evolution.\nSpeaker C: That should do some market tests once we have prototypes.\nSpeaker B: Well obviously.\nSpeaker C: And just some orders and things like that.\nSpeaker C: And test market it.\nSpeaker B: You might have to be throwing out on the market for people to get an idea to see.\nSpeaker A: So you can take the minimum two weeks.\nSpeaker A: Get there.\nSpeaker D: Get the maximum four weeks.\nSpeaker D: Minimum two weeks if we're going to develop prototypes and then try to take them to different places and see how people use.\nSpeaker D: It's not trivial tests.\nSpeaker A: Yeah, because we are not going to do it in all five three.\nSpeaker A: Okay, so we can get the product chosen.\nSpeaker A: No, no.\nSpeaker A: We definitely shouldn't do it in our factory.\nSpeaker A: So we will do it in the other place.\nSpeaker A: And I don't think it's take more than four weeks time.\nSpeaker A: Okay.\nSpeaker A: So then the real production we will start once we product evolution.\nSpeaker A: Okay.\nSpeaker A: Then it's approved from the technical team and your team.\nSpeaker A: Okay.\nSpeaker A: From the management.\nSpeaker A: Then we can launch in the market.\nSpeaker C: Any outstanding.\nSpeaker A: Any other questions or comments to be discussed?\nSpeaker B: No, I think we've pretty much covered everything.\nSpeaker A: Okay.\nSpeaker D: So then.\nSpeaker D: Did you have something?\nSpeaker D: I was just wondering about if we're going to do product evaluation, then what about time for redesign?\nSpeaker D: If the users come back and tell us, no, this is bad, this is bad, we want this done differently.\nNone: Okay.\nSpeaker A: Let's take like this, let's proceed with this model for the market introduction.\nSpeaker A: Okay.\nSpeaker A: So no more changes will be made in this, the basic design.\nSpeaker A: Okay.\nSpeaker A: So we will introduce this model and let's introduce in the market and let's take the feedback from the customers.\nSpeaker A: Then we can go for the.\nSpeaker A: Second generation.\nSpeaker A: Second generation.\nSpeaker A: Okay.\nSpeaker A: There is no end, there is no limit.\nSpeaker D: The problem is it may not be a second generation of the first generation flops for some silly reason.\nSpeaker A: Every customer, okay, they have the right idea.\nSpeaker A: They have the wrong taste.\nSpeaker A: Okay.\nSpeaker A: So there is no end, there's no limit.\nSpeaker D: No, but there's a difference between releasing a product that has been minimally tested and fine tuned to suit a general range of requirements versus releasing a product that we think will work, but we don't really have anything to back it up.\nSpeaker A: It was very specific.\nSpeaker A: Yeah, so that's the reason you're here for the design.\nSpeaker A: Okay.\nSpeaker D: I hope you made a good design.\nSpeaker D: Yes, but I'm not everybody.\nSpeaker D: I mean, the whole point of user evaluation is to see what real people need.\nSpeaker D: We have our own motivations in mind.\nSpeaker D: We have our own ideas in mind, but that doesn't mean that that's what's going to sell.\nSpeaker A: Yeah, but see, we have to take a few considerations.\nSpeaker A: Okay.\nSpeaker A: One is the price consideration, one is the future consideration.\nSpeaker A: Okay.\nSpeaker A: Like you can eat, you can order it more chill.\nSpeaker A: I can eat more chilli.\nSpeaker A: Okay.\nSpeaker A: So it depends on individual taste.\nSpeaker A: You know, so we have to balance somewhere.\nSpeaker D: Yeah, of course.\nSpeaker D: I'm just trying to point out that I think that your evaluation and redesign turnaround time is too short.\nSpeaker D: Well, you have no redesign, not you personally, but in the project, we have no redesign time.\nSpeaker C: Ed, do you know what season of the year or time of the year is the most important for TV remote control sales?\nSpeaker C: Would it be the Christmas season by any chance?\nSpeaker A: The sports time.\nSpeaker C: Sports season.\nSpeaker C: Right before the year.\nSpeaker B: The world cup.\nSpeaker B: The world cup soccer.\nSpeaker A: Yeah.\nSpeaker A: The world cup soccer.\nSpeaker B: They need those things.\nSpeaker B: They have a hand occupied and they need to be able to talk to the remote control.\nSpeaker C: So I think what we need to do is perhaps to synchronize the final, the launch of a user tested device with some special event.\nSpeaker C: Yeah, that's a good idea.\nSpeaker C: And then, so that gives us a little more time perhaps than we anticipated because I don't know when the world cup is, but I'm sure there's going to be one.\nSpeaker C: Or any major sports.\nSpeaker C: Or another major sports event.\nSpeaker C: Probably not the football game is coming up the end of January.\nSpeaker C: I think that might be a little too aggressive.\nSpeaker C: But so I'm just pointing out a strategy to do some additional user testing and then to launch at a major sports event or perhaps to also.\nSpeaker D: That's a good place to advertise it too.\nSpeaker C: And to work with motion pictures, there might be some motion pictures that are coming out that are coming out on DVD that they need to have a special remote control to work with it so we could maybe work out a campaign with Sony pictures for example.\nSpeaker C: Maybe some management has got relationships there.\nSpeaker C: We can leverage.\nSpeaker A: Yes, that of course I will convince the management to do that.\nSpeaker D: That's great.\nSpeaker D: It's just something to keep in mind because it's really, really important.\nSpeaker D: A lot of products have gone out there without being properly user tested and completely flopped when in fact it gets re-released a few years down the line with proper testing and it takes off like crazy.\nSpeaker C: Disposable diapers is an example of that in fact.\nSpeaker C: Really?\nSpeaker C: Yes.\nSpeaker C: One of the first consumer products that was launched about 30 years that was a disposable consumer product and people at the market hadn't really gotten on to the concept that you could use something and then throw it away because it wasn't.\nSpeaker C: But then when they relaunched them 30 years later, they were virtually the same design but people had gotten to throw away paper cups and napkin, all kinds of things that they hadn't.\nSpeaker C: So you're right, timing is very important but I think we've got a good product.\nSpeaker A: That's the reason Ed is here.\nSpeaker A: I think you can promote the brand value and the product value.\nSpeaker A: It's going to be very important to the company.\nSpeaker A: We are behind the scene and he's a friend screen so.\nSpeaker A: Yep.\nSpeaker A: I have no idea what he's doing.\nSpeaker B: Exactly.\nSpeaker B: Good luck Ed.\nSpeaker B: It's a flop, it's the marketer.\nSpeaker C: You look very relaxed.\nSpeaker C: Yes.\nSpeaker C: Considering the place on your shoulders.\nSpeaker A: Yeah.\nSpeaker A: Okay, so then let's come to the closing.\nSpeaker A: Celebration.\nSpeaker A: The cost within the budget and is the project evaluated?\nSpeaker A: Okay, so that will come soon.\nSpeaker A: Okay, but time being, thanks for all your efforts and great work and great design and let's leave it to the ad later for once production is over.\nSpeaker A: And I mean that I'm let's celebrate.\nSpeaker A: So let's meet up this evening to hang up for some party.\nSpeaker C: Sounds good.\nSpeaker B: It's good.\nSpeaker B: Thank you.\nSpeaker B: Nice working with you.\nSpeaker A: Thank you very much.\nSpeaker A: Thank you.\nSpeaker A: Thank you again for all.\nSpeaker A: See you in the evening.\nSpeaker B: Good to see you.\nSpeaker B: Okay, see you later on.\nSpeaker B: Thank you.\nNone: Indeed, goodround.\nNone: Oh!\n"}, {"meeting_id": "Bmr019", "summary": "The group discussed efforts to train and test the Aurora group's HTK-based recognition system on ICSI's digits corpus. Members also discussed efforts to produce forced alignments from a selection of Meeting Recorder data. Performance in both tasks was adversely affected by the manner of recording conditions implemented and difficulties attributing utterances to the appropriate speakers. While debugging efforts resulted in improved forced alignments, dealing with mixed channel speech and speaker overlap remains a key objective for future work. The group is additionally focused on a continued ability to feed different features into the recognizer and then train the system accordingly. ", "dialogue": "Speaker G: Okay, we're on. Okay. So I mean everyone who's on the wireless check that they're on.\nSpeaker G: Okay, our agenda was quite short. Oh, could you please add on two items which was digits and Possibly stuff on on forced alignment, which Jane said that was an and Ray's headed information on but they didn't. I guess the only other thing\nSpeaker H: Okay Okay, so there's digits alignments and I guess the other thing Which I came up prepared for Is to just see if there's anything anybody wants to discuss about the Saturday meeting right?\nSpeaker H: so I mean digits and alignments\nSpeaker D: but Aligning people schedules. Yeah\nSpeaker G: Yeah, I mean right first alignment the people schedule\nSpeaker H: With whatever was a month and a half or something ahead of time the only time we could find in common roughly in common was on a Saturday\nSpeaker D: Yeah\nSpeaker C: I Thought about having a conference call to include him in more in more of the meeting I mean, I don't know if we had if he had a telephone on the table. No, but he probably has to go do something right?\nSpeaker D: Right so we have we make for interesting noise background noise. Yeah\nSpeaker H: So we have to equip him with And we'd have to put lots and lots of digits\nSpeaker A: Real car noise\nSpeaker G: So anyway, I can talk about digits. Did everyone get the results or shall I go over them again? I mean that it was basically the only thing that was even slightly surprising was that the lapel did so well and And in retrospect, that's not a surprising as maybe it shouldn't have been as surprising as I as I felt it was The lapel mic is a very high quality microphone and as Morgan pointed out that there are actually some advantages to it in terms of breath noises and clothes Wrestling if no one else is talking. Yeah so\nSpeaker H: Well, it's yeah, so the breath noises in the mouth clicks and so forth like that though pal is gonna be better on The lapel is typically worse on the enclosed wrestling, but if no one's wrestling their clothes\nSpeaker G: Right, I mean a lot of people are just sort of leaning over and reading digits. So it's it's a very\nSpeaker A: Different tasks than sort of neutral such during reading digits. Yeah, right\nSpeaker H: Right, right so in the digits in most most cases there weren't other people talking\nSpeaker D: The lapel likes to make a direction. They're typically don't know because I suppose you could make some that have sort of that you have to orient\nSpeaker G: They have a little bit, but they're not noise-cancelling so\nSpeaker H: They're intended to be on the directional right and and because you don't know how people are gonna put them on right now\nSpeaker G: So also and grace on that one the the back part of it should be right against your head and that will keep it from flopping or up and down as much\nSpeaker H: Yeah, yeah, we actually talked about this in the friend-in meeting this morning too I think anything and it was I mean, they're the point of interest to the group was primarily that The the system that we had that was based on HTK that's used by you know All the participants in Aurora was so much worse than they than the SRI and the interesting thing is that even though Yes, it's a digits task and that's relatively small number of words and there's a bunch of digits that you train on It's just not as good as having a very large amount of data and training up a nice good big HMM also you had the adaptation in the SRI system which we didn't have in this so\nSpeaker D: No, or if you did I didn't think I think Stefan had seen them so Yeah A couple I don't remember but there was there was a significant And And that was the founder by the patient and then there was a very small Like point one percent of the natives Adaptation to the recognition hypotheses And I try both So\nSpeaker H: But I think one thing is that I would presume have you ever have you ever tried this exact same recognizer out on the actual T.I. Digits test set This exact same recognizer though it might be interesting to do that it's my because my sense\nSpeaker D: I have tried but I would do even slightly better. Yes or I are actually working on digits I could and they are using a system that's You know is actually trained on digits But otherwise it's the same You know the coder the same training Methods and so forth and I could ask them what they do yeah, but although I think it'd be interesting to just take this exact\nSpeaker H: Actual system so these numbers were comparable and try it on on T.I. Digits. Yeah, yeah, yeah Because our sense from the other from the Aurora task is that I mean because we were getting sub 1% Numbers on T.I. Digits also with the tandem thing so one so there were a number of things we noted from this one is yeah The SRI system is a lot better than the HTK This you know very limited training HTK system But the other is that the digits recorded here in this room with these close mics Actually a lot harder than the studio recording the digits I think you know one reason for that Might be that the still even though it's close talking there still is some noise and some room acoustics And another might be that I would presume that in the studio uh, uh, situation recording red speech that if somebody did something a little funny or not something a little funny or made a little\nSpeaker G: They didn't include it. They didn't include it. They made a good use. Whereas I took out the ones that I noticed that were blatant that were correctable Yeah, so that if someone just read the wrong digit I corrected it and then there was another one where Jose Couldn't tell whether I couldn't tell whether he was saying zero or six and I asked him and he couldn't tell either So I just cut it out You know, so just edit out the first word of the utterance um, so there's a little bit of correction But it's definitely not as clean as t i digits So my expectations is t i digits would especially I think t i digits is all American English right? So we probably do even a little better still\nSpeaker D: On the s r i system but we could give it a try Um, I was I thought that maybe that's actually\nNone: good thing Because it gets in some of the The noises Um, hello Um,\nNone: The Um, speech And um, I suspect that to get sort of the last Been out of these higher quality recordings you would have the These models that are portraying on lighter band data and\nSpeaker H: What's t i digits? I thought it's wide band. Yeah, it is wide band. We looked it up and it was actually 20 kilohertz sample. Oh, that's right\nSpeaker G: I did look that up. I couldn't remember whether that was t i digits or one of the other digit tasks\nSpeaker D: But I would yeah\nSpeaker G: So Morgan you're getting a little breath noise you might want to move the mic down a little bit\nSpeaker D: One issue with that is that The system has this The notion of a speaker Which would use an app patient\nNone: And Um, You know And also The VPL\nNone: So Yeah, I noticed the script that extracted it\nSpeaker D: So does Um, the t i digits database have speakers? Yep, there are now. Yep. And is there is there enough data or that I don't know\nSpeaker G: A terrible amount of data to the I don't know what we have in our acquaintance. I don't know how many speakers there are. Yeah, and how many speakers per\nSpeaker H: Outerance. Well, the other thing would be to do it without the adaptation and compare it to the Sembers without the adaptation\nSpeaker D: Right, but I'm not so much worried about the application actually Um, uh, the, uh, VPL Right, if you have only one other speaker, my next reason is The Morgan\nSpeaker G: I strongly suspect that they have more speakers than we do Right, so, uh,\nSpeaker D: We might have speakers to the numbers and see what the data first speaker\nSpeaker G: Right, so we could probably do an extraction that was roughly equivalent Um So although I sort of know how to run it, there are a little Few details here and there that I'll have to dig out\nSpeaker D: Actually, it's the speaker ID from the white farm names. Right, I saw that There's a script and that is actually all in one script. So there's one script that parses white farm names And the text is like the speaker ID or something that can stand in as a speaker ID. So I have to modify that script to recognize the, you know, uh, uh, uh,\nSpeaker G: Speak as, uh, yeah, I did it. Right, and that uh\nSpeaker D: Or you can, you can, uh, Make names for these white forms that are that relevant and that we use here for that. Right.\nSpeaker G: I might have to do that any way to, to do, because we may have to do an extract to get the amount of data per speaker about right.\nSpeaker G: The other thing is isn't TI digits isolated digits? Or is that another one?\nSpeaker G: I looked through a bunch of the digits to corporate, corporate, and now they're all blurring.\nSpeaker G: Because one of them was literally people reading a single digit.\nSpeaker G: And then others were connected digits.\nSpeaker H: I mean, we had a bell cord corpus that we were using.\nSpeaker H: It was a bell cord. That's that was isolated digits.\nSpeaker D: All right.\nSpeaker D: I think we can improve these numbers.\nSpeaker D: It can improve the improvement by not starting with the switchboard models.\nSpeaker D: By taking the switchboard models and doing supervised adaptation on the small and the digit data collected.\nSpeaker D: Yep.\nSpeaker D: And then we would adapt your models to the room acoustics and for the fun, if I could, you know, to the noise.\nSpeaker D: And that should really improve things further.\nSpeaker D: And then you use those adaptive models, which should not speak for adaptive, but sort of channel adapted.\nSpeaker D: You use that as the starting models for your feedback adaptation.\nSpeaker H: Yeah. But the thing is, I mean, when you, depends whether you're just using this as a starter task for, you know, to get things going for a conversation or for really interested in connected digits.\nSpeaker H: And I think the answer is both.\nSpeaker H: And for connected digits over the telephone, you don't actually want to put a whole lot of effort into adaptation because somebody gets on the phone and says the number.\nSpeaker H: And then you just want it.\nSpeaker D: You don't say it's better.\nSpeaker D: But, you know, my impression was that you actually just a minute of the positive.\nSpeaker D: So you want to, you want to, that's the obviously you can try.\nSpeaker D: Right?\nSpeaker D: Right.\nSpeaker D: You don't have any, that's where the most requested list matches between the correlation and models.\nSpeaker D: Right.\nSpeaker H: So that's how you're doing stuff.\nSpeaker H: Yeah, so that'd be another interesting data point.\nSpeaker H: I mean, I guess I'm saying I don't know if we're going to do that with this.\nSpeaker G: It clips over your ears. There you go.\nSpeaker C: If you have a strong, if you have a strong preference, you could use this.\nSpeaker C: I think it has some spikes.\nSpeaker C: So we use that.\nSpeaker C: But you could if you wanted to.\nSpeaker C: Anyway, I don't know if you want to.\nSpeaker H: Yes, your microphones a little bit low.\nSpeaker H: Yeah. I don't know if we use that as the...\nSpeaker C: Give it.\nSpeaker D: So if you see a picture like this.\nSpeaker D: And then you have to...\nSpeaker G: I already adjusted this a number of times.\nSpeaker G: Yeah, I think these mics are not working as well as I would like.\nSpeaker H: Yeah, I think there's too many adjustments.\nSpeaker H: Yeah.\nSpeaker H: Anyway, what I was saying is I think I probably wouldn't want to see that as sort of like the norm that we compared all things to.\nSpeaker H: To have all of this at all this adaptation.\nSpeaker H: But I think it's an important data point if you...\nSpeaker H: Yeah.\nSpeaker H: The other thing that, of course, everybody was looking at was just that, the near versus far.\nSpeaker H: And yeah, the adaptation would get to some of that.\nSpeaker H: But I think even if there was only a factor of two or something, like I was saying in the email, I think that's a big factor.\nSpeaker H: So...\nSpeaker G: Liz, you could also just use the other mic if you ever promise with that one.\nSpeaker C: Yeah, this would be okay.\nSpeaker C: We think that this has spikes on it, so it's not as good acoustically.\nSpeaker D: You're such a big...\nSpeaker D: I mean, mine are two.\nNone: Everybody's here.\nSpeaker F: Hello.\nSpeaker C: Well, if you'd rather have this one then.\nNone: Okay.\nSpeaker G: Yeah.\nSpeaker G: So to get that pivoted this way, it pivots like this.\nSpeaker G: Yeah, there you go.\nSpeaker G: And there's a screw that you can tighten.\nSpeaker F: Right. I already tried to get it good.\nSpeaker G: So if it doesn't bounce around too much, that's actually a good placement.\nSpeaker G: That's good.\nSpeaker G: But it looks like it's going to bounce a lot.\nSpeaker H: So where were we?\nSpeaker H: Digits.\nSpeaker H: So, I think that's an adaptation, an adaptation, factor two.\nSpeaker H: Oh, yeah, I know what it's going to be.\nSpeaker H: Oh, no, no.\nSpeaker H: That we were saying, you know, well, is how much worse as far the near, you know.\nSpeaker H: I mean, it depends on which one you're looking at, but for the everybody, it's the London factor two.\nSpeaker H: Yeah, I know what I was thinking was that maybe we could actually try, at least looking at some of the...\nSpeaker H: The archivic capillary speech from a far microphone, at least from the good one.\nSpeaker H: I mean, before I thought we'd get, you know, 150% error or something, but if we're getting 35%, 40% or something...\nSpeaker F: Actually, if you run though on a close talking mic over the whole meeting during all those silences, you get like 400%.\nSpeaker H: Right, I understand.\nSpeaker H: But doing the same kind of limited thing...\nSpeaker H: Yeah, sure. Get all these insertions. But I'm saying if you do the same kind of limited thing as people have done in switchboard evaluations, or as...\nSpeaker F: You know who the speaker is and there's no overlap.\nSpeaker F: Yeah.\nSpeaker F: You do just the far, like, for those regions.\nSpeaker H: Yeah, the same sort of numbers that we got those graphs.\nSpeaker G: We do exactly the same thing that we're doing now, but do it with a far-field mic.\nSpeaker G: Yeah, do it with a mic.\nSpeaker G: Because we extract the times from the near-field mic, but you use the acoustics from the far-field mic.\nSpeaker F: Right, I understand that.\nSpeaker F: I just meant that... So you have three choices. There's... You can use times where that person is talking only from the transcripts, but the segmentations were synchronized.\nSpeaker F: Or you can do a forced alignment on the close talking to determine that within this segment, these really were the times that this person was talking in, and the cells were in the segment. Other people are overlapping and just running those pieces, or you can run on the whole data.\nSpeaker H: But how did we determine the links that we're testing on in the stuff we reported?\nSpeaker F: In the HLT paper, we took segments that are channeled, time aligned, which is now being changed in the transcription process, which is good.\nSpeaker F: We took cases where the transcribers said there was only one person talking here, because no one else had time any words in that segment.\nSpeaker H: They called that non-overlapping.\nSpeaker H: And that's what we were getting those numbers from. Right?\nSpeaker F: The good numbers. The bad numbers were from the segments where there was...\nSpeaker H: Well, we could start with the good ones, but anyway, so I think that we should try it once with the same conditions that were used to create those, and in those same segments just use one of the PCMs.\nSpeaker H: And then, you know, I mean, the thing is, if we were getting what, 35, 40% something like that on that particular set, does it go to 70 or 80, or does it use it so much memory we can't decode it?\nSpeaker F: I might have said that then, which speaker it is on how close they are to the PCM? I don't know how different they are to each other to be best.\nSpeaker G: For this particular digit ones, I just picked that one.\nSpeaker H: So, we would then use that. This is kind of central, you know, but I pick that one. It will be less good for some people than for other, but I'd like to see it on the same exact same data set that...\nSpeaker G: I actually shouldn't pick a different one, because that could be why the PDA is worse, because it's further away from most of the people reading digits.\nSpeaker H: That's probably one of the reasons.\nSpeaker F: Well, yeah, you could look at, I guess, that, the PCM.\nSpeaker H: But the other is, I mean, even though there's, I'm sure, the SRI friend has some kind of pre-ampuses, it's still picking up lots of low frequency energy.\nSpeaker H: So, even discriminating against it, I'm sure some of it's getting through.\nSpeaker H: But yeah, you're probably part of it, it's just the distance.\nSpeaker F: And aren't these pretty bad microphones?\nSpeaker H: Yep.\nSpeaker H: Well, they're bad.\nSpeaker H: But I mean, if you listen to it, sounds okay, you know.\nSpeaker G: Yeah, when you listen to it, the PCM and the PDA, yeah, the PDA has higher sound flow, but not by a lot, it's really pretty.\nSpeaker F: It is, remember, you've got them to be cheap on purpose.\nSpeaker F: Cheap on terms of their quality.\nSpeaker H: Well, they're one at the 25 cents or so.\nSpeaker G: To be typical of what would be in a PDA, so they are, they're not the PCM $300 type. They're the 25 cent.\nSpeaker G: Buy them in tax of thousand times.\nSpeaker H: But I mean, the things, people use those little mics for everything, because they're really not bad.\nSpeaker H: I mean, if you're not doing something ridiculous like feeding it to a speech recognizer, they, you know, you can hear the sound.\nSpeaker H: Hear the sounds just fine, you know.\nSpeaker H: I mean, it's more or less the same principles as these other mics are built on.\nSpeaker H: There's just that there's less quality control.\nSpeaker H: They just turn them out and don't check them.\nSpeaker H: So, so that was, yeah, so it was interesting to me, so like I said, the front end guys are very much interested in this as well.\nSpeaker D: So, but where is this, what's, what do we want from, yeah?\nSpeaker G: Yeah, that was going to be my question.\nSpeaker H: I think what we want to do is we want to, we talked about this in other contexts, we want to have the ability to feed it different features.\nSpeaker H: And then from the point of view of the front end research, it would be substituting for HTK.\nSpeaker H: I think that's the key thing. And then if we can feed it different features, then we can try all the different things we're trying there.\nSpeaker H: And then also Davis is thinking about using the data in different ways to explicitly work on reverberations, starting with some techniques that some other people have found somewhat useful.\nSpeaker D: So, the key thing that's missing here is basically the ability to feed, you know, other features, and also try to fix it.\nSpeaker D: Right.\nSpeaker H: And I don't know where I try to pull it back, but that's exactly what he's, he's sort of back, but he drove for 14 hours and wasn't going to make it in today.\nSpeaker D: I think that's one of the things that he's doing.\nSpeaker D: Right.\nSpeaker D: So, that means probably for the foreseeable features that you have to dump out, you know, if you want to use some features, you have to dump them into little files and give those files to the right person.\nSpeaker G: We tend to do that anyway.\nSpeaker G: So, although you can pipe it as well, we tend to do it that way because that way you can concentrate on one block and not keep redoing it over and over.\nSpeaker G: Yeah, sorry.\nSpeaker G: So, that's exactly what the defile is for.\nSpeaker D: Yeah.\nSpeaker D: The thing is that you actually have to dump out little files of each segment that you want to recognize. That's not the separate file.\nSpeaker H: Cool. Okay. So, the next thing we had in the agenda was something about linements.\nSpeaker F: Yes, we have. Did you want to talk about it? I was just telling this to Jane. We were able to get some definite improvement on the forced linements by looking at them first and then realizing the kinds of errors that were occurring.\nSpeaker F: Some of the errors occurring very frequently are just things like the first word being moved as early as possible in the recognition, which is a...\nSpeaker F: I think it was both a pruning problem and possibly a problem with beating constraints on word-boundations.\nSpeaker F: And so we tried both these things. We tried saying, I don't know, I got this wacky idea that just from looking at the data that when people talk their words are usually chunked together.\nSpeaker F: It's not that they say one word and then there's a bunch of words together.\nSpeaker F: There might say one word and then another word, far away if they were doing just back channels. But in general, if there's like five or six words and one word far away from it, that's probably wrong on a error.\nSpeaker F: Then also the pruning, of course, was two severe...\nSpeaker D: It's actually interesting. The pruning was the same value that we used for recognition.\nSpeaker D: And we had to lower that. We had to use tighter pruning up to Liz Ratz, who is showing that a run slower and there's no...\nSpeaker F: It was better with slightly better, it was the same with type.\nSpeaker F: So for a free recognition, the lower pruning is better. Probably because the recognition is just bad at a point where it's bad enough that you don't lose that.\nSpeaker D: But it turned out to get accurate alignments, it was really important to open up the pruning significantly.\nSpeaker D: Because otherwise it would do greedy alignments in regions where there was no real speech yet from the front-end speaker.\nSpeaker D: So that was one big factor that helped me do things. And then the other thing was that, as I said, we put in force the fact that the foreground speech has to be continuous.\nSpeaker D: And you cannot have a background speech, hypothesis, or the middle of the foreground speech. You only have background speech at the beginning at the end.\nSpeaker F: But I mean, it isn't always true. And I think what we really wanted, whenever way to do this, from the data, or maybe some hand corrected alignments from transcribers, that things like words that do occur just by themselves alone.\nSpeaker F: Like, back tunnels are something that we did allow to have background speech around. Those would be able to do that, but the rest would be constrained.\nSpeaker F: So I think we have version of it. It's pretty good for the native speakers. I don't know yet about the native speakers.\nSpeaker F: And we basically also made noise models for the different group, some of the mouth noises together.\nSpeaker F: So then there's a background speech model. And we also, there was some neat or interesting cases. Like, there's one meeting where Jose is giving a presentation.\nSpeaker F: And he's talking about the word mixed signal. And so I wanted to understand that you were saying mixed. I think Morgan.\nSpeaker F: And so your speech was saying something about mixed signal. And the next turn was a lot of people saying mixed. Like he means mixed signal. Or I think it's mixed.\nSpeaker F: And the word mixed in this segment occurs like a bunch of times and chucks on the lapel here. And he also says mix, but it's at the last one. And of course the aligner aligns it everywhere else to everybody else's mix.\nSpeaker F: Because there's no adaptation yet. So there's, I think there's some issues about. We probably want to adapt at least the foreground speaker.\nSpeaker F: But I guess I just tried adapting both the foreground and a background generic speaker. And that's actually a little bit of a funky model. Like it gives you some weird alignments just because off the background speakers match better to the foreground and the foreground speaker.\nSpeaker F: So there's some things there, especially when you get lots of the same words.\nSpeaker D: I think you can do better by cloning. So we have a reject phone and you and what we wanted to try with.\nSpeaker D: Once we have this paper in one time. Cloning that reject mom and then one copy of it would be adapted to the foreground speaker to capture the rejects in the foreground by fragment of the other.\nSpeaker F: Right. I mean, in general, we actually right now the words like partial words are reject models. And you normally allow those to match any word. But then the background speech was also a reject model.\nSpeaker F: And so this constraint of not allowing reject can be to be you know, it needs to differentiate which is just sort of working through a bunch of debugging kinds of issues.\nSpeaker F: And another one is turns like people starting with well, I think, and someone else is well, how about so the word well is in this segment, multiple times.\nSpeaker F: And as soon as it occurs, usually the aligner will try to align it to the first person who says it. But then that constraint of sort of proximity constraint will push it over to the first person who really is.\nSpeaker G: Is the proximity constraint a hard constraint or did you do some sort of probabilistic way in distance?\nSpeaker D: I know it's a glitch. It's straightforward to actually just have a panel here that doesn't allow this. But we just didn't have time to play with the level.\nSpeaker D: And really the reason we can't do it is that we don't have a round truth. So we would need a hand mark wood level alignments or at least sort of the boundaries of the speech between the speakers. And then use that as a reference and to the parameters of the volume that you often hear.\nSpeaker H: I was going to ask anyway how you assessed things were better.\nSpeaker F: I looked at them as two days.\nSpeaker F: I was painful because the thing is the alignment share a lot in common. So you're looking at these segments where there's a lot of speech.\nSpeaker F: A lot of them have a lot of words. Not by every speaker but by some speaker.\nSpeaker F: I mean, if you look at these individual segments from just one person you don't see a lot of words. But altogether you'll see a lot of words up there.\nSpeaker F: So the reject is also mapping. So I looked at them all in ways and just lined up all the alignments. At first it sort of looked like a mess.\nSpeaker F: And then the more I looked at it, well it's moving these words left word. And it wasn't that bad. It was just doing certain things wrong.\nSpeaker F: But I don't have time to look at all of them. And it would be really useful to have a transcriber who could use ways.\nSpeaker F: Just mark the beginning and end of the foreground speakers real words. The beginning of the first word, the end of the last word.\nSpeaker C: Okay, I have to ask you something which is first of all, does it have to be waves? Because if we could benefit from what you did, incorporate that into the present transcripts that would help.\nSpeaker C: And then the other thing is I believe that I did hand. So one of these transcripts was gone over by a transcriber and then I hand marked it myself so that we do have the beginning and ending of individual utterances.\nSpeaker C: I didn't do a word level but in terms so for one of the NSA groups. And also I went back to the original one that I first transcribed and did it utterance by utterance for that particular one.\nSpeaker C: I think you do have, if that's a sufficient unit, I think that you do have hand marking for that. But it'd be wonderful to be able to benefit from your wave stuff.\nSpeaker F: Okay, you used to transcriber in it.\nSpeaker F: If you want to, well, Dan and I were, in terms of the tool, talking about this, I guess Sue had had some reactions, you know, interface wise if you're looking at speech, you want to be able to really wear the words.\nSpeaker F: So we can give you some examples of what the sounds that looks like and see if you can maybe incorporate it into the transcriber tools way or.\nSpeaker C: Well, I'm thinking just incorporating it into the representation. I mean, if it's, if it's, if you have start points, if you have like time tags, which is what I assume is not what you, well,\nSpeaker D: that would be you as, I mean, we've been related to this format that the miss scoring code on the CTM conversation in that part. And then that's that's what I think transcriber output CTM.\nSpeaker G: I think so.\nSpeaker C: It seems like she's, if she's moving time marks around since our presentation in transcriber uses time marks, it seems like there should be some way of using that benefit.\nSpeaker F: The advantage would just be that when you brought up a bin, you would be able, if you were zoomed in enough in transcriber to see all the words, you would be able to like have the words located in the time.\nSpeaker H: So, so if we even just had a, it sounds like we, we, we almost do.\nSpeaker H: We're two, yeah, just trying out the alignment procedure that you have on that, you could actually get something, the good objective measure.\nSpeaker F: You mean not on the hand mark.\nSpeaker F: Yeah.\nSpeaker F: So, we only have, I only looked at actually alignments from one meeting that we chose, and our four is randomly, and...\nSpeaker D: Actually, not randomly.\nSpeaker D: Not randomly.\nSpeaker F: We know that it has these things in sort of a virus.\nSpeaker F: That's sort of average recognition performance and a bunch of speakers and meeting the point of view.\nSpeaker F: But yeah, we should try to use what you have.\nSpeaker F: I did re-run recognition on your new version of our one.\nSpeaker F: Good. The one with Dan.\nSpeaker F: Yeah, exactly. Yeah, yeah.\nSpeaker E: I don't think that was the new version.\nSpeaker F: Yeah, actually it wasn't the new, it was the medium view, but we would do the latest.\nSpeaker E: Did you adjust the utterance times for each channel?\nSpeaker C: Yes, I did. And furthermore, I found that there were a certain number where not, not a lot, but several times I actually moved an utterance from Adam's channel to Dan's or from Dan's to Adam.\nSpeaker C: So there was some speaker, and the reason was because I transcribed that at a point before we had the multiple audio available.\nSpeaker C: So I couldn't switch between the auto.\nSpeaker C: I transcribed it off of the mix channel entirely, which meant an overlap.\nSpeaker C: I was at a terrific disadvantage.\nSpeaker C: In addition, it was before the channelized possibility was there.\nSpeaker C: And finally, I did it using speakers of my, you know, off the CPU on my machine because I didn't have a headphone.\nSpeaker C: So it was like, I mean, in retrospect, it would have been good to have got, I should have gotten the headphone.\nSpeaker C: But in any case, this was transcribed in a less optimal way than the ones that came after it.\nSpeaker C: And I was able to, you know, and this meant that there were some speaker identifications.\nSpeaker E: When are there speaker labeling channels after interrupting this?\nSpeaker E: So you're referring to...\nSpeaker E: Oh, well.\nSpeaker E: For example, you're running down the stairs.\nSpeaker E: I remember this meeting really.\nSpeaker E: Yeah, I'm pan-pan-pan.\nSpeaker E: I'm very well-planted with this.\nSpeaker E: You can just read it like a play.\nSpeaker E: Yeah, I can use it.\nSpeaker E: And she said, and I know.\nSpeaker E: So there's one point where you're running down the stairs.\nSpeaker E: And like, there's an interruption.\nSpeaker E: You interrupt somebody, but there's no line after that.\nSpeaker E: For example, there's no speaker identification after that line.\nSpeaker E: Is that what you're talking about?\nSpeaker E: Or were there mislabelings as far as like, Adam was...\nSpeaker C: That was fixed before...\nSpeaker C: I can't or said that pretty thank you for mentioning.\nSpeaker C: I know that, that, I think went away a couple of reasons.\nSpeaker C: But you're actually saying that's a...\nSpeaker C: I speak of herbicide.\nSpeaker C: Yeah, so with, under...\nSpeaker C: Listening to the mixed channel, there were times when, as surprising as that is, I got Adam's voice confused with dance and vice versa, not for long utterances, but just a couple of places.\nSpeaker C: And I bet it in Overlapse.\nSpeaker C: The other thing that was interesting to me was that I picked up a lot of back channels, which were hidden in the mixed signal, which, you know, you can not too surprising.\nSpeaker C: But the other thing that I hadn't thought about this, but I thought I wanted to raise this with respect to also a strategy, which might help with the alignments potentially.\nSpeaker C: But that's when I was looking at these back channels.\nSpeaker C: They were turning up usually very often, and I won't say usually anyway, very often, I picked them up in a channel, which was the person who had asked a question.\nSpeaker C: So like, someone says, and have you done the so-and-so?\nSpeaker C: And then there would be back channels, but it would be the person who asked the question.\nSpeaker C: And then people weren't really doing much back channeling.\nSpeaker C: And you know, sometimes you have, yeah, I mean, it wouldn't be perfect, but it does seem more natural to give a back channel when you're somehow involved in the topic, and the most natural way is for you to initiate the topic asking questions.\nSpeaker D: I think what's going on is back channeling is something that happens in two party conversations.\nSpeaker D: If you ask someone a question, you're essentially initiating a little two party conversation.\nSpeaker D: Well, actually, when you look at this...\nSpeaker D: It's a back channel because the person is about to be involved in everything.\nSpeaker C: Exactly, my point.\nSpeaker C: And so this is the expectation thing, just the diet, but in addition, you know, if someone has done this analysis himself and isn't involved in the diet, but they might also give back channels to verify what the answer is at this point.\nSpeaker H: I tell you, I say a half-long.\nSpeaker H: Well, people are talking to each other.\nSpeaker F: There you go, but I think there are fewer ah-haves.\nSpeaker F: I mean, just from, we were looking at word frequency lists, to try to find the cases that we would allow to have the reject words in between and doing the line, the one that we want to be constrained to be next to the word.\nSpeaker F: And ah-hah is not as frequent as it sort of would be in pitch-work.\nSpeaker F: If you looked at just word frequency lists of one word, short utterances.\nSpeaker F: And yeah, it's way out there, but not ah-hah.\nSpeaker F: So I was thinking, it's not like you're being encouraged by everybody else to keep talking in the meeting.\nSpeaker F: That's all I'll stop there.\nSpeaker F: Well, that's right. And that would make sort of-\nSpeaker C: And what you say is the other side of this, which is that, you know, so there are lots of channels where you don't have these back channels.\nSpeaker C: When a question has been asked, and these-\nSpeaker F: It's just probably less back-telling. So this good news, really.\nSpeaker F: Even if you consider every other person altogether one person in the meeting, but we'll find out anyway.\nSpeaker F: I guess the other thing we should say is that we're going to compare this type of overlap analysis to switchboard.\nSpeaker F: And call home where we have those sides.\nSpeaker F: We can try to answer this question of, you know, is there really more overlap in meetings or is it just because we don't have the other channel in switchboard?\nSpeaker F: And we don't know what to do.\nSpeaker F: Try to create a paper out of that.\nSpeaker H: Yeah, I mean, you folks have probably already told me, but we're intending to do a your speech to the mission.\nSpeaker F: You want to do tomorrow?\nSpeaker F: Yeah.\nSpeaker F: Yeah, we're still like writing the scripts.\nSpeaker F: We're doing the research.\nSpeaker F: You will, yes, we're going to try.\nSpeaker F: And I was telling Don, do not take this as an example of how.\nSpeaker H: Do what I say, don't do what I do.\nSpeaker G: You'll try to probably be a little late.\nSpeaker G: It is different.\nSpeaker G: In previous years, your speech only had the abstract to do by now, not the whole paper.\nSpeaker G: Right.\nSpeaker G: And so all our timing was off.\nSpeaker G: I've given up on trying to do digits.\nSpeaker G: I think that what I have so far makes a your speech paper.\nSpeaker F: Well, we may be in the same position.\nSpeaker F: And I figured, we'll try because that'll at least get us to the point where we have, we have this really nice database format that Andreas and I were working out that.\nSpeaker F: It's not very fancy.\nSpeaker F: It's just a asking line by line format, but it does give you information.\nSpeaker F: It's the spurred format.\nNone: Yeah, we're calling these spurts after, Chase.\nSpeaker F: I was trying to find what's a word for a continuous region with causes around it.\nSpeaker H: Yeah, I know that the telecom people use, use spurred for that.\nSpeaker H: Yes.\nSpeaker H: And that's, I mean, I was using that for a while when I was doing the greatest speech stuff because I, because I looked up and some books and I found, okay, I want to find a spurred in which, and because, because it's another question about how many pauses they put in between them, but how fast do they do the words within the spurred?\nSpeaker H: Right.\nSpeaker E: That's what we were doing.\nSpeaker E: First is used also.\nSpeaker E: Spurt.\nSpeaker G: Spurt has the horrible name overloading with other, with hard here in Mexico.\nSpeaker H: Just very locally, yeah.\nSpeaker F: I think it was Chase or somebody had the word spurred originally.\nSpeaker F: And so I, that's good.\nSpeaker C: Well, I know Sue wrote about spurts of development, but in any case, I think it's a good term.\nSpeaker F: So we have spurts and we have spurredify.\nSpeaker C: And maybe, maybe, maybe, Chase did.\nSpeaker C: I know, Chase dealt with, Chase speaks about intonation units.\nSpeaker C: Yeah, right.\nSpeaker C: Maybe he speaks about spurts as well, I guess.\nSpeaker D: I've heard your stuff, so.\nSpeaker D: Maybe someone has some ideas about how to do better, but we, there were taking these alignments from the individual channels.\nSpeaker D: We're, from each alignment, we're producing one of these CTM pod, which is actually has, is just a linear sequence of words, which we begin times for every word in the duration.\nSpeaker F: It looks like a wave's label file on all this.\nSpeaker F: Right.\nSpeaker D: So this is one of the first column has the meeting name, so it could actually contain several meetings.\nSpeaker D: And the second column is the channel.\nSpeaker D: Third column is the, start times of the words, the fourth column is the duration of words.\nSpeaker D: And then we're, okay, then we have a messy alignment process where we actually insert into this sequence of words, the tags for like, where were sentence and sentence, question marks, various other things.\nSpeaker F: And these are things that we had done, the done sort of, propagated the punctuation from the original transcribers, so whether it was like question marks, a period, or, you know, comment and things like that, and we kept, and this fluency dashes, we kept those in because we sort of want to know where those are relative to that.\nSpeaker D: So those are actually, so retrofit into the time alignment.\nSpeaker D: And then we merge all the alignments from the various channels and we sort them by time.\nSpeaker D: And then there's a process where you now determine the spurts.\nSpeaker D: That is, actually, you know, you do that before you merge the various channels, so you identify by some criteria, which is pause length, but identify the beginnings and then the spurts.\nSpeaker D: And you put another set of tags in there, and keep those straight.\nSpeaker D: And then you merge everything and you know, linearizing the sequence space on the time marks.\nSpeaker D: And then you extract the individual channels again, but this time, you know, where are the other people starting to end talking, you know, where their spurts start to end.\nSpeaker D: And so you extract the individual channels once spurred by spurred, and you're aware.\nSpeaker D: And then inside the words, or between the words, you now have begin and end tags for overlaps.\nSpeaker D: So you basically have everything sort of lined up in a form where you can look at the individual speakers and how their speech relates to the other speakers.\nSpeaker F: I mean, I think that's actually really useful also because even if you weren't studying overlaps, if you want to get a transcription for the carb-filled mics, how are you going to know which words, from which speakers, occurred at which times, relative to each other, you have to be able to get a transcript like this anyway, just for doing carb-filled recognition.\nSpeaker F: So, you know, it's sort of, hey, it's just an issue we haven't dealt with before, how you timeline things that are overlap.\nNone: That's one thing.\nSpeaker G: Well, I never thought about that.\nSpeaker G: Yes, I mean, when I came up with the original data, it suggests a data format based on the transcription graph.\nSpeaker G: There's capability of doing that sort of thing in there.\nSpeaker F: So you can't see it directly from the transcription.\nSpeaker D: Yeah, this is like a command for the form, and it's a command script, not a command script.\nSpeaker D: There's lots of little things, like the 12 different scripts, which you can run at the end of the year, you have what you want.\nSpeaker D: But, at the last stage, we saw a way, the actual time information, all we care about is whether that this assertive word was overlap by someone else's word.\nSpeaker D: So you sort of, at that point, you discretize things into just having overlap or no overlap.\nSpeaker D: Because we figure that's about the level of analysis that we want to do from paper.\nSpeaker D: But if you wanted to do a more frangering analysis and say, how far into the word of the overlap, you could do that.\nSpeaker D: Yes.\nSpeaker D: It's required more.\nSpeaker C: What's interesting is, it's exactly what, in discussing with Sue about this, she indicated that that's very important for all of us.\nSpeaker C: It's nice to know, and also, I think, as a human,\nSpeaker F: I don't always hear these in the actual order that they occur. So I can have two foreground speakers in a Morgan, and Adam, Jane, could all be talking, and I can align each of them to be starting their utterance at the correct time, and then look where they are relative to each other, and that's not really what I heard.\nSpeaker C: And that's another thing she said.\nSpeaker C: This is Bevers effect.\nSpeaker C: Where, in the second linguistics, you have these experiments, where people have perceptual biases as to what they hear.\nSpeaker F: Yeah, you move things around until you get the best low information point that you can bring in the other person.\nSpeaker F: So it's actually not even possible, I think, for any person to listen to a mix signal, even equalize and make sure that they have the words in their line.\nSpeaker F: So I guess we'll try to write the zeros of each paper.\nSpeaker F: We will write it, whether they accept it later or not.\nSpeaker F: And the good thing is that we have, it's sort of the beginning of what Don can use to link the prosodic features from each level to each other.\nSpeaker H: Yeah, that's a good thing about these papers.\nSpeaker H: You know, I might as well...\nSpeaker D: I don't know if I can read the frame too.\nSpeaker D: I mean, Jane likes to look at data, maybe.\nSpeaker D: You could look at this format and see if you find anything interesting.\nSpeaker H: Yeah.\nSpeaker H: Oh, it's a good thing about these paper deadlines and class projects, and things like that, because you're looking.\nSpeaker D: The other thing is that you usually don't tell your graduate students that these deadlines are actually not that strict.\nSpeaker D: Oh, now it's out in the public.\nSpeaker H: This is secret information.\nSpeaker G: No.\nSpeaker G: No.\nSpeaker D: That's what I'm going to do.\nSpeaker D: We have an interesting guy in the office.\nSpeaker D: Right.\nSpeaker D: Yeah.\nSpeaker D: Thank you, one.\nSpeaker C: It's a strict one.\nSpeaker C: Which sometimes means little ones with a string.\nSpeaker H: By the way, this is totally unfair.\nSpeaker H: You may feel, but the morning meeting folks actually have an extra month or so.\nSpeaker G: Yeah.\nSpeaker G: Yep.\nSpeaker G: There's a special Aurora.\nSpeaker H: There's a special Aurora session, and the Aurora people involved in Aurora have to really may or something to turn in their paper.\nSpeaker F: Well, then you can just name them a submitter.\nSpeaker I: Yeah.\nSpeaker G: Oh, I can submit that to Aurora.\nSpeaker G: Well, it's not a Aurora.\nSpeaker H: No, it's not a Aurora.\nSpeaker H: It's not the Aurora.\nSpeaker H: I mean, it's actually a very specific.\nSpeaker F: Well, maybe it won't be after this.\nSpeaker D: But the people, I mean, stand in the middle of it.\nSpeaker D: The paper that is not an Aurora would probably be more interesting at that point because I'm already so sick and tired of Aurora task.\nSpeaker G: Well, I thought you meant this was just the digit section.\nSpeaker G: I didn't know you meant it was Aurora digits.\nSpeaker D: Well, no.\nSpeaker D: If you have this, if you discuss some relation to the Aurora task, like if you use the same...\nSpeaker H: This is not the Aurora task.\nSpeaker H: So they just do a little grab for it.\nSpeaker D: We are not sitting in the middle of it.\nSpeaker D: Well, in relation to other than the case.\nSpeaker G: Anyway.\nSpeaker G: Well, I know.\nSpeaker G: You could do a paper on what's wrong with the Aurora task by comparing it to other ways of doing it.\nSpeaker D: Well, that's an oral answer to the rule.\nSpeaker D: No, a digit is collected in a different way.\nSpeaker H: Yeah.\nSpeaker H: Maybe.\nSpeaker H: It's pretty hoax.\nSpeaker H: I think it's a little far-fetched.\nSpeaker H: I mean, the thing is that Aurora is a pretty close community.\nSpeaker H: I mean, you know, the people who are involved.\nSpeaker H: The only people who are allowed to test them are people who made it above a certain threshold in the first round.\nSpeaker H: Even in 99.\nSpeaker H: And it's sort of...\nSpeaker D: Well, that's what you wanted.\nSpeaker D: Well, not that they have a company system.\nSpeaker D: I mean, I mean, seriously.\nNone: No, I'm sorry.\nSpeaker D: I mean...\nSpeaker D: In the particular system, I met this HTK backhand.\nSpeaker D: Oh, you don't like HTK?\nSpeaker D: I don't have any stock.\nSpeaker H: No, I mean, this is the HTK that is trained on a very limited amount of data.\nSpeaker D: But maybe you should, you know, do you have more data or...?\nSpeaker H: Oh, yeah, I really think that that's true.\nSpeaker G: But they had something very specific in mind when they designed it.\nSpeaker G: Well, so you can argue about maybe that wasn't the right thing to do.\nSpeaker G: But, you know, they...\nSpeaker H: But one of the reasons I have is Chuck's messing around with the backhand that you're not supposed to touch.\nSpeaker H: I mean, for the evaluations, yes, we'll run a version that hasn't been touched.\nSpeaker H: But one of the reasons I have messing around with it, because I think it's sort of an open question that we don't know the answer to.\nSpeaker H: I always say very glibly that if you show improvement on a bad system, that doesn't mean anything, because it may not be show...\nSpeaker H: Because, you know, it doesn't tell you anything about the good system.\nSpeaker H: And I've always sort of felt that that depends.\nSpeaker H: You know, that if something...\nSpeaker H: If you actually are getting at something that has some conceptual substance to it, it will report.\nSpeaker H: And in fact, most methods that people now use were originally tried with something that was not their absolute best system at some level.\nSpeaker H: Or sometimes it doesn't... not important. So I think that's an interesting question.\nSpeaker H: If we're getting 3% error on English native speakers using the ROR system, and we do some improvements and bring it from 3 to 2, do those same improvements bring the SRI system from 1.3 to 0.8.\nSpeaker H: Well, you know, so that's something we can test.\nSpeaker H: So, anyway, I think we've covered that one up.\nSpeaker H: It's simply well.\nSpeaker H: Okay.\nSpeaker H: So, yeah, so that's...\nSpeaker H: Well, you know, maybe you guys will have one.\nSpeaker H: You and Dan have a paper that's going in.\nSpeaker H: That's pretty solid on the segmentation.\nSpeaker H: Yeah.\nSpeaker H: Still send you the final roadstrip.\nSpeaker H: And there are folks here who will definitely get something in on a rare.\nSpeaker D: So, there's another paper from the US speech paper.\nSpeaker D: But it's on digits.\nSpeaker D: So, a colleague has to realize that it includes version of M&E training.\nSpeaker D: And he tested it mostly on digits.\nSpeaker D: So, you know, it takes weeks to try it.\nSpeaker D: Right.\nSpeaker D: So, I think it's a very impressive result.\nSpeaker D: It's been a very noisy environment.\nSpeaker D: I'm not sure about the order of make from like 10% to 8% or from the 8.1%.\nSpeaker D: Yeah, it got better.\nSpeaker G: Yeah.\nSpeaker G: Hey, that's the same percent relative.\nSpeaker G: Yeah.\nSpeaker G: 20% relative gain.\nSpeaker I: Yeah.\nSpeaker H: Let's see.\nSpeaker H: I think the only thing we had left was some of those...\nSpeaker H: Well, there's a couple things.\nSpeaker H: One is anything that anybody has to say about Saturday.\nSpeaker H: Anything we should do in prep for Saturday.\nSpeaker H: I guess everybody knows about...\nSpeaker H: I mean, Mary was asking, was trying to come up with something like an agenda and sort of fitting around people's times a bit.\nSpeaker H: But clearly, when we actually get here, what things around this, as we need to.\nSpeaker H: But, okay.\nSpeaker H: So, we can't absolutely count on it.\nSpeaker H: Yeah.\nSpeaker F: Are we leaving in here?\nSpeaker H: Yeah.\nSpeaker H: That was my thought.\nSpeaker H: I think this is...\nSpeaker F: We wanted an microphone.\nSpeaker H: No, I hadn't intended to.\nSpeaker H: We wanted... I mean, there's going to be Jeff Katrin, Mary, and two students, so there's five.\nSpeaker H: And Brian.\nSpeaker H: And Brian's coming, so six.\nSpeaker H: And plus all of us.\nSpeaker F: That's how fast we can.\nSpeaker G: It seems like too much coming and going.\nSpeaker G: Well.\nSpeaker D: Because it would be a different kind of meeting.\nSpeaker H: Well, I hadn't really thought of it.\nSpeaker D: Maybe they got the whole day, but just...\nSpeaker D: Maybe part of it.\nSpeaker G: Maybe part of it.\nSpeaker G: And make everyone read digits.\nSpeaker G: At the same time.\nSpeaker G: At the same time.\nSpeaker G: Yeah.\nSpeaker H: I don't know.\nSpeaker F: Is there an initiation in July?\nSpeaker F: Into our heart at Colter.\nSpeaker D: Maybe the sections that I know, right after...\nSpeaker D: I have to lie.\nSpeaker D: I have to stop watching.\nSpeaker D: Yeah.\nSpeaker F: Okay.\nSpeaker F: So, when you...\nSpeaker F: I know the schedule.\nSpeaker H: You know, I just...\nSpeaker H: Okay, yeah.\nSpeaker H: I guess I sent it around a little bit.\nSpeaker H: Is it changed now?\nSpeaker H: I hadn't heard back from Mari after I...\nSpeaker H: I brought up the point about...\nSpeaker H: about the enderesis schedule.\nSpeaker H: So, maybe when I get back, there'll be some mail from her.\nSpeaker H: So...\nSpeaker C: It's a bit more to saying representation.\nSpeaker C: That they...\nSpeaker F: We see that.\nSpeaker F: The two meetings from...\nSpeaker F: I mean, I know about the first meeting.\nSpeaker F: But the other one that you get, the NSA one, which we hadn't done because we weren't running with the initial knowledge.\nSpeaker F: Because the non-A to speak it.\nSpeaker F: Five-nine to say.\nSpeaker F: But it would be useful for the...\nSpeaker F: to see what we get.\nSpeaker F: Great.\nSpeaker C: It's said 2011-21-1000.\nSpeaker E: Yeah, three.\nSpeaker F: Great.\nSpeaker C: I sent email when I finished the...\nSpeaker C: that one.\nSpeaker C: Yeah, that's right.\nSpeaker C: That's what I said.\nSpeaker C: I know the thing.\nSpeaker H: That part's definitely going to confuse somebody who looks at these later.\nSpeaker H: I mean, this is...\nSpeaker H: We're recording secret NSA meetings.\nSpeaker H: Not that NSA.\nSpeaker H: It's network services and applications.\nSpeaker H: There.\nSpeaker F: Out there.\nSpeaker F: No idea what they're talking about.\nSpeaker D: Yeah.\nSpeaker D: The other good thing about the lineman system.\nSpeaker D: It's not always the machines falling if it doesn't work.\nSpeaker D: So you can actually find...\nSpeaker D: Persons falling.\nSpeaker H: It's Morgan's fault.\nSpeaker H: It's always Morgan's fault.\nSpeaker H: It's always Morgan's fault.\nSpeaker A: It's always the transcripts.\nSpeaker F: Oh.\nSpeaker F: Yeah.\nSpeaker F: I guess there are some cases where the wrong speaker...\nSpeaker F: these cases.\nSpeaker F: Not a lot, but where the wrong person...\nSpeaker F: the speech is attached to the wrong speaker and you can tell that when you run it.\nSpeaker F: Or at least you can get clues to it.\nSpeaker F: I guess it does.\nSpeaker F: I'm the early transcripts that people did on the next single.\nSpeaker C: It also raises the possibility of using that kind of representation.\nSpeaker C: I don't know if this would be something we want to check, but maybe using that representation for data entry and then displaying it on the channelized representation.\nSpeaker C: I think that my preference in terms of looking at the data is to see it in this kind of musical score format.\nSpeaker C: And also, you know, soos preference as well.\nSpeaker C: Yeah.\nSpeaker C: But I mean, this is a better interface for making these kinds of, you know, local changes.\nSpeaker C: I'd be kind to...\nSpeaker C: I have no idea.\nSpeaker C: I think there's something that would need to be checked.\nSpeaker H: The other thing I actually had was...\nSpeaker H: I didn't realize this till today.\nSpeaker H: But this is Jose's last day.\nSpeaker H: He's missing.\nSpeaker H: Yeah.\nSpeaker G: Been acting a day tomorrow?\nSpeaker G: About meetings.\nSpeaker G: Oh, that's right, tomorrow.\nSpeaker G: Because if...\nSpeaker B: In the Sunday, I will come back to Spain.\nSpeaker B: Yeah, so I...\nSpeaker B: We would like to say thank you very much to all people.\nSpeaker B: Yeah, it was good having you.\nSpeaker B: At least because I enjoyed very much.\nSpeaker B: I'm sorry by the result of overlapping.\nSpeaker B: I have a good result, Jett, but I...\nSpeaker B: I pretend to continue at Spain during the following months, because I have other ideas.\nSpeaker B: But I have enough time to...\nSpeaker B: Since Maun is not enough to research.\nSpeaker B: I'm mainly for the topic, because it's all difficult.\nSpeaker H: Yeah, maybe somebody else will come along and will be interested in working on it and start off from where you are also.\nSpeaker H: They make use of what you've done.\nSpeaker B: Yeah.\nSpeaker B: But I would like to recommend that...\nSpeaker B: It's funny, but the following to school or sleep...\nSpeaker B: Will be here more time.\nSpeaker B: My opinion is for us to spend more time here and to work more time in the topic.\nSpeaker H: Yeah, it's a very short time.\nSpeaker G: Yeah, it's once a hard and a year.\nSpeaker G: Yeah.\nSpeaker B: It's difficult, a lot better.\nSpeaker B: You are lucky and you find a solution in some few months.\nSpeaker B: Okay, but I think it's not the beginning.\nSpeaker B: Anyway, thank you very much.\nSpeaker B: I've been discharged to...\nSpeaker B:...20 years with you.\nSpeaker B: I hope if you need something for us in the future, I will be at Spain to help you.\nSpeaker H: Thank you, Sunsai.\nSpeaker E: Thank you.\nSpeaker E: Yeah.\nSpeaker H: Okay, I guess...\nSpeaker H: Did it?\nSpeaker H:...as something else, but we'd read our digits.\nSpeaker H: We'll do our last three or Jose's digits.\nSpeaker H: I'm sorry?\nSpeaker B: You prefer to eat chocolate at the coffee break or you prefer now...\nSpeaker B: We have a time.\nSpeaker C: Well, we have a time.\nSpeaker H: You're just people from that end of the table.\nSpeaker G: We've got to wait until after we take the mic off.\nSpeaker G: So we're going to do digits simultaneously or what?\nSpeaker H: We're going to do digits at the same...\nSpeaker B: That's nice.\nSpeaker B: Wow.\nSpeaker C: Well...\nSpeaker C: Very nice.\nSpeaker H: That looks great.\nSpeaker H: Oh, it's all in the interest of getting to the...\nSpeaker H: Yeah.\nSpeaker H: You said yes.\nSpeaker G: It's the rest of the digits.\nSpeaker G: The rest of the digits are very clean without a lot of background noise.\nSpeaker G: So I'm just not sure.\nSpeaker F: What chocolate you're eating because they might make you feel sad.\nSpeaker H: I'm not sure for all because...\nSpeaker H: Actually, I'm actually kind of careful because I have a strong allergy to nuts.\nSpeaker H: So I have to figure out one without it.\nSpeaker H: It's hard to say.\nSpeaker H: I don't know.\nSpeaker F: This is a different kind of...\nSpeaker H: I may hold off.\nSpeaker H: Maybe I'll get some later.\nSpeaker H: Well, he's worried about a ticket.\nSpeaker H: Why don't we do a simultaneous one?\nSpeaker G: Remember to read the transcript number, please.\nSpeaker F: You left it me to the first time.\nSpeaker H: I did, and now I love it so much.\nSpeaker F: Okay, everyone ready?\nSpeaker F: You have to sort of...\nSpeaker F: I'm saying you haven't done this yet.\nSpeaker H: Wait, wait, wait.\nSpeaker H: We want it synchronized.\nSpeaker C: You've done this before, haven't you?\nSpeaker C: Did it together with us?\nSpeaker C: You mean at the same time?\nSpeaker C: Oh, you haven't done this either.\nSpeaker C: The groupings are important.\nSpeaker C: So you're supposed to pass between the groupings.\nSpeaker H: So the groupings must be synchronized.\nSpeaker F: Synchronized digits.\nSpeaker F: We'll give everybody the same sheet.\nSpeaker G: What a good idea.\nSpeaker G: We could do the same sheet for everyone.\nSpeaker F: Have them all read them at once.\nSpeaker G: Or it's just same digits.\nSpeaker H: There's so many possibilities.\nSpeaker H: Okay, why am I going?\nSpeaker H: One, two, three, go.\nSpeaker G: No, 40 transcripts.\nSpeaker H: 911, 911, 911 45 Justice Services.\nSpeaker A: 911 Known.\nSpeaker A: 911, turncase.\nSpeaker A: 2, 1, 2, 1,ga-eral.\nSpeaker H: 3, 2, 3, 2, 3, 2, 5, 2, 1, 4, 3, 2, 1, 6oints, 8, 6 Michigan, 7, 6, 8, 7, 6, 8,\nSpeaker A: If the groupies are counted, then even community groups are captured.\nSpeaker A: So in turn, that's right.\nSpeaker A: 485-793-589-284-3455198-5984-843-1-277-5-0-8395-4397\n"}, {"meeting_id": "Bmr025", "summary": "The meeting discussed the progress of the transcription, the DARPA demos, tools to ensure meeting data quality, data standardization, backup tools, and collecting tangential meeting information. The team was making good progress on the transcription but was still concerned with correcting some of the data. Besides that, they were working on adapting the THISL GUI for their project and figuring out visual tools for meeting participants to help them know when their recording equipment was failing. The team also discussed collecting additional information, like laughter and breath data as well as meeting notes.", "dialogue": "Speaker D: Yep.\nSpeaker D: So if we can't, we can.\nSpeaker D: But we're trying to make this abbreviated meeting because the next documents are pushing for it.\nSpeaker D: So.\nSpeaker D: So, gendith, according to this, is transcription status, DARPA demos, XML tools, disks, backups, et cetera, and crosspads.\nSpeaker F: Does anyone have anything to add to the agenda?\nSpeaker D: OK.\nSpeaker D: We're just going order transcription status.\nSpeaker D: That's probably.\nSpeaker D: I'm going to get quickly.\nSpeaker B: I have several more transcribers and making more.\nSpeaker B: Seven?\nSpeaker B: Several.\nSpeaker B: Oh.\nSpeaker B: I'm sorry.\nSpeaker B: And I've been finishing up the double checking.\nSpeaker B: I hope to have that done by today, but it's going\nSpeaker F: to take one more week. As a somewhat segue into the next topic, could I get a hold of the data, even if it's not really corrected yet, just so I can get the data formats and make sure the information retrieval stuff is working?\nSpeaker F: So can just, oh, it is.\nSpeaker F: OK, just so transcripts.\nSpeaker F: Is the sub directory?\nSpeaker F: Yes.\nSpeaker F: OK.\nSpeaker F: So I'll probably just make some copies of those rather than use the ones that are there.\nSpeaker F: And then just we'll have to remember to delete them once the corrections are made.\nSpeaker B: OK.\nSpeaker C: I also got a shoulder mark to transcription.\nSpeaker C: I just processed the first five e-meaning since they are chunked up.\nSpeaker C: So they probably can be sent to IBM whenever they want them.\nSpeaker C: Well, the second one of those is already at IBM.\nSpeaker E: But the other one that we're waiting to hear from them on.\nSpeaker B: These are separate from the ones that.\nSpeaker E: They're the IBM set.\nSpeaker E: Yeah.\nSpeaker E: Excellent.\nSpeaker E: So as soon as we hear from Brian that this one is OK, and we get the transcript back, and we find out that hopefully there are no problems matching the transcripts with what we gave.\nSpeaker E: Then we'll be ready to go and we'll just send them the next four as a big back.\nSpeaker E: I'm going to work on that.\nSpeaker F: And so we're doing those as disjoint from the ones we're transcribing here?\nSpeaker E: Yes.\nSpeaker E: Exactly.\nSpeaker E: We're sort of doing things in parallel by way we can get as much time.\nSpeaker F: Yeah, I think that's the right way to do it, especially for the information retrieval stuff.\nSpeaker F: Anything else on transcription says?\nSpeaker D: OK.\nSpeaker D: There are pedagogues where you had the sub meeting the other day.\nSpeaker F: Right.\nSpeaker F: Which, so I've been working on using the Thistle tools to do information retrieval on meeting data.\nSpeaker F: And the Thistle tools are, there are two sets.\nSpeaker F: There's a back end and a front end.\nSpeaker F: So the front end is the user interface and the back end is the indexing tool and the querying tool.\nSpeaker F: And so I've written some tools to convert everything into the right phyiformats and the demand line version of the indexing and the querying is now working.\nSpeaker F: So at least on the one meeting that I had the transcript for, conveniently, you can now do information retrieval on it, do type in a string and get back a list of start end times for the meeting of hits.\nSpeaker E: And then what does that look like?\nSpeaker E: The string that you type in?\nSpeaker E: Are you the keywords?\nSpeaker E: Keywords.\nSpeaker F: Right.\nSpeaker F: And then it munges it to pass it to the Thistle YR, which uses an SGML-like format for everything.\nSpeaker D: And then does it play something back?\nSpeaker F: Right now, I have a tool that will do that on a command line using our standard tools.\nSpeaker F: But my intention is to do a prettier user interface based either.\nSpeaker F: So that's the other thing I want to discuss is, well, what should we do for the user interface?\nSpeaker F: We have two tools that have already been written.\nSpeaker F: The soft sound guys did a web-based one, which I haven't used, haven't looked at.\nSpeaker F: Dan says it's pretty good, but it does mean you need to be running a web server.\nSpeaker F: And so it's pretty big and complex.\nSpeaker F: And it would be difficult to port to Windows because it means porting the web server to Windows.\nSpeaker F: The other option is Dan did the Tickle TK Thistle GUI front end for broadcast news, which I think looks great.\nSpeaker F: I think that's a nice demo.\nSpeaker F: And that would be much easier to port to Windows.\nSpeaker F: And so I think that's the way we should go.\nSpeaker B: Can I ask a question?\nSpeaker B: So as it stands within the channel trans-interface, it's possible to do a fine play.\nSpeaker B: You can find a search string in play.\nSpeaker B: So you're adding like, I don't know, are they fuzzy matches or are they?\nSpeaker F: It's a sort of standard text retrieval based.\nSpeaker F: So it's term frequency inverse document frequency scoring.\nSpeaker F: And then there are all sorts of metrics for spacing.\nSpeaker F: How far apart they have to be in things like that.\nSpeaker F: So it's a lot more sophisticated.\nSpeaker F: It's like doing a Google query or anything else like that.\nSpeaker F: So it produces an index ahead of time so you're not doing a linear search through all the documents.\nSpeaker F: Because you can imagine if we have the 60 hours worth, you wouldn't want to do a search.\nSpeaker F: You have to do a pre-indexing.\nSpeaker F: And so these tools do all that.\nSpeaker F: And so the work to get the front end to work would be porting it, well, to get it to work on the Unix systems our side is just rewriting them and modifying them to work for meetings.\nSpeaker F: So that it understands that they're different speakers and that it's one big audiophile instead of a bunch of little ones and just sort of things like that.\nSpeaker E: So what does the user see as the result of the query?\nSpeaker F: On which tool?\nSpeaker F: The Thistle.\nSpeaker F: The Thistle GUI tool, which is the one that Dan wrote, TIGLTK.\nSpeaker F: You type in a query and then you get back a list of hits.\nSpeaker F: And you can type on them and listen to them.\nSpeaker F: Click on them rather with a mouse.\nSpeaker E: So if you typed in small heads or something.\nSpeaker E: Right.\nSpeaker E: You get something that will make you feel like a system audio where that's going to work.\nSpeaker E: You get a little step.\nSpeaker F: Beep.\nNone: That was a really good look.\nSpeaker D: It's too early.\nSpeaker D: It didn't get to work.\nSpeaker D: It didn't get to work.\nSpeaker G: Guess who I practice on.\nSpeaker B: It's not what we're going to do to say what that private joke is.\nSpeaker D: Yeah.\nSpeaker D: And then again, maybe not.\nSpeaker D: So that sounds reasonable.\nSpeaker D: Right.\nSpeaker D: My recollection of it's a pretty reasonable.\nSpeaker D: That sounds really nice.\nSpeaker F: That sounds really nice.\nSpeaker F: It'd be minimal effort to get it to work minimally.\nSpeaker F: And then we want to add things like query by speaker and by meeting and all that sort of stuff.\nSpeaker F: Dave Gilbert expressed some interest in working on that.\nSpeaker F: So I'll work with him on it.\nSpeaker F: And it's looking pretty good.\nSpeaker F: The fact that I got the query system working.\nSpeaker F: So if we want to just do a video based one, I think that'll be easy.\nSpeaker F: If we want to get it to Windows, it's going to be a little more work because the Thistle I are the information retrieval tools.\nSpeaker F: I had difficulty just compiling them on Solaris.\nSpeaker F: So getting them to compile on Windows might be challenging.\nSpeaker E: You were saying that there's that set of tools, the signal tools.\nSpeaker F: It certainly helps.\nSpeaker F: I mean, without those, I wouldn't even attempt it.\nSpeaker F: But what those do is provide sort of the BSD compatibility layer so that the normal Unix function calls all work.\nSpeaker F: But the problem is that the Thistle Tools didn't use anything like AutoConf.\nSpeaker F: So you have the normal porting problems of different header files and some things are defined and some things aren't.\nSpeaker F: And different compiler work around and so on.\nSpeaker F: So the fact that it took me a day to get a compile under Solaris means it's probably going to take me significantly more than that to get it to compile under Windows.\nSpeaker F: Well, I'm going to run under free BSD.\nSpeaker F: Free BSD would probably be easier.\nSpeaker H: All you need to do is say to Dan, gee, it would be nice if this worked under Underconf and it'll be done in a day.\nSpeaker H: That's true.\nSpeaker F: Actually, no, I should check because he did port it to SproutCore.\nSpeaker H: So you might have done that already.\nSpeaker H: I wouldn't be surprised.\nSpeaker F: So I'll check it.\nSpeaker D: But I would serve both purposes as you contact him and ask him if he's already done it.\nSpeaker D: Yeah, right.\nSpeaker D: If he has, then you learn, if he hasn't, then he'll do it.\nSpeaker D: Right.\nSpeaker E: I hope he never listens.\nSpeaker F: I've been corresponding with Dan and also with SoftSound guy, Blinking on his name.\nSpeaker F: Tony Robinson.\nSpeaker F: Do I mean Tony?\nSpeaker F: I guess I do.\nSpeaker F: Or Steve Reynolds.\nSpeaker F: Steve Reynolds.\nSpeaker H: Do I mean Steve Reynolds?\nSpeaker H: Steve Reynolds is not a soft sound, is he?\nSpeaker F: No.\nSpeaker F: My brain is not working.\nSpeaker H: I don't remember who I've been corresponding with.\nSpeaker H: Steve Reynolds wrote this whole IR.\nSpeaker F: Bennett Steve Reynolds.\nSpeaker F: So just getting documentation and format.\nSpeaker F: So that's all going pretty well.\nSpeaker E: And I think we'll be able to do this.\nSpeaker E: So this is a playing sound file between the two platforms.\nSpeaker F: We have, well that's a good point too.\nSpeaker H: Here's a crazy idea.\nSpeaker H: I don't know.\nSpeaker H: Actually, why don't you try and merge Transcriber and this will IR?\nSpeaker H: Well, this is one of the interfaces.\nSpeaker F: This is one of the reasons that I can have Dave Gilbart having him volunteer to work on I just a really good thing because he's worked on the Transcriber stuff.\nSpeaker F: Right.\nSpeaker H: And he's more familiar with that.\nSpeaker H: So Windows media playing for free.\nSpeaker F: Well, that's snack, not transcriber.\nSpeaker H: Right.\nSpeaker H: But the point is that the transcriber uses snack and then you can use a lot of the same functionality.\nSpeaker F: Yeah, I mean, I think this will, this will gooey probably use a snack.\nSpeaker F: And so my intention was just to base it on that.\nSpeaker H: Yeah.\nSpeaker H: Well, my thought was, it would be nice to have the running transcripts from a speaker to speaker.\nSpeaker H: Right.\nSpeaker H: You have a speaker mark here and a speaker mark.\nSpeaker F: Right.\nSpeaker F: We'll have to figure out a user interface for that.\nSpeaker H: Well, my thought was if you had like multi-trans or whatever, do it.\nSpeaker F: Yeah.\nSpeaker F: It might be fairly difficult to get that to work in the little short segments we'd be talking about and having the search tools and so on.\nSpeaker F: We can look into it.\nSpeaker D: But the thing I was asking about with the FreeBSD is that it might be easier to get PowerPoint shows running in FreeBSD than to get the other package running in.\nSpeaker F: Yeah, I mean, we have to, I have to sit down and try it before I make too many judgments.\nSpeaker F: My experience with the GNU compatibility library is really, it's just as hard and just as easy to port to any system.\nSpeaker F: The Windows system isn't any harder because it looks like a BSD system.\nSpeaker F: It's just, you know, just like all of them, the include files are a little different.\nSpeaker F: The function calls are a little different.\nSpeaker F: It might be a little easier, but it's not going to be a lot easier.\nNone: Okay.\nSpeaker D: So there was that demo, which was one of the main ones.\nSpeaker D: And then we talked about some other stuff which would basically be showing off the transcriber interface itself.\nSpeaker D: And as you say, maybe we could even merge those in some sense.\nSpeaker D: But in part of that, we're showing off with the speech non-speech stuff that TLo has done.\nSpeaker D: Yeah.\nSpeaker B: Looks like.\nSpeaker B: So with the answer, you end up with somewhat prioritized.\nSpeaker F: Right.\nSpeaker F: Excellent.\nSpeaker G: So another idea I've just now actually for the demo was whether it might be of interest to show some of the prosody work that Don's been doing.\nSpeaker G: Actually show some of the features and then show, for instance, a task like finding sentence boundaries or finding turn boundaries.\nSpeaker G: You can show that graphically sort of what the features are doing.\nSpeaker G: It doesn't work great, but it's definitely giving us something.\nSpeaker G: I don't know if that would be of interest.\nSpeaker D: At the very least, we're going to want something illustrative of that because I'm going to want to talk about it.\nSpeaker D: And so if there's something that shows it graphically, it's much better than me just having a bullet point pointing at something.\nSpeaker G: I mean, you're looking at this now.\nSpeaker G: Are you looking at waves?\nSpeaker G: I'm at it.\nSpeaker A: I'm starting to.\nSpeaker A: Yeah, we can probably find some examples of different type of prosotic events going\nSpeaker D: on. So when we hear having this demo meeting, what we're sort of coming up with is that we want to have all these pieces together the first order by the end of the month.\nSpeaker D: And then that will give us a little bit of this month or next month.\nSpeaker G: Oh, do you mean like today?\nSpeaker G: Oh, sorry.\nSpeaker G: Next month.\nSpeaker F: Next month.\nSpeaker F: This is it.\nSpeaker D: That will give us a week or so to port things over to my laptop and make sure that works.\nSpeaker D: I think I mean, we're not be here.\nSpeaker G: Yeah, if Don can sort of talk to whoever's because we're doing this anyway as part of our, the research, visualizing what these features are doing.\nSpeaker G: And so either might not be integrated, but it could potentially be in it.\nSpeaker D: Yeah, this is to an audience of researchers.\nSpeaker D: So I mean, let's go ahead and let them know what it is we're doing.\nSpeaker G: I don't think anyone's done this on meeting data, so it might be neat.\nSpeaker G: Yeah.\nNone: Yeah.\nSpeaker D: Done with that?\nSpeaker F: XML tools?\nSpeaker F: So I've been doing a bunch of XML tools where we're sort of moving to XML as the general format for everything.\nSpeaker F: And I think that's definitely the right way to go because there are a lot of tools that let you do extraction and reformatting of XML tools.\nSpeaker F: So yet again, we should probably meet to talk about transcription formats in XML because I'm not particularly happy with what we have now.\nSpeaker F: I mean, it works with transcriber, but it's a pain to use in other tools because it doesn't mark starting end of each utterance.\nSpeaker F: So it's implicit in there, but you have to do a lot of processing to get it.\nSpeaker F: And so, and also I'd like to do the indirect timeline business.\nSpeaker F: But regardless, I mean, that's something that you and me and Jane can talk about later.\nSpeaker F: And I've installed XML tools of various sorts and various languages.\nSpeaker F: And so if people are interested in doing extracting any information from any of these files, either information on users because the user databases that way, I'm converting the key files to XML so you can extract various sort of information on individual meetings and then also the transcripts.\nSpeaker F: And so let me know there.\nSpeaker F: It's mostly Java and Pearl, but we can get other languages too if that's desirable.\nSpeaker G: Quick question.\nSpeaker G: Do we have the seat information in the key files?\nSpeaker F: The seat information is on the key files for the ones which it's been recorded.\nSpeaker F: For the new one.\nSpeaker F: Okay.\nSpeaker F: Seat?\nSpeaker D: Yeah.\nSpeaker D: Where you're sitting.\nSpeaker D: Oh, not the quality, you're right.\nSpeaker F: Right.\nSpeaker F: It's pretty soft and squishy.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: That might just be me.\nSpeaker F: All right.\nSpeaker G: I'm just trying to figure out, you know, when we use the keys on someone's microphone.\nSpeaker G: Are they next to him or are they crushed?\nSpeaker E: Where is it in the key file?\nSpeaker F: The square bracket.\nSpeaker G: Because I mean, I haven't been putting it in and you haven't been putting it in.\nSpeaker G: Right.\nSpeaker D: I have not.\nSpeaker F: And some of these are missing, aren't they?\nSpeaker G: Yeah.\nSpeaker F: So we can go back and fill them in for the ones we have.\nSpeaker G: I mean, they're on the right, these, but I just hadn't ever been putting it in the key files and I don't think Chuck was either.\nSpeaker G: I never knew we were supposed to put it in.\nSpeaker F: I had told you guys about it.\nSpeaker E: Oh, sorry.\nSpeaker E: Sorry.\nSpeaker F: So this is why I want to use a tool to do it rather than the plain text because with the plain text, it's very easy to skip those things.\nSpeaker F: So if you use the edit key or key edit, I think it's edit key.\nSpeaker F: Command, did I show you guys that?\nSpeaker F: I did show it to you, but I think you both said no.\nSpeaker F: You'll just use text file.\nSpeaker F: It has it in there.\nSpeaker F: I placed a fill it in.\nSpeaker E: Oh, okay.\nSpeaker F: Yeah.\nSpeaker F: And so if you don't fill it in, you're not going to get it in the meetings.\nSpeaker F: Right.\nSpeaker G: Well, I just realized I hadn't been doing it in probably.\nSpeaker G: So.\nSpeaker F: And then the other thing also that TLo noticed is on the microphone on channel zero, it says handheld mic or a crown mic.\nSpeaker F: You actually have to say which one.\nSpeaker G: I know.\nSpeaker G: Yeah, I usually delete the.\nSpeaker G: Oh, okay.\nSpeaker G: Maybe I forgot to.\nSpeaker G: It's almost.\nSpeaker G: Yeah, that's because.\nSpeaker F: I know.\nSpeaker F: I know.\nSpeaker G: I know.\nSpeaker G: I know.\nSpeaker G: I was like, it's Chuck's like, oh, what did Chuck do?\nSpeaker G: Okay, I'll do that.\nSpeaker F: And then also in a couple of places, instead of filling the participants under participants, they were filled in under description.\nSpeaker F: And so that's also a problem.\nSpeaker F: So anyway, we will do better.\nSpeaker F: Oh, also I'm working on another version of this tool, the one that shows up here that will flash yellow if the mic isn't connected.\nSpeaker F: And it's not quite ready to go yet because it's hard to tell whether the mic's connected or not because the best quality ones, the crown ones, are about the same level if they're off or if they're on and no one's talking.\nSpeaker F: These ones are much easier.\nSpeaker F: There's a bigger difference.\nSpeaker F: So I'm working on that and it sort of works.\nSpeaker F: And so eventually we will change to that and then you'll be able to see graphically if your mic is dropping in or out.\nSpeaker A: That also included batteries dying?\nSpeaker A: Yep, just anytime the mic's putting out zeros basically.\nSpeaker A: Yep.\nSpeaker F: Now, we're off the screensaver too.\nSpeaker F: The thing is, as I've said before, it is actually on the thing.\nSpeaker F: There's a little level meter, but of course no one ever pays attention to it.\nSpeaker F: So I think having it on the screen is more easy to notice.\nSpeaker B: It'll be nice if these have little light indicators.\nSpeaker F: Buzzer.\nSpeaker D: Bam, bam.\nSpeaker D: Small shocks.\nSpeaker D: Yeah.\nSpeaker D: It's a leaf.\nSpeaker D: Minister to the.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker F: Dispack up the center.\nSpeaker F: I spoke with Dave Johnson about putting all the meeting recorder stuff on non-backed of disk to save the overhead of backup.\nSpeaker F: And he pretty much said, yeah, you could do that if you want, but he thought it was a bad idea.\nSpeaker F: In fact, what he said is doing the manual one, doing a NW archive to copy it is a good idea and we should do that and have it backed up.\nSpeaker F: He's a firm believer in lots of different modalities of backup.\nSpeaker F: I mean, his point was well taken.\nSpeaker F: The state cannot be recovered.\nSpeaker F: And so if a mistake is made and we lose the backup, we should have the archive.\nSpeaker F: And if the mistake is made and we lose the archive, we should have the backup.\nSpeaker D: Well, I guess it is true that even with something that's backed up, it's not going to, if it's stationary, it's not going to go through you increment.\nSpeaker D: It's not going to burden things in the incremental.\nSpeaker D: Just the monthly full.\nSpeaker D: Yeah.\nSpeaker D: So the monthly full will be a bearer.\nSpeaker F: He said that we shouldn't worry too much about that.\nSpeaker F: That we're getting a new backup system and we're far enough away from saturation on the full backups that it's probably okay.\nSpeaker F: And so the only issue here is the timing between getting more disks and recording meetings.\nSpeaker D: So I guess the idea is that we would be reserving the non-backed up space for things that took less than 24 hours to recreate or something like that.\nSpeaker F: Things that are recreatible easily and also, yeah, basically, things that are recreatible.\nSpeaker F: They expanded files and things like that.\nSpeaker F: They take a lot more room anyway.\nSpeaker F: Yeah.\nSpeaker F: But we do need more disk.\nSpeaker F: So we can get more disk.\nSpeaker F: Yeah.\nSpeaker F: And I think I agree with him.\nSpeaker F: His point was well taken.\nSpeaker F: If we lose one of these, we cannot get it back.\nSpeaker D: Okay.\nSpeaker F: I don't think there was any other, et cetera, there.\nSpeaker D: Well, that was allowing someone else to come up with something related that they had done.\nSpeaker H: I thought you guys were going to burn CDs.\nSpeaker F: Unfortunately, we could burn CDs, but first of all, it's a pain because you have to copy it down to the PC and then burn it and that's a multi-step procedure.\nSpeaker F: And second of all, the right ones, burners, as opposed to a professional press, don't last.\nSpeaker F: So I think burning them for distribution is fine, but burning them for backup is not a good idea.\nSpeaker F: I see.\nSpeaker F: Because they, they, they all after a couple of years.\nSpeaker A: All right.\nSpeaker B: I want to ask, I know that, that T.L., you were bringing the channel trends in your face onto the Windows machine.\nSpeaker B: Yeah, it's basically, it's done.\nSpeaker B: It's all done.\nSpeaker F: It's wonderful.\nSpeaker F: Great.\nSpeaker F: And it runs on it basically, things will just work.\nSpeaker C: Yeah, it was just from with the SNEC version and the transcribed version, but it's soft.\nSpeaker B: So.\nSpeaker B: And that, does that mean?\nSpeaker B: I think I shouldn't have this, but I don't.\nSpeaker B: Does this mean that this could be ported to a think pad note or is some other type of?\nSpeaker C: Yeah, basically, I didn't install it on my laptop and, yeah, you were.\nSpeaker C: Hmm.\nNone: Good.\nNone: Crosspads?\nSpeaker F: I got an email from James Landay, who basically said, if you're not using them, could you return them?\nSpeaker F: So he said he doesn't need them.\nSpeaker F: He just periodically, at the end of each term, sends out email to everyone who has recorded this having them and asks them if they're still using them.\nSpeaker D: So we've never used them.\nSpeaker F: We used them a couple times, but yeah, we have two.\nSpeaker F: So my opinion on it is, first, I never take notes anyway, so I'm not going to use it.\nSpeaker F: And second, it's another level of infrastructure that we have to deal with.\nSpeaker B: And I have, so my feeling on it is that I think in principle, it's a really nice idea and you have the time tags, which makes it better than just taking raw notes.\nSpeaker B: On the other hand, the downside for me was that I think the pen is really noisy.\nSpeaker B: So you have a couple of, and I don't know if it's audible on the, I sort of thought it was a disadvantage.\nSpeaker B: I do take notes, I mean I could be taking notes on these things.\nSpeaker B: And I guess the plus, the crosspads would be the time markings, but...\nSpeaker C: What is it, crosspads?\nSpeaker D: So it's a regular pad, just a regular pad of paper, but there's this pen, which indicates position.\nSpeaker D: And so you have time and position stuff stored so that you can, you have a record of whatever it is you've written.\nSpeaker F: Okay.\nSpeaker F: And then you can download it and they have OCR and searching and all sorts of things.\nSpeaker F: So if you take notes, it's a great little device, but I don't take notes.\nSpeaker D: And one of the reasons that it was brought up originally was because we were interested in higher level things, not just the microphone stuff, but also summarization and so forth in the question is if you were going to go to some gold standard of what was it that happened in the meeting, where would it come from?\nSpeaker D: And I think that was one of the things, right?\nSpeaker D: And so it seemed like a neat idea.\nSpeaker D: Well have a scribe, have somebody take good notes and then that's part of the record of the meeting.\nSpeaker D: And then we did it once or twice.\nSpeaker D: Yep, and I'm just sort of dying out.\nSpeaker D: I've been chose to write, but...\nSpeaker D: Yeah, that's right.\nSpeaker B: But I guess the other thing I'm thinking is we wanted that kind of thing.\nSpeaker B: I wonder if we lose that much by having someone be as scribe by listening to the recording afterwards, taking notes.\nSpeaker B: I mean we're transcribing it anyway.\nSpeaker E: Because that's summary.\nSpeaker G: I think there's also, there's this use that...\nSpeaker G: Summarize it from the transcript.\nSpeaker G: Well what if you're sitting there and you just want to make an X and you don't want to take notes and you just want to get the summary of the transcript from this time location like, you know, and then while you're bored you don't do anything in once in a while.\nSpeaker G: Maybe there's a joke and you put an X.\nSpeaker G: But in other words you can use that just to highlight times in a very simple way.\nSpeaker G: Also with, I was thinking, I know Morgan disagrees with me on this, but suppose you have a group in here and you want to let them note whenever they think there might be something later that they might not want to distribute in terms of content.\nSpeaker G: They could just sort of make an X near that point or a question mark that sort of alerts them that when they get the transcript back they get some red flags in that transcript region and they can then look at it.\nSpeaker G: So I know we haven't been using it, but I can imagine it being useful just for sort of marking time periods, would you then get back in a transcript?\nSpeaker D: So what makes one thing is maybe we should actually schedule some periods where people go over something later and put some kind of summary or something, you know, to be some scribe who would actually listen, who would agree to actually listen to the whole thing, not transcribe it, but just sort of write down things that struck them as important.\nSpeaker D: But then you don't have the time reference that you have if you had it live.\nSpeaker G: And you don't have a lot of other cues that might be...\nSpeaker F: I mean, that was one of the issues we talked about originally and that's part of the difficulty is that we need an infrastructure for using the time, the cross-pads.\nSpeaker F: So that means synchronizing the time, you know, you want to pretty close and there's a form on a skew because it's a handheld unit with a battery.\nSpeaker F: And so you have to synchronize at the beginning of each meeting all the pads that are being used so that it's synchronized with the time on that.\nSpeaker F: And then you have to download to an application and then you have to figure out what the data formats are and convert it over if you want to do anything with this information.\nSpeaker F: And so there's a lot of infrastructure which...\nSpeaker F: There is an alternative.\nSpeaker B: I mean, still, you know, your point stands about there being need to be an infrastructure, but it doesn't have to be synchronized with a little clock timer on it.\nSpeaker B: When I did it, I synchronized it by voice by whispering 1, 2, 3, 4 into the microphone.\nSpeaker F: Well, but then there's the infrastructure at the other end, which someone has to listen to that and find that point and then mark it.\nSpeaker F: So...\nSpeaker G: Well, could we keep one of these things for another year?\nSpeaker G: I mean, is there a big problem?\nSpeaker F: We keep up.\nSpeaker G: Just in case we...\nSpeaker G: Just in case we...\nSpeaker G: Even maybe some of the transcribers who might be wanting to annotate...\nSpeaker G: There's a bunch of things that might be neat to do, but it might not be the case that we can actually synchronize them and do all the infrastructure, but we can at least try it out.\nSpeaker D: Well, one thing that we might try is on some set of meetings, some collection of meetings, maybe EDU is the right one or maybe something else, we get somebody to buy into the idea of doing this as part of the task.\nSpeaker D: I mean, part of the reason...\nSpeaker D: I think part of the reason that Adam was so interested in the speech quarter sort of idea from the beginning is he set the beginning, he had you taking notes and so forth.\nSpeaker D: And Jane is more into it, but I don't know if you want to really do this all the time.\nSpeaker D: So I think the thing is to get someone to actually buy into it and have at least some series of meetings where we do it.\nSpeaker D: And if so, it's probably worth having one.\nSpeaker D: The problem with the more extended view, all these other, you know, with equibbling about particular applications of it is that it looks like it's hard to get people to routinely use it.\nSpeaker D: I mean, it just hasn't happened anyway.\nSpeaker D: Maybe if we can get a person.\nSpeaker G: Yeah, I don't think it has to be part of what everybody does in a meeting, but it might be a useful, neat part of the project that we can, you know, show off as a mechanism for synchronizing events in time that happen that you just want to make a note of, like what Jane was talking about, with some later browsing just as a convenience, even if it's not a full-blown note-taking substitute.\nSpeaker H: If you wanted to do that, maybe the right architecture for it is to get a PDA with a wireless card.\nSpeaker H: And that way you can synchronize very easily with the meeting because you'll be synchronized with the Linux server.\nSpeaker G: And so what kind of input would you be?\nSpeaker G: So, I mean, if you're not worried about it.\nSpeaker H: You just be pressing like, well, you have a PDA and you could have the same sort of X interface or whatever.\nSpeaker H: I mean, you'd have to do a little bit of coding to do it.\nSpeaker H: But you could imagine, I mean, if all you really wanted was you didn't want this secondary note-taking channel, but just sort of being able to use markers of some sort, a PDA with a lot of wireless card would be probably the right way to go.\nSpeaker H: I mean, even buttons you could do, sort of, I mean-\nSpeaker F: I mean, for what you've been describing buttons would be more convenient than anything else,\nSpeaker G: right? That would be fine too.\nSpeaker G: I mean, I don't have, you know, grandiose ideas in mind, but I'm just sort of thinking, well, we're getting into the next year now.\nSpeaker G: We have a lot of these things worked out in terms of the speech.\nSpeaker G: Maybe somebody will be interested in this.\nSpeaker D: I do like, if you have a couple buttons where one button was, oh, another button was that's great.\nSpeaker G: Or like, this is my, I'm supposed to do this kind of button.\nSpeaker G: I better remember it.\nSpeaker G: Action item.\nSpeaker G: Yeah, something like that or.\nSpeaker F: I mean, I think the cross-border idea is a good one.\nSpeaker F: It's just a question of getting people to use it and getting the infrastructure set up in such a way that it's not a lot of extra work.\nSpeaker F: I mean, that's part of the reason why it hasn't happened is that it's been a lot of extra work for me.\nSpeaker B: It's also, it has this problem of having to go from an analog to a digital record, too,\nSpeaker F: isn't it? Well, it's digital, but it's in a format that is not particularly standard.\nSpeaker B: If you're writing notes in it, it can't do handwriting recognition.\nSpeaker D: No, no, but it's just storing the pixel information, information, it's how digital.\nSpeaker B: I guess what I'm thinking is that the PDA solution, you have it already, without needing to go to pixelization to that.\nSpeaker D: Right.\nSpeaker H: The transfer function is less errorful.\nSpeaker H: Yeah.\nSpeaker G: Yeah.\nSpeaker G: Well, and also it's maybe realistic because people are supposed to be bringing their PDAs to the meeting eventually, right?\nSpeaker G: That's why we have this little, I don't know what I don't want to cause more work for anyone, but I can imagine some interesting things that you could do with it.\nSpeaker G: And so if we don't have to return it and we can keep it for a year, I don't know.\nSpeaker F: We certainly don't have to return it.\nSpeaker F: As I said, all he said is that if you're not using it, could you return it if you are using it, feel free to keep it?\nSpeaker F: The point is that we haven't used it at all, and are we going to?\nSpeaker D: So we have note, I would suggest you return one.\nSpeaker D: Yeah.\nSpeaker D: Because we haven't used it at all.\nSpeaker D: Why would probably-\nSpeaker G: Maybe we could do like a student project, maybe someone who wants to do this as their main project for something would be cool.\nSpeaker F: And if we had them out and sitting on the table, people might use them a little more.\nSpeaker D: Are there any incidents and meetings and press a button when somebody laughs?\nSpeaker D: Well, yeah, that's not a bad.\nSpeaker G: Jeremy is going to be- he's a new student starting on modeling breath and laughter, actually, which sounds funny, but I think it should be cool.\nSpeaker F: Sounds breathy to me.\nSpeaker F: Ha-ha.\nSpeaker F: Ha-ha.\nSpeaker F: Oh dear.\nSpeaker F: Um, that reminded me of something, oh well, too late.\nSpeaker F: It flipped out.\nSpeaker F: Are you going to tease me?\nSpeaker F: Okay.\nSpeaker F: Well, I'm always going to do that.\nSpeaker F: We ordered more wireless, and so they should be coming in at some point.\nSpeaker F: And then at the same time, I'll probably rewire the room as per Jane's suggestion, so that the first end channels are wireless, are the closed talking in the next stand, are far-field.\nSpeaker D: You know what he means, but isn't that funny, something?\nSpeaker D: We ordered more wireless.\nSpeaker D: It's like wires are the thing, so you're ordering- you're- you're- you're- we ordered more absence of the thing.\nSpeaker G: That's a very philosophical- It's sort of an acronym, am I right?\nSpeaker G: That's great.\nSpeaker F: Should we do digits?\nSpeaker F: Do we have anything else?\nSpeaker D: Yeah.\nSpeaker D: I mean, there's all this stuff going on between Andreas and- and Dave and Chuck and others with various kinds of runs.\nSpeaker D: Recognition runs trying to figure things out about the features, but it's all sort of in process, so it's not much to say.\nSpeaker D: Right now.\nSpeaker D: When we start with our esteemed guest.\nSpeaker H: Okay.\nSpeaker H: All right.\nSpeaker F: So just a transcript number, and then the-\nSpeaker H: Yes, this is number two for me today.\nSpeaker D: The idea to do is go away to move way up and-\nSpeaker G: You can do simultaneous. Initiate him.\nSpeaker G: Should we do simultaneous?\nSpeaker G: Well, I'm just thinking, are you going to try to save the data before this next group comes in?\nSpeaker F: Yeah, absolutely.\nSpeaker G: Yes, we might want to do a simultaneous.\nSpeaker D: Right, so- Okay, so let's do a simultaneous one.\nSpeaker D: We might need to do that, actually.\nSpeaker D: Okay.\nSpeaker D: Are you ready?\nSpeaker D: You have to plug your ears by the way.\nSpeaker D: Why?\nSpeaker D: You start laughing.\nSpeaker D: Okay.\nSpeaker D: One, two, and three.\nSpeaker D: Transformation.\nSpeaker D: Transformation.\nSpeaker D:formation.\nNone: Form 6400 West six four 7 Four seven five four seven five zero five of two coach six six six nine five six five five two store five four five set of 10 four service zero one six nine six seven four five twenty six five three two two four five three five five four five four five five six five five six seven four five seven three three four five seven three four four five seven seven five four four five seven four four five five\nNone: seven six four five six five seven five seven seven seven seven seven seven pode seven seven eight nine eight n jagu 426-744-2-426-499-666\nSpeaker D: OK, battle, take 5.\n"}, {"meeting_id": "IS1005c", "summary": "The meeting was mainly about the conceptual design of the remote control. Firstly, marketing introduced market research and fashion trends around the world. While the user interface emphasized that easy to use was more important. Then the industrial designer talked about the overall requirements from the engineering point of view, including the size of the device and source of energy. The group members spent time figuring out the real size of the TFT screen and energy problem. After that, they discussed the shape of the remote control and decided to make it like a banana. Based on this shape, they talked about the detailed design of the appearance and user interface.", "dialogue": "Speaker D: Hello, this meeting should be focused on conceptual design of the remote control.\nSpeaker D: The goal of this meeting is to reach a decision at the end on the concept and I think the last time we talked about a lot and we had a rough idea of what it is going to be but at the end of this meeting we have to reach a decision.\nSpeaker D: So we will have again three presentations from all of you and I hope it will be fast because I would like to have time to follow long description and decisions.\nSpeaker D: So who wants to start?\nSpeaker B: I think.\nSpeaker B: Okay.\nSpeaker B: No, no.\nSpeaker A: You get a start.\nSpeaker A: Okay, how's that?\nSpeaker A: Can you open my presentation please?\nSpeaker A: I'm number four.\nSpeaker D: Train this one?\nNone: Yep.\nSpeaker A: Can you pass the mouse please?\nSpeaker A: Okay, so I'm basically just want to present to you some recent results we've had from looking at some remote control market research and some fashion trends around the world.\nSpeaker A: Fashion trends we got from talking to our contacts in the fashion industry based in Paris and Milan.\nSpeaker A: So from our market research basically we've come to the conclusion that a fancy look and feel as opposed to a functional look and feel is our number one priority.\nSpeaker A: Fancy is the highest priority.\nSpeaker A: Secondly that our remote control needs to be technologically innovative.\nSpeaker A: So this is number two priority but it's two times less important as the fancy criteria.\nSpeaker A: And thirdly the easiness of use is important as well but again two times less important as the technologically innovativeness of the remote control.\nSpeaker A: From our fashion people in Paris and Milan we've discovered that this year the fruit and vegetable motif will be the most important thing in clothes, shoes and furniture.\nSpeaker A: So I'm pretty confident that our remote control fits into the furniture category.\nSpeaker A: And also the feel of material this year is expected to be spongy.\nSpeaker A: So hopefully our remote control reflects that in some way.\nSpeaker A: What does it mean spongy?\nSpeaker A: Sort of squishy.\nSpeaker A: Like soft or something.\nSpeaker A: Yeah soft like a sponge.\nSpeaker A: So in conclusion our remote control needs to be something that's really fancy.\nSpeaker A: Has lots of technology in it.\nSpeaker A: Somehow it would be good to have it related to fruit and vegetables with the spongy feel.\nSpeaker A: And that's easy to use.\nSpeaker A: And from our last meeting our you know Fabian told us that you know one of the requirements is that we have to reflect the look and feel of the real reaction company.\nSpeaker D: That's it.\nSpeaker D: Easy to use is it as much importance and technology offensive and it's less important\nSpeaker A: right? Less important.\nSpeaker A: Fancyness first and then two to you know half as important as that is technology.\nSpeaker A: And half important as technology is easy to use.\nSpeaker D: So Ahmed can you?\nSpeaker B: Yeah.\nSpeaker C: I think the biggest struggle will be the easy to use feature.\nSpeaker C: Yeah.\nSpeaker D: One number three.\nSpeaker D: This one?\nSpeaker B: Yes.\nSpeaker B: Just could you please check if it is the first one or the second?\nSpeaker B: No no no it's the first one.\nSpeaker B: The second one.\nSpeaker D: And this one.\nSpeaker B: Okay.\nSpeaker B: So I'm going to talk about a little bit about how this remote control should be appear to be more easy to use.\nSpeaker B: I think I think the future is easy to use is more important than being fancy but okay we can discuss about it later.\nSpeaker B: Generally this remote control should be something in my opinion.\nSpeaker B: The first feature is just to be easy to use.\nSpeaker B: So the more frequent buttons should be larger.\nSpeaker B: They should be placed in a good position inside the remote control.\nSpeaker B: And I can conclude like this that we shouldn't need to learn how to use it.\nSpeaker B: It should be we shouldn't need to open a book and start reading and learning how to use this remote control.\nSpeaker B: Okay so what I found out that as I said I think it's better to put more frequent buttons which are used more in the middle of the remote control and they should be bigger in size.\nSpeaker B: The shape of remote control should be in a way which should be taken easily in hand.\nSpeaker B: It should not be completed like a cube.\nSpeaker B: It should have round edge so then it's easier and maybe just like some toys some duosick which is easier to take inside the hand.\nSpeaker B: And also because customer doesn't like to buy lots of battery it should not consume lots of energy.\nSpeaker B: Okay and my personal preference is as I said just putting these buttons in special places and cover some buttons which are not used that much like settings buttons like mobile phone usually some mobile phone cover the dialling number part.\nSpeaker B: So we can cover these buttons which are not used or number buttons for the channels and just put volume change or channel change buttons in the remote control and if the user needed to do some more complex tasks he can open the cover and then change settings or something like this.\nSpeaker B: And also I think if we put some buttons in size of the remote control it can be used easier not on remote control.\nSpeaker B: I don't know if I can explain well but just in size for example sliding or rolling stuff if we put it in size then we can easily manipulate it with thumb.\nSpeaker B: So it can be another preference.\nSpeaker B: And I don't know but I think usually just speech recognizer can be good.\nSpeaker B: I know that it consumes lots of energy but if we do it in some way that it sleeps when there is no sun and when it detects some sun it may consume less energy.\nSpeaker B: And I think it's good because it's something new and usually young people like something new so it may not be very useful but because it's new people may buy it.\nSpeaker B: I personally think there should be a big difference between something otherwise they prefer to buy something which is coming from a famous company.\nSpeaker B: That's fine.\nSpeaker C: Okay, good news for me from Hamet but then here's the remote obviously because it's pointy design and I don't like it as an engine but okay so could you please go down and open it?\nSpeaker C: Yeah.\nSpeaker C: First and two and which one probably the first one I'm not sure but check the first one.\nSpeaker C: Most of the things I have to write myself on this board but yeah that's it.\nSpeaker C: It's only this slide.\nSpeaker C: Yeah.\nSpeaker C: This is just one thing I wanted to mention and show you that I just found this that our company developed seven fingers or yeah seven inch TFT screen which is good news for us since we wanted to include the display there.\nSpeaker C: So I probably draw it down overall scheme.\nSpeaker C: This is the stuff that I can use.\nSpeaker C: So this will be like the overall scheme or overall requirements from the engineering point of view.\nSpeaker C: First thing is this will be the overall shape.\nSpeaker C: Now I'm not speaking about the real shape of the device but the shape of the inside of the device.\nSpeaker C: So there will be some circuit for the power.\nSpeaker C: So say power circuit here.\nSpeaker C: The main energy will be taken from batteries that we can and if we decide to use the speech recognition stuff there we must use additional source of energy which I found the best is to use the solar cells which can supply everything.\nSpeaker C: I was computing all the things related to the speech recognition and it's okay to use just two batteries and solar cells.\nSpeaker C: So no problem.\nSpeaker C: There can be also solar cell.\nSpeaker C: The main board with all the circuits will take at least 7 to 7 centimeters so this is my like hard requirement for the guys from the design.\nSpeaker C: So at least 77.\nSpeaker C: It depends where you put your screen because the screen is 7 inch so it depends when you where you put it doesn't matter.\nSpeaker C: It's just one cord from the main board to the screen which can be elsewhere.\nSpeaker C: So this will be TFT.\nSpeaker C: And on the main board we have the interface to the microphone which is somewhere.\nSpeaker C: Say here.\nSpeaker C: Interface to the microphone then the graphical card for the TFT and the third unit is the IR.\nSpeaker C: The good news is that we can we decided to use the infrared unit because our company has also developed the chip for communication by the infrared including all the stuff inside so it will be very cheap for us.\nSpeaker C: So infrared here.\nSpeaker C: So once again the overall requirement 7 to 7 centimeters for the board which can be this pointy but has to be this size and the TFT which is 7 inches.\nSpeaker C: I have to check what I wanted to.\nSpeaker C: From my point of view I don't care about the material used for the overall device.\nSpeaker D: Do you fit any TFT or any electronic device in a sponge thing or is there any problem for that?\nSpeaker D: For example, put the electronic card on a sponge thing.\nSpeaker D: I can imagine it could be a problem.\nSpeaker C: All these things in this box are okay to put in any shape basically.\nSpeaker C: But we have to take care of the TFT.\nSpeaker C: Wow.\nSpeaker C: It's a good feature since it takes if it's around the TFT.\nSpeaker C: And it's good because it just keeps it safe.\nSpeaker C: I don't know.\nSpeaker B: Maybe you can have two shells, a hard shell inside and a sponge.\nSpeaker D: Maybe put electronic in a box and a sponge thing around.\nSpeaker D: Where is that?\nSpeaker C: Well, it's maybe related to the UI.\nSpeaker C: Can I ask a question?\nSpeaker C: Yeah, that's all for me.\nSpeaker A: This 7 inch TFT screen.\nSpeaker A: How big is it in reality?\nSpeaker C: Well, 7 to 7 inches.\nSpeaker A: Sort of like that.\nSpeaker C: Yes.\nSpeaker A: That's quite big.\nSpeaker A: We decided that we're going to use this TFT screen.\nSpeaker D: I don't think it's 7 by 7.\nSpeaker D: I think it's 7 by 7.\nSpeaker D: Usually when they say 7 inches.\nSpeaker D: I mean, even that is still likely to speak.\nSpeaker D: Yeah.\nSpeaker D: I don't know.\nSpeaker D: I don't know.\nSpeaker D: One each one.\nSpeaker D: Yeah, I should think.\nSpeaker C: Honestly speaking, I was thinking that it was 7 centimeters initially.\nSpeaker C: It's 7 inches.\nSpeaker C: And I think we can cut it.\nSpeaker A: You can cut the TFT screen.\nSpeaker A: Yeah.\nSpeaker A: No problem.\nSpeaker C: Because then the size of the graphic card will be one fourth.\nSpeaker A: So, you have no problem to meet your card screen.\nSpeaker A: Okay.\nSpeaker C: So for the same price, we have four screens.\nSpeaker A: Okay.\nSpeaker D: Technica engineers.\nSpeaker A: So what's the size of the device?\nSpeaker C: Well, this is almost nothing.\nSpeaker C: 7 to 7, 2 at least, well some 3 millimeters or something.\nSpeaker A: Even from my perspective, 7 centimeters by 7 centimeters is still.\nNone: Yeah.\nSpeaker C: But we wanted to be spot- Can you hold that?\nSpeaker C: Stuff like that, you know.\nSpeaker C: Because if it's too small, we can lose it at home, you know.\nSpeaker D: What user wants?\nSpeaker D: They want a small remote control.\nSpeaker D: Yeah.\nSpeaker A: It's a good control.\nSpeaker A: They can hold it in the hands of something.\nSpeaker A: But it's something that 7 centimeters square.\nSpeaker A: Easy to hold.\nSpeaker C: Well, I think so.\nSpeaker C: If the roller buttons are on the side, you don't have to catch it like that.\nSpeaker C: But just like this and, you know, follow, follow.\nSpeaker C: Well, that's, does not task for me.\nSpeaker D: But, well, so maybe you can finish your presentation and after we discuss these experiences.\nSpeaker C: Okay.\nSpeaker C: Thanks.\nSpeaker C: That's it.\nSpeaker C: Okay.\nSpeaker D: So, so I think we have a lot.\nSpeaker D: We have to take decision today.\nSpeaker D: So I think we have to take some work to finalize our idea and take decisions.\nSpeaker D: First, I think energy is a key problem because it depends what can we have as feature if we use only batteries, for example, or something like that.\nSpeaker D: Because can we have LCD and speech recognition with battery and it's also related to the size of the, oh, still.\nSpeaker C: Just a point to do energy things.\nSpeaker C: If we use the batteries and the additional solar cell, then it's okay for speech recognition and LCD.\nSpeaker C: So no problem in energy.\nSpeaker C: So, thing.\nSpeaker C: Okay.\nSpeaker C: But we have to use the solar cell.\nSpeaker D: Otherwise not.\nSpeaker D: Like, but using how many batteries, for example, or what?\nSpeaker D: Maybe what is the size of the batteries?\nSpeaker C: Yeah.\nSpeaker C: So, just common A-A cells.\nSpeaker C: Okay.\nSpeaker C: So like one or two, five centimeters.\nSpeaker C: I don't know exactly what.\nSpeaker A: So if we use solar cells, where is the sun?\nSpeaker A: If someone's watching TV inside.\nSpeaker C: Oh, there's a need to be.\nSpeaker C: It's just the daylight, you know.\nSpeaker C: To come on eyes.\nSpeaker C: I suppose that this remote control won't be in the room like this, whereas there is light only.\nSpeaker D: And there are people, but at least when there is TV, you can get light from the TV.\nSpeaker C: Yeah.\nSpeaker D: I don't think so.\nSpeaker D: I don't know.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker C: Well, it's a compromise, you know.\nSpeaker D: At least it's new and maybe technology.\nSpeaker D: Yeah, the toy I want to introduce.\nSpeaker C: You can put the speech recognition because you want it all in the same.\nSpeaker A: And if you watch TV outside, it's very useful.\nSpeaker D: I think before talking about the other thing, it's important thing.\nSpeaker D: It's the case.\nSpeaker D: What are going to be the size?\nSpeaker D: Because it's with drives, what we are going to use as features and so on.\nSpeaker D: For example, for the LCD.\nSpeaker D: If we choose to have a small device, we cannot use this as such.\nSpeaker C: Oh, the screen is okay about the board.\nSpeaker C: That's a problem.\nSpeaker C: Well, what would you guess as a shape or what would be the shape?\nSpeaker B: I think the bank, large or small is not important.\nSpeaker B: The only important thing is to be able to take it inside hand easily.\nSpeaker B: So let's say an average size, okay, and it should not be very heavy also.\nSpeaker B: And I prefer to, it shouldn't have a uniform shape.\nSpeaker B: So in the middle, it should be a little bit thinner maybe.\nSpeaker B: So it's like some joysticks.\nSpeaker B: You can take, some joysticks, you can take inside hand easier because it's designed for your taking into account your finger shape and your palm shape.\nSpeaker B: So the general shape should be like this.\nSpeaker B: I think 7 cm by 7 cm is a little bit large.\nSpeaker B: So 7, not 7, but let's say 5 by 10.\nSpeaker B: I think it's just my opinion.\nSpeaker B: Which is the same area.\nSpeaker A: Yeah.\nSpeaker A: Could you redesign your board?\nSpeaker C: 5 to 10.\nSpeaker A: Oh, 5 centimeters by 10 centimeters.\nSpeaker C: Yeah, right.\nSpeaker C: Yeah, I think it's feasible.\nSpeaker C: Well, one, how could we do it?\nSpeaker C: We could put the board next to, well, under the LCD.\nSpeaker C: And for example, make the LCD be totally unrelated to the thing that you hold in your hand, like holding something and the LCD should be just on top of it, you know, somehow.\nSpeaker C: But maybe let's stick to the spongy thing, like one unit.\nSpeaker C: 5 to 10.\nSpeaker D: So I think it's just thing could be to have a smaller LCD if it's possible.\nSpeaker C: Okay, so 5 to 10, I think it's feasible.\nSpeaker D: So we have to be with a smaller.\nSpeaker B: Or a 3 to 5 by 10.\nSpeaker B: All right, I don't know, but I don't want to now invent something new because we didn't discuss about it.\nSpeaker B: So using some LCDs, we can touch, so we can remove keys and just having, I don't know, the name.\nSpeaker B: I'll see the response into fingers touching the screen.\nSpeaker B: Some touch.\nSpeaker B: Some touch.\nSpeaker B: So, but for now, if we don't want to use such kind of screen, I think we're using a\nSpeaker D: smaller screen is better because the problem is we have a limited amount of time, so we cannot do something very new.\nSpeaker D: So that's both for small LCD.\nSpeaker C: Yeah, so just give me the 5 by 10 numbers that you find the best and send me.\nSpeaker B: So 5 by 10.\nSpeaker D: Okay.\nSpeaker D: So what about, so the case we talked about?\nSpeaker D: Something easy to use.\nSpeaker D: But how does that fit in a fashion way like with fruit and vegetable and about the color and logo of the company and so on.\nSpeaker D: Now, can we do that?\nSpeaker D: My first idea is because our color is more yellow and it should be easy to take in your hand, I thought about banana or something like that, which is fruit and 7 to 10 banana.\nSpeaker C: A big banana.\nSpeaker C: More yellow.\nSpeaker C: Rather mango or something.\nSpeaker A: Well, it's definitely obvious choice with the color of our company.\nSpeaker A: I mean, what are the other fruit vegetables?\nSpeaker A: Just an idea.\nSpeaker A: I don't know what you feel about.\nSpeaker A: Do you know any other fruit and vegetables that are yellow?\nSpeaker D: Yeah.\nSpeaker D: I don't know if you can fit with the technology you want to specialist.\nSpeaker C: I don't know if you want to put that in banana.\nSpeaker C: Well, but if it's, if the banana is big enough, then yes.\nSpeaker C: So if you want to look at the screen at banana, you know.\nSpeaker D: Yeah.\nSpeaker D: I think this is the screen that has to be square or it can be like a shape quite.\nSpeaker C: Well, it can be whatever you want.\nSpeaker C: But if it's square, then we get four screens out of one by cutting just.\nSpeaker C: But if you want some shape, then we can only get like two screens out of 7 to 7 inches.\nSpeaker C: So it's like more, more expensive to have shaped like that.\nSpeaker C: But I don't care.\nSpeaker C: You know, if we, if we, if we're quite a man.\nSpeaker A: I'd like a shaped screen.\nSpeaker A: I think that's more important than saving a bit of money on, on the TFT screen.\nSpeaker C: Maybe the banana could be like a bit fatter than the car, the car coming in one.\nSpeaker D: Yeah, you should remember banana, but it doesn't have to be like banana.\nSpeaker D: Like more than banana, okay.\nSpeaker D: Where is the size and exactly the shape of banana.\nSpeaker C: Okay.\nSpeaker A: Yeah, banana is good.\nNone: Okay.\nSpeaker D: So, this is the last point we decided it's in far-right.\nSpeaker D: Yes.\nSpeaker D: Yes, I think about the concept.\nSpeaker D: You have also thing to add to this point.\nSpeaker D: No.\nSpeaker D: So, about the user interface.\nSpeaker D: So we are going to use NCD.\nSpeaker D: In the last meeting we talked about hidden buttons.\nSpeaker D: I don't know what we're going to do with that.\nSpeaker D: You talked about healing buttons on the side.\nSpeaker B: Yes, that's banana.\nSpeaker C: It would be cool.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Feeling of the banana, you know.\nSpeaker C: Yeah.\nSpeaker C: Discover the other buttons.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And you mean the first layer would be spongy.\nSpeaker D: Yeah.\nSpeaker C: First layer, open the spoon.\nSpeaker C: So, it's like silly, but the people will appreciate it.\nSpeaker D: Is it possible to do that?\nSpeaker D: It would be a great idea, but possible technical.\nSpeaker D: Like doing a spongy layer of the banana.\nSpeaker A: If we have a spongy layer on the outside of the banana, then it's easy to manipulate that to be a cover that you can pull off.\nSpeaker B: Something like plastic cover covering some interface, USB interfaces in digital cameras.\nSpeaker B: You see it's like pulling you open a plastic cover and you see some USB interfaces, some interfaces for adapter.\nSpeaker B: So, buttons can be covered like this with a plastic cover.\nSpeaker B: And when you open this cover, it's like pulling a banana.\nSpeaker B: So, something like this.\nSpeaker D: Yeah, but do you see that as rigid thing or like a banana?\nSpeaker D: Something very soft. You can open a banana.\nSpeaker C: Is it possible to make it soft?\nSpeaker C: Yeah.\nSpeaker B: It's like a plastic cover.\nSpeaker B: So, I think it's so.\nSpeaker D: I don't know what you think, Bob, but it would be great for users, I think.\nSpeaker D: I think for sure.\nSpeaker A: Definitely the softer the better.\nSpeaker A: Yeah.\nSpeaker A: Cool.\nSpeaker C: Oh, it's the speaker.\nSpeaker C: I cannot imagine it so far, but it will be terrible.\nSpeaker D: Sitting buttons hidden in a...\nSpeaker D: Over remarks.\nSpeaker D: Something we didn't talk about, yes.\nSpeaker D: I think we're almost there.\nSpeaker D: Maybe how can we, if we have a soft thing like this, and to open it we have to attach it somewhere.\nSpeaker D: I don't know how to do that.\nSpeaker A: I could use Velcro.\nSpeaker A: Yeah, maybe magnetic.\nSpeaker B: Yes, it's a good idea.\nSpeaker B: Magnetic.\nSpeaker D: Magnetic?\nSpeaker B: Yeah, it can be a plastic cover sticking to the oldest stuff, like magnetic materials in the border.\nSpeaker B: So, it sticks like a refrigerator door completely.\nSpeaker B: And when you try to open it, it will be opened easily.\nSpeaker B: So, you can be sure that it will not be opened while you're commonly using buttons under the manner.\nSpeaker D: It could be the material of the first layer, I mean, like soft plastic.\nSpeaker A: Yeah, I imagine some sort of vinyl thing, in a sort of a... with a suede sort of feel.\nSpeaker C: Well, since the things like the screen and the solar cell has to be kept in shape, we cannot.\nSpeaker C: So, we need that appealing.\nSpeaker C: So, if the solar cell won't be visible while the banana is closed, then the material must be able to put the light inside, you know, so that we can reach the energy out of it.\nSpeaker A: So, your sign is possible to have a solar cell and this removable cover.\nSpeaker C: It is possible, but...\nSpeaker C: And a banana.\nSpeaker C: I don't know how to do it, if it's better to put the solar cell to the peeling, or inside, but then it must be some window.\nSpeaker D: And something we forgot maybe about the speech recognition system, now we're going to use it.\nSpeaker A: I think it's important, I think, one of our...\nSpeaker A: It's very big, something very bright.\nSpeaker D: So, we have the designs, a good shape, new and so on, and we have also the technology thing, it won't be.\nSpeaker D: And even the easy to use thing, so it will be perfect.\nSpeaker D: So, we had also the speech recognition device.\nSpeaker D: So, that's great, we have decided everything and I think, wow, in a good way.\nSpeaker D: So, the next meeting we start in 30 minutes.\nSpeaker D: So, the industry and manager will work on the look and feel design.\nSpeaker D: The user interface manager will work on the user interface device and the marketing expert will do the first product evaluation.\nSpeaker D: So, I hope you can do that in 30 minutes.\nSpeaker D: And, yeah, I think you should work together, you have made a picture to work in the first prototype, and more about the shape and so on, and how you can manage to put everything together.\nSpeaker D: And, as usual, you will receive some specific instructions and so on.\nSpeaker D: Thank you everybody and see you in 30 minutes.\nSpeaker A: Okay, thank you.\nNone: Thank you.\n"}, {"meeting_id": "ES2013c", "summary": "This meeting was a conceptual design meeting. In this meeting, the group mainly discussed the component, energy source, buttons, colours and styles of the remote control. They decided to use the kinetic as the energy source because of its convenience, light weight, fancy design and the good selling point of the environment, but they still needed more cost research. They also reached a final consensus to apply a hard plastic inner shell with a spongy plastic for the component, and push buttons for the type of interface due to their common features of convenience and user-friendliness.", "dialogue": "Speaker D: No, get this thing.\nNone: No it was up there and not here.\nSpeaker D: Now it's here and not there.\nSpeaker D: One more time, it should come up the right way around.\nSpeaker D: I'll just carry on.\nSpeaker D: This is the beginning of the third meeting, the conceptual design meeting.\nSpeaker D: Our agenda should be that we're opening the meeting.\nSpeaker D: I have the minutes from the previous meeting are on the shared drive at this point and we should each have a presentation to make.\nSpeaker D: We have certain decisions to make and we have 40 minutes total.\nSpeaker D: It's 25 after 2 at the moment.\nSpeaker D: So 40 minutes is 5 after 3, which I'll be keeping an eye on the clock for us.\nSpeaker D: They're the decisions we have to turn to, but we'll come back to them in a minute after I take us to the minutes of the previous meeting.\nNone: Right.\nSpeaker D: As we remember, I opened the meeting before us were present.\nSpeaker D: The first meeting's minutes were reviewed and approved.\nSpeaker D: You presented a marketing research report which pretty much represented that 15 to 35 year olds has to be hand held power channel volume number keys, possibly your speech recognition.\nSpeaker D: Then Steph did the second presentation that those functions plus streamlining them with big user friendly keys that were easy to use.\nSpeaker D: I think all of us agreed with those things.\nSpeaker D: Kate presented a working design of going after going over the basics on the whiteboard that it should be a simple mass produced device because of the 12 and a half pence cost.\nSpeaker D: But we did talk about possibly using rechargeable batteries and having a docking place as a selling point and the new requirements that it would be for TV only and that it includes the slogan and color of the corporate design being included.\nSpeaker D: The corporate image.\nSpeaker D: So we agreed that the target market would be 15 to 35 with more money than cents that were decision makers alternatively would be a manufacturer to enclose it with TV but it still should meet those parameters.\nSpeaker D: And the function we agreed was volume power numbers, enter channels, a way to move between channels, easy to use and hand held.\nSpeaker D: At that point we agreed that Sarah would look at the current cost of competition.\nSpeaker D: The current one is self-worth and Steph was going to look at ergonomics.\nSpeaker D: Kate was going to look at cost and feasibility of the various possibilities that we discussed and I was to type up these minutes and work on the final report.\nSpeaker D: Is this a fair presentation?\nSpeaker D: Yes.\nSpeaker D: Our last meeting was.\nSpeaker D: Yes.\nSpeaker B: Okay.\nSpeaker D: Right.\nSpeaker D: So we're ready to close that and go back to our that one.\nSpeaker D: Right.\nSpeaker D: We're up to the point of the go back.\nSpeaker D: The three presentations.\nSpeaker D: So we're going to pull the plug on me and turn to Sarah.\nSpeaker D: Is that okay?\nSpeaker D: Is that right with everybody else?\nSpeaker D: Yep.\nSpeaker D: Especially since Kate asked to be last.\nSpeaker D: Sarah, I'm sorry if I misspelled your name.\nSpeaker D: I didn't know what this S.A.R.A. or S.A.R.H.\nSpeaker D: You respond to whatever you get.\nNone: Okay.\nSpeaker D: Did you do your yes to it?\nSpeaker D: Ah, there it is.\nSpeaker A: Ta-da.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: I'm glad to see the marketing budget is being so well spent.\nSpeaker A: I think I appreciate the time to figure out Joyce has to go in.\nSpeaker A: Okay.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker A: I think it's close.\nSpeaker C: I think that's a good idea.\nSpeaker D: Don't you?\nSpeaker D: Yes.\nSpeaker D: It sounds like the covers that they use for the remote,\nSpeaker A: you know, your cell phone. Exactly.\nSpeaker A: I think that's something that's not a good idea.\nSpeaker A: It's something that's not a good idea.\nSpeaker A: Hmm.\nSpeaker A: They actually make the use of spongy.\nSpeaker A: They can be in the midst of a fact.\nSpeaker A: They can start it.\nSpeaker A: And breathe.\nSpeaker A: Breathe.\nSpeaker A: Breathe.\nSpeaker A: Breathe.\nSpeaker A: Breathe.\nSpeaker A: It's an active thing for me.\nSpeaker A: It's a lot easier to get in and the remote can go.\nSpeaker A: It's a little bit easier to get in the window.\nSpeaker A: It's spongy, it's a little bit further in mind.\nSpeaker A: And the answer is, while the function has been with everyone from outside.\nSpeaker A: What we're talking about is changing the concept of everyone's happy to be the lead.\nSpeaker A: But then we are in the fact that we need a lead.\nSpeaker A: We are in the fact that we need a lead.\nSpeaker A: Okay.\nSpeaker A: That's the common concept of the concept of a lead.\nSpeaker D: Hmm.\nSpeaker D: Hmm.\nSpeaker D: Um.\nSpeaker D: In most families don't, isn't the remote use a remote?\nSpeaker A: But I think that this is a lot of things that have to be written in the past.\nSpeaker A: Hmm.\nSpeaker A: So when your dad's sitting there overriding your decision going, no, we're going to watch this.\nSpeaker A: You can bring out your own remote and be like,\nSpeaker B: is that? No, we're going to watch this.\nSpeaker B: Well, actually some households do have three and four\nSpeaker D: TVs and they would have a remote for each one.\nSpeaker A: So this is a good idea. I think that's the reason that the open sound is being a little bit spongy.\nSpeaker A: Yeah.\nSpeaker C: I think we're going to have some trouble when we get down to the component design.\nSpeaker C: Spongy is going to be difficult, I'm afraid.\nSpeaker B: Hmm. And as for, as for, um, well, budgeting as well. If we're going to have lots of different interchangeable components.\nSpeaker C: Is this ear to the market? Yeah. I mean, basically we can make these things out of wood, titanium, plastic or rubber. I suppose rubber is the closest to spongy.\nSpeaker C: I was thinking titanium.\nSpeaker B: I was thinking, it's just, I have been influenced by pictures of\nSpeaker A: the slides and they're also minimalistic, shiny.\nSpeaker D: Are we talking about the device itself or the, or a cover for the device that be an interchangeable cover as a separate product?\nSpeaker A: Well, it would be a very simple product.\nSpeaker D: Perhaps that's, that particular suggestion, you need to go back to management and perhaps go to another group to actually design as a separate product.\nSpeaker D: Let's delegate.\nSpeaker D: Would that be agreeable? Yeah.\nSpeaker C: Would it be helpful if I describe the components a bit because I think it would give you, um, maybe bring this discussion back to earth of what we can actually physically do.\nSpeaker D: Yeah, maybe. Do you want it if you next or you want Kate to go next?\nSpeaker B: I think possibly I might be more useful. Kate went next.\nSpeaker B: Okay.\nSpeaker B: You can even have them in different flavors as well.\nSpeaker B: So if you just want to sit there and chew on the remote, it could be like\nSpeaker A: a pair of flavors. Yeah. Scratch and sniff.\nSpeaker C: Right. Well, I think some of this, um, you're going to be a little disappointed with some of the things I have to tell you, but I'm afraid this is the real world.\nSpeaker C: So, um, I've been looking at the, the basics of how these devices actually have to work in order to operate.\nSpeaker C: I've had some discussions with the manufacturing division who have told me what's actually available, what the current state of the art in components is and some of the exciting new things they've got, but I'm not sure of it.\nSpeaker C: It's quite, um, this isn't a very good overhead, but this is just to show you.\nSpeaker C: This is the earrings of a remote control.\nSpeaker C: I really need a pen or something, but I don't know how that works now.\nSpeaker C: Um, oh yeah. Can you see my little mouse pointer?\nSpeaker C: This is, this is a remote that's been opened up.\nSpeaker C: And that's the, the back of the interface.\nSpeaker C: And this is a push-button one. So you see these little, little buttons here, little rubber, rubber buttons that go through onto the board of the back. And they push these buttons here.\nSpeaker C: Um, and we, that's the basic construction that we've got to, got to accommodate. We've got to have something that pushes the little buttons that, um, talk to the chip that encodes the message that sends the message to the receiver.\nSpeaker C: So, um, I want, I want to go through not, not just addressing the, um, the points that you made, Sarah, but, um, doing my presentation in the order I wrote it.\nSpeaker C: So, first of all, um, I want to talk about what possibilities we've got for the energy source. Um, we can have your bog standard double A batteries, you know, replaceable little compartment. We can have a hand that's a wind up, a wind up, yeah. Which I think is quite an interesting concept for a, a remote control, but it maybe doesn't quite go with the, um, the fruit and veg.\nSpeaker C: Alright.\nSpeaker C: One, one that I think is quite interesting is the connection in the energy source where, um, you, you actually get the energy by moving the device, which is kind of like a nice neat one. You have to, it means that if it's sitting there for a long time, it probably won't work, but you have to sort of throw it between your hands every now and then. Or we, we had talked about solar power, but I think we agreed that that's not so good in the dark.\nSpeaker C: Now, I'm afraid this is the options we've got on, on the case. It can be made of plastic rubber, would if you like.\nSpeaker C: Or titanium. Um, and the nearest we've got to sponge you there, I guess is rubber, but, um, I'm going to come back to the advantage of titanium. And basically it can have a flat surface, a curved surface, or a double curved surface, but I think if we want to use standard components, we're going to have difficulty with anything much beyond that.\nSpeaker C: Okay, what does the interface look like? Um, well, push button, that's, that's the one we're all familiar with. Um, we can have scroll buttons and the scroll button can incorporate a push, so you can have it like on a mouse, where you can use it to go up and down, change volume or channel or something like that, and push to select something. Um, you can have multiple scroll buttons. Um, this is maybe getting a little bit complicated, but um, it's, if the technology is there. And we can also incorporate an LCD display in the remote, but this will increase the cost.\nSpeaker C: Um, the electronics actually makes the device work. We basically got three simple regular advanced and the price goes up, as we choose each of this. If we want the nice cheap one, the simple, then we can only have push buttons. All the other fancy interface designs go out the window, right? Um, pay a little bit more for a regular chip and you can have scroll buttons. If you want the advanced chip, it obviously costs more, but that's what you need if you want the LCD display. And the manufacturing division tell me that they have recently developed um, sample sensor, sample speaker devices. Now, I don't know what that is, but I think they think it's quite important and we might want to incorporate it somewhere. Um, our real expertise is in push buttons, I have to say, but maybe you think that's old technology. And well, I think we've got two options. We can either go for a really cheap model, people if we cost down, um, which means a flat plastic case with an ordinary battery and simple push buttons, or we can have something that looks a bit nicer, I think. Um, it might look like a pineapple, but um, they all may not have.\nSpeaker C: Um, but it could have an LCD screen and it could have multiple scroll buttons and it could have the company's new development of the sample sensor and speaker.\nSpeaker A: So, thank you.\nSpeaker A: Yeah, I'm\nSpeaker C: afraid I don't have that information available. Um, manufacturing can't actually give attention to any prices to any of this in the front.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Bare in mind it has to be the colors and styles of the company. So,\nSpeaker B: what I had been thinking was something chunky and yellow and plastic with black buttons with a logo on it. But, um, I don't know how important that is to keep it exactly the colors of the new reaction company.\nSpeaker B: It's just what I had understood. We've been going, that's before I heard about all this hot, hot, tips for the future fruit rice. We could, we could do a double curved\nSpeaker C: rubber one which would allow, um, say, a banana. But, um, unfortunately I see from my notes that if we do that we have to have a push pull as the interface. We can't do anything else here.\nSpeaker B: Shall we wait till I, so I've showed you what, well, my extensive presentation on what sort of interfaces are available.\nSpeaker B: Okay.\nSpeaker D: Thank you, Kate.\nSpeaker B: You did seem to include just in more detail what I've got though.\nSpeaker B: So, it's a bit pointless.\nSpeaker B: Right.\nSpeaker B: And, uh, yes.\nSpeaker D: Okay.\nSpeaker A: I don't have a minor.\nSpeaker C: Do it again.\nSpeaker D: Keep doing it until you get it in both.\nSpeaker D: You get it there, you get it yours.\nSpeaker D: And then you get it for both.\nNone: Just do it again.\nSpeaker B: Yeah.\nSpeaker D: Okay.\nSpeaker D: This time I should come up with this.\nSpeaker B: Right then.\nSpeaker B: I don't actually have a huge amount of different information.\nSpeaker B: Can I say just that my method was to look at the, well, my, my inspirations, which was a big collection of lots of different models of remote controls and other things that use a similar sort of thing, including MP3 players.\nSpeaker B: Like, you know, high-five robots, not just television, these things.\nSpeaker B: I mean, I think about the feature of each one is what functions we actually need and then how to group and arrange these on the actual, the actual plastic or rubber handheld piece.\nSpeaker B: I've been especially interested in the iPod style scroll wheel, which, right, I couldn't find the picture of the iPod on the link to web browser, I couldn't copy and paste it.\nSpeaker B: But it had a similar thing to this thing on the right.\nSpeaker B: It has scroll wheels without a display, but they scroll like a computer in the house.\nSpeaker B: I was thinking that a really simple interface was just a couple of scroll things on it.\nSpeaker B: And then instead of a display, the display could appear on the television screen.\nSpeaker B: Like, I guess, an existing sky or cable one does.\nSpeaker B: Like, you know, you press enter and it comes up with what's showing on that channel at the moment.\nSpeaker B: And you can do, you can scroll along and it will show you what's on the next half hour.\nSpeaker B: And you scroll up and you can see what's on the other channels while you're watching.\nSpeaker B: The same channel on the screen.\nSpeaker B: But I'm also eagerly taken with this chunky plastic kitty remote, which is really nice and fun and good to hold and nice big, easy buttons to press, but still quite simple and quite cost effective.\nSpeaker B: So what seems kind of ideal for me would be to just have a fairly simple, not fancy, but not totally minimalist.\nSpeaker B: I mean, just pretty simple plastic.\nSpeaker B: Probably I was thinking yellow and black just because that's the company's colours with very, very few buttons, but that would correspond to a screen that would appear on the television screen.\nSpeaker B: Like, you know, just small along the bottom.\nSpeaker B: Instead of having the iPod style display screen on the actual remote, which is quite expensive.\nSpeaker B: And when you've got a screen sitting there in front of you, you might as well have it appearing on the screen in front of you.\nSpeaker B: It doesn't obscure much of the actual picture you're watching.\nSpeaker B: And so on that you can just match in the same way as an existing sky remote.\nSpeaker B: Scroll along, scroll up and down.\nSpeaker B: And I definitely support these plastic and all rubber.\nSpeaker B: I mean, titanium would be great, but I think it might be a bit too expensive.\nSpeaker B: And too, kind of spacey. If we're trying to keep it colourful and slightly organic, then titanium is too futuristic.\nSpeaker C: It's difficult to make fancy shapes in it as well.\nSpeaker C: We can't pretty much have a double curved case in titanium unfortunately.\nSpeaker B: Whereas the plastic's so cheap and easy and mouldable and everything else.\nSpeaker C: Then the having the screen on the television screen I think is a good idea because otherwise we're putting up the price, not only for putting building the LCD in, but for the electronics as well.\nSpeaker B: And it's a little bit pointless as well I think.\nSpeaker B: When you've got the screen there, it doesn't have to be anything fancy, just a little menu showing.\nSpeaker B: Yeah, a menu, going to a menu, it can have different options whether you want to change the settings, or the information about programme that's on at the moment.\nSpeaker B: I don't think that's too taxing to engineer.\nSpeaker B: But as for actually arranging them, let me go back to the picture of the kitty one.\nSpeaker B: And I quite like it because it's just smooth and hand-held and it's got these easily reachable buttons that quite nicely spaced out, so something a bit like that where buttons arranged in a circle up down left right.\nSpeaker B: Do you want to scroll buttons in there as well?\nSpeaker B: I was thinking not actually scroll like a mouse scroll, but you know, I can't quite describe it.\nSpeaker B: I think right, you see on the one on the right down at the bottom, the mouse, yeah, see where the mouse is, like this style thing where you have up down left and right.\nSpeaker B: Right.\nSpeaker B: And enter in the middle, so you pick your menu and then your different options and when you click on each one, you can go into a new menu for that and get a bit specific here.\nSpeaker B: Really would have to use something to show you in nice part of design.\nSpeaker C: It's playing towards strengths which are in push-backing technology.\nSpeaker C: Here we are.\nSpeaker B: I think I can get up on the screen.\nSpeaker B: Here we go.\nSpeaker B: Right.\nSpeaker B: Well, the iPod spinning wheel is really complicated.\nSpeaker B: It does scroll, but it is hugely complicated.\nSpeaker B: But that's a good answer.\nSpeaker B: But they all have this feature of this.\nSpeaker B: It's not quite a scroll wheel, but it's a selection.\nSpeaker B: It's a kind of selection in this circle which I think is a really good idea, but we can implement it with simple push-backing.\nSpeaker B: Yeah.\nSpeaker B: It's just much cheaper, like up down left and right.\nSpeaker B: Which is good.\nSpeaker B: And then, yeah, so I mean either channel up and down, volume up and down, next up here in programs up and down.\nSpeaker B: And then also when you get to the different menus on the screen, it's got things like settings and contrast.\nSpeaker B: So all you really need is these four buttons with the enter in the middle.\nSpeaker B: Okay.\nSpeaker B: That's it.\nSpeaker C: In fact, look at this one.\nSpeaker C: Oh, that's really nice.\nSpeaker B: Almost impossible to miss place or lose.\nSpeaker A: Might take up your whole living room.\nSpeaker A: Possibly.\nSpeaker C: We could actually do it in rubber instead of plastic if you rather.\nSpeaker A: Or call it a banana.\nSpeaker B: Okay.\nSpeaker B: Is there something else I need to bring to your attention?\nSpeaker B: Have you seen these on a remote?\nSpeaker B: This is actually the volume up and down, but they both say V on them.\nSpeaker B: Which, when you first look at it, you expect that to be the down because it looks like a downward pointing arrow.\nSpeaker B: But it's actually the V to indicate that it's the volume.\nSpeaker B: So we do need to avoid little ambiguities like these.\nSpeaker C: Yeah.\nSpeaker B: But we wouldn't have a specific volume up and down.\nSpeaker B: If we're having this scroll, then it's going to double up as all the other up and down functions.\nSpeaker B: Okay.\nSpeaker B: That's complicated.\nSpeaker D: I want to thank you all for all your presentations.\nSpeaker D: We have about 10 minutes left in which time we have to make some decisions on the remote control concepts.\nSpeaker D: And I think you all have been self stimulating of working together.\nSpeaker D: We need to come up with some specifics of the components, the materials, things like making the decision on the energy and the case and the interface type things.\nSpeaker D: So let's.\nSpeaker D: Right.\nSpeaker D: They want us to decide what form of energy are we going to use, which the choices that we've discussed were pretty much battery, kinetic or solar.\nSpeaker D: This is where Kate's expertise comes in.\nSpeaker D: And our decision making will be a little bit guided by Kate at this point.\nSpeaker D: The you were saying that the kinetic would be useful.\nSpeaker D: That is you just have to move it so many you'd be able to use it as opposed to battery that you have to either put it in a recharger or keep replacing batteries.\nSpeaker D: Or solar that you have difficulty with it.\nSpeaker D: If it's a dark day that will die on you and no way to do it.\nSpeaker D: That's the day you want to use the TV.\nSpeaker D: So what's our pleasure here? What would be the cost consequences of each of the three?\nSpeaker C: Well, unfortunately I don't have constant information.\nSpeaker C: Okay.\nSpeaker C: In terms of workability, I think the two front runners are the standard battery or the kinetic.\nSpeaker C: Okay.\nSpeaker C: I'm going to show you. I don't have to figure out how to put some cost.\nSpeaker A: What's the feeling for the group on the kinetic?\nSpeaker B: It sounds great. I've never come across it before, but it sounds fantastic.\nSpeaker B: It sounds like it could be really good, economical.\nSpeaker B: It would make the whole thing a lot lighter.\nSpeaker A: Hmm.\nSpeaker A: Yeah.\nSpeaker B: It's really quite attractive thought, isn't it?\nSpeaker D: Yeah.\nSpeaker D: So the consensus seems to be used to kinetic, gifted to all possible.\nSpeaker D: But it does depend how much it costs.\nSpeaker B: I mean, how much it costs and how much more development and research it needs.\nSpeaker C: And how much you do have to keep it moving.\nSpeaker C: I'm conscious that watching TV can be called as a disability.\nSpeaker B: Do you exercise as well? You're watching the TV.\nSpeaker B: Yeah.\nNone: More environmentally friendly.\nSpeaker A: Yeah.\nSpeaker D: So I just think it is a thanks for the monger for five minutes to finish.\nSpeaker D: The next thing that they wanted us to do was look at chip on print as a decision.\nSpeaker C: Oh, right. Okay.\nSpeaker C: That's something I maybe should have covered.\nSpeaker C: Chip on print is just a manufacturing technique.\nSpeaker C: Oh.\nSpeaker C: And I would certainly recommend it, I think, because I'm not sure I have an alternative.\nSpeaker C: Okay.\nSpeaker C: What about the way we progress?\nSpeaker C: Oh, the way we progress.\nSpeaker D: Oh, okay.\nSpeaker D: What about the case?\nSpeaker D: I think they're talking there about do we want wood plastic titanium or rubber?\nSpeaker D: And I think we've discussed not having titanium one is too expensive.\nSpeaker D: And second, it won't do this double curves.\nSpeaker D: So we've got a lot of aluminum with sort of eliminated wood.\nSpeaker D: You said plastic or rubber.\nSpeaker C: What's the pleasure?\nSpeaker C: Well, if there's key spongy feel, the rubber that we're talking about is the same as you have in those little stress bones.\nSpeaker B: So what about you?\nSpeaker B: Do you not think it might need a kind of hard plastic in our shell with the actual delicate workings inside?\nSpeaker B: And then a kind of spongy cover.\nSpeaker B: Yeah, it's thick, spongy cover.\nSpeaker B: So it feels like the whole thing is spongy, but actually you're not damaging anything by plastic in size.\nSpeaker C: Because I mean, I carried away with the texture to get about whether that's actually possible, but I agree it sounds like a nice idea to produce.\nSpeaker B: You didn't get that carried away with things that attack tile.\nSpeaker B: Hmm, it's still good.\nSpeaker D: The next part they want is the user interface concept.\nSpeaker D: I'm sorry to push you, but we only have a couple minutes to finish with.\nSpeaker D: Okay, and it says interface.\nSpeaker D: What type and what supplements?\nSpeaker B: I'll just copy the one on the left.\nSpeaker B: Well, like four buttons up, down, left and right, with enter in the middle,\nSpeaker C: that will correspond to a menu on the screen. Which I think technically is just push button.\nSpeaker C: Yeah, I'd like to push button.\nSpeaker C: It's something we're going to accompany the next one.\nSpeaker D: So push bones.\nSpeaker D: Okay.\nSpeaker D: This is going to sound weird, but the next meeting starts 30 minutes.\nSpeaker D: And they actually want a look and feel design user interface design.\nSpeaker D: You can look and see this as well as I can marketing.\nSpeaker D: They want product evaluation.\nSpeaker D: Oh, oh my, I'm sorry.\nSpeaker D: Oh, okay.\nSpeaker D: Sorry about that.\nSpeaker D: Missed that one.\nSpeaker D: This ought to be fun trying to get this thing to work.\nSpeaker B: Ah, ta-da.\nNone: Ah.\nSpeaker C: Yeah, I think the company is the next project should design a better overhead device.\nSpeaker D: Yes.\nSpeaker D: Because I can't even see mine.\nSpeaker D: Next meeting starts in 30 minutes.\nSpeaker D: These are the individual actions.\nSpeaker D: Yeah, right.\nSpeaker D: The look and feel design is for Kate.\nSpeaker D: Steph gets the user interface design.\nSpeaker D: You give product evaluation.\nSpeaker D: The two of you get to play with the plot modeling play to do a prototype.\nSpeaker D: And everybody gets individual instructions in the usual way.\nSpeaker D: Anything else we need to do?\nSpeaker D: Go to it.\nSpeaker D: And that's the end of this meeting.\nSpeaker D: That's for her benefit.\nSpeaker C: No.\nNone: No.\nNone: No.\nNone: No.\nNone: No.\nNone: No.\nNone: No.\n"}, {"meeting_id": "ES2012c", "summary": "This is the third meeting of the design group. Project Manager raised the question left over from the last meeting. That is, what kind of energy source should be used. Industrial Designer preferred a combination of solar cells with a back-up basic battery, but Project Manager rejected and decided to use a long-term battery in light of the production cost. Then Industrial Designer described the advantages and disadvantages of custom-designed chips and chips off the shelf. Due to the significance of technological innovation, the team decided to choose the former. After discussion, the team agreed that they needed buttons for channel change, power, menu and volume, and proposed to add a slide button. In terms of the appearance design, Marketing proposed to follow the fashion and use a spongy material. In order to keep up with the changes in fashion, the team recommended interchangeable covers.", "dialogue": "None: All right.\nSpeaker A: Thank you very much for coming to this third meeting of the design group.\nSpeaker A: I think we made some definite progress to the last one.\nSpeaker A: And come up with some interesting suggestions for our new remote control.\nSpeaker A: Again, very quickly, just present some notes of that meeting.\nSpeaker A: Now, the problem with existing remote controls we felt was that they were ugly and that people who prepared premium for something better.\nSpeaker A: There's got lots of buttons on them that people don't use and find difficult to learn and people lose them.\nSpeaker A: We thought that for our new remote control that everybody will want to rush out and buy, that we should look at speech recognition rather than buttons.\nSpeaker A: And that if you have any buttons, there should be very few of them and only for those functions that are actually identified that the people use.\nSpeaker A: We want to go for a long lasting battery that we guarantee for the life of the product and a shape that will be instantly recognizable, A, as a trendy remote control, and B, as a real reaction product so that when people are happy with that, they will want to buy everything else from us.\nSpeaker A: So, again, we'll have the three presentations from the three of you and then we'll make a final decision.\nSpeaker A: And the decision that we need to make today find NER, what energy source we want to use, whether it is practical to use a long lasting one.\nSpeaker A: I think our discussion was around the fact that if we're going to go for a long lasting power supply then basically it's sealed for life.\nSpeaker A: If anybody has managed to run one down, we'll give them another one.\nSpeaker A: It will be prominently displayed as part of the advertising literature that it's for life, guaranteed for life.\nSpeaker A: Now, the internal chip, and this is where I need Kate's expert advice and given that this has to go to market as quickly as possible, do we go for a custom-designed chip or do we buy one off the shelf and program it ourselves?\nSpeaker A: I mean, I'm not an expert on these things, but presumably there must be loads of them already on the market that we can modify, but that's your area of expertise.\nSpeaker A: And then the overall design of the case is Kendra's field.\nSpeaker A: We had some discussions last time as to how we might go forward, and we'll finalise those today.\nSpeaker A: This is all linked in with the user interface, whether we go for voice buttons or a bit of both.\nSpeaker A: And then for the next meeting, Kate will be looking at the look, feel, and design, Kendra, how the user actually uses it, and of course the product evaluation.\nSpeaker A: Kate and Kendra will be producing a model for us to look at.\nSpeaker A: So if we can have the three presentations again, please, and perhaps we'd like to start.\nSpeaker B: Kate, what was I saying?\nSpeaker E: Okay, I'll just be talking about the components design.\nSpeaker E: Okay, basically I just looked at what exactly do remotes do.\nSpeaker E: Basically they wait for you to press a key or give a voice command, and then this is translated into light signals, which are then seen by the TV.\nSpeaker E: The materials we're going to need to look at.\nSpeaker E: The two big ones are the integrated circuit chip and the battery, and the integrated circuit chip works in conjunction with a diode, a transistor, a resonator, two more resistors, and a capacitor.\nSpeaker E: And the battery works in conjunction with a resistor and a capacitor.\nSpeaker E: Basically what happens is you'll press a number or give a voice command, and this creates a connection within the remote.\nSpeaker E: And it allows the chip then senses this connection and produces a signal in the Morris code format.\nSpeaker E: This signal is sent to the transistor, which amplifies it, and then sends it on to the light emitting diode, which is then changed into infrared light, which is sent to the TV and sort of seen by the TV, and which changes the channels.\nSpeaker E: So as for how we should end up using this in our remote, a couple of main questions are the buttons.\nSpeaker E: The fewer buttons you have, I guess the few internal connections and internal codes you're going to need.\nSpeaker E: However, to not have buttons, or to use a voice command instead of buttons, might make these connections more difficult and raise the production cost.\nSpeaker E: That's something we should think about.\nSpeaker E: Also, we have to work within the company constraints, and the company has informed me via email that there are experts at push buttons, and that seems to be the most cost-effective way of producing it.\nSpeaker E: Also, with battery connections, the company has some limits on the batteries we can use, so I was thinking perhaps a combination of solar cells with a backup basic battery, and somehow between the combination of that two, we might be able to come up with something that will last the lifetime, or the five to ten years, and we could still keep that original idea.\nSpeaker E: We also need to look at the chips, custom-divine versus off-the-shelf, and the custom design will give us much more flexibility and enable us to incorporate the voice function that we all seem to have agreed upon.\nSpeaker E: However, that's going to cost more, but the off-the-shelf is going to be cheaper, and it's going to be allowed us to produce it quicker and get it out there faster, but it's going to be less flexible with the features, especially things like voice activation, which haven't really been used in the company.\nSpeaker E: There's not really chips out there, that would be easy to convert, so if we were definitely going to go with the voice option, we'd probably have to design our own chip.\nSpeaker E: That pretty much sums it up.\nSpeaker A: Sorry, can you just put that one back up again, please?\nSpeaker E: I mean inevitably a custom design chip is going to be more expensive. Do we know by hand-notch?\nSpeaker E: I don't actually have any price information.\nSpeaker A: Do we know how long it will take to develop a custom chip?\nSpeaker E: A lot longer than an off-the-shelf chip. The problem is the voice technology is not really highly developed, it's still in an experimental form, so it's hard to predict the time.\nSpeaker A: Right, I think we need to make a decision here, given that the company wants us on the market quickly and cheaply, that would appear to effectively constrain us to an existing chip, unless they're for conventional button technology.\nSpeaker A: Before we go around, does anybody have anything to say about that?\nSpeaker C: I just have a question about that. Does it make a difference if there are just a few commands?\nSpeaker C: For example, if you can pre-program in numbers 1 through 10 and pre-program say 9 channels and then just use the voice recognition to say channel 1, and then you've programmed in, say, BBC4 as your channel 1 as your favorites.\nSpeaker C: I'd like to have a certain number of favorites.\nSpeaker E: To incorporate the voice activation in it is sort of the trick. Once you've got the whole voice chip in there, then it's pretty much the world, the sky is your limit, but the big step is to actually get the voice activation chips in there and working.\nSpeaker A: I find that it's surprising, given that mobile phones incorporate voice activated dialling.\nSpeaker A: This information is from the internal information.\nSpeaker A: Of course, mobile phones do tend to be more expensive.\nSpeaker A: Because mobile phones are expensive, you can't say it's the voice recognition that these, but we don't know.\nSpeaker A: Given that the technology is not well developed and given that it's never been done before, with the double risk, perhaps we ought to stick to two buttons.\nSpeaker A: Since the last thing we want to do is present a project that doesn't work.\nSpeaker B: Thoughts?\nSpeaker D: Another thought, oh sorry, go ahead.\nSpeaker D: I was just going to say that maybe we can cut cars somewhere else to bring them to the local cost.\nSpeaker A: Do we think the voice technology is fundamental to the project?\nSpeaker E: It's something we've discussed since the beginning.\nSpeaker E: I think in our minds it's fundamental, but I don't know that the upper echelons of the company would necessarily agree with that.\nNone: Thank you.\nNone: Thank you.\nSpeaker A: Thank you.\nSpeaker A: Thank you.\nSpeaker C: I think we're going to have the voice recognition for the whole thing.\nSpeaker D: I think it's a second most important aspect to you, so it should be technologically innovative.\nSpeaker A: If that means we've got four buttons, but I mean the same question, do we need the five buttons for channel change, up down, volume up down, and on off, just as a backup?\nSpeaker E: I would say we do.\nSpeaker A: Okay, sorry, did you want to say anything?\nSpeaker C: Yes.\nSpeaker B: It's going to work.\nSpeaker C: Okay, so I did some research on the internet and what, you know, the user interface are just the aspects that are seen by users, commands and mechanisms for the operation, and they're just kind of a variety of choices.\nSpeaker C: So, a lot of times they tend to look cluttered, and these were just a couple examples of different kinds that are a little bit more unusual. There are some special ones available, like this one right here, which is marketed towards children, different designs, and one of the things that we need to watch out for is a V in volume because people, some of the other ones, I think a little picture of what I thought ours could look like.\nSpeaker C: So just kind of minimize the clutter, avoid too many buttons, and also one of the things that people have used is a slide button, like you have on a mouse, that possibly we could use that on the sides for volume, for example, have a slide button on the side, and then you can pre program the channel's voice recognition, and then the voice response sample, okay, to interrupt, okay, do you understand anything about slide controls? I mean, I think the reason they're used is push buttons is that they're simple, cheap, and I think they're about the same cost, really.\nSpeaker E: I mean, I think it's just sort of, there's a lot of slide buttons out there, I think it's pretty much the same sort of connection.\nSpeaker B: Fair enough, fine.\nSpeaker C: Just because I'm, for example, if I'm using a mouse, I like to be able to slide it up and down, so I thought it might be good for volume to just be able to kind of roll it, and then have the up and down, and then the...\nSpeaker A: So there's three buttons in the slider.\nSpeaker A: Three buttons, the channel up there.\nSpeaker D: If you've got a channel up down, you'd have a slider on that as well, because if you know what's on the thing, it kind of like it's got kind of...\nSpeaker D: It kind of like sticks, I mean, up like one unit, if you're saying me?\nSpeaker D: Yeah.\nSpeaker D: So it kind of goes up one, then you can keep rolling up, but it's like a cog or something, so you kind of take it up one at a time.\nSpeaker C: The only advantage I was thinking of to having the buttons on one side for the channel, and then the slider is that if you're just holding it in your hand and you pick it up, it's easy to know, okay, this is just a volume.\nSpeaker C: That's quite a lot of what you're thinking.\nSpeaker E: But it doesn't like pick up remote, so you're meant to change the channel and turn the volume, or the rest of it, so it would be kind of good to have and feel completely different.\nSpeaker E: You'd know what you were fiddling with.\nSpeaker E: Yeah, I think she'd have a bit of a mouse with that.\nSpeaker A: I mean, this is what we have to come up with, is the actual shape that people instantly pick it up and know what it's going to do.\nSpeaker A: Yeah.\nSpeaker A: Okay, so we're looking at sliders for both volume and channel change.\nSpeaker C: Well, I was thinking kind of just for the volume, but what do you guys think?\nSpeaker D: I don't know, depending on the final shape of it, because you could have like, I don't know, where it is there, it looks like you can control the volume with your thumb, and then you can control the buttons.\nSpeaker E: Yeah, in that kind of position, the fingers would be better for pressing.\nSpeaker A: Yeah, I mean, it seems to me that it also has the advantage that the two are clearly different.\nSpeaker A: Yeah.\nSpeaker A: And there's no possibility of confusing the two.\nSpeaker A: So, okay.\nSpeaker A: That's just a long.\nSpeaker A: Right.\nSpeaker A: So, that's, sorry, that's all you.\nSpeaker A: Yes.\nSpeaker A: Okay, fine.\nSpeaker C: Yeah.\nSpeaker D: Oh, that's it.\nSpeaker D: There we go.\nSpeaker D: Yeah, this is my important trend watching.\nSpeaker D: The digital is from Internet from Executive Summary for us on the top three things once it's back.\nSpeaker D: You can see my, I've got reports from a past plan on new fashions.\nSpeaker D: And the most important aspect is the look, it has to look fancy looking fuel.\nSpeaker D: Instead of the current functional looking fuel, this is, this is a twice as important as the second aspect, which is the 10, logically innovative side of it.\nSpeaker D: So, on the third being easy to use, it's probably a given before you have to try to incorporate.\nSpeaker A: So, well, I mean, I think that's what we're suggesting to address all three of those.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nNone: Nice.\nSpeaker D: Yeah, sorry.\nSpeaker D: Yeah, and from the fashion watches from on, I have said, for vegetables are an important thing for clothes, furniture, shoes, etc.\nSpeaker D: So, clothes, shoes, furniture.\nSpeaker D: And a spongey material to be used on the outside.\nSpeaker A: I'm not sure that's different, so, yeah.\nSpeaker D: But I was going to say, yeah, for vegetables, important to this year, important to furniture.\nSpeaker D: I just can't say, like, fashion, we're going for the, it looks fancy in fashion.\nSpeaker D: So, obviously, the way we're going to be going through it.\nSpeaker D: But, but, fashion don't last very long.\nSpeaker E: Well, that can kind of tie into our changing face things, like we could have vegetable themed this year.\nSpeaker E: Yeah.\nSpeaker E: Whatever happens next year, we can have the space.\nSpeaker D: Yeah.\nSpeaker D: A spongey skin on it.\nSpeaker C: And then we can just whip that off and...\nSpeaker C: Yeah, like the kind you get on my canned wipes, you know, that kind of spongey.\nSpeaker E: Yeah, weird, I don't know what that is, but yeah.\nSpeaker A: Yeah.\nSpeaker A: Also, we need to drop it without damaging it.\nSpeaker A: Yeah.\nSpeaker C: Yeah, that's true.\nSpeaker D: Yeah, we could, if we could save, depending on the cost of the product itself, you know, could we have a cheaper, a cheaper power source, and then just have it annually renewed?\nSpeaker D: Instead of having a ten year guarantee with interchangeable covers, we're going to buy a new one every year, a new one-wide new fashion's coming.\nSpeaker A: I mean, it's, it's an interesting idea.\nSpeaker A: It's like the old swatch watch, where only batteries have a lot changed in those, because people just bought a new one when it went to fashion.\nSpeaker A: It's just never been seen as a fashion item before.\nSpeaker A: Yeah.\nSpeaker A: Yeah, so if they're made in sufficient quantity.\nSpeaker E: I think it's easier sometimes to have them buy changeable covers for it than to buy a whole new one, because you don't feel like you're investing in a whole new product.\nSpeaker A: I mean, we can... I mean, my feeling is that current power sources are such that for relative to little cost, you can make it last, you know, a long time.\nSpeaker A: But if you also make it cheap enough that people are there changing the cover every year or even buying new one every year, then it's even better.\nSpeaker C: What if we included the batteries in the cover?\nSpeaker C: Oh, yeah.\nSpeaker E: So, that also encourages them to buy new covers.\nSpeaker C: Yeah, so can I see that?\nSpeaker C: And then just this is the example.\nSpeaker A: Yeah, I don't know.\nSpeaker A: The immediate thing that comes to mind with that is that you then got the connection and you've immediately got a source of unreliability.\nSpeaker A: Whereas the advantage of having it plumbed in is that the whole thing is all completely soldered together.\nSpeaker A: Yeah, I guess that's...\nSpeaker A: And it takes a reliability.\nSpeaker A: But I mean, I know what you're saying and understand where you're coming from.\nSpeaker D: But like more than just the battery, like a complete different, like you've only got like this bit to the bit you keep.\nSpeaker D: And this is expensive, but this is like the chip and this is the microphone.\nSpeaker D: And then this is the power source and the... but everyone sees.\nSpeaker D: And then...\nSpeaker A: I think if you're going to get it down that route, then we're talking about the cost slightly more than that.\nSpeaker A: Just building the whole thing in one.\nSpeaker A: Yeah, you're putting cheap production costs.\nSpeaker A: And giving people the option of buying a new, a complete new thing.\nSpeaker A: The advantage of a replaceable cover is that anything that doesn't have a cover on at all, it will still work totally.\nSpeaker A: And people lose the cover.\nSpeaker A: I mean, they might be well inclined to go out and spend however many euros on a new one.\nSpeaker A: Rather than a complete new...\nSpeaker E: Well, that's just it with the covers, you're sort of tricking them into continuing to spend their money on our products without making them feel like they're being ripped off having to buy a new product.\nSpeaker A: Yeah, I mean, it's up to...\nSpeaker A: It's up to...\nSpeaker A: It's up to our marketing people to make turn it into a fashion item.\nSpeaker A: And as external fashions change, then we get new covers on the market and readily available.\nSpeaker E: And that's the sort of thing, once you get the mold set, you can just whip out different colors.\nSpeaker C: If they have a fresh, fresh, fresh, fresh, fresh fruit and animal print and colors.\nSpeaker A: Yeah, okay.\nSpeaker A: Sorry, you can't finish the door.\nSpeaker D: I don't watch that.\nSpeaker D: I was just going to say, yeah, we're going to make this part of the fashion supply to furniture, so we're going to make this part to furniture.\nSpeaker A: It's the sort of thing that we want people to have prominently displayed on their coffee table to say, this says something about me.\nSpeaker A: This is fashionable.\nSpeaker A: This is fashionable.\nSpeaker A: I'm with it.\nSpeaker A: I'm up to date.\nSpeaker A: The design that I've got, then it could be a homemade design.\nSpeaker A: This says this is not just a television remote control. This is a fashion accessory.\nSpeaker A: So, I mean, the basic shape is what we give them that we've got.\nSpeaker A: One on off button, presumably two buttons for channel change and one slider.\nSpeaker A: And basically nothing else.\nSpeaker A: We need to decide on the, just the basic shape of the thing.\nSpeaker A: Now, whether we go down the fruit and veg roots, and I don't suggest we make it look like a banana.\nSpeaker A: That's what the organic curved look, the two deliberately get away from the square look of most currents, remotes.\nSpeaker A: Whether the fixed part of it is the corporate yellow.\nSpeaker A: Because there's something in there, the corporate logo needs to be prominently displayed so that people look at it and say, right, that's a real reaction.\nSpeaker A: Remote control, I want one of those.\nSpeaker E: I don't know that we should make the whole thing.\nSpeaker E: You'd have a yellow circle with the RR and it's on it.\nSpeaker A: Yellow seems a bit of a strong color to make me feel like the thing, no, but we should like wipe the gray black or some sort of black color.\nSpeaker A: We're simply required to incorporate the corporate logo prominently.\nSpeaker A: Well, this is the whole point.\nSpeaker A: Yes, I've got a real reaction remote control.\nSpeaker A: Here's my real reaction, copy maker or whatever.\nSpeaker A: And then people demand more real reaction stuff.\nSpeaker C: Well, I was sort of picturing maybe a shape that's almost like a mouse.\nSpeaker C: When they hold it, because a mouse is pretty comfortable to hold in your hand, maybe we could make it a slightly different shape so they could hold on to it.\nSpeaker C: But that way they can have the volume on the side and then the channel buttons, the power, wherever.\nSpeaker A: I mean, now it's just an idea that I had.\nSpeaker A: Well, as I said, my idea was probably a bit fatter than this, but that sort of shape.\nSpeaker E: You want to narrow it a mouse, because a mouse you're kind of just resting on it.\nSpeaker E: You want something you can definitely grip.\nSpeaker A: It's sort of a combination.\nSpeaker A: The mouse but held.\nSpeaker A: So you sort of hold it, you hand like that.\nSpeaker A: Fiddle around it.\nSpeaker C: Yeah, so yeah, maybe a little wider than this because this is pretty comfortable to hold.\nSpeaker E: Maybe almost like a hairbrush, you'd get the width of that end of the hand and it widens up top.\nSpeaker C: And then it would have a wider thing to have the light, the infrared light at the TV and just kind of change channels and adjust the volume.\nSpeaker C: The power could be wherever up at the top or something.\nSpeaker C: Cool.\nSpeaker C: What do you guys think about that?\nSpeaker D: Yeah, so it's okay, yeah, I'm just suddenly realizing that, yeah, we're discussing how much how good it's going to be to change channels.\nSpeaker D: We still can just tell it to.\nSpeaker D: We're starting to get away from the voice functionality of it.\nSpeaker C: Oh, yeah.\nSpeaker C: That's a very good point.\nSpeaker A: That is a very good point.\nSpeaker A: I think we decided that we were actually going down both routes.\nSpeaker A: Yeah, we need the manual controls and that they should be of that form.\nSpeaker A: Yeah.\nSpeaker E: Do you know this shape also kind of lends itself to voice-ness because if you're holding it and you're fiddling, which you can also bring it up like that and it's like a blessing.\nSpeaker D: Yeah, maybe we could, like we were leaving out the buttons for the forestry kind of used stuff.\nSpeaker D: We could incorporate that into the voice.\nSpeaker A: Honestly, my personal view is that if it's not there, people wouldn't use it anyway.\nSpeaker A: I suppose.\nSpeaker D: The areas for the off-chance, you know, the brightness is rough on your TV.\nSpeaker D: The contrast is changing.\nSpeaker A: It's certainly possible.\nSpeaker A: I mean, but we're going beyond, given the state of the technology, we want something that we know will work.\nSpeaker E: We can still incorporate the voice with less buttons.\nSpeaker E: I mean, if the power button was also somehow like a menu button, you could press that or do voice commands.\nSpeaker E: And the volume thing could also be to scroll through other options.\nSpeaker E: And you could scroll through brightness.\nSpeaker E: Yeah, I suppose you can minimize the buttons and still have those, you know, brightness and tint and stuff.\nSpeaker A: I mean, I'm getting a clear message that we think that we should have those facilities available.\nSpeaker C: We could have the channels and the power and then a menu button and then the volume.\nSpeaker C: So have four buttons and the volume instead of three buttons on the volume.\nSpeaker A: Okay, if we're going down that route, then we need some sort of display.\nSpeaker A: Do we need some sort of...\nSpeaker A: Yeah, that's the turn.\nSpeaker E: Yeah, that usually could probably be the menu.\nSpeaker E: And then hit menu and menu will come up on television and have like tint brightness.\nSpeaker D: And you use the scroll through it.\nSpeaker D: On to like a mouse, the scrolling button is actually a button as well. You could press it and press that.\nSpeaker C: Yeah, press that.\nSpeaker C: Yeah, that might be where you saw that work though, but yeah.\nSpeaker C: Yeah, it's like the mouse where you just kind of click it. You just press it.\nSpeaker C: Yeah, and you could just click that.\nSpeaker C: So if you had like the menu, then you could scroll through and then click it to select.\nSpeaker C: Yeah, you know what I mean?\nSpeaker A: Yeah.\nSpeaker A: Okay, we've got five minutes to go in this meeting.\nSpeaker A: So I think we've actually very conveniently just come to a good point to sum up.\nSpeaker A: So the Kate and Kendra now go away and play with a bit of plasticine or play dough or whatever it is on the other side of the Atlantic.\nSpeaker A: And actually what we've discussed into something, I was going to say concrete.\nSpeaker A: That's a slightly inappropriate word to something that we can see.\nSpeaker A: And Andrew can go away and think about how we can actually market this as a concept.\nSpeaker A: So not just a simple remote control.\nSpeaker D: Can I just get some things clear as front of you?\nSpeaker D: Yeah, of course.\nSpeaker A: Energy source is going to be long term.\nSpeaker A: I think we decided that we're going to, for simplicity of manufacturing and maintenance that we will go for a long term battery source.\nSpeaker A: On the basis that if you're going through making a fashion statement, then people are more likely to change it anyway before it runs out.\nSpeaker A: And making assumption that we can aim for a battery that will last most people for, I'll say at least five, ten years, and that we will guarantee it for five years.\nSpeaker A: And we're having a custom chip.\nSpeaker A: We're having a custom chip, but given we've cut the functions down, that will hopefully not be too problematic.\nSpeaker A: But given the technological innovation is important, then we need to say it again technologically innovate.\nSpeaker A: And we must resist any efforts to try and water that down.\nSpeaker D: And in the tangible case.\nSpeaker A: Interchangeable case seems to be important to the concept.\nSpeaker A: It should be cheap if we avoid any electrical connections.\nSpeaker A: If we can come up with some way of allowing people to actually personalize it to whatever they want, then this is totally new.\nSpeaker A: We don't know whether to what extent people do it or not, but if they've at least got a good selection of covers that they can use anyway.\nSpeaker A: And if we can keep them rolling, then get them in the supermarket when they go down to any of the famous supermarkets I wouldn't mention any names.\nSpeaker A: It's good for the supermarket, it's good for those, and it hopefully makes them feel better.\nSpeaker D: And are we going to have it being illuminated from inside onto the button?\nSpeaker A: No, because we've got to say a few buttons that it actually makes that redundant.\nSpeaker C: Yeah, and especially if we're making them so different to feel.\nSpeaker A: And that solves one of the problems of battery life, because that would pop up the battery life.\nSpeaker A: I mean, given the nature of the buttons we're having, it's actually unnecessary, I think.\nSpeaker D: Are we having it at any angle?\nSpeaker A: Or is it just as wide?\nSpeaker A: Certainly wider angle than current, so that if you're holding it anyway, like your 92.\nSpeaker A: And it will work most of the time.\nSpeaker A: Not like my mum who points it at the ceiling and why the television doesn't work.\nSpeaker A: I mean, I sort of had visage that if this was the thing, then sort of the hold of the top would be the infrared score.\nSpeaker A: I mean, the reality is people are they're going to be looking at the television whilst they're using it, the chances are.\nSpeaker A: So, it's all clear where we go from here.\nNone: So,\n"}, {"meeting_id": "Bmr009", "summary": "Postdoc C suggested a possible hypothesis that energy is increased as more people are speaking. New forms were discussed, as these will include more details that may have to be completed by analysts through interviews with participants. The Multitrans update by the graduate student seems promising, and has been sent for review. ", "dialogue": "Speaker B: starts no no there's another I don't know it starts with a p or something I forget the word for it but it's it's Typically when you but starting around 40 for most people It starts to harden and then it's just harder for the lens to shift things and the the symptom is typically that you You have to hold stuff further away to see in fact My brother's a Dermatological psychologist and he he came up with in a body-aged test Which gets down to sort of only three measurements that are good enough statistical predictors of all the rest of it and one of them is is the distance\nSpeaker A: We're alive by the way, so we got a good intro here. Oh, yeah, but oh, I am okay. Yeah, we can edit that out if you want No, it's okay. Okay, so This time the form\nSpeaker B: Discussion should be very short, right? It also should be later. Okay, because Jane is not here yet good point and That should be most interested in that It's probably least involved in the signal processing stuff so maybe we can just just I don't think we should go through an elaborate thing but Jose and I were just talking about the Speechless energy thing and I we didn't talk about the derivatives But I think you know the if I can if you don't mind my speaking for you for a bit Right now he's not really showing any kind of distinction, but we discussed a couple of the possible things that he can look at and one is that this is all in log energy and log energy is basically compressing the distances between things another is that he needs to play with the different temporal sizes He was he was taking everything over 200 milliseconds and he's gonna vary that number and also looking moving windows as we discussed before and The other thing is that the doing the Surracting of the mean and variance in the divided by the standard deviation and log domain may not be the right thing to do\nSpeaker A: Hi Jane, we just started. Did you take that might there?\nSpeaker B: Over the whole means of what between no between neither it's between the pauses for some segment and so he is He's making the constraint it has to be at least 200 milliseconds and so you take that and then he's Measuring at the frame level still the frame level what and then and then just normalizing with that larger amount And but one thing he was pointing out is when he he looked at a bunch of examples in log domain It is actually pretty hard to see the change and you can sort of see that because I was just putting on the board that You sort of have log x plus log x that's the log of x plus the log of two and it's just you know it diminishes the\nSpeaker E: Effective having two of them. Could you do like a CDF there instead? I mean we don't know that the distribution here is yes\nSpeaker B: Right so so what I was suggesting to him is that actually a pdf, but you know, but either way yeah Yeah, yeah Yeah, but I think also I think a good first indicator is when the the researcher looks at Examples of the data and cannot see a change in how big the the signal is when the two sweet then that's a problem right there So I think you should at least be able doing casual looking and get the sense Hey, there's something there and then you can play around with the measures and when he's looking in the log domain is not really seeing it So and when he's looking in the straight energy is that's a good place to start So that was that was the discussion we just had The other thing actually we can't had a question for Adam in this when you did the sampling Over the speech segments are sampling over the the individual channels north to the the Applicaticalization did you do it over just the entire everything in the mic channels?\nSpeaker A: You didn't try to find speech no, I just took over the entire entire entire channel\nSpeaker B: Sampled 10 minutes randomly right okay, so then that means that someone who didn't speak very much would be largely Represented by silence and someone who is who would be so the normalization factor probably is\nSpeaker A: Yeah, this was quite quick and dirty and it was just for listening okay for listening it seems to work really well\nSpeaker B: Yeah, yeah, yeah, but that's but it's not yeah, not a good matter Okay, so yeah, there there's a good chance then given that different people do talk different amounts Yeah, there is there there's still a lot more to be gained from gain normalization Yeah, that's absolutely if we can figure out a way to do it But we were agreed that in addition to that there should be stuff related to pitch and harmonics and so forth So we didn't talk at all about the other derivatives, but again, just just looking at I Think Liz has a very good point that in fact it would be much more graphic just to show Actually, you do have some distributions here for these cases you have some histograms and They don't look very separate\nSpeaker C: Separate this is the first debate to real free energy. Yeah, we thought to any kind of normalization Yeah, log energy. Sorry. These are the first pretty men's we come in to\nSpeaker A: Except it's hard to judge this because the they're not normalized. It's just number of frames. Yeah But yeah, you and so\nSpeaker E: What I mean is even if you use linear raw measure like raw energy or whatever maybe We shouldn't make any assumptions about distribution shape I'm just use the distribution to model the The mean you know rather than mean take some yeah, but and so Denise he's got that he's got some pictures\nSpeaker B: But he doesn't he doesn't do that he just in the derivatives, but not in the But he doesn't know what they look like\nSpeaker E: But he didn't have it for the energy for the drivies. Yeah, I mean there might be something there\nSpeaker F: Yeah\nSpeaker B: Interesting. Oh, yeah, that's a good did you have this sort of thing for just the just the The the unnormalized No, log energy. Okay. Yeah, so she's right. That's No, but it looks like yeah, that's that's because I mentioned scatter plots before but she's right I mean even before you get the scatter plots just looking at a single feature\nSpeaker C: Yeah, looking at the distribution is a good thing to do is cut to combine in different possibilities for the parameters I mean the the scatter plot combine in different True combination. Yeah, but what she's saying is which is right is young\nSpeaker B: I mean, let's start with the before we get complicated. Let's start with the most basic thing Which is we're arguing that if you take energy If you look at the energy that when two people are speaking at the same time usually There'll be more energy than one one is right. That's that's sort of hypothesis and the first way you'd look at that She's you know, absolutely right is that you would just take a look at the distribution of those two things much as you've plotted them here You know, but just but just just do it well in this case you have three of the silence and that's fine So with three colors or three shades or whatever just just look at those distributions and then given that as a base You can see if that gets improved you know or or or worsened by looking at regular energy looking at log energy We were just proposing that maybe it's you know tired of see with the log energy and Also these different normalizations do is a particular choice of normalization make it better But I had maybe made it too complicated by suggesting Early on that you look at scatter plots because that's looking at a distribution in two dimensions Let's start off just in one With this feature. I think that's probably the most basic thing for anything very complicated And then we I think we're agreed that pitch related things are are going to be a really likely candidate to help but since Your intuition from looking at some of the data is that when you looked at the regular energy that it did in fact usually go up When people were talking that's You know you should be able to come up with a measure which will match your intuition and she's right that having Having having this table with a whole bunch of things with the standard deviation the variance of it It's it's harder to interpret than just looking at the same kind of picture you have here\nSpeaker C: It could use but I found it in the in the missus fight That's several several times You have an speaker talking along We say high level of energy in the middle a zone of overlapping with Less energy and Going with another speaker with high energy and the overlapping zone has Less energy. Yeah, so there'll be some cases for which\nSpeaker B: But the so so there'll be This is one point to visual things that I mean that there'll be time there'll be overlap between the distributions The question is if it's a reasonable feature at all there's some separation. Yeah, especially locally\nSpeaker A: And the other thing about I was just gonna say that that right now we're just exploring Well, you would imagine eventually is that you'll feed all of these features into some Discirmative system. Yeah, and so even if if one of the features Does a good job at one type of overlap another feature might do a good job at another type of overlap?\nSpeaker B: Right, I mean the the reason I see just a scatter features is I used to do this a lot when we had 13 or 15 or 20 features to look at Because something is a good feature by itself You don't really know how it will behave in combination and so it's nice to have as many as many together at the same time as possible In some reasonable visual form there's cool graphic things that people have sometimes to put together three or four and some funny Funny way, but it's true that you shouldn't do any of that unless you know that the individual ones at least have have some\nSpeaker E: Normalizing I mean it's really important to pick a normalization that matches the distribution for that feature and it may not be the same For all the types of overlaps or the windows may not be the same I mean actually I was wondering right now you're taking all of the speech From the whole meeting and you're trying to find points of overlap But we don't really know which speakers overlapping which speak away So I mean another way would just be to take the speech from just say Morgan and just Jane and then just their overlaps by hand by cheating and looking at you know if you can detect something that way because if we can't do it that way There's no go-pro that we're gonna be able to do it that you know there might be something helpful and cleaner about looking at just Individuals and then that combination plus I think it has more elegant Then the right model will be easier to see that way so I don't know if you go through and you find Adam because he has a lot of overlaps and some other speaker who also has enough speech and just sort of look at those Three cases of Adam the other person and the overlaps maybe And just look at the distributions maybe there is a clear Pattern but we just can't see it because there's too many combinations of people That can overlap I had the same intuition last last week\nSpeaker D: I can't start with it's just your idea of simplifying starting with something that you can see you know without the extra layers\nSpeaker E: Right because if energy doesn't matter there like I don't think\nSpeaker C: To study individual to study individual\nSpeaker D: But just simple case and rid of one that has a lot of data so\nSpeaker E: Because what if it's the case and I don't think that was a great overlap by the way What if it's the case that when two people overlap the equate or you know There's a conservation of energy and everybody please talk more soft. I don't think\nSpeaker D: Or what equipment what if the equipment adjust somehow?\nSpeaker A: Yeah, well, but I think that's why I was saying about different types of overlap\nSpeaker E: There there are different Types and within those types like as Jose was saying that sounded like a back channel overlap meaning the kind that's a friendly Encouragement like great. Yeah, and it doesn't take you to take the floor But some of those as you showed I think can be discriminated by the duration of the overlap so It actually this new student Don who Adam has met and he was at one of our meetings He's getting his feet wet and then he'll be starting again in mid-January He's interested in trying to distinguish the types of overlap I don't know if he taught me yet, but in sort of home name\nSpeaker C: No, you don't consider that possibility This is a general study of the overlapping\nSpeaker E: Yeah, so it might be something that we can help by categorizing some of them and then I would actually still recommend that he do the overall\nSpeaker B: Thing because it would be the quickest thing for him to do he could he already has all this stuff in place He has his to Graham mechanism has the stuff that subtracted and all he has to do is change it From from log to plane energy and plot the histogram and look at it And then he should go on and do the other stuff\nSpeaker E: Yeah, no, I didn't mean that that for you to do that, but I was thinking if If Don and I are trying to get categories and we made some data for you and we say this is what we think is going So you don't have to worry about it and here's three types of overlap we'll do the little thing for you\nSpeaker C: Consume different class of overlap\nSpeaker E: Yeah, that we will be working on anyway Then maybe you can try some different things for those three cases\nSpeaker C: This is a thing I commend with you before that we have a great variation of the situation of overlapping and The behavioral for energies it locked the energy is not the same all the time\nSpeaker B: But I guess I was saying that right now From the means that you gave I don't have any sense of whether Even you know there are any significant number of cases for which there's distinct and I would imagine there should be some You know, there should be the distributions should be somewhat separate And I would still guess that if they're not separated at all That there's some there's most likely something wrong in the way that we're measuring But For instance, I mean I wouldn't expect that it was very common Overall that when two people were talking at the same time that it was that it really was lower Although sometimes as you say would Yeah, no that was that was the joke\nSpeaker E: Yeah No, it's where you would never know that unless you actually go and look yeah, right\nSpeaker A: Yeah, let if I turn that light off the flickering is annoying me. Okay\nSpeaker E: They might be the case though that the significant energy just as I was saying comes in the non-back channel cases Because it that most people when they're talking don't change their own Energy when they get a back to because they're not really predicting the back channel and sometimes it's a nod and sometimes it's them Mm-hmm and the um-hmm is really usually very low energy So maybe those don't actually have much difference in energy, but All the other cases might and the back channels are sort of easy to spot in terms of their words\nSpeaker B: And again what the what difference there was would kind of be lost in taking the log so\nSpeaker E: Well, it would be lost no matter what you do\nSpeaker B: No, if it's if it's\nSpeaker E: Even if you take the log you can your model just has a more sensitive\nSpeaker A: Sure, but tone might be very Yeah, your mm-hmm tone is gonna be very different Right you could imagine doing specialized ones for different types of back channels if you could if you had a good model for it\nSpeaker B: You're if you're I guess my point is if you're doing essentially a linear separation taking the log first doesn't fact make it harder to Right so it's so if you so if they're close things it does it's a non-linear operation It doesn't fact change the distinction if you're doing a not if you're doing some fancy thing then and yeah And right now we're essentially doing this linear thing by looking across here and saying we're gonna cut it here Um, and that's the indicator of getting but anyway yet we're not disagreeing on any of this we should look at it more more finally but I think that This often happens you do fairly complicated things and then you stand back when you realize you haven't done something simple so If you generated something like that just for the energy and see and then as as as says when they have Uh a smaller a more coherent groups to look at that would be another interesting thing later And then that should give us some indication between those she give us some indication of whether there's anything to be achieved from energy at all And then you can move on to the more pitch-related stuff Not consider the local yeah, but then have you started looking at the pitch-related stuff?\nSpeaker B: Pitch-related I'm an assidine. I am preparing the program parents. I don't I don't be\nSpeaker C: Because I saw you remain and yeah, I agree with you. It's better to I suppose it's better to consider the energy this kind of parameter\nSpeaker B: Oh, that's not what I meant. No, no, I I well we certainly should see this but I I think that the harm I certainly wasn't saying this was better than the Harmonicity and pitch-related things. I was just saying I won't with the yeah, okay\nSpeaker C: Yeah, I was just saying understood that I had to finish my the moment with I'm concentrating my energy\nSpeaker B: Okay, okay, but I think like all these derivatives and second derivative is now these are the very fancy things I think I just sort of look at the energy and then get into the harmonicity Okay, so maybe we're trying to compress the meeting I know Adam had some forum stuff you wanted to talk about and did you have some?\nSpeaker D: I wanted to ask just something at the end of this topic so when I presented my results about the Distribution of overlaps in the speakers and the profiles of speakers at the bottom of that I did have a proposal and I planned to go through with it of Coating the types of overlaps of people were involved in just with reference to speaker style So you know with reference and you know, I said that in my Yeah, you know, I so I was like people may have different amounts of being overlapped with or overlapping But that in itself is not informative without knowing what types of overlap they're involved is So I was planning to do a taxonomy of types of overlaps to that So but I you know, it's like it sounds like you also have something in that direction is we have nothing\nSpeaker E: You know basically we got his environment set up. He's he's a doubly You know, so It's mostly that if we had to label it ourselves we would or we have to get started But it it would be much better if you can do it you'd be much better at doing it also because you know, I I'm not I Don't have a good feel for how they should be sorted out and I really didn't want to Go into that. I didn't have to so if you're willing to do that it would be interesting though to talk\nSpeaker A: Maybe not the meeting but at some other time about what are the classes?\nSpeaker E: I think that's a research effort in and of itself. Yeah, it would be interesting But I don't know how it'll turn out and\nSpeaker D: Oh, you know, so we also with reference to a purpose too that we\nSpeaker E: Still have some funding for this project like probably if we had to hire some like an undergrad because Don is being covered half time on something else. I mean You were not paying him the full our a ship for all the time so If it got to where we want it we needed someone to do that. I don't think there's really enough data\nSpeaker D: Where we see this is a prototype to use the already transcribed meeting is I think another problem\nSpeaker C: There we we can consider is the duration I know besides the class with overlap the duration because it's possible some Some classes has a type of duration duration very short And we have a rapping with speech is possible to have and it's interesting I think to consider the the window to the normalization conversation window because if we have a type of a kind of overlap Like channel overlap with a short duration if possible And to normal that if we normalize with a consider only the window and by the left Side and the right side of the rapping with a very small window If the effect of the normalization is bigger In that's a rapping sound very short I mean that you have you have a channel you have a rapping sound very short and you consider A All the chance to normalize this very short For example, hmm And the energy is not a height I think if you consider all the chance to normalize and the Chinese bigger Compared with the with the rapping duration the effect is stronger I mean the the effect of normalization with the mean and the at the very end is different that if you consider only a eh window to compare with the the duration of the rapping\nSpeaker B: You want it around the overlapping part to include something that's not in the overlapping But I don't know\nSpeaker E: Well, it's a sliding window right so if you take the the measure in the center of the overlap Yeah, you know they're better but something but if your window is really huge then yeah Yeah, this is the portion of the\nSpeaker C: This is the thing to consider only the more windowed Yeah, so\nSpeaker E: Yeah, they're the overlapping so you know you shouldn't be more than like you should definitely not be three times as big of your And your back channel then you're gonna have a wash and hopefully it's more like I'm not sure that's necessarily true\nSpeaker D: It isn't empirical because because it because\nSpeaker B: Again if you're just compensating for the gain you know the fact that this this gain thing was crude And the game with someone is speaking relatively consistent level just to give an extreme example All you're doing is compensating for that And these still and then if you look at the frame with respect to that It still should should Change\nSpeaker E: Yeah, it depends how different your normalization is as you slide your window across\nSpeaker A: I mean we're also talking about a couple different things I mean one is your analysis window and then the other is any sort of normalization that you're doing And they could be quite different\nSpeaker B: Yeah\nSpeaker E: Or last week anyway, we have to look at that be great if if you're marking those and That it is definitely true that we need to have the time marks and I was assuming that will be inherited because If you have the words and they're roughly aligned in time The Then you know this student and I would be looking at the time Marks coming classifying all the frames inside those as whatever label\nSpeaker D: So it wouldn't be I wasn't planning to label the time marks. I was thinking that would help think you need to engineering side\nSpeaker E: Yeah, that should be linked to the words which are linked to times\nSpeaker A: Well, we're not anytime soon gonna go first alignment so Um, it's not hand-marked then we're not gonna get the times\nSpeaker E: Well, we wouldn't be able to do any work without a fourth lemon anyway So somehow if once he gets going we're gonna have to come up with one Yes, I mean, whatever you would do a very bad one with broadcast news\nSpeaker B: Well again for the close mic stop That might be good enough to come up take this take the switchboard system. Yeah\nSpeaker A: Maybe worth a try it would be interesting to see what we get\nSpeaker E: There's a lot of work you can't do without that. I mean how would you You'd have to go in and measure every starting stop point next real word Yeah, you're interested in anything before it Anyway, so that'd be great\nSpeaker B: Yes, something we should talk about later maybe not just now but should talk about our options as far as the Transcription\nSpeaker A: Yep, if IBM doesn't we have to turn\nSpeaker E: We'll do that later. I can't keep recording\nSpeaker F: Yeah, right\nSpeaker B: I'll talk about later. So Forms forms\nSpeaker A: Next iteration of forms. Oh good So it's two pages per person. Nope one's a digit form one's a speaker form. Oh So one is one time only speaker form and the other is the digits. So don't fill these out This is just a suggestion for what the new forms would look like So he incorporate the changes that we talked about\nSpeaker D: And why did you switch the order of the date and time fields? I'm which one on the new one time comes for us in then date, but I thought oh you mean on the digit form\nSpeaker A: It's rather a low-level question because the user fills out the first three fields and I fill out the rest So it wasn't intentional. It's an interesting observation, but it wasn't intentional Because the date is when you actually read the digits and the time and excuse me the time is when you actually read the digits But I'm filling out the date beforehand if you look at the form in front of you that you're going to fill out when you read the digits You'll see I've already filled in the date, but not the time\nSpeaker D: I always assumed so the time is supposed to be pretty exact because I just been taking the beginning time. Yeah, I mean\nSpeaker A: Yeah, I've noticed that the forms the the reason I put the time in is so the person who's extracting the digits meaning me We'll know where to look in the meeting. Oh Dear, we've been messing up your forms. I know\nSpeaker B: Yep\nSpeaker D: Well, I'm saying we started the meeting 30 I put 230 and I guess everything was going to 30\nSpeaker A: No, it's about 5050 actually it's about one thirty each about one third of them are blank about one third of them are when the digits are read and about one third of them are when the meeting starts\nSpeaker D: So this would be a radical suggestion, but I could put instructions Yeah, either that or maybe you could maybe write down one people Start reading did is not at the meeting. I can't do that. Yeah, he's been setting stuff up and going away\nSpeaker A: I see some reason he doesn't want to sit through every meeting. Yeah, but that is the reason name email and time are where they are And then the others are later on\nSpeaker C: Okay The seat is number\nSpeaker E: For official use only\nSpeaker D: Actually, you could very professional. Well, that does raise another question, which is why is the professional use only line not higher? Why doesn't it come in at the point of date and see? Oh, why is we're feeling in other things? Well because If you're your professional use you're going to already have the date and this what which form are you talking about?\nSpeaker D: Well, I'm comparing the new one with the old one. This is the digit form. Oh, you're talking about the digit form. The digit form doesn't the digit\nSpeaker A: No, that's all right. The digit form doesn't have a four official use only line. It just has a line Which is what you're supposed to read So in the digits form everything above the line is a fill-in form and everything below the line is digits that the user reads\nSpeaker D: Okay All right, I didn't mean to do real discussion. You really wanted to start with this other form. No, either way is fine\nSpeaker A: I just you just started talking about something and I didn't know which form you were for it\nSpeaker D: I was comparing so this is so I was looking at the change first So it's like we started with this now. We've got a new version of it with with the So the digit form we had one already now the the fields are slightly different\nSpeaker B: So the main thing that the person fills out is the name and email and time right\nSpeaker A: You do the rest yep just as as I have for all the others and there's an addition of the native language\nSpeaker D: Which is a very redundant this one has native language and this one does too\nSpeaker A: That's because the one the digit form that has native language is the old form not the new form\nSpeaker D: I'll catch up here I see\nSpeaker B: South Midland North\nSpeaker A: Yeah, this is the problem with these categories. I picked those categories from timmit. I don't know what those are\nSpeaker E: Actually the only way I know is from working with the date of a minute of figure it out. So I was like\nSpeaker B: Kansas and North's midland like like Illinois, but yeah, I mean so what accent are we speaking?\nSpeaker A: Western\nSpeaker C: Hmm, but yeah for me Hollywood if language is spani Spanish The regino is the center of Spain the\nSpeaker A: Yeah, I mean you could call whatever you want for the foreign language we couldn't Classify every single one. So I just left it blank and you can put whatever\nSpeaker C: It's different the span spani language from the north of Spain to south of the west and\nSpeaker A: So I'm not sure what to do about the region field for English variety You know when I wrote I was writing those down I was thinking you know these are great if you're a linguist Actually, I don't know how to I don't know how to categorize\nSpeaker E: This wasn't developed by this region for it if you're a T.I. RMIT\nSpeaker B: Yeah 1985\nSpeaker A: So I guess my only question was if if you were a south midland speaking Region person would you know it is that what you would call yourself?\nSpeaker B: I don't know yeah, you know, I think if you're talking if you're thinking in terms of places as opposed to names different names people have given to different ways of talking I would think North Midwest and South Midwest would be more common saying midland, right? I mean I yeah\nSpeaker E: The usage we can give him like a little map with the regions and they just No, I'm sorry Because it takes less time and it's sort of cute The side that's a little you know, it doesn't have to have all the detail\nSpeaker B: But what if you move five times and\nSpeaker D: Well, I was thinking you could have a great multiple one and then the amount of time so roughly so if you could say you know 10 years\nSpeaker A: Well, five years in the west coast or something. I think we don't want to get that level of detail At this form. I think that's all right if we want to follow up\nSpeaker E: But I mean I as I said, I don't think there's a huge benefit to this region thing it it gets The problem is that for some things it's really clear and usually listen to it You can tell right away if it's in New York or Boston accent but New York and Boston are two Well, I guess they have NYC, but New England has a bunch of very different dialects and and so does\nSpeaker A: Yeah, so I picked these regions because we had talked about timet and those are right from timet\nSpeaker E: These would be satisfying like a speech Research community if we release the database, but as to whether subjects know where they're from. I'm not sure because um, I know that they had to fill this out for switchboard. This is I almost exactly the same as switchboard Regents or very close. Yeah Um, and I don't know how they feel better But if midland Yeah, midlands the one that's Difficult, I guess Also north west so you got Oregon Washington, they're Oregon now which You people don't know if it's western or northern yeah, I certainly don't mean that I'm saying I don't even know what I speak\nSpeaker A: Oh, what is speaking am I speaking western?\nSpeaker E: Oh, what's what's northern I think originally it was north Northwest west\nSpeaker A: But yeah, so this is a real problem. I don't know what to do about it\nSpeaker D: I would not a characterized mine either and and so I would think I would say I've got a mix of California and Ohio\nSpeaker A: I think at the first level for example, we speak the same our our dialects Or whatever you region are the same, but I don't know what it is\nSpeaker E: So you have a like techno speak a techno speak a geek region I mean you can sort of identify You're deepening, but maybe that maybe we could Leave this and see what people see what people choose and then\nSpeaker D: Let them just fill in I mean, I don't know what else we can do. I'm wondering about a question like where are you from mostly?\nSpeaker B: But I'm not even actually though. I really am confused by northern I really am I mean if if you're in New England that's north if you're if you're\nSpeaker D: I'm not sure\nSpeaker B: Scandinavian the Minnesota areas north yeah, that's that's also North Midland right and yeah, of course that's very different from like Michigan and Oregon, Washington are our western but they're also northern I don't hope\nSpeaker E: Well, there are No, probably a few people in there hardly any subjects from Sorry\nSpeaker B: I should put a little map and say put an X really yeah really\nSpeaker E: We could ask where they're from yeah, we went Well, we sort of we\nSpeaker A: People end up saying that it you know, well, I like the idea of asking what variety of English do you speak as opposed to where you're from? Because if we start asking where we're from again, you have to start saying well is that the language you speak or is that just where you're from?\nSpeaker E: I mean it gives us good information. I'm where they're from but that doesn't tell us anything I mean so so I would say Germany\nSpeaker A: You know am I speaking with the German accent? I don't think so I'm thinking where are you from mostly?\nSpeaker D: Okay, yeah, you know then you have some some kind of subjective amount of time factored into it\nSpeaker A: Yeah, I guess I could try to put squeeze in a little map. I mean there's not a lot of room So yeah lost in New York City the South and regular I think of those Northern is the only one that I don't even know what they're meeting\nSpeaker E: So let's make it up. I mean who cares right we can make up our own so we can say Northwest Rest of west or something West\nSpeaker A: Yeah, I don't think the Northwest people speak any differently than my dad\nSpeaker B: You come from the Louisiana purchase\nSpeaker C: Usually here people here and know what is the kind of Lamb English language that that's exactly what we're arguing about we don't know\nSpeaker E: It's harder in America than anywhere\nSpeaker A: It's some of them are very obvious if you if you talk to someone speaking with a southern draw you know Yeah, or Boston or Boston can a\nSpeaker E: Washington and those people if you ask them to self-identify their accent. They know they know very well They know they don't speak the same as the and they're proud of it. Yeah, it's Boston New England Well, depends for you\nSpeaker A: I guess that's the problem with these categories\nSpeaker D: Why can't we just say character right? Characterize your accent if you can and and so I would say I don't know Right, which probably would be a question with this really strong coloration would know and it's actually good main or something\nSpeaker E: I was thinking something along that line because if you don't know Then you know ruling out the fact that you're totally in after something if somebody doesn't know it probably means their accent is a very strong compared to I mean it was\nSpeaker B: That long ago that we had somebody here who was from Texas who was absolutely sure that he didn't have any accent And and had he had a pretty\nSpeaker A: Okay, so I propose take out northern ad don't know\nSpeaker D: Oh, yeah, I would say more or sweepingly. How would you characterize your accent?\nSpeaker A: So you want to change the instructions also not just say region\nSpeaker D: Well, I think this discussion has made me think that's something to consider\nSpeaker A: I don't know if I if I read this form I think they're gonna ask it They're gonna answer the same way if you say what's variety of English do you speak region?\nSpeaker A: As if you say what variety of region region do you speak please characterize your accent?\nSpeaker D: They're gonna answer the same way. Well, I was not sure that I so I was suggesting not having the options just having them\nSpeaker A: Oh, I see what we talked about with that is is so that they would understand the granularity\nSpeaker D: Yes, but if as Liz is suggesting people who have strong accents know That they do\nSpeaker A: I mean that's what I had before and you told me to list the region To listen to\nSpeaker B: Yeah last week last week I was sort of arguing for having a wide open and everybody said No, but then we're really hard to interpret because some people say things and I mean I had it wide open last week\nSpeaker A: And then said Tim it and that's what the other is for no, I mean what if we put in both ways of asking them\nSpeaker E: So one is region and another one is if you had to characterize yourself your accent won't they answer the same thing? Well, they might not answer only one of the questions\nSpeaker D: But if you know they might say other region because they don't know what category to use, but they might have something because it is easier to learn from what they say\nSpeaker E: as to which ones a better way to ask it because I really don't know this is just a small thing but\nSpeaker B: Says variety and then it gives things that have American is one of the choices but then says region but region actually just applies to\nSpeaker A: US right right I mean that's why I put the other in what we thought about okay\nSpeaker D: We just we sort of thought yes, I mean last me my collection was that we felt people would have the less that there's so many types of varieties of these other languages and we are not going to have that many subjects from these different Language groups and that is a huge waste of space\nSpeaker A: Okay, so I mean I the way I had it last time was region was blank. It just said region calling And I think that's the best way to do it because because of the problems we're talking about But what we said last week was no put in a list so I put in a list so should we go back to It's a little smaller while certainly dropping northern I think it's right because none of us know what that is and keeping other\nSpeaker E: And then maybe this north midland we call it north midwest south midwest Yes, I think so yeah, that would help me yeah, unless you're from midland\nSpeaker B: Or midland middle is midwest one word Texas or midland can There's a town\nSpeaker E: But yeah, so Kansas would be south midland right yeah and wouldn't yeah in Colorado Yeah The dropping north so it would be Western it's just one big Of course you have a huge variation in dialects, but that's true\nSpeaker A: So I mean the only one yeah Well, I shouldn't say that I have no clue. I was gonna say the only one that doesn't have a huge variety is New York City But I have no idea whether it does or not. It does seem I mean\nSpeaker D: I would think that these categories would be more would be easier for an analyst to put in rather than the subject himself\nSpeaker A: I think that that was what happened with timid\nSpeaker B: Was that it was a new York west of west of New York City and Pennsylvania and That's new in No, it's not\nSpeaker D: I think one of the midlands no no pencil vanes. It goes under other definitely under other\nSpeaker E: Well, you know Pennsylvania has a pretty strong diet\nSpeaker B: Yeah, Pennsylvania is not new England and New Jersey is not new England and Maryland is not okay\nSpeaker A: So none of those are the sound other suggestion Rather than have the circle fill in forms say region open pern eg southern comma western close pern colon\nSpeaker E: Okay, that's good\nSpeaker B: I like that. That's just to see what we hear on the subject Tired of this that we were green No, I think you like it\nSpeaker A: Actually, maybe we do one non-English one as well Southern Cockney Yeah, is that a real accent? Sure. Yeah, how do you spell it?\nSpeaker A: Cockney\nSpeaker B: Any why? Yeah, I live her\nSpeaker E: River Pudley and yeah, actually Liverpool. Yeah, it's\nSpeaker A: Okay, we'll do it that way. Actually, I like that a lot Because that gets at both of the things we were trying to do the granularity and the person can just self-assess and we don't have to argue about what these regions are\nSpeaker D: And see on the subjects now I have one suggestion on the next section So you have native language you have a region and then you have time spent in English between country Now I wonder if it might be useful to have another open field saying which one parenthesis s closed parenthesis Because if they spent time in and Britain and America Yes, it doesn't have to be so at all exact just in the same open field Yep, just which one I think it's fine with an s which one optional s\nSpeaker B: Okay We uh Yeah, yeah, okay Um any any other\nSpeaker A: Open mic topics I should go right to the digits um, did you guys get my email on the multi-trans? The okay, isn't that wonderful? Yeah, so I have a version also which actually displays all the channels\nSpeaker D: But it's hideously slow so you this is dance patches\nSpeaker A: What the ones I applied that you can actually do are dance because it doesn't slow it down Just uses a lot of memory Slow does that mean to you know the one that's installed is fine. It's not slow at all I wrote another version which instead of having the one pain with the one view It has multiple paints with the views But the problem with it is the drawing of those waveforms is so slow that every time you do anything it just crawls It's really bad As you play as you move as you scroll just about anything and it was so slow it was not usable And this will be so that's why I didn't install it in the pursuit\nSpeaker D: Because in terms of like disentangling overlaps and things\nSpeaker A: So I think the one that Dan has is usable enough It doesn't display the others it displays just the mixed signal But you can listen to any of them\nSpeaker D: He also has version control, which is not a nice\nSpeaker A: Hey, so you know he suggested that but he didn't it's not installed\nSpeaker E: I thought it was in those patches. No, okay So is there any hope or actually displaying\nSpeaker A: Um, not if we're gonna use tickle tk At least not if we're gonna use snack and you would have to do something ourselves\nSpeaker D: Well, or use the one that crawls\nSpeaker E: Well, I probably would be trying to use the Whatever there\nSpeaker A: Well, we see how Dan's works and if it if we really need the display\nSpeaker E: I mean, I'm just wondering if we can display things other than the waveforms So suppose we have a feature feature stream and it's just you know, you know, dimensional feature varying in time And we want to plot that instead of the whole waveform that might be faster right\nSpeaker A: We could do that but that would mean changing the code I mean, this isn't a program we wrote. This is a program that we got from someone else and we've done patches on\nSpeaker E: Well, I'll talk to you about it. See, but it's definitely great to have that one\nSpeaker B: If there was some is there some way to have someone write patches in something faster and link it in or something or not easily\nSpeaker A: Yes, we could do that you could you can write widgets and see yeah and try to do it that way But I just don't think it let's try it with Dan's and if that isn't enough We can do it otherwise. I think it is because when I was playing with it the mix signal has it all in there And so it's really it's not too bad to find places in the in the stream Where things are happening and so I don't think it'll be bad also case that that\nSpeaker D: doesn't multi wave thing is proposed to the so Dan proposed it to the transcriber central people and it's likely that So and they responded favorably because it will be incorporated in the future version They said that the only reason they hadn't had the multi the parallel Stream one before it was simply that they hadn't had time to do it and So it's likely that this this may be energy to the child this central they may well not have had much demand for it\nSpeaker E: Well, that's that's true. Yeah, so you mean they could they could do it and it would be\nSpeaker D: Oh, no, I mean that is that that his so this one that we now have Does have the status of potentially being incorporated likely being incorporated in central code now that now if we develop further\nSpeaker A: Then I don't I mean I think if one of us sat down and coded it so that it could be displayed fast enough I'm sure they would be quite willing to incorporate it, but it's not a trivial task. Yeah\nSpeaker D: I just like the idea of it being something that's you know tied back into the original so that people can However, I also understand that you can have widgets that are very useful for the purpose\nSpeaker A: Anyway, show you digits\nSpeaker B: Yeah, let's do digits And then we'll turn off the mics and then I have one other thing to discuss\nSpeaker E: Like you I go first or I can't stay for the discussion Okay, well should we you want to go do digits or do you want to just skip digits? No, I can do digits then But I don't want to bud in all right you go ahead But if there's something on the rest of the I'll be around I So I Or we can talk about it. Why don't you read the digits? Yeah, right? Oh, this is the new one Yeah, all right This is transcript 2751 2770 92544 04016 05907 17908382 3407 3 0 64377 689 8495 0 0 810634 42559 6707 80533\nSpeaker C: transcript 2711 that's 273 75768 7000 97444 01019 0017 22934 1745 701 566 429 506 780 756 159 044 05 228 102 718\nSpeaker A: Transcript 2651 2670 4 71365 8308 96056 0 1 2 0 2 4382 52077 676 22137 255 3571538 48110\nSpeaker B: transcript 2671-2690 5 6 70830 922 0360704 06 1 2710124 3 40672 6478 74830 9 0 10899 319 42775 5752\nSpeaker D: transcript 2691-2710 6978 9 0 0268224 1408 2333 3 4924506 5 606 84134 934 086\nNone: 0 1 2 3 5271848 6662071 8\nNone: 8 8 8 8\n"}, {"meeting_id": "IS1004c", "summary": "Project Manager started meeting on the conceptual design of the remote control. Marketing gave a presentation on the preference of users which was easy to use with a fancy look. Also, Marketing suggested the remote control to be the shape of cool fruits with fewer buttons because they should target the young user group that would be keen on new device innovation. Next, User Interface presented on the sample of sensor speaker but didn't recommend it because it didn't add too much functional design and it was not mature to use it as a speech recognition engine. The group agreed with User Interface to have two scroll wheels with turbo button and switch on button. Then, Industrial Designer gave a presentation on the component design. The group decided to use a power source with spongy design and double curve chip for easy handling. Lastly, Project Manager helped conclude the features on the upcoming prototype presentation and asked group mates to prepare individually on it.", "dialogue": "None: Fright thataturdays WAIT I could it MMMMadd Oh Okay, good afternoon. Hope you have good lunch.\nSpeaker B: Yeah, three-hat for Luffle.\nSpeaker E: Oh, nice.\nSpeaker E: And you?\nSpeaker C: Yes, I had something similar, but non-vegetarian.\nSpeaker E: Okay, so today is our third meeting.\nSpeaker E: This will be about the conceptual design.\nSpeaker E: If I come back to the minutes of the last meetings, we decided not to go for speech recognition technologies, because of some reasons. And we are not decided about the use of the screen on the remote control because of costs.\nSpeaker E: So maybe we will be able to clarify this question today.\nSpeaker E: At the end of the meeting, we should take a decision on that point.\nSpeaker E: So I hope that your respective presentations will help us.\nSpeaker E: So each of you have some presentation to perform. Who starts?\nSpeaker A: Okay, we are starting.\nSpeaker E: Marketing. So you saved your presentation somewhere?\nSpeaker E: Yeah.\nSpeaker E: So you're four?\nSpeaker A: Four, yeah.\nSpeaker E: Which is trend watch.\nSpeaker E: Okay, Mr. Marketing Experts.\nSpeaker A: Yeah, that's me.\nSpeaker A: So good point.\nSpeaker A: Well, I investigate the preference more.\nSpeaker A: I investigate deeper the preference of the users.\nSpeaker A: Also the current investigation, sorry, the current.\nSpeaker A: Then current trends.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Doesn't sound.\nSpeaker A: Okay.\nSpeaker A: Well, what I found. Can you?\nSpeaker E: Next slide.\nSpeaker A: Yeah. Thank you. What I found in order of importance, from less to more important is that people want an easy to use device.\nSpeaker A: After they want something new, technologically speaking.\nSpeaker A: But the most, what they find more interesting, more important is a fancy look and feel instead of the current trend, which was the functional look and feel.\nSpeaker A: So now, more cool aspect, more, a cooler aspect rather than a device with many functions and many buttons with, instead of, instead of a device which can do many things, a device which is pleasant to watch to see.\nSpeaker E: Okay.\nSpeaker A: Also, well, in Europe, in Paris and Milan, the, in Paris and Milan, the current trend of clothes, furniture and all this fashion, it's, it's free.\nSpeaker A: And the, the theme is free time, vegetables.\nSpeaker A: And also, in the, in the USA, the, the current, the more, the most popular feeling, it's, it's a spongy.\nSpeaker A: Spongy means, a pongy.\nSpeaker A: So maybe we should, we should think in, in this direction.\nSpeaker C: What do you mean by fruit and vegetables and spongy?\nSpeaker A: What do you mean, it's like, fruit vegetables is the, the new, have you seen the last, exposition of clothes in Milan?\nSpeaker C: No, I missed that one.\nSpeaker A: Yeah, I, I didn't miss an, I didn't miss an, I saw that the fruits, there are many pictures of fruits and vegetables in the world.\nSpeaker C: Oh, okay. So they're not like dressed as a carrot, they just have, like, not yet.\nSpeaker C: Pictures of fruit on, okay.\nSpeaker C: So we're not going to have a remote control in the shape of, of a banana just, maybe.\nSpeaker A: Yeah. Visit all the textures and all this can,\nSpeaker E: Dreams of bananas.\nSpeaker C: Okay. And, there. Yeah.\nSpeaker B: But what's your suggestion, how we can have some shape like that on the, one?\nSpeaker E: One, so this is the next slide, so don't you?\nSpeaker A: No, no, it's not. It's not.\nSpeaker A: So which fruit are you thinking of?\nSpeaker C: Mmm.\nSpeaker A: Neither. I haven't thought of any particular fruit, but the general aspect of the, of the remote control may, may, could remain some kind of vegetable, some kind of, instead of vegetable, some natural, natural, object of something.\nSpeaker E: But yeah, it depends on the, you can display a banana on the whole city.\nSpeaker C: I think you want the remote control to be the shape of a fruit or you want just some kind of like fruit logo on the, this buttons are, the shape of fruits.\nSpeaker B: Yeah, maybe the shape, the shape buttons are in the shape of fruits or something.\nSpeaker B: Apple, banana, something like that.\nSpeaker A: Not, not, not, not too much focus.\nSpeaker A: Not too, not too, not too similar to a fruit because next year the, the trend, the trend will be different.\nSpeaker A: So we shouldn't be really attached to, to the trend.\nSpeaker C: So something that looks half like a fruit and half like an elephant?\nSpeaker A: For instance, yeah.\nSpeaker A: African or Asian or African?\nSpeaker B: Let me just discuss that.\nSpeaker C: Okay, I'm not, I'm not really sure if that would really appeal to everyone though.\nSpeaker C: Just to fashion gurus.\nSpeaker C: Like maybe just like a little bit, a little fruit picture somewhere in the corner.\nSpeaker C: But I don't know about a, I don't know how ergonomic an origin.\nSpeaker A: Maybe we should first specify what target are we focusing.\nSpeaker A: I think in my opinion, we should focus on, on young people because they are more open to new devices.\nSpeaker A: And also, yeah, according to the marketing report, 95% of young people was, was able to, to buy a, a cooler remote control.\nSpeaker C: But this is, it's very cool.\nSpeaker E: What?\nSpeaker E: That's a question.\nSpeaker C: What?\nSpeaker C: It's very cool.\nSpeaker A: Yeah?\nSpeaker A: What? It's the new trend of the...\nSpeaker C: Well, I guess, you know, Apple has their iPod so...\nSpeaker C: In my, just because they have an Apple on the, on the product doesn't mean fruit is cool.\nSpeaker A: No, I think we should think about a shape with it.\nSpeaker A: A device with the shape of some...\nSpeaker C: Okay, but it has to be easy to, to use though and to hold, you know, you know, on a pair or a watermelon.\nSpeaker A: Then you think we can find a, the shape of a fruit which is handy?\nSpeaker A: Well, probably the only thing is a banana.\nSpeaker C: I don't know.\nSpeaker C: I don't think of a cucumber.\nSpeaker C: Or maybe it's log.\nNone: Maybe.\nSpeaker C: It's a green.\nSpeaker C: Maybe.\nSpeaker C: So, but I mean, you also have to, you also have fit all the buttons and, you know, it's, it...\nSpeaker C: Yeah.\nSpeaker C: The thing is you have to, normally with, with buttons, they have to be at some point attached to a circuit board.\nSpeaker C: So if you're going to have things like on a cylindrical kind of device, it may be difficult to kind of, to build.\nSpeaker A: Yeah, but I like, I like your idea that we shouldn't have a lot of buttons.\nSpeaker A: Yeah.\nSpeaker E: And you will not have play, a lot of plays to put LCD on a banana also.\nSpeaker A: Yeah.\nSpeaker A: Oh, do you want an LCD with 25 euros?\nSpeaker E: Well, you're a marketing expert.\nSpeaker E: You should tell us if it's too much more.\nSpeaker A: Well, according to the, to the report, people are more interested in a fancy look and feel than in a technological innovation.\nSpeaker A: So, I will give more importance to the look and feel than rather than the...\nSpeaker E: So you, you...\nSpeaker E: You need to...\nSpeaker A: And also it's, I'm not convinced about this LCD because you need internet connection.\nSpeaker A: You need more things.\nSpeaker A: It's not just buying a new controlled remote.\nSpeaker A: You need buying control remote, buying more things.\nSpeaker E: So you are simply looking to...\nSpeaker E: A remote controlled that looks like a banana.\nSpeaker A: For instance.\nSpeaker A: Yeah.\nSpeaker A: For, for, given an example.\nSpeaker E: Okay.\nSpeaker E: So maybe you can go ahead.\nSpeaker A: Yeah, no, it's what I already said.\nSpeaker E: Okay.\nSpeaker E: Thanks.\nSpeaker E: Okay.\nSpeaker E: I'll give the floor.\nSpeaker E: So you're using the big guy.\nSpeaker E: Okay.\nSpeaker E: So you're three.\nSpeaker C: Yeah.\nSpeaker E: And it's... this one.\nNone: Yep.\nSpeaker E: Go for it.\nSpeaker E: Yep.\nSpeaker C: Okay.\nSpeaker C: So...\nSpeaker C: Next slide.\nSpeaker C: Okay.\nSpeaker C: So I received an email around lunchtime.\nSpeaker C: Let me know that the brilliant minds that our technology division have developed an integrated programmable sample, center sample speaker unit, which is a way for you to have a conversation with your coffee machine and or remote control.\nSpeaker B: It's just a speaker, right?\nSpeaker C: It's no, what it is.\nSpeaker C: It's not a microphone.\nSpeaker C: It has a microphone, has a speaker, it's got a little chip.\nSpeaker C: And last year...\nSpeaker B: Actually, I'm not reading microphone there, so that's why you can't have conversations.\nSpeaker B: Well, it's a sample.\nSpeaker C: It's a big deal.\nSpeaker C: It's a sensor sample speaker.\nSpeaker C: It means that it can recognize, it can do like a match on a certain phrase that you speak and then can play back a phrase in response to that.\nSpeaker C: But there's no kind of understanding of the phrase.\nSpeaker C: Okay.\nSpeaker C: So, I mean, you know, I guess you could build that in.\nSpeaker C: You could link the recognition of a certain phrase to some function on the remote control.\nSpeaker C: But basically the thing is we have this technology available.\nSpeaker C: Enos.\nSpeaker B: Enos.\nSpeaker C: Enos.\nSpeaker C: So, but the thing is obviously there's still going to be a cost if you decided to integrate that because you still have to pay for the production of the components.\nSpeaker C: So, it, it, but it basically means we, we can kind of consider this from a, you know, a theoretical or usability kind of viewpoint without worrying too much about, you know, how to develop it because we have this already done.\nSpeaker C: And, past, you know, some people might get annoyed if we, if we just dump it.\nSpeaker E: Something that I unclear really understanding.\nSpeaker E: Is this a technology that recognize keywords?\nSpeaker E: Speech keywords?\nSpeaker C: It's, it's not, well, it's, it will recognize, I guess keywords, but, you know, keywords in a certain order, like a phrase.\nSpeaker C: You train it for a certain, for a certain phrase, you say the, the example they said that they have up and running with their prototype is, although they're actually integrating to the, into the coffee machine that, that web producing is, you can say good morning.\nSpeaker C: The coffee machine, and it can recognize that phrase and it'll play back.\nSpeaker C: Good morning.\nSpeaker C: How would you like your coffee?\nSpeaker E: It's just to play back something.\nNone: Yeah.\nSpeaker C: So actually that was a bad example because it doesn't actually ask how do you want your coffee because they can't really understand the response.\nSpeaker E: Yeah.\nSpeaker E: So this is not really to do, to, to do control.\nSpeaker C: Only, like, only in the sense that it can recognize the set, a set target kind of way.\nSpeaker C: This is just more like a, it's designed, it's designed as a fun kind of thing, but I guess you could use it as a, as a way to implement a.\nSpeaker E: It's easy, a fancy thing that you can bring to, we can bring to the remote control that we not have any.\nSpeaker C: Completely pointless, yeah.\nSpeaker E: Yeah.\nSpeaker E: Completely pointless.\nSpeaker E: From the interaction point of view.\nSpeaker E: Yeah.\nSpeaker C: Unless you know you like having conversation with your remote control.\nSpeaker A: Okay.\nSpeaker A: Can we use it for saying, okay, channel 50, channel 20?\nSpeaker C: I mean, if you can, but you have to, I think it's a fairly simple design.\nSpeaker C: So you would have to record into the device every possible combination.\nSpeaker C: You have to train it to learn channel 15, that whole thing, not just the word channel and the word 15.\nSpeaker C: It doesn't have that kind of logic in it.\nSpeaker E: So this is much more than taking this technology, bringing it to the remote control.\nSpeaker E: Yeah.\nSpeaker C: There would be some development work.\nSpeaker E: So this is out of discussion.\nSpeaker E: Yeah.\nSpeaker E: So if it is something that we can bring easily and to put it into the banana remote control.\nSpeaker A: Mando.\nSpeaker A: Banana Mando.\nSpeaker A: Banana Mando.\nSpeaker E: Banana Mando, yeah.\nSpeaker E: That, it could be cool.\nNone: Yeah.\nNone: Okay.\nSpeaker B: Let's go ahead.\nSpeaker C: I, I, I, I think it's worth it.\nSpeaker C: I think it doesn't really add much to the functional design.\nSpeaker C: It's, it's, it's not mature enough to use as a speed tracking engine.\nSpeaker C: So, um, yeah.\nSpeaker C: So if we can just move on to the next slide, I've just done a quick mock up of, uh, some of the features of our potential front-key looking.\nSpeaker C: Yeah.\nSpeaker C: It doesn't look like a banana at all.\nSpeaker C: Well, you see, I was, I was unaware at this point of, of the fruit focus.\nSpeaker C: Yeah.\nSpeaker C: Um, so at the moment it looks like a box focus.\nSpeaker A: You see now you can, I dropped, I dropped it on fruit.\nSpeaker A: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Well, this is actually, this could be a genetically engineered fruit that's designed to be, you know, square so that it packs tighter in the boxes.\nSpeaker C: But, um, I've just indicated here we could have actually two scroll wheels.\nSpeaker C: Because I think the scroll wheel is a fairly, um, key part of, you know, I think everyone has the degree that it could be quite a useful, um, thing.\nSpeaker C: But I think it's important, you know, to have two scroll wheels because, you know, you want one for, for the channel, but you also want one for the volume because it's, it's, the volume is, you know, it's very handy for it to have a instant kind of feedback and response.\nSpeaker C: So, but, um, I also included this turbo button because I think, you know, every design should have a turbo button.\nSpeaker C: What's the turbo button?\nSpeaker C: Well, you see, this is, you know, a unique problem with, with televisions is that if you have this scroll wheel for the television, the, the tuner and the TV is not going to be able to switch between stations as fast as you can scroll.\nSpeaker C: So, you know, the, the person might want to have a, uh, might want to be able to scroll past television stations without seeing what's on them, which gets it just waits until you stop scrolling and then, you know, displace that station.\nSpeaker C: Or they might want to scroll and have a quick glimpse of it, even if it lags behind what they're doing.\nSpeaker A: If you control the speed.\nSpeaker C: Yeah, so with the turbo button, you can say skip over channels if, uh, you know, if I'm, if I'm going, if I'm scrolling past them and, you know, it's, um, you could have a little red light that comes up when they press it so they feel, you know, it's really going fast or whatever.\nSpeaker C: So, yeah, that's, um, this is the two important features, I think we need on the right, but I mean, we can discuss about what other kind of buttons we need.\nSpeaker C: Um, you know, it could be, you know, if we, if we want to have like a very cheap kind of device, I mean, we could even consider that maybe we want to sell this as a very, if it's going to be a banana, you know, as a pretty gimmicky kind of thing that doesn't have that much functionality is just, you know, it's going to be a little bit more functionality is just, you know, a couple of scroll wheels in the button because it's hard to get so many buttons on the banana and it's still very, it may even be for most, for some people more functional than their current remote, but if they have these scroll wheels so, um, you know, what other buttons do we want?\nSpeaker C: I mean, we could have, well, I guess you need an on and off switch, but you could, you could turn it on by taking the top off the banana maybe.\nSpeaker C: It's kind of like a spike and a flick. Sounds crazy. I like this idea. That's why you're a marketing. Yeah.\nSpeaker E: So it looks like we're going completely to forget about the LCD thing.\nSpeaker C: Well, that's the thing is every decided that we can only spend, uh, 25 year. I think we'll not spend, but, you know, charge.\nSpeaker A: I think we could use somehow the coffee machine dialogue interface or so.\nSpeaker E: No, we can't, we can't use that.\nSpeaker A: You, we can't, we can't use that to.\nSpeaker E: To command, to communicate is just to fit.\nSpeaker A: Yeah, we can say 2025. No. No.\nSpeaker C: But then you have to have a template for every channel for 100 channels you have to be able to recognize.\nSpeaker C: It's not allowed to 100 templates. Well, I, I think it's probably more than our chicken handle because it's designed for a coffee machine, you know, to say hello in the morning.\nSpeaker A: Ah, it's designed for a coffee machine.\nSpeaker C: Well, that's its current application. No, it's kind of, they wouldn't design it to handle 100 things.\nSpeaker A: Yeah. So maybe you could ask your, you could ask the engineering department.\nSpeaker E: Yeah. Good. Good. Good thing. You want to move to your slides? You have finished?\nSpeaker C: Well, I just, I just made the point. I don't, I don't know that speech recognition is, you know, even if we can do it, I think it's not really appropriate for a.\nSpeaker C: Yeah, I think so.\nSpeaker C: Television environment. But I did have one thing from a previous meeting you were talking about being able to find the remote control.\nSpeaker C: And I was talking about extending, being able to extend the remote control by having, you know, a base station that can control other things as well.\nSpeaker C: It might be useful to have some kind of base station, even if it's just, you press on the button on it and the, and the remote control starts beeping.\nSpeaker C: You know, this is a way of finding the remote.\nSpeaker C: In that case, maybe the, maybe the speech recognition, the speech thing could be useful just to say, I'm here.\nSpeaker C: But it's probably a bit of overkill if you could just have a, so it's a speech.\nSpeaker B: So it's a, this is kind of thing. Something else. It's, it's just a spoken old.\nSpeaker C: It's, it's speech in, in speech kind of, not really speech recognition, but kind of pattern.\nSpeaker C: That's matching.\nSpeaker E: Well, good idea. Yeah.\nSpeaker E: Make it.\nSpeaker E: Okay. Let's move on.\nSpeaker E: So you too?\nSpeaker E: That's right.\nSpeaker B: Okay.\nSpeaker B: So this is going to be about the component design.\nSpeaker B: So first thing is we need power source for the remote control.\nSpeaker B: So I was of the idea that we can have two kind of power supplies. One is the usual batteries, which are there.\nSpeaker B: They could be chargeable batteries if there's a base station kind of thing.\nSpeaker B: And on top of that, we can have solar cells. When the lighting conditions are good, they can be used.\nSpeaker B: So it'd be pretty innovative kind.\nSpeaker B: Then we need plastic with some elasticity so that if your, if the remote control falls, it's not broken directly into pieces.\nSpeaker B: There should be some flexibility in.\nSpeaker C: I guess that fits in with the spongy kind of design philosophy.\nSpeaker B: Yeah.\nSpeaker B: So there should, we should think of something like that. And then it should be double curve the sides for the ease of handling.\nSpeaker B: And there are some other issues why we need double curve.\nSpeaker B: Then controls for the traditional use, traditional users, we can have the push buttons so that they don't feel that it's an alien thing for them.\nSpeaker C: So this one second, when you say double curve, what do you actually mean? You're looking like drawers, I think, on the whiteboard.\nSpeaker B: Double curve is you have curves on both the sides if I'm right.\nSpeaker B: So it's a metrical kind of thing, whatever it is.\nSpeaker B: Okay, but like, kind of convex.\nSpeaker B: It could be curve. So it could be convex, convex, convex, depending on what we want.\nSpeaker B: Okay.\nSpeaker B: So there are flats, there are single curve and there are double curves.\nSpeaker B: These are the three things. And there are different materials with plastic.\nSpeaker B: You can have double curve, but with certain other materials, we cannot have double curve.\nSpeaker B: So there were many other materials like wood, titanium and all those things, but plastic is, I think, is the most appropriate one.\nSpeaker B: It'll bring the cost down. And anyway.\nSpeaker C: Also, you know, wood could be quite a stylish option.\nSpeaker C: If you take like, nice quality kind of wood that's got a nice grain and you kind of put some, some varnish on it.\nSpeaker E: But there is no elasticity. Which wood in pieces.\nSpeaker C: Well, it depends. I mean, you have, the wood itself is not going to break.\nSpeaker C: So you don't have to worry too much about the case being broken inside.\nSpeaker C: Yeah, but inside, you know, you can still have some kind of cushioning that's not visible to the user.\nSpeaker E: Very too expensive to do.\nSpeaker E: Yeah.\nSpeaker C: And I mean, you can also, you can have just a very thin veneer of wood as well.\nSpeaker E: Yeah, but it's more easier to do a banana in plastic than in wood.\nSpeaker C: That's true. But are we set on the banana idea?\nSpeaker E: Actually, I was thinking that the shape of a banana is not, it's not really handy.\nSpeaker E: Yes, it is.\nSpeaker A: I don't know the name in English.\nSpeaker A: It's not a free. It is a vegetable.\nSpeaker C: It's like a pumpkin or?\nSpeaker E: Pumpkin.\nSpeaker A: Green.\nSpeaker A: Green.\nSpeaker E: Green.\nSpeaker E: I'm...\nSpeaker E: Yes, I see.\nSpeaker E: What is it?\nSpeaker A: And you put in the salad.\nSpeaker A: I don't know.\nSpeaker C: Oh, yeah, what's it in French?\nSpeaker C: It's a...\nSpeaker E: Okay, so it capsicum all of pepper.\nSpeaker E: Pepper.\nSpeaker E: But...\nSpeaker A: It also sweets with the double cure for easier.\nSpeaker C: I don't know, it seems a little bit kind of bulky for me.\nSpeaker A: No, I mean, in a...\nSpeaker C: It's not true.\nSpeaker E: Like, you're a fancy and fancy and fun.\nSpeaker E: You think that young people are...\nSpeaker E: I'm sure it's fun.\nSpeaker E: More than a banana?\nSpeaker A: But banana is not so handy.\nSpeaker A: I think that's...\nSpeaker B: It's more handy.\nSpeaker B: It's more than a rest compared to this, I think, and to capsicum.\nSpeaker C: But, you know, like a banana, you can be holding like this and have the scroll wheel kind of on top and just...\nSpeaker C: Roll it back and forth like that.\nSpeaker C: It's kind of...\nSpeaker E: I don't know how you would hold a capsicum.\nSpeaker E: It's really ergonomic.\nSpeaker E: It fits in the hand.\nSpeaker E: You had a lot of soft face to...\nSpeaker E: To put the control.\nSpeaker E: Okay, let's move.\nSpeaker E: Yeah, all right.\nSpeaker E: So, Stami is running.\nSpeaker B: Okay, so push buttons for the traditional user so that they don't feel they are alienated.\nSpeaker B: Just...\nSpeaker B: And a scroll button with push technology for channel selection volume control and teletext browsing.\nSpeaker B: These are the three scroll buttons which are already available with us.\nSpeaker B: In the company and we can go ahead with that.\nSpeaker B: Okay.\nSpeaker B: Yeah, we can go to the next slide.\nSpeaker B: Then there are different kind of chips.\nSpeaker B: One is the basic chip and then regular chip and then advanced chip.\nSpeaker B: So we can have regular chip for control.\nSpeaker B: Pricing is a factor for us.\nSpeaker B: That's why we'll go for the regular chip.\nSpeaker B: And regular chip support, speaker support.\nSpeaker B: So this functionality could be used for tracing the mobile phone which has been missed.\nSpeaker C: So, when you say speaker support, you mean it just has some output pins which kind of...\nSpeaker C: It could be a beep kind of thing.\nSpeaker C: Okay, but the speaker is actually attached to the chip in some way or is it just the signal?\nSpeaker B: Yes, that's right.\nSpeaker B: It's on to the chip.\nSpeaker B: Most probably.\nSpeaker B: Not 100% shareable.\nSpeaker C: So, are there any issues where we place this chip to make sure you can actually hear the speaker from the outside of the banana?\nSpeaker B: That will be the volume control I think which a user...\nSpeaker B: It should be already predefined.\nSpeaker B: It should be whatever be the case, the chip is always going to be sitting inside.\nSpeaker C: Yeah, but the speaker is actually on the chip.\nSpeaker C: Then if it's too far away from the casing or if the casing is too thick, then you may not hear the speaker.\nSpeaker B: So we can have it at one of the boundaries so that things are slightly better.\nSpeaker B: As far as hearing is concerned, we can have some gap at some place.\nSpeaker C: So there's only have to keep in mind with the actual physical designers to keep the speaker close enough to the outside.\nSpeaker B: Yeah.\nSpeaker B: So these were the components selection in these things.\nSpeaker B: We can go to the next slide.\nSpeaker B: And these were the findings which I saw with the web.\nSpeaker B: That user wants to have control more than one device, wants to control more than one device from the same remote control.\nSpeaker B: So our TV remote can have little extra things to support additional devices like VCR DVD players which are usually attached with the TV because users are like this that they don't want to have one remote control for everything.\nSpeaker B: With this additional little, we might be having slightly better market for us.\nSpeaker C: It depends.\nSpeaker C: If we are concentrating on a fruit design, then maybe we want to sell a collection of fruit.\nSpeaker C: You have a different fruit for each device.\nSpeaker C: Because sometimes people like to collect more things that are similar.\nSpeaker A: I think that would be funny at the beginning, but after one month, you will be tired of being surrounded of fruit.\nSpeaker C: Well, you still only wanted to do fruit in the first place.\nSpeaker A: No, but I think just one fruit to control everything.\nSpeaker C: Like a power fruit.\nSpeaker A: A power, a power, a mando, a super mando fruit.\nSpeaker B: And as well as I could see on the web, there is crawl button is becoming really a hot thing.\nSpeaker B: And we should have it on the remote.\nSpeaker A: I didn't understand very well this trace speaker lost control.\nSpeaker B: So you are having a base station.\nSpeaker B: Usually your remote sits on that.\nSpeaker B: That's why it can have chargeable batteries.\nSpeaker A: Now let's say you have to buy two things.\nSpeaker A: The banana and the base station.\nSpeaker E: Base station is with the thing.\nSpeaker E: You bought the war thing.\nSpeaker B: It's like a telephone handset is there and the base station for the telephone handset is there.\nSpeaker B: So now what user gets additionally, he doesn't have to buy batteries.\nSpeaker B: They are rechargeable batteries.\nSpeaker B: So all the period of time he will recover the cost.\nSpeaker B: So you are having the base station and there is a button.\nSpeaker B: If you press that button, wherever the remote it will start beeping.\nSpeaker B: So you know where the remote is.\nSpeaker C: I think it's a pretty handy feature.\nSpeaker C: Yeah.\nSpeaker C: I think it's kind of people would find it worth it even if it wasn't a recharging station.\nSpeaker C: Even if they didn't have to buy extra batteries.\nSpeaker C: Yeah.\nSpeaker A: I'm worried about the budget.\nSpeaker B: This is base station is nothing more just it's a wire which is coming from the main cable.\nSpeaker B: And you are having one socket on which the thing sits.\nSpeaker C: I thought you do need to include RF kind of.\nSpeaker C: That's right.\nSpeaker B: But all these things are usually in house so we don't have much problems.\nSpeaker B: So component cost is going to be the least.\nSpeaker B: Anyway, we are not using very advanced technology.\nSpeaker B: LCD has already been ruled out.\nSpeaker B: ASR has been ruled out.\nSpeaker B: So it's the basic thing but very trendy and very user friendly.\nSpeaker C: I'm just wondering actually because you know this whole fruit thing with the banana.\nSpeaker C: It's like a first seems a bit kind of niche.\nSpeaker C: Many people would really want a banana.\nSpeaker C: But what if it was kind of a stylised banana?\nSpeaker C: Rather than having it kind of yellow and really looking exactly like a banana.\nSpeaker C: It could make it kind of silver.\nSpeaker C: And to give you kind of the idea of a banana but without it looking completely kitsch.\nSpeaker C: One for better word.\nSpeaker E: You're using a yellow as kitsch.\nSpeaker C: I don't know how many people.\nSpeaker E: You make some things that look like a banana.\nSpeaker E: You can have the colour of a banana.\nSpeaker B: Otherwise you don't get anything.\nSpeaker B: Maybe like that.\nSpeaker B: It's neither of a banana nor a.\nSpeaker C: Yeah, like this colour.\nSpeaker C: Maybe, you know, maybe like still in the shape of a banana.\nSpeaker E: Roughly.\nSpeaker C: No, exactly.\nSpeaker C: Exactly.\nSpeaker C: But you know, just maybe not exactly the same texture as a banana.\nSpeaker C: And just kind of, you know, because the thing is it's going to be a little bit difficult to make.\nSpeaker C: To give like the texture of a banana anyway and to have the exact shape.\nSpeaker C: I think if you're going to not be able to do it properly you may as well do it in a stylised way.\nSpeaker C: That just looks a bit more kind of, you know, 21st century rather than 60 or 70s.\nNone: Okay.\nSpeaker E: Let's move on.\nSpeaker E: Going to the last slide.\nSpeaker E: Before ending the meeting I'd like to draw some sketch about the future prototype.\nSpeaker D: Go for it.\nNone: Okay.\nSpeaker E: Well, no, not you.\nSpeaker E: You can finish your slides.\nSpeaker B: Okay.\nSpeaker B: So anyway, users will be, so the findings is users will be very interested in our locator divide to find their misplaced amounts.\nSpeaker B: Okay.\nSpeaker B: So that was very, I thought it's a very suggestion by everybody.\nSpeaker E: Okay.\nSpeaker E: That's all.\nSpeaker E: Okay.\nSpeaker E: So we're done for the presentations.\nSpeaker E: So we need to take some decisions about what we're going to do.\nSpeaker E: So I propose that you go to the whiteboard.\nSpeaker E: Yeah.\nSpeaker E: And we're going to report all the ideas we had during these presentations just to draw some sketch about what will be the final product and our super mango banana.\nSpeaker E: And with the basis extra functionalities such as wheels, the speaker you need, well, not too lost the device.\nSpeaker E: I don't remember what you call it.\nSpeaker E: That's right.\nSpeaker E: The basis station.\nSpeaker E: Basis station, yeah.\nSpeaker E: So we're going for a stylish banana shape.\nSpeaker C: Yeah. So I guess you want to hold like the way the end of the banana you want to kind of hold.\nSpeaker C: You may want to kind of hold like a gun rather than because you don't want to point kind of towards the floor.\nSpeaker C: Right.\nSpeaker C: So, you know, so if you have like.\nSpeaker A: What about what are these shape more or less?\nSpeaker B: Well, there's less space on this to put the buttons.\nSpeaker E: Yeah, if it has really more of a banana, you could the starting is good, but it could it should have more of a shape of a banana if you want to point really.\nSpeaker E: I just think if you don't want to do that that movement, which is which is difficult, if you don't have to do it, in fact, it's better.\nSpeaker E: So that time is running. We have to move forward.\nSpeaker E: So let's skip to this this this this idea.\nSpeaker C: Yeah. Okay. So.\nSpeaker E: So we have this we have a basis. How do you call it?\nSpeaker E: The base station. A base station. We'll have a base station extra on the side.\nSpeaker C: Okay. So I guess we need, you know, something that can fit a banana shaped object.\nSpeaker E: Yeah. Yeah. We have a RF for for beating.\nSpeaker E: That's right. Yeah.\nSpeaker E: For beating. We need RF to be.\nSpeaker E: So, so that means we need a button on the basis.\nSpeaker B: Yeah. Basis station.\nSpeaker E: Basis station. Thank you.\nSpeaker C: All right. So we need.\nSpeaker C: Okay.\nSpeaker E: Can you go quickly please? Okay.\nSpeaker E: So we are going to add also your as you suggested, the wheels, some wheels to control the volumes and channels.\nSpeaker E: Yeah. And your turbo, turbo.\nSpeaker C: Yeah. I think it's it's probably best actually on the on the underneath of the device.\nSpeaker E: Yeah. So you have maybe here and the wheel at the level of the thumb, for instance.\nSpeaker C: Yeah. So you have the thumb kind of here and you have two wheels.\nSpeaker C: So yeah. You need one one here and one on the other side.\nSpeaker C: Okay. Right. So you get volume and channel. Good. And so no LCD.\nSpeaker E: No LCD. Okay. Great.\nSpeaker C: Very good. Okay. Oh, we need we need a power on off switch as well.\nSpeaker B: Yeah. For the remote. Yeah.\nSpeaker E: Just don't have power. No, no for the TV for the TV.\nSpeaker B: Okay. So you know that will be controlled by the buttons will be there already.\nSpeaker B: Where? Means on the side.\nSpeaker B: Because remote is going to have both the interfaces, scroll as well as buttons.\nSpeaker B: They are not going to cost you much.\nSpeaker B: Everything is in house and now you don't want the traditional users to be.\nSpeaker C: Apprehensive of this. Well, I don't know if the traditional user is going to buy a.\nSpeaker C: Up and on a remote. Oh, the first place. Yes.\nSpeaker C: That's that's another issue. I mean, you need to kind of keep it.\nSpeaker B: But you know, our targets are very high means 50 million euros is the profit.\nSpeaker C: Which we want to talk about. Yeah. How many of these to be one or so?\nSpeaker B: I can't remember. 25. 25 is the profit on one.\nSpeaker C: Yeah. But how many units did we need to sell?\nSpeaker B: 44 million. 4 million. 4.4 million.\nSpeaker C: 0.4 million. That's a lot of fruit.\nSpeaker E: Yeah. In the market. What about so?\nSpeaker E: Well, no time is running. We have to close the meeting in a few minutes.\nSpeaker E: So, okay. The next step you can come back to sit.\nSpeaker E: The next step is to go for is to go to building a prototype based on this.\nSpeaker E: Okay. Okay. So next meeting you guys have to prefer the fully things.\nSpeaker E: You have to work on the look and feel design and you have to work on the user interface.\nSpeaker E: In fact, you two, you have to work together to model the first prototype.\nSpeaker E: Marketing expert has to go to product evaluation.\nSpeaker A: Okay. What about adding this key word spot in recognition saying volume up, volume down?\nSpeaker A: It's too difficult. It's too difficult. But people like innovation.\nSpeaker A: And that's really innovative. And I don't know if it would cost a lot.\nSpeaker A: There's a few, five words.\nSpeaker E: It's not possible to implement it for the next prototype. So it's in the next prototype.\nSpeaker B: So let's keep it. Okay. For the future of prototypes.\nSpeaker E: Yeah. Maybe. If it works well, we will go for the range of banana plus plus.\nSpeaker E: Yeah. First, yeah. Yeah. All other fruit.\nSpeaker E: Thank you. Thank you.\nSpeaker E: Thank you. Thank you.\nSpeaker E: Thank you.\nSpeaker C: Thank you.\nSpeaker C: Thank you.\nSpeaker B: Thank you.\nSpeaker B: Thank you.\nSpeaker B: Thank you.\nSpeaker B: The problem is after all this meeting that is down.\nSpeaker B: It's a meeting.\nSpeaker B: It's a meeting.\nNone: Thank you.\n"}, {"meeting_id": "IS1009d", "summary": "Project Manager first recapped the previous meeting, concluding that the remote should be compact, user-friendly and could function like speech recognition. Next, the presentation of the bright blue snail shaped prototype was made by Industrial Designer and User Interface, introducing the appearance and features of the remote. Then, feedback was given regarding the prototype, which was mainly positive and speech recognition and cute appearance were decided as its star feature. Afterwards, the team discussed the financing of the remote, which should be within the budget of twelve fifty euros. Finally, the meeting ended with a project evaluation, which everyone was satisfied with the product and dynamic of the team, and team building.", "dialogue": "Speaker C: Okay, okay.\nNone: I'll let the arches go first.\nSpeaker B: Okay, oops.\nNone: How did I do this before?\nSpeaker A: Well, hi everyone again.\nSpeaker A: Hello.\nSpeaker A: Like before, I have to reduce the meetings from the minutes from the last meeting.\nSpeaker A: And so here we go.\nSpeaker A: It was discussed in the last meeting, which was opened by the presentation from the interface designer that looks would be very important on this new remote.\nSpeaker A: And it is to send messages, of course, to the TV.\nSpeaker A: Switch to control features, color, contrast, sharpness, etc.\nSpeaker A: You would have a memory switch, a mute button, in case the telephone rings or something.\nSpeaker A: Beach recognition is one of her very favorite personal features she would like to be integrated in this new remote.\nSpeaker A: It should be a child-friendly design with few buttons, colorful, maybe with star-shaped or other-shaped buttons.\nSpeaker A: She presented also an oversized remote, which she guarantees nobody will ever be able to lose.\nSpeaker A: She was challenged on that point.\nSpeaker A: But her very personal favorite really, she would very much like to see a speech recognizer integrated in this remote.\nSpeaker A: The industrial designer presented her thoughts on the issue.\nSpeaker A: She would like a special case made out of plastic that is very strong, not using any harmful materials.\nSpeaker A: It should be recyclable and should be colorful.\nSpeaker A: It should have an integrated circuit board that's highly sophisticated and temperature-resistant.\nSpeaker A: She would like to see a timer and or alarm facility integrated.\nSpeaker A: Technically, this thing would also have a resistor and a capacitor, a diode, a transistor, resonator, and, if possible, a rechargeable battery.\nSpeaker A: And, of course, the circuit board and how it would work, you press the button, the chip is more code-related to the generator amplification.\nSpeaker A: The circuit board is very inexpensive to build.\nSpeaker A: And so she thinks this is a great feature to consider.\nSpeaker A: This whole thing should be push buttons with a simple chip.\nSpeaker A: The scrolling method is more expensive and not that practical anymore.\nSpeaker A: It should be battery-operated and, of course, should have the special cases.\nSpeaker A: The marketing expert who has to finally come up with to market this product has been watching the competition, has done some research on the internet, and also has used personal observations to come up with the fact that such a remote should be small, easy to use, and it should be eye-catching.\nSpeaker A: From her point of view, of course, one of the most important facts is that we should get to market before our competition does.\nSpeaker A: To do that, maybe one or two features should be developed on which we could dwell on, or in other words, on which our campaign could be built on.\nSpeaker A: Too many new features or too many points would only confuse matter.\nSpeaker A: So we prefer to have one or two features that can be really driven home.\nSpeaker A: It should have a fruit and vegetable design and should have a soft feel.\nSpeaker A: She feels that's really what people want today.\nSpeaker A: The decision that we took last time was that the special feature we would like to see is a speech recognizer.\nSpeaker A: The energy should be battery, should be on a chip, should be drain design, compact and strong, and should have buttons.\nSpeaker A: And that concludes the presentation from the last minutes, from the last meeting.\nSpeaker A: Now, we are ready for the presentation of the prototype.\nSpeaker D: Just the look-alike, the buttonpad I'll explain.\nSpeaker E: So this is what we have made. This is a model of the remote controller which we are going to build.\nSpeaker E: This is in a snail shape. It is attractive and it's blue in color, bright.\nSpeaker E: It has yellow buttons and all the different color buttons.\nSpeaker E: So it is beautiful and compact in shape.\nSpeaker E: It will be easy to fit into the hands and you can access all the buttons easily.\nSpeaker E: You used to have all the buttons.\nSpeaker E: The material which we are going to use for the case is plastic which is strong.\nSpeaker E: The material is plastic and for the buttons it is soft rubber.\nSpeaker E: Oh, that's good.\nSpeaker E: That's awesome.\nSpeaker E: You will be touching the buttons more so it is soft when you touch it.\nSpeaker E: For the light emitting diode it is a fluorescent green and it is a bulb like an ordinary infrared.\nSpeaker E: And the button's part will be explained by Francina.\nSpeaker D: We decided upon including certain features are remote.\nSpeaker D: Now these features include the signal emitting signal.\nSpeaker D: It's the LED or LED infrared.\nSpeaker D: Now we have included the switch on and off button.\nSpeaker D: Now we have included another feature that is the mute button on the side of the model.\nSpeaker D: Then we have included one to nine buttons for controlling the programs, the different channels.\nSpeaker D: We have also included two buttons for increasing or decreasing the volume.\nSpeaker D: And we have also included two buttons for scrolling up and scrolling down the program channels.\nSpeaker D: Our model also contains a button which is called as a menu button.\nSpeaker D: A menu button at the center.\nSpeaker D: We have included a button which is fluorescent green color and this is the menu button which will control the color sharpness, brightness of the screen of this picture.\nSpeaker D: We have also included a button which is called as the swapping button.\nSpeaker D: Now this is a special feature which we have included.\nSpeaker D: Now this button is an elongated shaped button and this is slightly flexible.\nSpeaker D: So if it is turned towards the right it will take to the previous channel.\nSpeaker D: If it is turned towards the right it will take to the next channel.\nSpeaker D: It will take the user to the previous and the next channel.\nSpeaker D: So this is a swapping button.\nSpeaker B: The next channel in the numeric pattern or?\nSpeaker D: No, swapping is if example you are watching the second channel and then you go to the tenth channel.\nSpeaker D: And if you want to go back to the second channel you can swap this button.\nSpeaker B: Okay.\nNone: Yes.\nSpeaker D: And at the end this remote has inbuilt voice recognizer which will recognize the user's voice and then it will act accordingly.\nSpeaker D: So this is a proposed model.\nSpeaker D: The marketing expert has to give her suggestion and it will be sellable.\nSpeaker D: It will request a vector.\nSpeaker B: Okay.\nSpeaker B: Well, what I really like a lot about it is that you can reach the whole thing with one thumb.\nSpeaker B: Yes.\nSpeaker B: You can really hold it in one hand.\nSpeaker B: You don't need two hands and it's easily reachable even for somebody with a small hand.\nSpeaker A: Yes.\nSpeaker A: The buttons are all raised.\nSpeaker B: And if you hold it in the center of your hand you can even reach it over here.\nSpeaker B: Right.\nSpeaker B: So you don't have to turn it around, turn it upside down, move it up, up and down.\nSpeaker B: I really like that.\nSpeaker B: You really did a good job on that.\nSpeaker B: My little designers.\nSpeaker B: And I like the idea that the on off button is in a really prominent place.\nSpeaker B: That's a really good thing.\nSpeaker B: Yes.\nSpeaker A: And it sort of sticks up.\nSpeaker A: So you don't have to first go like, oh yeah, here it's on.\nSpeaker B: The color is very attractive.\nSpeaker B: These buttons around here are the mute.\nSpeaker D: No, these the front buttons here are the mute buttons.\nSpeaker D: On both sides?\nSpeaker B: Yes.\nSpeaker B: So you can push either one.\nSpeaker B: Yes.\nSpeaker A: Okay.\nSpeaker A: So if you left handed or right handed you the button.\nSpeaker B: And this brings the menu up on the screen.\nSpeaker B: Yes, yes.\nSpeaker B: And the orange ones are.\nSpeaker D: These two are the to increase or decrease the volumes.\nSpeaker D: Okay.\nSpeaker D: And these two are to scroll the program channels.\nSpeaker B: Okay.\nSpeaker D: All right.\nSpeaker D: Roll up was scroll down the channels.\nSpeaker B: Looks like something I can sell.\nSpeaker B: Okay.\nSpeaker B: And now I'm supposed to.\nSpeaker A: I have one question.\nSpeaker A: Will there be anything written on the buttons like that people know or they have to learn that from a piece of paper which button does what?\nSpeaker D: Yes.\nSpeaker D: These buttons will have the numbers and all the rest of the buttons will have symbols.\nSpeaker D: Will have symbols so that that can be easily.\nSpeaker A: That you really know, you know, it doesn't have to.\nSpeaker B: First point because we need the symbols because we're going into an international market.\nSpeaker B: We can't have anything with language dependent.\nSpeaker A: But anyway, it has to have some kind of symbols or text or something.\nSpeaker A: Yes.\nSpeaker E: We can have on the case.\nSpeaker E: That's right.\nSpeaker A: We can be printed on the case.\nSpeaker E: Okay.\nSpeaker D: I just want to make sure of that.\nSpeaker D: And one more feature is we have a holder for this remote which is an oyster shape.\nSpeaker D: A shell shape.\nSpeaker B: For the snail.\nSpeaker B: Right.\nSpeaker B: We have the snail shell.\nSpeaker B: He goes right back into his shell.\nSpeaker B: Right.\nSpeaker B: Well, you know, I think we could do something really funny with this too because the snail is known to be slow.\nSpeaker B: And we could have some sort of little comic effect on our marketing about how this is a rapid snail or something like that.\nSpeaker B: You know, that would really work.\nSpeaker A: Now, what are our special features for the marketing that's really the voice recognition that's really unusual?\nSpeaker B: I think voice recognition is our big selling point because nobody else seems to have that in this price range.\nSpeaker A: And then the other thing would basically be shape or practicality or fuse.\nSpeaker A: Yeah.\nSpeaker B: Well, I think that everybody is going to say their remote control is practical.\nSpeaker B: I think we have to do well on the appearance.\nSpeaker B: You're really going to have the best, the cutest remote control on the block.\nSpeaker B: So I think we have to play with the image, play with the snail image, play with the visual, and then the voice recognition.\nSpeaker B: I think those are the two things to push.\nSpeaker B: The look and the voice recognition, they're going to be our two selling points.\nSpeaker A: Okay.\nSpeaker A: Now, having said that.\nSpeaker A: I'm supposed to make a little presentation on it.\nSpeaker A: No, now this was our evaluation criteria which we just have done.\nSpeaker A: Now we're going to talk about financing.\nSpeaker C: Ah.\nSpeaker B: But in my instructions, I think it's how I was supposed to go to the board and do something.\nSpeaker A: Well, there's a production evaluation, is that you?\nSpeaker A: Yeah, that's me.\nSpeaker A: But that's after the financing.\nSpeaker B: Oh, okay.\nSpeaker B: Sorry.\nSpeaker B: Sorry.\nSpeaker B: Sorry.\nSpeaker A: Okay.\nSpeaker A: We had looks and voice recognition.\nSpeaker A: Okay.\nSpeaker A: Now, on the financing, we bring up the...\nSpeaker A: There it is.\nSpeaker A: Okay.\nSpeaker A: Energy source, we say that's battery, right?\nSpeaker A: That's right.\nSpeaker A: Okay.\nSpeaker A: Now.\nSpeaker A: So we...\nSpeaker A: I guess we use one.\nSpeaker D: Yeah, yes.\nSpeaker D: What?\nSpeaker A: The seller chart, you're trying to change this project.\nSpeaker A: Oh, that's nice.\nSpeaker A: She told me I could just change it here and then it would...\nSpeaker A: It doesn't work.\nSpeaker A: Hmm.\nSpeaker B: Can you just fill it in the yellow boxes or...?\nSpeaker B: Oh, okay.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Let's see.\nSpeaker A: Okay.\nSpeaker A: One.\nSpeaker A: Okay.\nSpeaker A: I'll go away.\nSpeaker A: Kinetic solar.\nSpeaker A: So that's in the energy source, that's all we need.\nSpeaker A: Electronic simple chip on print, is that what we're using?\nSpeaker A: One of those.\nSpeaker A: Come on.\nSpeaker A: Okay.\nSpeaker A: One.\nSpeaker A: Regular chip on print?\nSpeaker A: No.\nSpeaker A: No.\nSpeaker A: That's all we need.\nSpeaker A: The one case.\nSpeaker A: Uncurved flat, single curved, double curved.\nSpeaker B: Oh.\nSpeaker E: This is a single curved.\nSpeaker A: I guess it's double curved.\nSpeaker A: Double curved.\nSpeaker A: One of those.\nNone: Mm-hmm.\nSpeaker A: Case materials supplements.\nSpeaker A: Plastic, we said, right?\nSpeaker A: Plastic.\nSpeaker A: Would rubber?\nSpeaker B: Rubber because we're going to have the soft buttons.\nSpeaker A: Uh, but yes, but...\nSpeaker E: I think that is a...\nSpeaker E: That's just for the case materials.\nSpeaker A: Okay.\nSpeaker A: Special color zone.\nSpeaker A: We're having that, right?\nSpeaker A: Yes.\nSpeaker A: And then we have the interface push buttons.\nSpeaker A: Mm-hmm.\nSpeaker A: Scroll wheel no, integrative scroll wheel LC display.\nSpeaker B: No.\nSpeaker B: We didn't put the clock in it after all, right?\nSpeaker B: No.\nSpeaker A: Uh, button supplement special color.\nSpeaker A: Yes.\nSpeaker A: Special form.\nSpeaker A: Yes.\nSpeaker D: Yes.\nSpeaker D: We do have special form.\nSpeaker A: And special material rubber would, yes.\nNone: Yeah.\nSpeaker A: Okay.\nSpeaker A: Total 7.6, whatever that means.\nSpeaker E: I think that's the...\nSpeaker A: That's the price.\nSpeaker A: Mm-hmm.\nSpeaker A: You should just...\nSpeaker A: 8.2.\nNone: Nice points.\nSpeaker D: Okay.\nSpeaker A: Yes.\nSpeaker A: Okay. Yes.\nSpeaker A: 8.2, right?\nSpeaker A: So we...\nSpeaker A: Looks like we are well within budget.\nSpeaker A: Mm-hmm.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: I guess I should save this, I suppose, huh?\nSpeaker B: Yeah.\nSpeaker C: Oops.\nNone: Ha-ha-ha.\nSpeaker D: And the desktop?\nSpeaker A: I just tried that.\nSpeaker A: My documents computer.\nSpeaker A: My computer.\nSpeaker A: I don't know.\nNone: Okay.\nSpeaker A: Fine.\nSpeaker A: Save.\nSpeaker A: Okay.\nNone: Good.\nSpeaker A: So that's the good news.\nSpeaker A: You're going to be popular.\nSpeaker A: Mm-hmm.\nSpeaker A: Um...\nSpeaker A: So that, I think, financing was pretty simple.\nSpeaker A: Mm-hmm.\nSpeaker A: Now we would like to have a presentation by the marketing expert.\nSpeaker A: On production evaluation.\nSpeaker A: Okay.\nSpeaker B: I'll take my file down so you can bring it up.\nSpeaker B: Okay.\nSpeaker B: Should be able to get it now.\nNone: Okay.\nSpeaker B: Why don't you move just to the next slide right away?\nSpeaker B: You want to go to the next slide?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Well, obviously my method for the marketing of this thing is first asked the big question.\nSpeaker B: Will it sell?\nSpeaker B: And I think we should show this prototype to people from various age and socioeconomic groups and see about any fine-tuning that maybe little things that we haven't thought of.\nSpeaker B: We can't accept every suggestion, of course, but maybe we just need to get a few.\nSpeaker B: And show the prototype to consider research groups who want somebody to suddenly come to us and tell us that this button is toxic and some child will swallow it and then we won't sell any.\nSpeaker B: So we have to get some input from those people.\nSpeaker B: And then after that we just have to go with our best intuition.\nSpeaker B: And we like it.\nSpeaker B: We think it's good.\nSpeaker B: We're going to get behind it and sell it.\nSpeaker B: Next slide, please.\nSpeaker B: Okay.\nSpeaker B: Now the things that I was thinking and my wish list has really been realized in this prototype.\nSpeaker B: I wanted the shape to be biomorphic.\nSpeaker B: I didn't want anything with angles on all square.\nSpeaker B: I wanted it to be comfy and roundy.\nSpeaker B: So we've got that.\nSpeaker B: The size is small.\nSpeaker B: The color is bright and warm, which is what we wanted.\nSpeaker B: We wanted the feel to be as soft as possible.\nSpeaker B: We'll have the soft buttons.\nSpeaker B: And the way this is shaped, even though it's going to be hard plastic, it feels good in your hands.\nSpeaker B: So that's nice.\nSpeaker B: And functionality I put last on my list because people aren't going to use it before they buy it.\nSpeaker B: So paradoxically the other features, in other words, the look, the feel, and the shape, that's what people are going to get in the store.\nSpeaker B: They don't have a television in the store.\nSpeaker B: They can't play with it.\nSpeaker B: So they'll be our main selling points.\nSpeaker B: So those have been fulfilled by your prototype.\nSpeaker B: And go ahead to the next slide, please.\nSpeaker B: Okay.\nSpeaker B: So the shape, I think, is a one.\nSpeaker B: That's really excellent shape.\nSpeaker B: The size is small.\nSpeaker B: And these points are in the importance for the marketing.\nSpeaker B: These aren't in how I feel.\nSpeaker B: I think that it's plenty small enough to sell, but I think the scale is one to seven.\nSpeaker B: I think we're right in the middle as far as other competitors.\nSpeaker B: And our color, I think, is great.\nSpeaker B: The colors are bright and warm, and we really do a great job there.\nSpeaker B: And given the constraints that we had, I think we got it as soft as possible.\nSpeaker B: And then functionality, I think you did a really good job on functionality.\nSpeaker B: Obviously we could have added different functions, but then we disturbed something else.\nSpeaker B: So I would say that we got to a five out of seven on functionality.\nSpeaker B: So I think that basically we've got a great product, and we can get off and running with it.\nSpeaker A: I just realized one thing.\nSpeaker A: In the financing, there was no room for our voice recognizer.\nSpeaker A: And I don't know how we can evaluate that or how we can include that.\nSpeaker B: Well, we had what, eight euros, 20 as our cost?\nSpeaker B: Yes.\nSpeaker B: And so we've got four euros to go to spend.\nSpeaker A: And then in the same time, we have another 4.3 euros, I mean, 430.\nSpeaker A: But we have no way of presenting that to management as a finished product and saying, okay, we always recognize that costs so much.\nSpeaker A: So we just have to be aware of that.\nSpeaker A: And know whether the 430 will really cover that.\nSpeaker B: Well, as we know in today's technological world, you can do just about anything at any price, the problem is quality.\nSpeaker B: So we're just going to have to settle for whatever quality that will buy us.\nSpeaker B: And it may not be the greatest quality, but it may sell anyway.\nSpeaker B: Yeah.\nSpeaker B: As we've seen with so many of these kind of products.\nSpeaker A: I'm sorry to interrupt them, but I just remembered that there was no, that was not included, that there was no room for any special features.\nSpeaker A: Okay.\nSpeaker A: So to be aware of that, you want to go to the next slide, marketing expert?\nSpeaker A: Well, isn't this my last slide?\nSpeaker B: I don't know.\nSpeaker A: Yes, it is.\nSpeaker B: This is my last slide.\nSpeaker B: And I'm supposed to present the scale on the whiteboard.\nSpeaker B: And we're supposed to talk about those things as a team now.\nSpeaker B: So if you put my last slide back up there.\nNone: Oh.\nSpeaker B: I'm sorry.\nSpeaker B: I got to get.\nSpeaker A: Wow.\nSpeaker A: Why you need that up?\nSpeaker B: Well, because I can't remember what I put on there.\nSpeaker B: Okay.\nSpeaker B: Now I'm supposed to see how long my leash is here.\nSpeaker B: I think you can make it there.\nSpeaker B: Okay.\nSpeaker B: Alrighty.\nSpeaker B: So now we're all supposed to say what we think.\nSpeaker B: Okay.\nSpeaker B: So on shape.\nSpeaker B: I gave it one.\nSpeaker B: What would you, one being good and seven being the worst?\nSpeaker B: It was okay.\nSpeaker B: What do you think the shape is?\nSpeaker B: One.\nSpeaker B: One.\nSpeaker B: Okay.\nSpeaker B: And Betsy?\nSpeaker A: Yes.\nSpeaker A: I think shape is one.\nSpeaker A: Okay.\nSpeaker B: And how about on size?\nNone: On size.\nSpeaker B: You gave it a four.\nSpeaker B: I gave it a four.\nSpeaker B: Yeah.\nSpeaker B: I feel it's just average.\nSpeaker A: I don't know.\nSpeaker A: I think I would give it at least a two.\nSpeaker A: Okay.\nSpeaker A: One.\nSpeaker D: Mm-hmm.\nSpeaker E: Yeah.\nSpeaker E: You and I think it is one.\nSpeaker E: Okay.\nSpeaker B: It's quite small.\nSpeaker B: Okay.\nSpeaker B: You're the designer.\nSpeaker B: Of course you want to give it one.\nSpeaker B: And then how about how are we doing on color?\nSpeaker B: Color.\nNone: Color.\nSpeaker C: I gave it one.\nSpeaker B: I really like all those nice bright, warm colors.\nSpeaker A: I like the colors.\nSpeaker E: One.\nSpeaker E: One.\nSpeaker B: One.\nSpeaker B: One.\nSpeaker B: Okay.\nSpeaker B: And how about the feel?\nSpeaker B: Taking into consideration texture and comfort in the hand.\nSpeaker A: I think I will give it a two.\nSpeaker B: Okay.\nSpeaker B: I gave it a three.\nSpeaker B: Two.\nSpeaker B: I gave three.\nSpeaker B: Three.\nSpeaker E: Maybe two.\nSpeaker E: Two.\nSpeaker B: Okay.\nSpeaker B: And the next is functionality.\nSpeaker B: Where I admit I was a little hard on our team here.\nSpeaker B: But.\nSpeaker A: Well, it's also you can't really try it out.\nSpeaker A: Yeah.\nSpeaker A: The other things you have more, it's more tangible.\nSpeaker A: So from that point of view, but, um...\nSpeaker A: I'll give it a three.\nSpeaker B: Okay.\nNone: Two.\nNone: Two.\nNone: Okay.\nSpeaker B: Three.\nSpeaker B: Three.\nSpeaker E: Okay.\nSpeaker E: Well, um...\nSpeaker B: It looks like we've got ourselves a pretty good product.\nSpeaker B: Um, functionality is the only place where maybe we have to think about maybe having forbid having another meeting.\nSpeaker B: But, um...\nSpeaker B: Otherwise, I think we're ready to go with this product.\nSpeaker B: Anybody else have any other comments or any other things that we feel we should evaluate?\nSpeaker C: Uh...\nSpeaker A: Here is what we're looking at.\nSpeaker A: Uh, satisfaction on, for example, room for creativity.\nSpeaker A: Um...\nSpeaker A: Is there more room for creativity or are we absolutely happy?\nSpeaker D: We can always improve.\nSpeaker E: Yes.\nSpeaker E: Include some more buttons and the features.\nSpeaker E: Yeah, features.\nSpeaker E: We can make the buttons, few buttons smaller.\nSpeaker E: Uh, I think they are quite big.\nSpeaker E: So I think, I mean, we can just have small buttons and more buttons in that case if we want to have more features in that.\nSpeaker B: Well, then again, if we're gonna, um...\nSpeaker B: Do the speech recognition thing, we're gonna, there's gonna be some buttons that are gonna have to be added for that for the recording of the speech.\nSpeaker B: So that's where we're gonna have to do, maybe we can eliminate one of the mute buttons instead of having two mute buttons.\nSpeaker B: And, um, then maybe we can do something with the, um...\nSpeaker B: The volume control, maybe we can put that all on one button.\nSpeaker B: Um...\nSpeaker B: And a couple of other, maybe consolidate some of the usage and see what we can do with that.\nSpeaker A: Yeah, obviously, bearing in mind that right now we are, of course, well within the budget and that we still, you know, we probably can't...\nSpeaker A: With this particular item, we probably can't just add a whole lot of more things.\nSpeaker A: No.\nSpeaker A: Um, we need the...\nSpeaker A: You know, we need to leave space money-wise for the voice recognizer.\nSpeaker A: Mm-hmm.\nSpeaker A: So the question really is, how do we feel with the project process?\nSpeaker A: Um...\nSpeaker A: Are we happy with the creativity that has passed here, or we're not happy with the new product we created, or that was created?\nSpeaker A: Uh, I think, personally, I think, uh, I'm pretty happy.\nSpeaker A: I'm pretty happy with the show now.\nSpeaker A: Um...\nSpeaker A: That's something I think I can market.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: And then the next question is, are we happy with the leadership of this project?\nSpeaker A: Uh...\nSpeaker B: I think you've done a good job, Ms. Leedor.\nSpeaker B: Yes.\nSpeaker A: And I think teamwork, I think, was very, very good.\nSpeaker A: I think we were, I think, well, together as a team, yeah.\nSpeaker A: Uh-huh.\nSpeaker A: And, uh, I think we...\nSpeaker A: Are we happy with the means we used, we used whiteboard, we didn't use digital pens, I guess, are these things?\nSpeaker B: Yeah, maybe we could have used the whiteboard a little bit more.\nSpeaker A: Yeah.\nSpeaker A: We didn't use that enough.\nSpeaker A: Yes, we could.\nSpeaker A: Uh, it's maybe not in the best position in the room.\nSpeaker A: Uh, you know, like, sometimes it's positioned so that it's much better visible for everybody.\nSpeaker A: Uh-huh.\nSpeaker A: And I think from that point, if you, we sort of ignored it a little bit.\nSpeaker A: And we used the slide because it was better position.\nSpeaker B: Yes, I think so.\nSpeaker B: I think that's true.\nSpeaker A: I think that's true.\nSpeaker A: And, fortunately, we all had slides.\nSpeaker A: Presentation, which made it a little easier.\nSpeaker A: Uh-huh.\nSpeaker A: Um, did we find new ideas?\nSpeaker A: I think we did.\nSpeaker A: I think we were very good.\nSpeaker A: I think we did, uh, in more than one respect.\nSpeaker A: And, uh, so I think we did very well here.\nSpeaker A: Okay.\nSpeaker A: Are the costs within budget?\nSpeaker A: Yes, yes.\nSpeaker A: Uh, is the project evaluated?\nSpeaker A: Yes, yes.\nSpeaker A: Yes.\nSpeaker A: Uh, so...\nSpeaker A: And celebration.\nSpeaker A: And celebration.\nSpeaker A: If you want to see this.\nSpeaker B: Yes.\nSpeaker B: Today we have apple juice and after we sell a million of them, we have champagne.\nSpeaker A: So, I thank you all very much.\nSpeaker A: Um, I think this was very good.\nSpeaker A: And, um, I think we did come up with a new product that's, uh, feasible.\nSpeaker A: Visible from the production point of view and feasible from a marketing point of view.\nSpeaker A: Uh-huh.\nSpeaker A: So, okay.\nSpeaker A: Thank you.\nSpeaker A: Thank you.\nSpeaker A: Okay.\nSpeaker B: Watch, I have my cord behind you.\nSpeaker B: Okay.\nNone: Okay.\nSpeaker A: Oops.\nSpeaker A: I always get it on there, but getting it off is...\nSpeaker D: Oh, you have time to put me on there.\nSpeaker B: But we were told we could end the final meeting at any time.\nSpeaker B: Yeah.\nSpeaker B: Whenever we felt we were friends.\nNone: Yeah, because we were friends.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: I'm going to get my microphone out for my necklace.\nSpeaker B: Now there we go.\nNone: Okay.\n"}, {"meeting_id": "ES2004c", "summary": "After a brief summary of the last meeting, the team discussed the conceptual design for new remote control, which covered the properties, materials, user interface and trend watching. Marketing shared feedback about the design of the device. Industrial Designer gave a presentation on the design and availability of actual components, whereas User Interface led the discussion about customer needs and feasibility on design and competitors' study. A final decision was reached regarding the design of the remote control such as using kinetic charging and a rubber cover.", "dialogue": "Speaker D: I'll wait till you're all in.\nSpeaker D: Thank you.\nSpeaker D: Welcome back everybody.\nSpeaker D: I love you.\nSpeaker D: I had fun.\nSpeaker D: Right.\nSpeaker D: This is our conceptual design meeting.\nSpeaker D: I think we're starting to head over now from the last meeting.\nSpeaker B: Yeah.\nSpeaker B: I was getting the question.\nSpeaker D: I think, yeah.\nSpeaker D: This is where we talk about properties, materials, user interface and trend watching.\nSpeaker D: So I think we've touched on a few things to that already, but we'll just go over it.\nSpeaker D: The minute from last time, we had a couple of changes in our plans in that we couldn't use teletext.\nSpeaker D: It wasn't going to be a control for everything.\nSpeaker D: And that we had to incorporate the image of a company into it from here.\nSpeaker D: We have decided on leaving out the voice recognition.\nSpeaker D: We've decided on there being a flip design and a different shape from what's normal.\nSpeaker D: We were thinking a shell, but something along the sides, just a different shape from what's normal.\nSpeaker D: We were going to look into the rechargeable batteries.\nSpeaker D: Yep.\nSpeaker D: And we were going to think about the ports and the alarm for getting it long.\nSpeaker D: Things like that.\nSpeaker D: And our market was going to be young, business kind of range.\nSpeaker D: Okay.\nSpeaker D: So she was starting with the first presentation.\nSpeaker E: Yes.\nSpeaker E: Just connect up.\nSpeaker E: Okay.\nSpeaker E: So we decided on our market, and so this feedback from the marketing department is really about trend watching.\nSpeaker E: The methods we use to decide on current trends and so on, market trends, we did our traditional, our usual market research study with our hundred subjects.\nSpeaker E: The general feedback from them is that the most, I'm sorry, I'm slightly tongue tied after lunch, sorry.\nSpeaker E: We decided on the most important aspect required in a remote control device and we'll come to that later.\nSpeaker E: So that was one of the first things we did.\nSpeaker E: Also with our company being full runners and putting fashion into technology, we also looked at a fashion update using our consultants on fashion and design in Paris and Milan.\nSpeaker E: So the general findings from that was in market trends, the most important aspects for remote controls were people want a fancy look and feel rather than the current functional look and feel of remote controls.\nSpeaker E: They also want a remote control to be technologically innovative and of course as you predicted that it should be easy to use.\nSpeaker E: Now I should point out that the first of those findings, fancy look and feel is the most important, it's twice as important as the second technologically innovative which is in turn twice as important as being easy to use.\nSpeaker E: So possibly that the feature that we put first is actually third, it's still important but it's third in order of preference for the subject that we looked at.\nSpeaker E: Okay.\nSpeaker E: Now the function update which relates to very personal preferences among our subject group, we found from our consultants in Paris and Milan who attended all the design and fashion fears then, but fruit and vegetables are going to be the theme for clothes, shoes and furniture.\nSpeaker E: So should we be thinking of using something like that in our remote control design too?\nSpeaker E: There also seems to be a trend towards a spongy feel to materials again in contrast to last year.\nSpeaker E: So a lot of interesting feedback, they are both in our market research departments and from the people in Paris and Milan.\nSpeaker E: You're interested to see what our design people make of that.\nSpeaker D: Okay, thank you very much.\nSpeaker D: Let's start from the inside and work our way out.\nSpeaker E: Yeah, okay.\nSpeaker E: Unless anyone has any questions about that?\nSpeaker E: Not yet.\nSpeaker E: Okay.\nSpeaker E: Yes.\nSpeaker D: Thank you.\nSpeaker D: Thank you.\nSpeaker D: Yes.\nSpeaker D: Thank you.\nSpeaker B: Thank you.\nSpeaker D: I hate this bit of things.\nSpeaker D: I wish if you do them on too tight and you can't get the damage on them to it on the end.\nSpeaker A: Okay.\nSpeaker B: This is all about the design of the edge components and the availability of the components we have in stock at the moment.\nSpeaker B: Okay.\nSpeaker B: First, the into the method, the main idea of the whole thing is just so you just press buttons and it should activate things on the TV and we discussed last meeting.\nSpeaker B: It should be easy to find.\nSpeaker B: Also in this study, I've looked into the availability of some of the materials because some of the things that we looked at last time, unfortunately aren't available.\nSpeaker B: Findings.\nSpeaker B: I've got my nice little picture there.\nSpeaker B: This is the chip called the TA-1-835, which is what's used in pretty much every remote control because it sends out standard signals based on your input.\nSpeaker B: It's pretty much used for all TV remotes at the moment.\nSpeaker B: Then we're looking into battery options.\nSpeaker B: There's actually no rechargeable option available.\nSpeaker B: I saw the standard AA and AAA, which we thought were a bit too bulky at the moment.\nSpeaker B: I thought that might be silly to be honest.\nSpeaker B: It's basically like wind up radio.\nSpeaker B: So you wind up your remote control before you use it.\nSpeaker D: How long can you get out of that?\nSpeaker D: I mean, can you pick it up and then wind it for two minutes and then that's it for the night?\nSpeaker B: Yeah, I presume you wouldn't have to wind it very long, but I don't think it's really necessary when you think of the next two options like the solar charging because most people have the light on in the room anyway, so they could get when you've got TV.\nSpeaker E: Does light charges us somewhat?\nSpeaker D: Is it?\nSpeaker B: Is it?\nSpeaker B: Alright.\nSpeaker E: It has to be silly.\nSpeaker C: Any?\nSpeaker C: Regarding the sizes, which one you think will be like because we have to take into consideration size also, so maybe standard AAA might take less space and the dynamo might take more space.\nSpeaker B: Yeah, it would.\nSpeaker B: The dynamo would take more space because you'd actually need a physical sort of handle to wind up.\nSpeaker D: I'm pretty sure that solar is from the sun.\nSpeaker B: Yeah, okay.\nSpeaker D: I don't think it's parents' perspective.\nSpeaker D: But I mean, you don't want to limit your market.\nSpeaker D: I mean, people who live in base with that.\nSpeaker D: There's not that many people.\nSpeaker D: I know a different part of the world too.\nSpeaker E: If we're at the marketing internationally.\nSpeaker B: Yeah, and most people also watch TV in the night anyway.\nSpeaker D: But then it would be charging for the day.\nSpeaker D: I think the point is they charge us through the day and then you've got it charged.\nSpeaker B: But I think the next one's the best anyway.\nSpeaker B: The kinetic charging, which is like you get it in wrist watches and you don't even notice it.\nSpeaker B: The amount of probably moving around you'd be doing all the time would charge it up.\nSpeaker B: I don't think it ever need to actually physically start shaking it up to make it work.\nSpeaker C: And are this like a, what is the life of the kinetic battery?\nSpeaker C: Like it's a transfer long time?\nSpeaker B: Yeah, it charges into some form of, it's a smaller cell which it charges into.\nSpeaker B: And the size of a watch battery because they use them quite frequently in watches.\nSpeaker B: And that would last for, well I'm not entirely sure how long it would last but I've never seen one run out.\nSpeaker D: But then if you think about watch, it's on your whole body, it's on your body the whole time.\nSpeaker D: So you're walking around you're doing things.\nSpeaker D: It is moving a lot of the time.\nSpeaker D: Yeah, and you switch the TV on the inside then you pick it up to change it.\nSpeaker D: And you put it on the side.\nSpeaker D: Yeah, but then again.\nSpeaker B: I think it is because if you think about it, the watch, although it's only a tiny amount, it's always moving for the whole day.\nSpeaker B: And they don't run out overnight when you leave them on the side.\nSpeaker B: And for the same reason you're only using it for an incredibly short amount of time just to send the signal and then you've finished with it.\nSpeaker B: It's not withdrawn.\nSpeaker B: No, no, I don't think the...\nSpeaker E: Could I just start?\nSpeaker E: Referring back to solar charging, is that compatible with standard batteries?\nSpeaker E: I mean, could people put standard batteries in but they could leave it in the sunlight for solar charging?\nSpeaker E: No, the two things are not compatible.\nSpeaker B: Yeah, I think you could have a dual power thing.\nSpeaker B: But the thing with the solar cells is you would need to put them on the case.\nSpeaker B: And the other thing that's happening is that they're not only on a calculator, they're quite big and they all look at identical.\nSpeaker E: That effect, the idea of design.\nSpeaker D: They're expensive.\nSpeaker D: The price they were looking at through, I presume, the normal batteries were the cheapest.\nSpeaker B: Yeah, the normal batteries would be the cheapest.\nSpeaker B: I presume then it would actually be the solar charging ones.\nSpeaker B: But I think they're not very resistant to dropping.\nSpeaker B: Because if you drop stuff, yeah, if you have the size you'd need to charge a remote, it wouldn't be that big.\nSpeaker B: But I mean, if remotes always get thrown around and stuff.\nSpeaker B: So I think practically, I mean, calculators you don't really throw around a lot.\nSpeaker B: Whereas remotes do.\nSpeaker B: They can't.\nSpeaker D: Don't care if they just have a battery in the world.\nSpeaker B: Yeah, they do.\nSpeaker B: They've got dual things.\nSpeaker B: But the batteries are smaller.\nSpeaker C: Look at it.\nSpeaker C: So which one would last the longest?\nSpeaker C: Because we don't want customers to be charging a mobile phone every day.\nSpeaker C: A remote phone for like, so.\nSpeaker B: Yeah, so look at the life also.\nSpeaker B: Yeah, if you had something using the standard batteries and the solar charging.\nSpeaker B: I don't think you'd.\nSpeaker B: I think the it would well, you know how long the standard double A's would last in order to replace.\nSpeaker E: It would just detract from the attractiveness of the whole feature.\nSpeaker D: Yeah, I think it would.\nSpeaker D: Okay, can we add in attachment to clothing feature?\nSpeaker E: Okay.\nSpeaker D: Can we think about that?\nSpeaker D: Because if we're doing the kinetic thing.\nSpeaker D: Yeah.\nSpeaker E: Shouldn't we do some market research on that first before we add it in?\nSpeaker E: Because I.\nSpeaker E: I'd add it in to think about.\nSpeaker D: Right, okay.\nSpeaker D: Because.\nSpeaker D: I don't know.\nSpeaker D: If it is the kinetic thing.\nSpeaker D: And it is small and it is portable and it is a different shape.\nSpeaker D: And the kinetic is something people don't do.\nSpeaker D: Kind of.\nSpeaker D: We're doing something original and different.\nSpeaker D: But if you wore it, if it's something you could just click on your pocket.\nSpeaker D: Then you would have that less.\nSpeaker D: You wouldn't lose it.\nSpeaker D: Yeah.\nSpeaker D: So much.\nSpeaker D: But then maybe that's looking at someone who's just sitting on their own rather than the eternal battle for control of it.\nSpeaker E: It's not something that's come up in any of our focus groups and market research.\nSpeaker E: It's not a thing that people are looking for.\nSpeaker E: Okay.\nSpeaker E: Let me throw it open to the field here.\nSpeaker E: Okay.\nSpeaker E: But it's something to put on the side to think about then.\nSpeaker D: Okay, right.\nSpeaker B: Okay, well I'll move on.\nSpeaker B: My second part of my findings.\nSpeaker B: The most current remote used this silicon PCB board which printed circuit board.\nSpeaker B: Which basically has these contacts that are really close together.\nSpeaker B: And then when you press down on the rubber button, it will connect the circuit.\nSpeaker B: And each switch is connected to different legs on the chip and so sends a different message.\nSpeaker B: And that then gets translated by the chip into a code and then it's fired out of this LED in a sequence of on and off bursts.\nSpeaker D: What kind of things do we have to consider there?\nSpeaker D: Can we, what kind of size does it come in varying sizes or is it just one size?\nSpeaker B: Well, well, the thing about it is that they can be as big or as small as you want them to be because you can print circuit board like that.\nSpeaker B: It's simply spaced like that so you can fit the size of the nine buttons in.\nSpeaker B: If you see how thin the tracks are, you could put them virtually right next to each other and have a much smaller switch on each one if you wanted to.\nSpeaker B: But there is an option to do it like that or you could have some sort of array of switches which I'll speak about in the next bit.\nSpeaker B: So that.\nSpeaker B: And then to, yeah, so to conserve battery life, remote should be in a standby mode while not being used.\nSpeaker B: And what I was thinking about that was because sometimes when you have these things that got little lights on behind the buttons so you can see what all the buttons are like on a mobile phone.\nSpeaker B: They do it more often than on a TV remote, but you could have lights behind the buttons and after like five or ten seconds and not being used, I'd have that turned off if we decided to go for buttons that could light up.\nSpeaker B: The case material I've been sent what the factory can actually give us.\nSpeaker B: The there's the plastic which I think we were going to go for anyway as the main case case housing and the main problems with the way you could go for wood but I think of it impractical.\nSpeaker B: Titanium.\nSpeaker B: Very expensive.\nSpeaker B: Just the process which to make it is expensive.\nSpeaker B: And rubber where you're saying that people like this spongy feel this year so perhaps something made of rubber but I was thinking more of the buttons because the buttons which they've said.\nSpeaker B: So they've put across.\nSpeaker B: Are what is used in some stressful manufacture and it's meant to be anti RSI.\nSpeaker B: I'm not sure how that would work because surely your fingers would still be moving over the short distances but that's what they said.\nSpeaker B: The problem with the casing is that there's quite there's quite a few design restrictions that they've got on the shape of the case if you go for titanium they can only do a standard box shape.\nSpeaker B: Whereas if you come to plastic they can they can be a little more they can they can do sort of curved shapes but whether whether or not because we'd have it in two separate units.\nSpeaker C: So for the flip phone.\nSpeaker C: Yeah and second question is like mobile you can change the cover.\nSpeaker C: Yeah.\nSpeaker C: So in this case you'll be looking at like the custom making change of color like from green parrot green to chilli red or something like that.\nSpeaker B: Yeah in titanium I don't think it would be available at all really.\nSpeaker B: Well you could make it available in the titanium it was just it would be so expensive to buy a new case for it because of the expense of how much titanium is it's light and strong but I think it should be left for aircraft design rather than for.\nSpeaker E: It does work quite easily too if you let it fall I've got a mobile from myself which is titanium and it doesn't mark very easily.\nSpeaker E: Yeah I was just wondering if we make the basic mold out of plastic but have like a rubber cover rubber yeah that you give the sponge you feel that also allows us to kind of have different.\nSpeaker D: And you can peel them off yeah instead of the fascia that comes off being plastic the fascia that comes from a rubber sleeve almost yeah.\nSpeaker D: Like those pens that you get with the grip yeah right that you can that could be a good.\nSpeaker C: Okay it would be comfortable to hold on.\nSpeaker E: Very cheap way of changing the look of it and people can just buy a new one.\nSpeaker D: Yeah well that's been really popular with mobile phones so I don't see why not.\nSpeaker B: Also the just just going back to the interface designs with the buttons.\nSpeaker B: We were planning to do some sort of touch screen.\nSpeaker B: Now what I was saying before about instead of having you could have just a bare PCB circuit board and I'm sure you could probably get it in different colors and say just by touching it with your finger it would make the connection.\nSpeaker B: Over if you had them close enough that would be one option.\nSpeaker B: The second option they offer rubber buttons but I thought that an LCD type of screen because one they're incredibly thin.\nSpeaker B: And don't take on much space to they you can have them in a sort of a rain you could arrange the buttons on the screen in a sort of set thing.\nSpeaker B: You could have them like a touch.\nSpeaker C: So I just read the last part.\nSpeaker C: Oh what what just what do you say to adding the meaning of it completely.\nSpeaker B: Oh on the on the LCD screen you could you could fit it the problem with it basically is that it's flat and so you can't do lots of curve curve things with it.\nSpeaker B: But you can you wouldn't have with the LCD you'd have the wires coming off.\nSpeaker B: You wouldn't have that with the with the LCD you'd only have that with the printed circuit board with.\nSpeaker E: Also it's I don't see why the curve thing is a problem if we for example had a shell.\nSpeaker B: Yeah you can have a flat screen inside.\nSpeaker A: Yeah.\nSpeaker B: I'm just wondering whether we want an LCD screen inside it wouldn't be like full color it would just be black and white.\nSpeaker B: But there'd be touch touch buttons so you wouldn't be pressing down on them.\nSpeaker B: Or we could have rubber buttons which are made of this material which is anti-IS on.\nSpeaker D: I think it would be good to have a contrast between if the whole thing is going to be rubber saying it would be good to open it up and see something quite fancy with it inside.\nSpeaker D: Now how would you distinguish if you had it bare how do you distinguish where you had to press.\nSpeaker B: Yeah I hadn't I hadn't really thought so.\nSpeaker E: I don't know if it helps with that but just to do with the RSI is it possible just as an option when we open it up people can use their fingers to press the button.\nSpeaker E: Or we have inside like a small pointer thing.\nSpeaker B: Yeah you could you could have some sort of stylist that you could pull out but I think they could get a bit easily lost.\nSpeaker E: Yeah it's a little bit easy to replace as well.\nSpeaker C: I'm going to show this very often.\nSpeaker C: A person who switches channels very often or does use a particular function very often.\nSpeaker C: He'll find it very irritating to use earlier.\nSpeaker C: Yeah.\nSpeaker C: And he might spoil the touchpad very fast compared to button like yeah we keep punching with and point over.\nSpeaker E: Yeah just a thought they wouldn't actually need one in the case anything they've got couldn't they have pencil or pen so they wouldn't really need a.\nSpeaker D: Okay we'll talk about that so if you finish here.\nSpeaker B: Yeah that's yeah that's the end of the.\nSpeaker C: I haven't forgotten before like you're listening.\nSpeaker C: Regarding the circuit since we have having a flip top.\nSpeaker C: Yeah customize the two circuits for different type of buttons like we are keeping the standard buttons on the top.\nSpeaker C: Oh yeah.\nSpeaker C: We'll complex button in the distance so we can divide the circuit like that.\nSpeaker B: Yeah yeah that would be fairly simple.\nSpeaker B: I mean you'd actually have two separate.\nSpeaker B: You'd have two separate circuit boards but they'd be joined by wires or like some cabling between them.\nSpeaker B: Because in the in the actual flip that you'd have some linking.\nSpeaker E: I just raised another point well it occurs to me it kind of applies to both our designers here.\nSpeaker E: I'm not sure how it would fit in.\nSpeaker E: If we flip open now you know how you've got some mirrors in nightclubs that are mirrors when you turn to them.\nSpeaker E: And you turn away and it gives like an advertising display.\nSpeaker E: You see those?\nSpeaker E: No.\nSpeaker E: And they kind of respond to the turn of your body.\nSpeaker E: Is it possible that when we open our flip top shell it's a little compact mirror.\nSpeaker E: And when you press a button it then goes on to the phone display.\nSpeaker E: The art there, the remote control display thing.\nSpeaker D: We are marketing to guys as much as we are to women.\nSpeaker E: They don't look at themselves.\nSpeaker E: Just a thought.\nSpeaker D: Well it's a remote controller you were sitting watching TV.\nSpeaker D: Yeah.\nSpeaker D: Are you going to want to sort of open it and say.\nSpeaker D: I know what you mean.\nSpeaker D: So they are using the ideas to mobile.\nSpeaker E: So they are using the ideas to mobile phones.\nSpeaker E: We kind of got fact tracked onto that I think.\nSpeaker E: Okay.\nSpeaker E: I'm just trying to think of other features we can build in that we don't cost too much.\nSpeaker E: But one would be that one on the side.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker B: I mean you can't.\nSpeaker B: You could do it.\nSpeaker B: You could have a.\nSpeaker B: Because if you just put the full charge through an LCD display or completely blank it out.\nSpeaker B: But I don't know if you can get any mirror effect on it.\nSpeaker B: I think we forget about the mirror.\nSpeaker E: That was just a very quick passing.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: I would like to show something which I did here.\nSpeaker C: First thing is basically on design we just took the input from the previous meeting.\nSpeaker C: Especially from the marketing and industrial design to check on the customer needs and feasibility.\nSpeaker C: Second is we check into competitors.\nSpeaker C: The picture he shows what are the standard models offered by competitors here.\nSpeaker C: So you generally see there is not much of variety and like marketing team said people need trendy they are bored of black and white.\nSpeaker C: So you generally see rectangular shape very monotonous kind of designs here.\nSpeaker C: And second thing is there is too much of confusion here.\nSpeaker C: No particular remote is standard.\nSpeaker C: Like some people have here.\nSpeaker C: You see this.\nSpeaker C: This is on I found that the only common feature is the channel control and the volume control.\nSpeaker C: Rest other buttons they are in a very disorganized and they are not consistent with other models.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: And second as I already discussed with William we are going to have maybe a GUI interface in the middle of the flip top.\nSpeaker C: Sorry what does that sound like?\nSpeaker C: Graphic user interface.\nSpeaker C: Basically which is what we do in computer.\nSpeaker C: Have icons or touch pad or whatever.\nSpeaker B: If you have buttons that appear on the screen then this LCD screen.\nSpeaker C: Like you have an icon or something you have is a good example of GUI and graphic user interface.\nSpeaker C: Yes.\nSpeaker C: So basically a person you not point or click or press any particular device he just has to click on that particular icon.\nSpeaker C: That is simplified.\nSpeaker C: And on the top repetitive buttons which are like volume or channel changing and all could be on the pointing device that means user button.\nSpeaker C: And he not user pointer to keep changing the channel if a person is frequent circle.\nSpeaker C: So we are having a combination of both.\nSpeaker C: So on the simpler board on the top we have this button, rubber buttons to keep frequently changing the channels.\nSpeaker D: Can you swap that round?\nSpeaker D: The ones that we use all the time on the bottom part.\nSpeaker D: And what did you think about holding it?\nSpeaker D: You would be using your thumbs to press.\nSpeaker D: Just like a mobile phone.\nSpeaker E: You would flip it open and you would be, wouldn't you?\nSpeaker E: Isn't that the idea?\nSpeaker E: If we just use the shell as an example again you open that.\nSpeaker E: You have your LCD display there and you have the buttons there.\nSpeaker E: That is what I was saying.\nSpeaker D: But you can't do it with your thumb.\nSpeaker D: And then have the LCD at the top and then be able to touch that for the other person.\nSpeaker D: And you would have the volume and the program and things like that.\nSpeaker C: And the lower the side.\nSpeaker C: So the findings are too many cluttered buttons.\nSpeaker C: Repetition of certain patterns which I already explained.\nSpeaker C: Example the volume and channel control buttons.\nSpeaker C: All are confusing and inconsistent.\nSpeaker C: Okay we had a latest finding of voice recognition.\nSpeaker C: There was a mail which mentioned that our division has developed a new speech recognition feature.\nSpeaker C: We have to check into the financial feasibility whether we can incorporate this at low cost.\nSpeaker C: But for like we had response from the customers that they would like to have the feature of finding it.\nSpeaker C: So it could be like where is the remote and the remote answers I am here.\nSpeaker C: Some kind of thing or it gives a blip sound or some kind of sound.\nSpeaker C: And if this can be incorporated this would be more, you can say trendy also.\nSpeaker C: And technologically innovative also.\nNone: Yeah.\nSpeaker C: So we can check into the financial feasibility of incorporating this.\nSpeaker C: My personal preferences would be like as already highlighted by the marketing department they want something to do with fruit.\nSpeaker C: So I wouldn't say the design should be like a fruit.\nSpeaker C: But yeah we can take inspiration from fruit colors.\nSpeaker C: Like the vibrant colors.\nSpeaker C: Red chili.\nSpeaker C: Yeah.\nSpeaker C: Two mitre red or whatever.\nSpeaker C: And second thing is certain standard buttons we should have.\nSpeaker C: Like for example if you see the previous slide.\nSpeaker C: Wow.\nSpeaker C: Yeah.\nSpeaker C: I think most of the users would now be well acquainted with this central pattern.\nSpeaker C: Here.\nSpeaker C: Yeah.\nSpeaker C: So we would not change that particular pattern because this is the most consistent thing in all remote controls.\nSpeaker D: I can't see that.\nSpeaker D: That play in the store.\nSpeaker D: This is the central one.\nSpeaker C: The one you have volume and channel.\nSpeaker C: So keep that standard x because that's the most common feature across all models.\nSpeaker C: If you look at all the models it's here.\nNone: This.\nSpeaker C: And.\nSpeaker C: Voice recognition.\nSpeaker D: Okay.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: I'm not sure how long we've got left.\nSpeaker D: But we need to make a decision about.\nSpeaker D: Things we've discussed.\nSpeaker D: So.\nSpeaker D: We agree on.\nSpeaker D: Do we agree on the battery?\nSpeaker E: Kind of.\nSpeaker E: Kind of.\nSpeaker D: Yeah.\nSpeaker D: That means that there's no function for the port.\nSpeaker D: You know that it sits in the impression button then having.\nSpeaker D: So we could incorporate voice recognition.\nSpeaker A: Yeah.\nSpeaker D: I think.\nSpeaker D: I mean if it's like a gadget.\nSpeaker D: Yeah.\nSpeaker B: It is.\nSpeaker B: It is quite a cool feature to have.\nSpeaker B: And also if they've got it.\nSpeaker B: If they've got these parts already in stock.\nSpeaker B: Then it won't be cheap to remanufacture them.\nSpeaker B: It will be cheap though to remanufacture them.\nSpeaker B: So do you recommend a good idea that.\nSpeaker B: Where's the remote.\nSpeaker B: I'm here thing.\nSpeaker B: I think that would be quite fun.\nSpeaker D: I think the only.\nSpeaker D: I think the only.\nSpeaker D: Pitfall that we would have would be.\nSpeaker D: How much you've been.\nSpeaker D: Yeah.\nSpeaker D: That means we have to cut down somewhere else.\nSpeaker D: But I think pretty much.\nSpeaker D: We.\nSpeaker D: Is.\nSpeaker D: Yeah.\nSpeaker D: Better to be cheap than simple things.\nSpeaker B: Yeah.\nSpeaker B: LCD is not cheap.\nSpeaker B: Well it's.\nSpeaker B: Not as expensive as it would be if it was full color.\nSpeaker B: Because if it just got black and white one.\nSpeaker B: I mean they used them in calculators.\nSpeaker B: Yeah.\nSpeaker C: Maybe we could start with the black and white.\nSpeaker C: Yeah.\nSpeaker A: That would be good.\nSpeaker C: Yeah.\nSpeaker C: Upgrade later on up.\nSpeaker E: We talked about kinetic charging.\nSpeaker E: We seem to have decided on that.\nSpeaker E: Did we decide on double A or triple A batteries.\nSpeaker B: What?\nSpeaker D: You went like a backup.\nSpeaker E: No.\nSpeaker E: One of the decisions we had to make was whether we had double A or triple A batteries.\nSpeaker E: Because they've still got to be charged this way haven't they?\nSpeaker E: No the kinetic one.\nSpeaker B: Oh I see.\nSpeaker B: Come with the sort of watch.\nSpeaker B: A battery that goes in the watch.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: So it's a lot smaller.\nSpeaker B: So we.\nSpeaker E: You're on that okay.\nSpeaker E: Okay.\nNone: Okay.\nSpeaker D: Got battery.\nSpeaker D: The inside components is pretty standardized across the board isn't it?\nSpeaker D: Yeah.\nSpeaker D: So there's not really a decision to be made there.\nSpeaker D: The buttons.\nSpeaker D: What did you give us as a.\nSpeaker D: The bare board LCD.\nSpeaker D: Yeah.\nSpeaker B: I think we were going for the LCD on that one.\nSpeaker B: On the buttons.\nSpeaker B: On the.\nSpeaker B: On the top one.\nSpeaker B: On the top one.\nSpeaker D: Okay.\nSpeaker D: You've got the touch.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And then.\nSpeaker B: On the bottom we were going to have to rub.\nSpeaker B: The one that we were going to have to do is just to put the buttons on the bottom.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker E: Sorry could you repeat that last part?\nSpeaker B: Okay.\nSpeaker B: On the flip screen.\nSpeaker B: The flip screen.\nSpeaker B: The top one is going to be the LCD and the bottom one is going to be the rubberized.\nSpeaker B: Okay.\nSpeaker C: And for the.\nSpeaker D: No.\nSpeaker D: So the body design I think plastic.\nSpeaker C: Plastic.\nSpeaker C: Yeah.\nSpeaker B: Okay.\nSpeaker B: For the inside.\nSpeaker C: Yeah.\nSpeaker C: Rubberized padding or for the crypt something like to add to the design.\nSpeaker E: So we decided on a rubber casing for the plastic shell and variety of designs.\nSpeaker E: Okay.\nSpeaker E: And it is just a little bit rubberized and spongy.\nSpeaker E: Apart from that we're just going to go for sort of vegetable and fruit colors.\nSpeaker E: We're not going to try and make it actually a vegetable or fruit design or anything else.\nSpeaker E: I shall be discussed.\nSpeaker D: No.\nSpeaker D: We don't want to be tacky if we've got a different shape anyway.\nSpeaker D: The fact that the material that we would be using would be cheap.\nSpeaker D: So we could make it red.\nSpeaker D: We could make it psychedelic.\nSpeaker D: You know we could make it black and white.\nSpeaker D: They're prescribed.\nSpeaker D: But that's not really what we're focusing on.\nSpeaker D: What we're focusing on is the feel.\nSpeaker D: Yes.\nSpeaker D: So just pick anything.\nSpeaker E: Is that going to be the image I've got in my head of something kind of shaped like that.\nSpeaker E: Maybe about that size.\nSpeaker E: Made of plastic.\nSpeaker E: Fits into the power of the hand.\nSpeaker E: Rubberized cover that's spongy.\nSpeaker E: Is that really going to fit fancy looking feel which was the major thing that people wanted market research.\nSpeaker E: Is that something different?\nSpeaker D: It's just different.\nSpeaker D: It's just different from everything else.\nSpeaker D: I mean I'm trying to imagine clean looking houses sort of beige and black.\nSpeaker D: Do you either want something that goes with that which is what's on the market anyway or you want something that contrasts as you know like you get clocks now that are more of a talking point than an actual clock.\nSpeaker D: Yeah.\nSpeaker E: Because they're so interesting.\nSpeaker E: I'm just kind of pushing it that to see if you came out with anything else.\nSpeaker E: But I mean I totally agree.\nSpeaker E: We looked at those remotes.\nSpeaker E: I mean they are kind of anonymous very similar looking things.\nSpeaker E: This would definitely be different enough.\nSpeaker E: I just wanted if anybody could come up with something even more.\nSpeaker D: And you would just have a cross that you would have so many different options.\nSpeaker D: You could do.\nSpeaker D: You could have a plain black one.\nSpeaker D: You could have sparkly pink glittery ones.\nSpeaker E: The things that are upperized covers are going to be so relatively cheap to produce.\nSpeaker E: If in a years time we get feedback from the design fears that show something else is coming.\nSpeaker E: It's so easy for us to produce that that can be slipped on.\nSpeaker E: And what is the beauty of it?\nSpeaker D: The ones that have rubbery spikes.\nSpeaker D: You know you just go so far from there.\nSpeaker D: You could just take it wherever.\nSpeaker D: So I think that's quite a flexible.\nSpeaker C: And finally the body should be reducible.\nSpeaker C: Maybe you can change the body of the case kissing.\nSpeaker C: The case outside.\nSpeaker C: It should be molded with the design in such a way you can change it every time.\nSpeaker B: Yeah, yeah.\nSpeaker B: With these rubberized cakes.\nSpeaker B: Not the actual plastic outside case.\nSpeaker B: Just the rubber.\nSpeaker D: Okay, we've got five minutes.\nSpeaker D: That might have been up for a while.\nSpeaker E: So can I just recap, Sarah?\nSpeaker E: The decisions that we've made.\nSpeaker E: Kinetic charging.\nSpeaker E: The watch type batteries.\nSpeaker E: LCD display.\nSpeaker E: The top side of the flip top.\nSpeaker E: Rubberized buttons on the bottom side.\nSpeaker E: We're going to use fruit and vegetable colors for the rubber.\nSpeaker E: Cover the case itself as plastic.\nSpeaker E: That's how far we've got.\nSpeaker E: We have to add to our decisions here.\nSpeaker C: Fine.\nSpeaker C: We're talking of voice recognition also.\nSpeaker C: Because we have not addressed the issue of how to locate a remote control if it's lost.\nSpeaker C: Yeah.\nSpeaker E: It was just a cost issue with that.\nSpeaker E: But it's a good idea.\nSpeaker E: We just need to check on the cost.\nSpeaker C: Or maybe like the limit we're suggesting in the last thing.\nSpeaker C: Some devices you put on keychains.\nSpeaker C: Oh yeah, yeah.\nSpeaker B: There were so ones.\nSpeaker C: The whistle.\nSpeaker C: So we can, financially voice recognition is not feasible.\nSpeaker C: We could go for a whistle.\nSpeaker E: And incorporating the company logo.\nSpeaker C: Yep.\nSpeaker C: Have you?\nSpeaker C: I'm sorry.\nSpeaker C: But we'll be incorporating on the design.\nSpeaker E: Because I took it from the feedback you gave in the last meeting from your report.\nSpeaker E: It's not just a case of having a little RR hidden somewhere.\nSpeaker E: They do want it to be obvious that it's our product.\nSpeaker D: Well they do.\nSpeaker D: But I think we can.\nSpeaker D: Since it's the only one of its kind on the market.\nSpeaker D: Well it is.\nSpeaker D: I think you just addressed that with advertising.\nSpeaker D: You associate the name with the individual product that it is.\nSpeaker D: Okay.\nSpeaker D: And that does the work for you.\nSpeaker D: Obviously you'll have a logo on it just like everything.\nSpeaker D: But everything else has a logo on it.\nSpeaker D: The point is that it's got to stand out somehow differently.\nSpeaker D: I think it does.\nSpeaker E: I know at the last meeting we spoke about a beeper.\nSpeaker E: I think that was you talked about a beeper for a location.\nSpeaker E: We just rejected that and people voiced recognition.\nSpeaker D: Depending on how.\nSpeaker B: Yeah depending on the expense.\nSpeaker D: I mean we've got this memo saying that it's something that the company is looking into.\nSpeaker D: I've got in stocks.\nSpeaker D: That seems to make sense.\nSpeaker D: But if it turns out that it's hugely expensive we'd have to cut down somewhere else.\nSpeaker D: Then it's worth thinking about.\nSpeaker D: I think we'll find out more about that afterwards.\nSpeaker D: Right.\nSpeaker D: I'm going to wrap this up today.\nSpeaker D: I got our end meeting now.\nSpeaker D: Did it?\nSpeaker C: Yeah.\nSpeaker C: So I think we've probably got...\nSpeaker D: Instead of closing we have 40 minutes.\nSpeaker D: So I think we've probably got about that time before we come back.\nSpeaker D: See that our family had 40 minutes to discuss what we've discussed.\nSpeaker D: I don't know how long that took.\nSpeaker C: I think we're pretty going in a clear direction.\nSpeaker D: Is everyone happy?\nSpeaker D: Okay.\n"}, {"meeting_id": "ES2009d", "summary": "This was the last meeting of their work. The meeting was about a summary of the previous achievements and agreements, including the details of the design of the remote control, like issues of energy source and budget control. After talking about their ideal remote control, the group moved on to plan their group presentation. They discussed what the PPT should include, what they should and should not mention in the presentation and also compared their work with their competitors.", "dialogue": "Speaker C: Oh, I told me.\nSpeaker C: I didn't know where to go.\nSpeaker C: I was like, cool.\nSpeaker B: All right, it is PowerPoint time.\nSpeaker B: I've done more power points in this particular experiment than I've ever done in my life before.\nSpeaker B: I'm just kind of fun.\nSpeaker B: So here we have our detailed design meeting where we will look at the prototype and right.\nSpeaker B: So I finally figured out what this whole second bullet point is about in my that my coach was sending to me.\nSpeaker B: It means I'm supposed to read the minutes from the previous meeting.\nSpeaker B: Oh, really?\nSpeaker B: Okay, I think.\nSpeaker B: I don't know.\nSpeaker B: Otherwise it's just saying I'm the secretary and I'm there for I'm taking minutes.\nSpeaker B: So just to go just real briefly to go over minutes from the last meeting.\nSpeaker B: I will open them slowly.\nSpeaker B: No?\nSpeaker B: Wait for it.\nSpeaker C: Wait for it.\nSpeaker C: Yeah, that's not you.\nSpeaker B: Wait.\nSpeaker B: This is very high powered stuff here.\nSpeaker B: There we go.\nSpeaker B: So basically the moral of the story from our last minute last meeting was that we had meetings from we had presentations done by the industrial designer from Nathan and Ron and Sarah about what we can do here and what sort of limitations we're operating with.\nSpeaker B: Excuse me, limitations were operating under what kind of risk we'd be looking at with some of the various approaches we were discussing.\nSpeaker B: And we essentially came to the conclusion that we should develop a remote with their voice recognition that had a vaguely non remote like shape because you didn't really use it as a remote as you could just use your voice.\nSpeaker B: It would include some mostly just a simple design features for a television operation, but with a slider full now bay for more advanced functions for users.\nSpeaker B: And the the idea and the idea were asked to go ahead and start developing a prototype press to look at.\nSpeaker B: So that sort of that the main.\nSpeaker B: So, I'm going to go ahead and take it away guys.\nSpeaker D: Well, we have a simple our prototype.\nSpeaker D: What's to be said about it.\nSpeaker D: We took into account a lot of things that we went over in the last meeting.\nSpeaker D: And then we decided not to go for the touch screen, which you can see and opted for some very large buttons for the primary functions.\nSpeaker D: This is going to be the on and off button and we have these buttons to go through the channels and then to volume buttons down here.\nSpeaker D: And then for the more advanced functions, there's a slide out panel here.\nSpeaker D: You can see that there's a lot of things going on, but this actually can slide back in.\nSpeaker D: It's very nice aesthetic when it's all put away.\nSpeaker D: As far as the whole visible light thing, we decided to go with multiple colors.\nSpeaker D: Why not?\nSpeaker D: Of course, if that's annoying for some people that function can be turned off.\nSpeaker E: Now, it's important to we talked quite a bit about the interchangeable faces.\nSpeaker E: And what we've done here is come up with a bit of a natural look here.\nSpeaker E: We call it fruity.\nSpeaker E: Of course, that's interchangeable.\nSpeaker E: I think it would be desirable for the regular product in the first packaging to be something a little bit more subdued, but this is kind of something that can be done.\nSpeaker E: And as you can see on the television there, we have the voice detector device on the top.\nSpeaker E: So that will work quite well with regard to finding this contraption.\nSpeaker E: What other things do we see here?\nSpeaker E: Well, if you give it a touch, it does have actually a bit of a spongey feel.\nSpeaker E: So I think that will work well in regards to our market.\nSpeaker E: Let's see.\nSpeaker E: Clearly, there's going to be some more colors and what not available.\nSpeaker E: Do you have anything else to add to that?\nSpeaker D: I'm worried about the materials.\nSpeaker D: The entire thing is covered in rubber coating.\nSpeaker D: So it's very durable.\nSpeaker D: It's not going to break like some types of plastic that's dropped.\nSpeaker D: And of course, as you can see, if you touch it, it does have that nice squishy feel.\nSpeaker E: It's actually important to note that the television, if there's an earthquake or anything like that, it actually is edible inside.\nSpeaker D: I don't know if you noticed, but I wrote the company's name.\nSpeaker D: This is actually an apple on the inside.\nSpeaker C: Do we need to worry about rot factors?\nSpeaker E: It's encased in a new type of product.\nSpeaker D: We got a bit ahead of ourselves.\nSpeaker D: I know we're not talking about making televisions.\nSpeaker D: It's a way for the future.\nSpeaker E: It's a couple of years off at least.\nSpeaker E: But I think that sums up the main features of the remote.\nSpeaker E: I don't know if you guys have any questions or whether we need to worry about any other marketing areas or anything of that nature.\nSpeaker E: Did we come in under budget?\nSpeaker D: We did.\nSpeaker D: We did that.\nSpeaker D: We did that.\nSpeaker D: We didn't have to use as many microchips, which was quite nice.\nSpeaker D: That's how to keep the cost down.\nSpeaker D: We had a lot of people who were talking about the modern technology.\nSpeaker D: For example, the voice recognition.\nSpeaker D: In a lot of ways, it's just a simple remote.\nSpeaker E: I think if we shopped around for other manufacturers, we might be able to get it even cheaper.\nSpeaker E: On the back here, you'll notice this area here, which is actually the voice recognition console.\nSpeaker E: I think it's nicely designed into the overall look.\nSpeaker E: Basically, the voice recognition incorporates the latest designs that our research team has come up with.\nSpeaker E: It's quite similar to the coffee maker design that we were talking about earlier.\nSpeaker E: I think that has given a proven ease of use and whatnot.\nSpeaker E: It allows futures like the remote actually talking back to the user.\nSpeaker D: Cool.\nNone: Any questions?\nSpeaker B: No, no.\nSpeaker C: Do we have other, for lack of a word, skins covers a plane hour?\nSpeaker C: Are those going to be developed later?\nSpeaker C: Once we see how the couple we have go, do we stay on that yet?\nSpeaker E: Well, we didn't quite have enough material.\nSpeaker E: Oh, I was expecting a prototype.\nSpeaker A: I just don't know if you guys had me in mind.\nSpeaker D: As you can see, this is just a most superficial layer.\nSpeaker C: Something else.\nSpeaker E: Actually, this bottom red ring here just unclips and then you put a new plate on top of that.\nSpeaker E: We definitely priced out even spongier non-natural look materials, which I think worked out fine.\nSpeaker E: We also continued on with the ideas of following apples, color schemes with kind of the light orange and the fine.\nSpeaker D: It's not quite a faceplate.\nSpeaker D: It's more like a pseudo faceplate because it's simple enough that in the factory, we could very easily put a different one on it.\nSpeaker D: It walks into place such that it's pretty permanent.\nSpeaker D: At the same time, if we want to go the other way, it's just a matter of a couple of adjustments.\nSpeaker D: We could go the faceplate way, if you know what I mean.\nSpeaker C: Yep, it's still an option.\nSpeaker C: If we need it.\nSpeaker C: Very cool.\nSpeaker C: Nice job.\nSpeaker B: Right, thanks guys.\nSpeaker B: Very, very good work.\nSpeaker B: I like it. Brilliant.\nSpeaker B: What we need to discuss now is the finance of it.\nSpeaker B: You've provided a number that actually sounds quite nice.\nSpeaker B: The trouble is I was just given this by finance.\nSpeaker B: It's a spreadsheet of the parts and I just tend to put in what it's going to look like.\nSpeaker B: I'm just going to clear this out real quickly.\nSpeaker B: But it looks like...\nSpeaker B: If we can just itemize what's in here, it's a solar cell thing, right?\nSpeaker D: Right.\nSpeaker D: With a backup battery?\nSpeaker D: It's in there.\nSpeaker B: With the back.\nSpeaker B: Okay.\nSpeaker E: The voice recognition area actually doubles as the solar cell area.\nSpeaker B: Clever.\nSpeaker B: Well then.\nSpeaker B: So I guess that would mean we've got a bit of a speaker in a sensor at the same time, isn't it?\nSpeaker D: Yeah.\nSpeaker D: It's just making use of the same space and the same material.\nSpeaker B: In the case, it's more of a single curved case.\nSpeaker B: I guess that would be the general...\nSpeaker D: Yeah, one big curve.\nSpeaker B: I guess you could say.\nSpeaker B: We've got a rubber skin material basically throughout.\nSpeaker B: Push button interface with this other dropdown.\nSpeaker B: So maybe we've got two push button interfaces.\nSpeaker B: Yeah.\nSpeaker B: I guess we've got sort of a wood material, a rubbery type material.\nSpeaker B: Yeah.\nSpeaker B: Special.\nSpeaker D: Throughout.\nSpeaker D: I guess you'd have to mark special color in special form as well.\nSpeaker D: It is very unconventional.\nSpeaker B: Yeah, I like it.\nSpeaker B: So it looks like a bit over budget.\nSpeaker B: Just watch out this.\nSpeaker B: So what we could do perhaps a simple fix would maybe to switch away from the solar cells or take out the backup battery.\nSpeaker E: I mean, I think that if we were talking about it being one of her main selling features being environmental and without the batteries and whatnot.\nSpeaker E: Although it does still have a battery, so I'm not sure that what the cell is on that.\nSpeaker D: I mean, we could take the battery out of it.\nSpeaker D: You see, and it probably worked 99% of the time, but you're going to have to set up a call center for that 1% of the time when people are calling and saying, oh, look, my remote isn't working.\nSpeaker D: What am I going to do?\nSpeaker D: People will be upset.\nSpeaker D: I think in the long run, it's better to keep the battery.\nSpeaker D: It's hard to scrap the whole solar battery idea because that's so integral to the theme that we have.\nSpeaker A: Hmm.\nSpeaker B: Well, it's difficult. We have all these things integral to the design of it that we just can't back out of now. It has to be.\nSpeaker B: Seems like we'd have to go back to square one in a way.\nSpeaker B: If we were going to try to undo one bit, we'd probably have to undo most.\nSpeaker B: Yeah.\nSpeaker C: Although we don't want to get rid of the whole environmental.\nSpeaker C: I mean, obviously, the solar cell is a big piece of the way we're marketing this is like a natural new thing.\nSpeaker C: But honestly, if we cut that one piece out, we're actually coming in on the budget effect on my math correctly.\nSpeaker E: You might be able to sway me on the idea that our main selling point could be already this voice recognition thing.\nSpeaker E: I mean, that's what sets us apart, right?\nSpeaker C: Yeah, that's what's setting us into this young market.\nSpeaker C: And that's where we started from. So.\nSpeaker C: I don't know. And I mean, you know, perhaps when the cell technology comes down in price, we can bring that back into the game, but it looks like at this point that may be out of our league.\nSpeaker E: And the reality is, you know, for me from an ideological standpoint, I'd like to stick with the solar cell, but.\nSpeaker E: I kind of have to throw myself in the business structure and model here.\nSpeaker E: Yeah, I think either or I think that I think that we need to come to a compromise here and maybe move ahead with the project without the solar cell.\nSpeaker D: Yeah, I think I'm fortunately that's our best option. See, like, where we're going to get below.\nSpeaker D: Uh, cool.\nSpeaker B: Cause we can't remove the push buttons because they're.\nSpeaker B: Yeah, I can't get rid of the, I mean, removing the change in the case wouldn't be so much of a.\nSpeaker B: Nor would change in the case materials.\nSpeaker B: So, yeah, that looks like to be the only thing.\nSpeaker B: Yeah, so that would be a major change, but.\nSpeaker B: Yeah.\nSpeaker B: All right, so we're in agreement on that.\nSpeaker C: Unfortunately, I think we are.\nSpeaker E: I think that was a good compromise.\nSpeaker B: Right.\nSpeaker B: Moving along swiftly. So I guess now we just go to the project evaluation.\nSpeaker B: Which I will allow.\nSpeaker C: That will be me.\nSpeaker C: Of course, sorry.\nSpeaker C: No problem.\nSpeaker C: You reach.\nSpeaker C: It will be great.\nSpeaker C: I didn't even do that on purpose.\nSpeaker C: Okay.\nSpeaker C: Basically, I was just evaluating.\nSpeaker C: From what we know of how our products working right now with the criteria that we set at the beginning of these are the things we needed to do.\nSpeaker C: Like we feel they're important.\nSpeaker C: So I was looking at basic design things.\nSpeaker C: Does it fulfill its function as a remote?\nSpeaker C: Is the design what we wanted to do?\nSpeaker C: Our technology is up to where we'd hope they would be.\nSpeaker C: And does it fulfill the aesthetic qualities that our original market research was looking for?\nSpeaker C: Basic questions like, you know, does it turn on?\nSpeaker C: Does it respond to voice recognition?\nSpeaker C: And overall in general, it looks like it's coming up to par.\nSpeaker C: The only thing is with the pullout panel, that is going to take some adjusting because of the new interface.\nSpeaker C: That looked like it was coming up rough.\nSpeaker C: But then once you get used to it, it does make a lot of sense.\nSpeaker C: So I think overall we're heading in the right direction.\nSpeaker E: Like that sponge of you.\nSpeaker C: Yeah.\nSpeaker C: It looks like it's going over well.\nSpeaker C: The future function works well.\nSpeaker C: That's good to hear.\nSpeaker C: We're good.\nSpeaker C: Yeah.\nSpeaker C: I think eventually if we do branch out with this program, maybe when we do have higher budget options, and if it goes over with this model, we can look into wider range voice recognition, like from other rooms of the house and stuff.\nSpeaker C: But for now, what we've got is working in the range we need it for.\nSpeaker C: So it's all good.\nSpeaker C: Really?\nSpeaker D: That's it for me.\nSpeaker D: Losing the solar panel.\nSpeaker D: Yeah.\nSpeaker D: It's okay.\nSpeaker C: Yeah.\nSpeaker C: It is a setback, but.\nNone: Okay.\nSpeaker C: Do you need the cord back?\nSpeaker E: Yeah, we might have lost that granola market again.\nSpeaker E: Yeah.\nSpeaker B: Oh, they don't want to tell you anything.\nSpeaker B: That's true.\nNone: Right.\nSpeaker B: So this one's a bit unclear to me to be perfectly fair.\nSpeaker B: I got this slide from the coach, and I'm not sure what it's connected to.\nSpeaker B: So I guess we are going to discuss.\nSpeaker B: Project process.\nSpeaker B: And that is going to go into my report.\nSpeaker B: So I guess this is the point where we go.\nSpeaker B: Out of role.\nSpeaker B: It looks like and talk about our satisfaction for room for creativity and so forth.\nSpeaker B: Okay.\nSpeaker B: How that all worked, I guess.\nSpeaker C: As in within the team.\nSpeaker D: So yeah, right.\nSpeaker C: It is now.\nSpeaker B: But I trust that she would jump in if I was.\nSpeaker B: Okay.\nSpeaker B: Right.\nSpeaker B: So any thoughts?\nSpeaker D: We can start in these points here.\nSpeaker D: Yeah.\nSpeaker B: When did you guys feel about the process?\nSpeaker C: You know, I think in general, for a day's worth of work, we actually were relatively productive, considering the little amount of input we had going in.\nSpeaker C: And the technology has definitely been a help.\nSpeaker C: It's really been interesting to try out all this new stuff.\nSpeaker E: We didn't use the whiteboard at all.\nSpeaker C: No, no whiteboard.\nSpeaker C: We could now.\nSpeaker C: And I feel like if you guys have been designing in here, perhaps that would have changed, but because of room constraints, it doesn't really matter.\nSpeaker E: Also, had I not been intrigued about the pen, I don't think I would have used it at all.\nSpeaker E: I didn't write barely anything.\nSpeaker C: Yeah.\nSpeaker C: I think I was taking notes more often than usual.\nSpeaker C: Yeah.\nSpeaker C: That's true.\nSpeaker B: I'm not sure.\nSpeaker C: It was pretty cool.\nSpeaker C: I am disappointed.\nSpeaker D: I didn't get a note back for my personal coach.\nSpeaker D: Is he right?\nSpeaker D: Personal coach.\nSpeaker D: Yeah.\nSpeaker D: But I didn't get her.\nSpeaker B: See.\nSpeaker B: What if you get a response two or three months from now?\nSpeaker C: Well, what kind of coaching is that really?\nSpeaker E: What if I really needed something?\nSpeaker E: I think there was a lot of room for creativity.\nSpeaker E: We could do everything.\nSpeaker E: Basically, we wanted until the budget came down on us.\nSpeaker C: And even then, we did get a decent product turned out, although it's not everything we wanted it to be.\nSpeaker E: Yeah, with the natural look.\nSpeaker C: It's very natural.\nSpeaker B: Organic, really.\nSpeaker B: They had a peeler in here.\nSpeaker C: And highly resourceful teammates might add, which is always a class.\nSpeaker C: Yeah.\nSpeaker B: I thought it was really creative, actually.\nSpeaker B: Yeah.\nSpeaker C: I'm impressed.\nSpeaker E: The teamwork was good.\nSpeaker E: The proof that we weren't wasteful.\nSpeaker D: We didn't waste a single bit of play to waste every bit.\nSpeaker C: Nice.\nSpeaker C: All four of those little containers.\nSpeaker C: Yeah.\nSpeaker B: Including the multicolored wave pattern.\nSpeaker D: By one criticism is that we didn't have enough colors to work with.\nSpeaker D: We only had four.\nSpeaker D: Yeah.\nSpeaker C: You could have developed multiple skins, really.\nSpeaker C: I know.\nSpeaker C: More colors.\nSpeaker C: It would be amazing.\nSpeaker C: Oh, well.\nSpeaker B: What did you guys think about the roles?\nSpeaker C: They were good?\nSpeaker C: Yes.\nSpeaker D: What kind of fun?\nSpeaker D: It was, I think it was pretty clever because we were never able to get too far off track because the information came in at the right time and kind of filled the gaps enough.\nSpeaker D: At the same time, you had enough room to kind of just make things up.\nSpeaker D: Do you?\nSpeaker C: Do I just feel like the level of information dropped off severely over the course of the day?\nSpeaker D: Yeah.\nSpeaker C: Maybe it's just me, but I didn't actually get any information.\nSpeaker C: That's the presentation at all.\nSpeaker C: That's true.\nSpeaker C: I got the spreadsheet.\nSpeaker C: I didn't even get an email.\nSpeaker C: Like, that was it.\nSpeaker C: So, yeah, I feel like that was slightly lacking.\nSpeaker C: But then, you know, fill in the blanks on your own level of creativity up.\nSpeaker E: Well, I think that was an issue I kept finding with regard to what to do.\nSpeaker E: Well, no, but also, yeah, when I was reporting about what each of us was doing, I was often talking to people just as to what you were doing.\nSpeaker E: That was a good bunch.\nSpeaker E: And then I also felt like a lot of our discussion would center around specifically what my task was because that was kind of the interface version, which is what the whole project was about.\nSpeaker E: You're very much so.\nSpeaker E: But, and then in the end, I think our job's kind of melded together a little more.\nSpeaker D: That was fun.\nSpeaker D: I think it's fun.\nSpeaker D: The most helpful thing out of everything was getting the PowerPoint slides already put together.\nSpeaker C: So, we didn't have that.\nSpeaker D: There's no way we could have got all that done in time.\nSpeaker C: Very much so.\nSpeaker B: Cool.\nSpeaker E: And I think your leadership was quite good.\nSpeaker E: It was.\nSpeaker B: She said, she actually made a comment off on the boy.\nSpeaker B: You're getting into this.\nSpeaker B: And I really, it's true.\nSpeaker B: I did get it.\nSpeaker B: I felt like I got way too into it.\nSpeaker C: That's kind of a good thing.\nSpeaker B: I feel like I'm a lot.\nSpeaker C: You're the rest of us in the structure to work with.\nSpeaker C: So, hey.\nSpeaker E: So, is that the first time you're taking on that kind of rule?\nSpeaker B: The first time you've ever done anything like project management.\nSpeaker B: I usually organize crap.\nSpeaker B: It's one thing to do, you know, set up a party with your friends.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: A little different.\nSpeaker B: But you guys felt that you could keep the suspension of disbelief kind of like the role.\nSpeaker B: Yeah.\nSpeaker C: Except for a couple moments where it just got out of hand and I knew we were all like 14th.\nSpeaker C: Look at that.\nSpeaker E: I had to admit, as soon as we started, I mean, as soon as we got the play-doh.\nSpeaker E: I can only imagine.\nSpeaker E: You know, the whole concept of really trying to stick with reality went out the window.\nSpeaker E: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Maybe like us, you know.\nSpeaker B: Also like us too, like in a game of control or spaceship.\nSpeaker B: We should have spaceship like us.\nSpeaker B: Do you guys ever used to build a spaceship like us?\nSpeaker A: Yeah.\nSpeaker A: Still everybody knows.\nSpeaker B: Totally.\nSpeaker B: Best spaceships ever.\nSpeaker B: You guys felt like there was enough teamwork and all.\nSpeaker C: Yeah.\nNone: Thanks.\nSpeaker B: Yeah.\nSpeaker B: No, I don't know.\nSpeaker B: I don't know.\nSpeaker B: I was just.\nSpeaker C: No, I didn't actually.\nSpeaker C: I mean, other than minor discussion at meetings, there wasn't.\nSpeaker C: Except for the actual building.\nSpeaker C: Yeah.\nSpeaker C: But I feel like if this was a team project, there actually would have been much more of the collaborative like brainstorming used the board.\nSpeaker C: Well, and this would have been six months worth of work, not like three hours worth of meetings.\nSpeaker E: I mean, I think had the issue, being more serious, we probably would have brainstormed more during our meetings as a chair.\nNone: Yeah.\nSpeaker B: Of course, I'm unconscious of the idea of the project manager asking if you guys feel like there's a team.\nSpeaker B: Yeah, that's kind of like, like, yeah, that is kind of.\nSpeaker B: Yeah.\nSpeaker B: Interesting.\nSpeaker B: It's kind of fascinating, wasn't it?\nSpeaker B: I mean, the whole process is.\nSpeaker D: I wonder why, is there anything about the way that we got so much from it?\nSpeaker D: What was the kept us from going to the board?\nSpeaker B: I don't know.\nSpeaker B: I don't know if there was a right.\nSpeaker C: Mine was the mics.\nSpeaker C: I didn't feel like getting up and down and dealing with all these wires.\nSpeaker C: Yeah, that's it.\nSpeaker C: I was afraid of the break.\nSpeaker D: The mics are loose and each time you get up, it's a possibility of tripping over something.\nSpeaker D: You can tangle there.\nSpeaker E: Well, I don't know what I would have shown on that board.\nSpeaker C: True.\nSpeaker C: But it didn't even occur to me as an option.\nSpeaker C: I mean, I don't know that I would have, but I know that I consciously didn't.\nSpeaker E: I mean, it's just like the paper.\nSpeaker E: I don't know what I really need the paper for.\nSpeaker E: True.\nSpeaker E: Because I've got this laptop.\nSpeaker E: Yeah.\nSpeaker E: And I just use it because it was literally right in front of me.\nSpeaker B: I want to see the output files from the digital paper.\nSpeaker B: I know.\nSpeaker B: My handwriting looks like digitized because my handwriting is crap.\nSpeaker B: Yeah.\nSpeaker B: I mean, see what it looks like in PDF format or something.\nSpeaker D: Usually I would do a lot more to doing too.\nSpeaker D: I know.\nSpeaker C: I felt like I needed to be professional.\nSpeaker C: So I didn't like draw a paper and stuff.\nSpeaker C: Okay.\nSpeaker C: Well, not entirely, but still.\nSpeaker C: I do have less than I usually do.\nSpeaker B: I'm curious about what the debriefing is going to be like.\nSpeaker B: You know, like what is the, well, exactly we're looking for here.\nSpeaker E: So this is all I need to go through.\nSpeaker B: I don't know.\nSpeaker B: I'm not sure if the new idea is found.\nSpeaker D: It is about ideas.\nSpeaker C: Did it just say in an email that we need to discuss that?\nSpeaker B: Well, that's the thing.\nSpeaker B: I got in the email.\nSpeaker B: I got this PowerPoint file, but this slide was.\nSpeaker B: That slide was like that.\nSpeaker B: I didn't change this one at all.\nSpeaker B: Well, I guess one, the right track.\nSpeaker E: Any new ideas with regard to.\nSpeaker E: We're by no control concept.\nSpeaker E: No, I kind of like.\nSpeaker C: I think they still do their job.\nSpeaker C: Yeah, you're fine.\nSpeaker C: I am thinking outside the little square box though.\nSpeaker C: Literally in like form.\nSpeaker D: Yeah, maybe a circle would be all right.\nSpeaker D: Yeah, different.\nSpeaker B: It's kind of make you wonder.\nSpeaker B: I mean, how much can you do with the remote control?\nSpeaker B: It's like inventing a new car.\nSpeaker C: Yeah, yeah, you got to be technically car shanker won't fit on the road.\nSpeaker C: Yeah.\nSpeaker C: I don't know.\nSpeaker B: Okay.\nSpeaker B: I know.\nSpeaker B: I'll be back into.\nSpeaker B: I don't know.\nSpeaker B: I think this is.\nSpeaker C: Oh, how long was I mean, it's supposed to be?\nSpeaker B: 40 ish.\nSpeaker B: I wish you go on a bit.\nSpeaker B: Yeah, about the project you've out.\nSpeaker B: I don't know, but you guys, but I felt like.\nSpeaker B: A bit.\nSpeaker B: Understimulated on the whole thing.\nSpeaker B: Like what like, you know, what am I really doing?\nSpeaker B: You know, what is.\nSpeaker C: Yeah.\nSpeaker C: At the beginning, it started out and I felt actually like under pressure, like the first couple.\nSpeaker C: Yeah.\nSpeaker C: And I was like, had like all this brainstorming I was doing.\nSpeaker C: And then suddenly I was like, well, it's just another two minute presentation that you guys don't really care about.\nSpeaker C: Anyway, so.\nSpeaker C: I think it was the real.\nSpeaker C: You know what I mean?\nSpeaker C: Like we also knew when we were headed with it.\nSpeaker C: So it didn't feel.\nSpeaker C: Definitely.\nSpeaker D: When I first filled up the question, I was marking it probably higher in terms of how much I had to, how much I stressed over it.\nSpeaker D: And then by the time I got to the last, I was like, yeah, whatever.\nSpeaker D: Yeah.\nSpeaker E: I think it was also a realization of basically just copy and paste.\nSpeaker E: Let's give it to you.\nSpeaker E: Yeah.\nSpeaker E: Which wasn't so clear to me at the beginning.\nSpeaker B: I actually didn't do that at all though.\nSpeaker B: Every single one, all the presentations, I either added slides or edit them.\nSpeaker E: Oh, I added like five slides.\nSpeaker C: See, I only got blank ones.\nSpeaker D: I just got blank ones.\nSpeaker D: What?\nSpeaker C: Really?\nSpeaker C: My slides were all blank.\nSpeaker C: They'd have a title maybe.\nSpeaker B: Yeah.\nSpeaker B: And they were just empty.\nSpeaker B: Like mine, yeah, they didn't come like this.\nSpeaker B: Like this was what it looks like.\nSpeaker B: Like with those words are on it.\nSpeaker B: Literally just like that.\nSpeaker C: What do I hear is always so more complicated.\nSpeaker C: I just needed slides.\nSpeaker C: I think I added a slide one time.\nSpeaker E: I had many slides every time.\nSpeaker C: With the whole new background.\nSpeaker C: That was pretty cool.\nSpeaker D: That was class.\nSpeaker D: High moment.\nSpeaker A: Oh, really?\nSpeaker B: Any of the thoughts come to mind?\nSpeaker C: I want to know how our product was there.\nSpeaker C: I can't.\nSpeaker D: I think it would fail.\nSpeaker D: I think it would be expensive.\nSpeaker C: Especially like that.\nSpeaker A: Okay.\nSpeaker C: And Apple would have read about that.\nSpeaker C: He would have been high.\nSpeaker A: It was skeptical.\nSpeaker C: But you know the whole thing.\nSpeaker E: Even you.\nSpeaker D: you. We kind of designed it to look a little bit like face.\nSpeaker E: Here by happy face.\nSpeaker E: Actually, that looked a lot more like a tongue previous to build other design modifications.\nSpeaker E: I hope you appreciate the incorporation of some taint file from a random\nSpeaker A: pick at bar that had re-consumed by accident.\nSpeaker B: Interesting.\nSpeaker C: An interesting day, I don't know.\nSpeaker E: Yeah, I'd say so. So again, I reiterate my question of how different we are compared to the other groups, especially between culture groups.\nSpeaker E: I know.\nSpeaker B: I know.\nSpeaker B: I know.\nSpeaker B: It seemed like everything flowed pretty logically, you know, from the basics to the concept.\nSpeaker B: Although the whole concepts thing, the whole concepts phase, I don't think I really understood like the concept.\nSpeaker B: Well, the idea.\nSpeaker B: OK, connocean of functional item.\nSpeaker B: Yeah, I mean, it's like I have a concept of a mug's material.\nSpeaker B: It's just it is what it is.\nSpeaker B: Yeah.\nSpeaker B: You know, maybe rather than concepts, you should be thought of.\nSpeaker B: We should.\nSpeaker B: I thought of it.\nSpeaker B: I thought of rather than terms of concept, I thought of it in terms of.\nSpeaker B: Like.\nSpeaker B: It's a proposed idea and then the final would be like the actual specified prototype or whatever.\nSpeaker B: I don't know.\nSpeaker B: Bono is kind of interesting.\nSpeaker B: So we have more slides.\nSpeaker B: No, just this closing one.\nSpeaker B: I mean, if established at the cost weren't really within budget, but we could.\nSpeaker C: We got it to be.\nSpeaker B: We did the project evaluation based on.\nSpeaker B: But Sarah's evaluation of on offs with.\nSpeaker C: It was really technically an evaluation of the product, not the project.\nSpeaker C: In general, which is not sure is the same thing at the time that just made more sense, but I could see if they were really asking about.\nSpeaker C: Us.\nSpeaker B: Yeah.\nSpeaker B: Of course we did.\nSpeaker B: We had a fight think about it.\nSpeaker B: And it's all recorded.\nSpeaker B: And what I'm going to I'm going to put.\nSpeaker B: I'm supposed to do this final report thing at the end.\nSpeaker B: So I'll put all that into the final report as well, or as much as seems like maybe not like the articles and stuff like because and so forth.\nSpeaker B: But I'll put most of it in the cards.\nSpeaker D: It'd be so cool if we get a copy of the recording.\nSpeaker C: I've had to done.\nSpeaker C: I've done transcription before and it's really ridiculous how many words people say like just in the middle of their sentences like that that mean nothing.\nSpeaker B: The whole branch of psychology that looks into that cycle in the six.\nSpeaker B: What the guys are sitting here.\nSpeaker B: Yeah, he's studying.\nSpeaker B: Something.\nSpeaker B: Yep, they're called the disfluidities.\nSpeaker D: Disfluidities.\nSpeaker C: That's a good word for it.\nSpeaker B: Yeah, we like our fancy phrases and.\nSpeaker C: Just add some prefixes sounds class here.\nSpeaker B: Exactly.\nSpeaker B: I will save this into the.\nSpeaker D: I find myself hitting the send and receive button.\nSpeaker C: Yeah.\nSpeaker C: I can pull something.\nSpeaker B: Yeah, I know something.\nSpeaker B: Yeah.\nSpeaker B: Come on, give me some information.\nSpeaker B: Well, to be fair, we're hooked.\nSpeaker B: We're all hooked on the internet.\nSpeaker B: Yeah.\nSpeaker E: So I mean, we are addicts.\nSpeaker E: That's scary.\nSpeaker E: Yeah, it is scary.\nSpeaker E: I'll just throw on that either.\nSpeaker B: I know imagine we went the first 10, 15 years of our lives without the internet.\nSpeaker B: It's only the last 10 that were like what was the internet.\nSpeaker B: I mean, you know, just in the past five, we've gone from 28 eight modems to broadband all the time.\nSpeaker E: I think we've had internet for like 18 years.\nSpeaker B: No, we have, but not in the sense.\nSpeaker C: It's so, you know, my grandparents had an 80s and I got it in 94, but still it's crazy.\nSpeaker D: OK.\nSpeaker C: So in the 80s, my grandma does computer science back in the States.\nSpeaker C: And so they had an EDU and Gov Network.\nSpeaker C: There's basically the fundamental structures, but it wasn't household household yet because it hadn't been.\nSpeaker E: Yeah, it was to the like seven universities or something.\nSpeaker C: Yeah.\nSpeaker C: That was him.\nSpeaker E: You guys ready to celebrate?\nSpeaker E: Yeah, that's how I said that includes like\nSpeaker C: champagne or something exciting.\nSpeaker B: Should probably call that meeting to an end. I think that's a closer.\nSpeaker B: Cool.\nSpeaker D: It's been 40 minutes or whatever.\nSpeaker D: Because it's been long enough.\nSpeaker E: So great.\nSpeaker C: Where'd you find that?\nSpeaker C: I have no idea.\nSpeaker D: It's at the only song I have.\nSpeaker D: There's another one.\nSpeaker B: This is where I was media player.\nSpeaker B: Yeah, that's awesome.\nSpeaker B: The default track.\nSpeaker B: Yeah.\nSpeaker B: I thought it was David Burns looking to the eyeball.\nSpeaker B: That's when it seems to be shipping with her, or the ship with for a while.\nSpeaker B: Maybe this is a new version.\nSpeaker C: So is that a close?\nSpeaker B: Yeah, I guess we'll call that a day.\nSpeaker B: The end of the meeting.\nSpeaker B: Fab.\nSpeaker C: Thank you, gentlemen.\n"}, {"meeting_id": "ES2004b", "summary": "The team mainly made some adjustments to the remote control design to meet the new project requirements. The second meeting started with Project Manager who briefly went over the last meeting. Industrial Designer hoped to minimize the size of the battery. The team thus decided on a rechargeable one. User Interface desired to simplify the remote control style, cutting down unnecessary functions. According to the market research presented by Marketing, the team agreed on a trendy and user-friendly flip-top design. An alarm would be incorporated for detection in case the remote control was lost in the house.", "dialogue": "None: How?\nNone: Maybe.\nSpeaker B: Is that clear?\nSpeaker B: Huh.\nSpeaker B: That's green flat.\nSpeaker D: No, not really too hard.\nSpeaker B: Oh, my God.\nSpeaker B: Okay.\nSpeaker B: That's good.\nSpeaker B: Right, okay.\nSpeaker B: I'll go over what was decided on universal control.\nSpeaker B: One hand set for all TV videos.\nSpeaker B: That it was important that the product that built a wide range of consumers, wide age range, not limited in one.\nSpeaker B: It's very important to affect the company's image and our product, the profession, and electronics, and that kind of thing.\nSpeaker B: Our budget would have to affect, try not to reflect our budget that we might have a bit of...\nSpeaker B: Oh, you can do it.\nSpeaker B: The dissonance between what our budget was and what we wanted to look like.\nSpeaker B: We discussed the flip-o from design, reducing the size of control and an electronic panel.\nSpeaker B: The further features that we're programming, things like that.\nSpeaker B: Three presentations I've got written here, so...\nSpeaker B:...surely here for my team, though.\nSpeaker C: Is it okay if I postpone that to later? I just want to get access to a little bit more information.\nSpeaker D: Is that okay?\nSpeaker D: Okay, yeah, okay, first. Can I grab the...\nSpeaker B:...like me?\nSpeaker D: Thanks.\nSpeaker D: Okay, so, I think...\nSpeaker B:...and F.A.\nSpeaker B:...butching.\nSpeaker B: All right.\nSpeaker E: Maybe.\nSpeaker D: Okay, this is the working design presented by me, the industrial design extraordinaire.\nSpeaker D: Okay, this is where I went a bit mad with PowerPoint, so...\nSpeaker D: Well, the first thing question asked was, what are we trying to design?\nSpeaker D: Well, a device which basically just sends the signal to the TV to change it, and just say whether that be the power or the channel, or the volume, everything is just some sort of signal to change the state of the TV or other appliances, sending a signal to.\nSpeaker D: So, I decided that I'd have a look at what other people have designed and try and take some inspiration from that.\nSpeaker D: But although we will want to be taking ideas from other people, we want to make sure that our design stands out.\nSpeaker D: And I thought that was something that, well, it wasn't really in my area because I'm dealing with the inside, really.\nSpeaker D: I ran out of time so I couldn't do this one as fun as the last time.\nSpeaker D: And I found out that most controls use some form of infrared sent signals to the TV, presumably because of the cost issue of something like the same thing that can be used wireless, and you don't need to send very much information.\nSpeaker D: Most of them are powered by some form of battery.\nSpeaker D: Now, R1, I'm not sure whether we want to look at the size issue, because most of them are powered by AAA batteries, but those can be quite bulky.\nSpeaker D: So, I didn't know if you wanted to look at something else, so we could shrink down the size of the control.\nSpeaker C: Can I interject to ask the question?\nSpeaker C: Is that appropriate?\nSpeaker C: You're saying the AAA batteries are small or the current?\nSpeaker D: No, no, if you look at most remote controls, they're quite chunky, and that's because of the size of the batteries.\nSpeaker D: They have to be, obviously, this certain size to fit those batteries.\nSpeaker C: The AAA's are the smallest you can get.\nSpeaker D: Yeah, well, you can get the sort of circular round ones, but I'm just wondering about power consumption and how much you need to send the battery across, which leads on to the next point about being small and easy to carry around.\nSpeaker D: Now, the ones at the moment are small, but I was just wondering if we can look at something a bit smaller.\nSpeaker D: Now, the main components I came up with were obviously the power source for the batteries, because otherwise it's not going to work.\nSpeaker D: As I said about the which batteries we're going to choose, we can discuss that later, and then obviously you need something to decode the information that you're putting in from the controller and to turn it into an infrared signal, which we're going to use to send the information.\nSpeaker D: Now, these have a wireless range up to about five meters, which is so suitable for anyone who's watching their TV unless they're in a cinema, which not most people do say as we're applying for the most audience, that should be fine, and then I was just had a quick look at the external design, but I left that mostly to the interface designer.\nSpeaker D: And so this is what I had as the basic idea of what we want to do.\nSpeaker D: It's not a proper circuit.\nSpeaker D: I'm not sure if it works or not.\nSpeaker D: I'm not even confident that those are the real way you'd wire it up.\nSpeaker D: But anyway, we've just got the power going to the infrared bowl with the chip and the UI interface.\nSpeaker D: Which would basically, when you pressed anything, the chip would convert it into some form of IR data, which could be decoded by the TV receiver, which obviously means that we're going to have to conform to whatever form of communication that the TVs are already using.\nSpeaker D: And since that's another reason to use infrared, because that's what all TVs are doing.\nSpeaker D: And then finally, we want it to be available to a wide audience at a low cost.\nSpeaker D: So all the components that I've put forward are low-end costs, so that should be good.\nSpeaker D: It should be different enough from the alternative products to get a good consumer base.\nSpeaker D: We were talking about it before, and also just something that I was thinking about.\nSpeaker D: Because they're small, they're also easy to lose.\nSpeaker D: So if we could look into some way of, I don't know, some anti-going down the side of the sofa.\nSpeaker D: The thing that you could have, that was just sort of a general point there.\nSpeaker C: That's a very important point. It came up in our market research findings too, so I can refer to that.\nSpeaker D: I really like my tip, isn't it?\nSpeaker D: Yeah, that's what I came up with there.\nSpeaker B: Would you like to continue more from there?\nSpeaker B: It's completely different.\nSpeaker A: It's completely different.\nNone: It's completely different.\nSpeaker A: I think that's okay. I feel right.\nSpeaker D: It's a good function.\nSpeaker D: I think it's a good function.\nSpeaker A: It's a good function.\nSpeaker A: Actually, some of my points might overlap with what William just mentioned.\nSpeaker A: But basically, my method was like whatever brainstorming we did in the last movie, just took up valuable points and started developing on that.\nSpeaker A: There might be some missing loops in this thing, which I think will take a feedback from the marketing.\nSpeaker A: Because I haven't had some marketing data, and basically every product is market driven.\nSpeaker A: So purpose as William already said, I would put it to simplify the interaction with TV, to make it as simple as possible.\nSpeaker A: And to summarize, I would say it should be user friendly by being easy to use, rather than having a lot of complex buttons, because you can have an engineering driven, maybe having 100 buttons, and maybe having a remote control, which has the main features, like volume control or channel changing the channel or whatever.\nSpeaker A: But we have to make it unique so that people want to buy it, will punch in these two features together.\nSpeaker A: So what the concept is to have a flip top model, the main functions such as which are often used will be on the top, and the complex functions which say you can say like the young generation or trendy generation want to say program their favorite channels or whatever, can be put in the middle part of the flip top.\nSpeaker A: So it could be accessed by wider range of audience.\nSpeaker A: And we can punch in new features such as added features such as shockproof body, and maybe designed to appeal to a lot of people.\nSpeaker A: Finding most people prefer user friendly rather than complex remote controls, because there are times like people have used a remote control for, say, a year or something, and they are not used maybe 30 to 40% of the buttons.\nSpeaker A: So it is no use of punching in those, trying to put in those things on the top of the remote controls and try to confuse our user.\nSpeaker A: As we saw we have to make a profit also.\nSpeaker A: So we can maybe go for an economy of a higher production by 50 million we said.\nSpeaker A: With that I think we will be able to achieve the economies of scale also.\nSpeaker A: So we can add in more features and make it less costly.\nSpeaker A: And that is, so this is, if you ask me personally, I would make a flip top with a trendy design and maybe we should look at also like the buttons, whether they are like soft or a little hard because there are times where the buttons tend to be a bit hard after continuous usage and all that.\nSpeaker A: So in overall the simple and user friendly design, any comments?\nSpeaker B: I think we should ask for it at the end.\nSpeaker B: Okay.\nSpeaker B: So anyone wants to write something down, or want to put them up at the end?\nSpeaker C: I have to move it one time.\nSpeaker C: I need to push it a bit more.\nSpeaker C: A bit more?\nSpeaker C: Yep.\nSpeaker C: I do.\nSpeaker A: Okay.\nSpeaker D: I want to go a bit more.\nSpeaker A: Is that okay?\nSpeaker C: Thank you.\nSpeaker C: Just pull it closer a little bit.\nSpeaker D: You should be able to.\nSpeaker D: Okay.\nSpeaker C: Get it right over.\nSpeaker C: Thank you.\nNone: Okay.\nSpeaker C: I am going to look at the functional requirements from the marketing point of view.\nSpeaker C: And obviously our starting point with marketing is always market research.\nSpeaker C: So that's where we started.\nSpeaker C: We used our usability lab, the company's usability lab.\nSpeaker C: We did our usual selection methods to get a cross section of the general public male and female all age groups on 15 upwards.\nSpeaker C: And we observed them in the lab, just their general use of the remote control.\nSpeaker C: You can see we had 100 subjects there.\nSpeaker C: Our findings, lots of findings, I have just summarized some of them here.\nSpeaker C: The overall thing, which I have at the top there in italics, is that users dislike the look and feel of current remote control.\nSpeaker C: So that's obvious design implications there.\nSpeaker C: We found that 75% of users find most remote controls ugly.\nSpeaker C: 80% would spend more money when a remote control would look fancy.\nSpeaker C: We were quite surprised by that finding.\nSpeaker C: But that's quite a high portion of our international target group are prepared to spend more money for something that's a bit nicer.\nSpeaker C: But nice looking.\nSpeaker C: Current remote controls do not match well with the operating behavior of the user overall.\nSpeaker C: For example, you can see below there 75% of users zap a lot.\nSpeaker C: So you've got your person sunk back in the sofa channel hopping.\nSpeaker C: So again, there's power implications there.\nSpeaker C: 50% of users only use 10% of the buttons.\nSpeaker C: So again, a big design issue there.\nSpeaker C: And possibly we can also cut back on cost if we don't have so many functions actually on the remote control.\nSpeaker C: The biggest frustrations that people found was regard to personal preferences were something that you mentioned earlier.\nSpeaker C: Remote controls are often lost in the room.\nSpeaker C: It's the slipping down the back of the sofa type of thing.\nSpeaker C: 50% were particularly frustrated by that.\nSpeaker C: 34% of people said they take too much time to learn to use.\nSpeaker C: And I think that ties in with the previous finding of people only using 10% of the buttons.\nSpeaker C: They just can't be bothered to learn about the other functions.\nSpeaker C: Slightly more than a quarter of people said it was bad for repetitive strain injury.\nSpeaker C: You know, there's small movements of the remote control can lead to kind of shoulder and elbow problems.\nSpeaker C: The vast majority of the 35 and under age group would like a good crystal display and speech recognition.\nSpeaker C: Again, that was to aid, I think, in when they've lost the actual remote control, some kind of speech recognition.\nSpeaker C: Something we didn't put to them, but which I'm thinking of now is even if perhaps the lost control can give off a bleep every now and again till you find it or a flashing light possibly.\nSpeaker C: That trend reverses in the older age groups.\nSpeaker C: So 35s and unders who would like those two features.\nSpeaker C: That kind of evens out 35 to 45s and in the older age group it kind of reverses and not supported with this.\nSpeaker C: I had marvelous tables and things that I could show you, but I think I'll just keep it simple.\nSpeaker C: If there's any more information I can email you.\nSpeaker C: Extra details. Is that okay?\nSpeaker B: Right.\nSpeaker B: We have new project formats.\nSpeaker B: We're not going to be using teletext.\nSpeaker B: Partly consumers consider outdated for the internet now.\nSpeaker B: Our control is only going to be for TV. It's not going to be combined control, which limits all of the different things.\nSpeaker B: It limits the cost for us.\nSpeaker B: It also makes it easier to understand.\nSpeaker C: So can we not program a video with this remote control?\nSpeaker B: Says for TV only.\nSpeaker B: So it looks like this channel.\nSpeaker B: I think maybe Skye, things like that might be incorporated into it.\nSpeaker B: But I don't know what do you think it just said for TV only.\nSpeaker B: Would that imply a video using controls?\nSpeaker B: Yeah.\nSpeaker B: Do you do video as well?\nSpeaker D: I don't know.\nSpeaker B: Well, if you get combined TV and video.\nSpeaker D: Yeah. If you've got Skye, one of those Skye Plus boxes now you can record straight off the TV anyway.\nSpeaker D: On to the TV hard drive or something.\nSpeaker B: I think we've seen it still got small and soft.\nSpeaker B: Yeah.\nSpeaker B: So grabbing.\nSpeaker A: Anything about that on the market research or something regarding whether people want a combined something on that?\nSpeaker C: We didn't really look into that, but remember we found that finding that most people only use about 10% of the buttons.\nSpeaker C: I think that those did tend to be the basic channel hopping things and on and off of the video fast forwarding so on.\nSpeaker C: And I think it's sort of general knowledge that people do find programming their videos a night.\nSpeaker C: So I don't know if that's something we need to do.\nSpeaker D: Just as an idea on the speech recognition thing.\nSpeaker D: It would probably be quite expensive to incorporate an entire speech recognition thing and they're not that great.\nSpeaker C: I've just called up that table there.\nSpeaker C: We asked those two questions. The table relates to both questions so we didn't differentiate.\nSpeaker C: Would you prefer an LCD screen? That's a multifunctional mode.\nSpeaker C: And would you pay more for speech recognition and remote control so you can see how the yes, no sort of varies across the age group.\nSpeaker C: A substantial number of don't know is in the older age group.\nSpeaker C: I think that's just general fear of new technology.\nSpeaker D: But on that again I just thought because you can get those key chains now and you whistle.\nSpeaker D: And then it allowed for loud noise to let you know where it is.\nSpeaker B: So I thought that could be for us.\nSpeaker B: We do have a budget limitation that we can't control ourselves.\nSpeaker B: So I think when we can take a cheaper option which still does the same kind of thing.\nSpeaker B: Because it is something.\nSpeaker B: It needs to be the thing that you use to find it needs to be something you don't lose.\nSpeaker B: You're listening to whistling.\nSpeaker B: Maybe on the TV you could put like a pack on the TV or something.\nSpeaker B: So you can't see the remote. You go and press the button on top of the TV and it beeps.\nSpeaker B: And it beeps.\nSpeaker A: Yeah, okay.\nSpeaker B: It's over there.\nSpeaker B: It's a super idea.\nSpeaker B: But that sounds a lot cheaper too.\nSpeaker B: Yeah.\nSpeaker B: The third thing was that we have to make sure the corporate image is very much incorporated into the handset.\nSpeaker B: So we want logo.\nSpeaker B: We want fashionable trendy.\nSpeaker B: I mean what you were talking about with the marketing.\nSpeaker B: People paying more to loot goods.\nSpeaker B: We need to focus on that as well.\nSpeaker C: Further market research will be needed to focus on what that is.\nSpeaker C: It's going to be different for a 15 year old from somebody who's 60.\nSpeaker C: And also across the world of reaming the international market.\nSpeaker C: What is attractive to a trending New Yorker and what is attractive to a retired South African?\nSpeaker C: I don't know.\nSpeaker C: You get the idea.\nSpeaker C: It's going to vary around the world.\nSpeaker C: At the end of the day, the engineering design is one thing.\nSpeaker C: It's the user interface design that may sort of be a fashionable aspect of it.\nSpeaker C: We might have to change the different markets around the world.\nSpeaker A: So are we talking about a single model or maybe a classic design?\nSpeaker D: Well, yeah, you could have a number of different designs.\nSpeaker D: I mean inside that would be essentially the same.\nSpeaker A: We just could be same in the body, because we're not good at it.\nSpeaker A: So...\nSpeaker B: What about...\nSpeaker B: We were talking about the buttons.\nSpeaker B: The controls that are coming out now have kind of big rubber buttons.\nSpeaker B: Not tiny, but on big rubber buttons.\nSpeaker B: But what about...\nSpeaker B: I mean, because we've got to make it original.\nSpeaker B: What about...\nSpeaker B: You know with the touch screen, you do?\nSpeaker B: Yeah, that's what I was just...\nSpeaker B: Yeah.\nSpeaker B: So it's like a little panel that you touch rather than a button.\nSpeaker B: Which shouldn't wear out as much either.\nSpeaker D: Well, no, because you don't actually have to press.\nSpeaker B: I don't have to press it, you just have to put yourself onto it.\nSpeaker B: I think that might appeal to Japan thinking young office people trendy.\nSpeaker C: It will appeal to sections of the market.\nSpeaker B: But quite easily labelled so that anyone can...\nSpeaker B: Oh yeah, that's obvious what that's for.\nSpeaker B: And it's not daunting to maybe the older generation.\nSpeaker D: Well, say ergonomics, as we're saying with the different designs.\nSpeaker D: Ergonomics, there's sort of physically different things.\nSpeaker D: Because if you've ever seen the Xbox, they had to make two different sized controllers.\nSpeaker D: Because people in Japan wouldn't buy it.\nSpeaker D: Because the controller was physically too big.\nSpeaker D: Because they're cut just generally.\nSpeaker D: Japanese people have smaller hands.\nSpeaker D: So they can get around the controller.\nSpeaker D: So if you...\nSpeaker D: I don't know what, because you obviously can have bigger buttons for some countries or something.\nSpeaker D: Small ones rather.\nSpeaker A: Maybe to...\nSpeaker A: It was indicated that the risk for the replicated use, the injuries.\nSpeaker A: So a touchscreen could be a better option for them.\nSpeaker A: Yeah.\nSpeaker A: So a bigger...\nSpeaker A: Okay, so you're dividing designs based on not only segment age groups.\nSpeaker A: You're dividing it according to the countries also, the market.\nSpeaker A: Maybe for US and all you can have a slightly bigger remote menu.\nSpeaker A: And maybe in Japan and all you need to have is small.\nSpeaker B: I think we have to design one product.\nSpeaker B: And then the company can take it wherever they want to.\nSpeaker B: And then they can make it smaller, or they can make it bigger, or they can change features slightly.\nSpeaker C: The internal engineering design has got to remain the same.\nSpeaker B: It's going to be the same.\nSpeaker B: So we need to focus on just one thing.\nSpeaker B: You can get bogged down in lots of different possibilities.\nSpeaker C: I'm concerned when you raise the RSI issue again, the p additive strain injury.\nSpeaker C: I don't think just moving your finger around in a small screen is going to deal with that enough.\nSpeaker C: That is still a kind of question mark issue, how we deal with that.\nSpeaker C: RSI tends to be caused by repetitive small movements.\nSpeaker C: I really can't get my head round.\nSpeaker C: This one is me.\nSpeaker C: I have to be postponed to a future meeting, but it's something we should think about.\nSpeaker D: I was just thinking about how you can combat that.\nSpeaker D: Because without doing something where you have to move your arm.\nSpeaker D: I know.\nSpeaker B: And it becomes ridiculous.\nSpeaker B: Yeah.\nSpeaker B: I think that's the only way that you can avoid that issue.\nSpeaker C: Do we have to initially, you know, looking at the findings here, focus on a younger age group initially, and then broaden out the market later?\nSpeaker C: Do we really have to go for everyone right away?\nSpeaker A: Focus on the biggest market.\nSpeaker A: Yeah.\nSpeaker C: When we've been throwing up our ideas, we're automatically talking about business people, young people trying to do it.\nSpeaker B: We are talking about the type of company that we're working for as well.\nSpeaker B: They want it to be fashionable.\nSpeaker B: They want it to be trendy.\nSpeaker B: And you wouldn't automatically assume, associate that with the older generation.\nSpeaker B: Now, with the baby boomers, the older generation are actually larger.\nSpeaker B: They have a greater population than us young people.\nSpeaker B: But I don't think we're focusing on that.\nSpeaker B: I think we are focusing on a sort of mid-range business kind of class.\nSpeaker C: I'm just thinking of budgetary issues too.\nSpeaker C: For when it does get to the broad scale marketing stage, we want to waste money, not be profligate.\nSpeaker C: And, you know, focus on where the idea will be taken up.\nSpeaker C: First, it's more slightly to be taken up.\nSpeaker C: First of all, where the main purchasing power is coming from for a product like this.\nSpeaker B: Okay.\nSpeaker B: So, the remote control functions, we've got the TV, we've got the video.\nSpeaker B: Now, there's, I can't remember what it's called, the little code at the end of program details.\nSpeaker B: Video pass.\nSpeaker B: Yeah.\nSpeaker B: We could use that as an alternative to programming in times, things like that.\nSpeaker B: Yeah.\nSpeaker B: Is that, I always find that really easy when I discovered it.\nSpeaker B: Mm-hmm.\nSpeaker B: Because you've got your general record anyway.\nSpeaker B: There's a program on you once put the card on, that's fine.\nSpeaker B: But if you do want to take something in today's time, you're not sure if you're going to, you put the number in, and it's just a number.\nSpeaker B: It's not a date, it's not a time, it's not a channel, it's not when it finishes, it's not anything like that.\nSpeaker D: It's just a number.\nSpeaker D: And you wouldn't, you wouldn't need a whole host of extra buttons for that.\nSpeaker D: Yeah.\nSpeaker D: You just need to work, because you've already got numbers there.\nSpeaker B: You've already got numbers there.\nSpeaker C: I'm not going to cross that function, but it sounds wonderful.\nSpeaker B: It is after, if you look in the newspaper, a TV guide or any TV guide, there's a five, six digit number afterwards, and that's number you put in.\nSpeaker B: Huh?\nSpeaker B: And it's recorded, but it's going to be on on Tuesday, but ten o'clock, on the seventeenth, things like that, so you don't have to worry about dates, and you don't have to worry about times.\nSpeaker B: And it has been around for a long time.\nSpeaker D: It's been around for a long time.\nSpeaker D: It's just, it's not very well advertised.\nSpeaker B: No, it's not.\nSpeaker B: I don't know.\nSpeaker B: I think if awareness was kind of much more fun about that.\nSpeaker D: Yeah.\nSpeaker D: So if you've got something like Sky Anyway, you can just click on it, you can just press the button on the program once, and it will record that program.\nSpeaker D: Well, it's on it.\nSpeaker D: You press it on it twice, and it will record the whole series.\nSpeaker D: Excellent.\nSpeaker D: So that, yeah.\nSpeaker D: But just add that function.\nSpeaker D: Yeah.\nSpeaker D: It would be really good.\nSpeaker D: Okay.\nSpeaker C: Can I just run this past your oil that occurs to me?\nSpeaker C: I don't mean to quickly jump from one subject to another, but just discussing the different age groups and targeting the different age groups.\nSpeaker C: It occurs to minority producer on mobile phones.\nSpeaker C: That's kind of what led us on to comparing TB remote controls with their design features.\nSpeaker C: Chain companies like Carful and Warehouse, you can pop in anytime with a phone that you bought for them, if you've got any problems with it, and they'll fix it before the company, you can use their telephone.\nSpeaker C: Is it worthwhile with our retail outlets having a similar kind of service?\nSpeaker C: So that if older users were deterred from buying this, if they know they can just pop into one of our high street outlets, you know, which button is it I press for this?\nSpeaker C: That's a free aspect of our service.\nSpeaker C: Would that not make it more attractive to them?\nSpeaker D: Mm.\nSpeaker D: But the only problem is that with a mobile phone, you sign for a contract.\nSpeaker D: So the companies who you deal with have actually got an obligation to help you out and also, I mean, it's fair enough that I have some sort of help service, but I'm not sure how much the cost would be.\nSpeaker B: Right. The functionality of it in the sense that you're sitting there, you're pressing this button and your TB's not doing it.\nSpeaker B: All right. Taking your TB and your phone is saying, look, this is what I'm doing, it's not what the question I do.\nSpeaker C: If it was something as simple as you couldn't change the channel, but I mean, for a three anymore, or are we absolutely deaf and it's only going to be for TB in video?\nSpeaker C: Yeah, I mean, no.\nSpeaker D: Okay. Just a thought.\nSpeaker D: I mean, instruction, but I feel like I'm saying.\nSpeaker B: I mean, there's customer service.\nSpeaker B: There will be a customer service number thing that you can phone up and speak to.\nSpeaker B: Sure. And that way there's no call out charge. There's no extra person has to walk to your shop in the high street.\nSpeaker A: Yeah. I think I'm going to do much of an effort for a person to, for a phone, maybe he might walk down the street.\nSpeaker A: But for a remote, he will just...\nSpeaker C: Not perception simple functions because we're focusing on that. Yeah.\nSpeaker B: We should focus on making manuals user-friendly as possible because a lot of them are just trying to write in lots and lots of pages.\nSpeaker B: It puts people of reading them, so they just do the obvious.\nSpeaker B: Sure.\nSpeaker B: That should be something I'm thinking about.\nSpeaker B: What other functions that we need?\nSpeaker B: I'm not sure if we need a design decision now, but we should think about...\nSpeaker B: Because you've got the dilemma between old batteries, that's why I think you're back.\nSpeaker B: What about phone batteries? What kind of batteries are that?\nSpeaker D: That's specially made for the mobile phones.\nSpeaker D: But they come with the charger. I mean, you could bundle a charger.\nSpeaker B: They've been last quite a long time.\nSpeaker B: And if you had... The thing that you get with mobile headphones, you sit it in its charger when you're not using it or at night or something, but it doesn't really matter because it never really runs out because it lasts a long time when it is charged.\nSpeaker B: Something like that. Shoot with juice.\nSpeaker D: Yeah, if you had one of those, just coming back to your other point about pressing the button and setting off the bleeper and the rim, it could be on that.\nSpeaker B: Yeah, on that as well.\nSpeaker A: So, are we talking about the concept of a rechargeable?\nSpeaker D: Yeah, some sort of docking station.\nSpeaker B: May charge for a little bit of protection.\nSpeaker B: So, the rechargeable would be your...\nSpeaker D: Yeah, that would be fine. And also that would mean they wouldn't have to go out buying batteries all the time.\nSpeaker B: Yes, which is too.\nSpeaker B: The shape you've got trendy. I don't want a big box with lots of things.\nSpeaker B: You don't want a tiny sort of little thing either because then you have got the repetitive stranger, no matter how much you try and make it simple.\nSpeaker C: And we don't want so many remote controls look absolutely identical.\nSpeaker C: These long-grown things with the same coloured buttons or crammed in on the surface.\nSpeaker C: We definitely, an obvious thing, a very simple thing, is to get away from these brown rectangles. We don't want that.\nSpeaker B: Okay, so we've got a flip screen.\nSpeaker A: What I was thinking of a design which Nokia came up with almost six or seven years back.\nSpeaker A: Basically, we have a flat one. It looks like a box, like a chocolate. That's very strenuous because your thumb is slightly up.\nSpeaker A: So, they come up with something like this, curved up. So, here. And you don't have to bend your thumb too much.\nSpeaker A: So, it's like a banana cutter, kind of curved up with a boat.\nSpeaker A: So, what happens is you don't have to press your thumb too down.\nSpeaker A: So, it's already curved up. So, your thumb doesn't need it.\nSpeaker A: So, we can have a mold it according to your...\nSpeaker A: The way you hold up. It's kind of semi-circular in the bottom.\nSpeaker A: I wouldn't say exactly semi-circular.\nSpeaker B: So, when you look into the company logo, things that are associated with the company, in view to try and incorporate that into the design of the product?\nSpeaker B: I mean, for example, if it was a C or something like that, you could have it in a vague C shape that opens up kind of like a shell or something.\nSpeaker B: Something along those lines to be able to incorporate it quite obviously into the design, while also making it quite different from anything else you can say.\nSpeaker A: If you look at that, what we were talking is to make it more...\nSpeaker A: Other than like, C-shells logo could become...\nSpeaker A: You could put a logo in the corner of the model, rather than, you know, trying to make it like a C-shell or whatever you're telling it.\nSpeaker B: Well, we need to think about how it's going to look different.\nSpeaker C: I think that's... that sounds really attractive idea.\nSpeaker C: If not come across anything like that before, if it kind of...\nSpeaker C: Yeah, and that opens out into your flip-top, then it's nice to...\nSpeaker C: You can make it lightweight plastic, nice to hold in the palm of the hand.\nSpeaker C: And just because you're having to actually insert in between the two covers, that's going to take care of some of the additive strain injury, trying to prevent a lot of it, I think.\nSpeaker C: I think it's a really nice idea.\nSpeaker C: And plus, you can get... even though we're using plastic, you can still...\nSpeaker C: I mean, think of the design you can get in plastic, you can do a nice C-shell or a scallop shell exterior.\nSpeaker B: Or you could do different... like you get with mobile phones, different fashers.\nSpeaker B: You could have different kind of casings.\nSpeaker B: Yeah.\nSpeaker B: So you could have like psychedelic ones for younger people and sleek ones.\nSpeaker C: Less chance of it being lost, too. It's not like a chocolate brown-loseenge that's going to go down the sofa.\nSpeaker C: People might want to put it on their mantelpiece or whatever, as if it looks attractive enough.\nSpeaker C: As a kind of ornament, and they're not going to lose it so much at all.\nSpeaker C: It's easy to fit nice to handle.\nSpeaker A: Yeah, maybe we could come with some C-56 designs and then choose which one would have been some of those.\nSpeaker A: That could be the most common design.\nSpeaker A: Okay.\nSpeaker A: If you can think of C-56.\nSpeaker B: Right, I think we have to round it out.\nSpeaker C: Can I just quickly go over the new project requirements that I haven't missed anything?\nSpeaker C: There's no teletext. It's only for TV and by implication video.\nSpeaker C: Corporate image should be incorporated in their remote control.\nSpeaker C: Something about video plus.\nSpeaker B: That was in cutting down the number. That was kind of separate.\nSpeaker B: Cutting down the number of odd functions.\nSpeaker B: Right.\nSpeaker B: Making it simpler instead of having lots of things.\nSpeaker C: So when they press for programming, you can come up on the LCD reminder about using video plus.\nSpeaker C: So that they will look at the guidance.\nSpeaker C: Is this nothing to do with the project requirements? Is this just a thumbs up and add on feature?\nSpeaker B: It kind of takes the place of having a button to press for the date and having the right press for the channel and things like that.\nSpeaker B: So it's kind of doing away with the programming feature.\nSpeaker B: Yes.\nSpeaker B: Right.\nSpeaker B: It could be an advertisement feature rather than design feature.\nSpeaker B: You know, draw attention to that and also pay lip service in the instruction manual.\nSpeaker B: Because it's very simple. So putting it down in words should be helpful.\nSpeaker B: And they be chargeable batteries.\nSpeaker C: Yeah.\nSpeaker C: Was there anything else there that we...\nSpeaker C: Any new project requirements?\nSpeaker B: We've got the button that I think will work through that with the design event.\nSpeaker C: That's not at the moment a requirement. It's something we're looking at.\nSpeaker C: Yeah.\nSpeaker C: Preference. So we come around to market research at some point.\nSpeaker B: And we're talking about there being an alarm or something, a beeping or being lost.\nSpeaker B: Learning how to use it should be as simple as possible.\nSpeaker B: There aren't going to be that many functions incorporated into it.\nSpeaker B: So it should be fine. And the instruction manual.\nSpeaker B: I don't know if we deal with that.\nSpeaker B: I think...\nSpeaker B: Yeah. I'm definitely going to work that way.\nSpeaker B: Okay.\nSpeaker C: Yeah. I think that's us.\nSpeaker C: What would you specifically like marketing to look at before the next meeting?\nSpeaker C: Or should I just sort of generally look at all the issues?\nSpeaker B: I think you might get guided, but...\nSpeaker B: I...\nSpeaker B: In-sharked manuals.\nSpeaker B: Because there tends to be a demonized thing there.\nSpeaker B: Everyone's got like big pilots and button-on videos.\nSpeaker B: Okay.\nSpeaker B: Oops.\nSpeaker C: Questionnaireful.\nSpeaker C: Yes. Right.\nSpeaker C: Because we're at lunchtime and now I think...\nSpeaker B: Okay.\nSpeaker B: I think you can email me if there's any more questions.\nSpeaker B: Absolutely.\n"}, {"meeting_id": "Bmr020", "summary": "The main topics of the agenda were a paper submitted to Eurospeech and the organising of the recording transcriptions to be done by IBM. The results presented in the former show a significant percentage of overlapping speech even without counting in backchanneling. Regarding the transcriptions to be carried out by IBM, the discussion mainly concerned the format of the recordings that should be sent to them. Suggestions included sending only the channels with the dominant speakers for transcription, but it was finally agreed on sending the original files with minimal modifications, as there will be extensive in-house post-processing.", "dialogue": "Speaker E: Okay, we're recording.\nSpeaker C: Say the word zero.\nSpeaker C: I'm doing something.\nSpeaker C: We have brackets, coffee sipping.\nSpeaker C: It's not allowed, I think.\nSpeaker B: Carly brackets.\nSpeaker B: Is that voice?\nSpeaker B: Still a voice.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: Channel two.\nSpeaker E: Do you square brackets running?\nSpeaker E: I have.\nSpeaker E: He's poor transcribers.\nSpeaker H: Not right now.\nSpeaker H: I mean.\nSpeaker F: There's going to be some zeros from this morning's meeting because I noticed that, very, I think, maybe you turned your mic off before the digits were...\nSpeaker F: Oh, it's during digits also.\nSpeaker F: It doesn't matter.\nSpeaker G: So it's not that bad if it's at the end, but it's...\nSpeaker E: Yeah.\nSpeaker E:...and at the beginning it's bad.\nSpeaker E: You want to keep it warm so you get good noise.\nSpeaker E: Noise floors.\nSpeaker E: It's a beautiful meeting.\nSpeaker D: Oh, I know, I just showed a book.\nSpeaker D: Yeah, I did that.\nSpeaker B: Is there any way to change that in a software?\nSpeaker B: Change what?\nSpeaker B: Where, like, you just don't...\nSpeaker B: Like, if you...\nSpeaker B: If it starts catching zeros, like in the driver or something, in the card or somewhere in the hardware, where if you start seeing zeros on what cross-month channel, you just add some random noise floor, like a small noise floor.\nSpeaker E: And certainly we could do that, but I don't think that's a good idea.\nSpeaker E: We could do that in post-processing.\nSpeaker E: Yeah.\nSpeaker E: And the application needs a manual post-processing.\nSpeaker D: Well, I think we don't know what the default is anymore.\nSpeaker D: It's how we're using the front-end stuff, but for when we use the X-E, but it's argument.\nSpeaker D: There is an option.\nSpeaker D: An option, which...\nSpeaker D: When I first put it in, that means when I answer real things, I did actually put in a random address.\nSpeaker D: Okay.\nSpeaker D: Then I realized that putting in random address, you could go to the equivalent to adding a lot of spectrum.\nSpeaker D: Right.\nSpeaker D: And it was a lot faster than you would say, to add a constant to the spectrum.\nSpeaker D: To the certain thing.\nSpeaker D: Okay.\nSpeaker D: Calling random, or something.\nSpeaker B: Right.\nNone: So, this doesn't...\nSpeaker D: Gee, here we all are.\nSpeaker E: So, the only agenda items where Jane wanted to talk about some of the IBM transcription process.\nSpeaker E: I sort of condensed the three things you said into that.\nSpeaker E: And then, just...\nSpeaker E: I only have, like, this afternoon and maybe tomorrow morning to get anything done before I go to Japan for 10 days.\nSpeaker E: So, if there's anything that absolutely desperately needs to be done, should let me know now.\nSpeaker D: You just sent off their speech paper.\nSpeaker C: Right.\nSpeaker C: I hope they accepted both, as a submission and as a paper.\nSpeaker C: But...\nSpeaker C: Well, yeah, you sent in.\nSpeaker C: First, you have to read the first thing.\nSpeaker C: We actually exceeded the delayed deadline by another day.\nSpeaker C: So, oops.\nSpeaker D: They had some extension.\nSpeaker C: Well, yeah, Liz had sent them.\nSpeaker C: I know what's saying.\nSpeaker C: Could we please have another, I don't know, three days.\nSpeaker C: There's a three days.\nSpeaker C: They said yes.\nSpeaker C: And then, did I say three?\nSpeaker E: That was the same thing.\nSpeaker E: Dave Galbert sent me an email.\nSpeaker E: I think it's in the YouTube that there's a special topic section in Eurospeach on New Corp.\nSpeaker E: Corp.\nSpeaker E: Corp.\nSpeaker E: And it's not due until, like, May 15th.\nSpeaker E: Well, this isn't new or?\nSpeaker E: No, it's new.\nSpeaker E: No, it's new.\nSpeaker E: And I got this.\nSpeaker E: I thought it was a Jane as I thought being the most relevant person.\nSpeaker E: So, I thought it was highly relevant.\nSpeaker E: I'm interested to.\nSpeaker H: Have you looked at the URL?\nSpeaker H: I haven't gotten over to there yet, but what I just mentioned yesterday, yeah.\nSpeaker G: I want to send a message.\nSpeaker G: I think Christopher Throck was meant to.\nSpeaker E: I'll help, but obviously I can't really do most of it.\nSpeaker E: So, I don't need any help being that can certainly provide.\nSpeaker C: But there were some interesting results in this paper, though.\nSpeaker C: For instance, that Morgan accounted for 56% of the robustness meanings in terms of number four.\nSpeaker C: Wow.\nSpeaker C: In terms of what?\nSpeaker C: Number four.\nSpeaker H: It's just because he talks really fast.\nSpeaker E: Do you mean...\nSpeaker E: Oh, that's...\nSpeaker H: Is it partly correctly identified words?\nSpeaker E: No, well, according to the transcripts.\nSpeaker E: That worked well regardless.\nSpeaker E: I think it's he's in all of them.\nSpeaker C: We've been working by name, we just...\nSpeaker E: One participant.\nSpeaker E: Did you identify him as a senior member?\nSpeaker C: No, we identified him as the person dominating the conversation.\nSpeaker D: Yeah.\nSpeaker D: I think it's a RP thing, but it's a meal.\nSpeaker D: But, other than that, the like, what was the rest of the paper about?\nSpeaker C: Well, it was about...\nSpeaker C: In terms of the reactions, three kinds of results, if you will.\nSpeaker C: The one was that just the amount of overlap...\nSpeaker C: That's the ugly subject.\nSpeaker C: In terms of number of words, and also we computed something called a spurt, which is essentially a stretch of speech with no pause as exceeding five million seconds.\nSpeaker C: And we computed how many overlap...\nSpeaker C: Spurt's, there were, and how many overlap words there were, four different corpora, the meeting recorded meetings, the robust meetings, switchboard and call home.\nSpeaker C: And found, and sort of compared the numbers, and found that the...\nSpeaker C: You know, as you might expect, the meeting recorder, meetings are the most overlap.\nSpeaker C: But next, we're switched what and call home, which both had roughly the same, almost identical in effect.\nSpeaker C: And the robust meetings were at the least.\nSpeaker C: So one sort of unexpected result there is that two party telephone conversations have about the same amount of overlap, sort of, you know, order of magnitude wise, as face-to-face meetings were...\nSpeaker C: I have had that as a work-changing all my slides.\nSpeaker C: Yeah.\nSpeaker C: Also, in the Levenson, the pragmatics book, you know, textbook, I found this great quote where he says, you know, how people talk about how people are so good at turn-taking.\nSpeaker C: And so, they're so good that generally, the overlap speech does not is less than 5%.\nSpeaker C: So, this is way more than 5%.\nSpeaker B: Did he mean face-to-face or...?\nSpeaker C: Well, in real conversations, everyday conversations.\nSpeaker C: I've been studying these conversation L.S. have been studying for years.\nSpeaker H: Chris, no, it doesn't necessarily go against what he said, because he said generally speaking in order to go against that kind of line.\nSpeaker H: Well, he's made a claim on the big hand-basing.\nSpeaker G: Yeah, but 5% of time or 5% of work is going to ask that.\nSpeaker G: Well, it's time.\nSpeaker C: So, it's still...\nSpeaker C: It's not a discussion.\nSpeaker H: It just says that is it big bell curve and that you have something that has a nice range of...\nSpeaker C: Yeah, so there are differences in how you measure it.\nSpeaker C: But still, it's, you know, the difference between...\nSpeaker C: between that number and what we have in meetings, which is more like, you know, close to...\nSpeaker C: in meetings like these, you know, close to 20%.\nSpeaker C: What was it like saying in the room?\nSpeaker C: That's just meaning.\nSpeaker C: Robustness meeting?\nSpeaker C: It was about half of the...\nSpeaker C: So, in terms of number of words, it's like 17 or 18% for the meeting recording meetings and about half that for...\nSpeaker E: But I don't know if that's really a fair way of comparing between multi-party conversations and two-party conversations.\nSpeaker E: Yeah, I didn't have to.\nSpeaker E: I mean, that's just something.\nSpeaker F: Yeah, I just wonder if you have to normalize, but then I'm going to speak or normalize.\nSpeaker F: Yeah, I just need to look at that.\nSpeaker C: But it's obviously to see if there's a dependence on that number of participants.\nSpeaker E: I bet there's a weak dependence.\nSpeaker E: I'm sure it's not a real strong one.\nSpeaker E: There's not everybody talks.\nSpeaker E: Right.\nSpeaker E: You have a lot of, a lot of two-party subsets within the meeting.\nSpeaker E: Well, regardless.\nSpeaker E: It's an interesting result, right?\nSpeaker C: Right.\nSpeaker C: And we also computed this both with and without back channels.\nSpeaker C: So, you might think that back channels have a special status because they're essentially just...\nSpeaker E: We all said, aha, not at the same time.\nSpeaker C: But even if you take out all the back channels, so basically you treat back channels as non-speech, as pauses, you still have significant overlap.\nSpeaker C: You know, it goes down from maybe for a switchboard, it goes down from, I don't know, 14% of the words to maybe...\nSpeaker C: I don't know, 11% or something.\nSpeaker C: It's not a dramatic change.\nSpeaker C: So it's...\nSpeaker C: Anyway, so that was one side of results.\nSpeaker C: And then the second one was just basically the stuff we had in the HLT paper on how overlaps affect the recognition performance.\nSpeaker C: And we rescored things a little bit more carefully.\nSpeaker C: We also fixed the transcripts in numerous ways.\nSpeaker C: But mostly we added one number, which was, what if you basically score ignoring all...\nSpeaker C: So the conjecture from the HLT results was that most of the added recognition errors from insertions due to background speech.\nSpeaker C: So we scored all the recognition results in such a way that the...\nSpeaker E: By the way, who's on channel 4?\nSpeaker E: You're getting one red.\nSpeaker G: Yeah, I was just wondering.\nSpeaker B: That's me.\nSpeaker C: Well, that's been working hard.\nSpeaker C: That's right.\nSpeaker C: Okay, so if you have the foreground speaker speaking here and then there's some background speech, maybe overlapping it somehow.\nSpeaker C: And this is the time been that we used.\nSpeaker C: Then of course you're going to get insertion errors here and here.\nSpeaker C: Right, right?\nSpeaker C: So we scored everything, and I must say that this scoring tools are pretty nice for this, where you just basically ignore everything outside of the region that was...\nSpeaker C:...to be foreground speech.\nSpeaker C: And where that was, we had to use the first alignment results from the four.\nSpeaker C: So that's somewhat subject to error, but still we...\nSpeaker C:...donded some hand checking, and we think that based on that, we think that the results are valid, and of course some errors are going to be in there.\nSpeaker C: But basically what we found is after we take out these regions, so we only score the regions that were certified as foreground speech, the recognition error went down to almost the level of the non-overlapped speech.\nSpeaker C: So that means that even if you do have background speech, if you can somehow separate out or find where it is, the recognizer does a good job, even though there isn't...\nSpeaker E: Yeah, I guess that doesn't surprise me, because with the close talking mics, the signal will be so much stronger.\nSpeaker E: What sort of normalization do you do?\nSpeaker C: Well, we do...\nSpeaker E: You know, we do recognize it, and the SRI recognize it.\nSpeaker C: Well, we do VTL, MoCotractLengthNomization, and we make all the features have zero mean and unit variance over an entire utterance.\nSpeaker C: Over the entire...\nSpeaker C: Over the entire channel.\nSpeaker C: Now we didn't rerun the recognizer for this.\nSpeaker C: We just took the old...\nSpeaker C: So this is actually a suboptimal way of doing it, right?\nSpeaker C: The old recognition output, and we just scored it differently.\nSpeaker C: So the recognize didn't have the benefit of knowing where the foreground speech started.\nSpeaker D: Were you able to develop the...\nSpeaker D: Did the problems with the local way also...\nSpeaker C: Yeah, not completely, but...\nSpeaker C: Yes, dramatically.\nSpeaker C: So we have to...\nSpeaker C: I mean, still...\nSpeaker C: I should bring the table with results.\nSpeaker D: We can look at it.\nSpeaker D: I would presume that you still would have somewhat higher error with the local assertions.\nSpeaker C: Yes.\nSpeaker D: Because again, looking forward to the non-close mic case.\nSpeaker D: I'm not looking forward to it.\nSpeaker D: It's a high signal noise ratio.\nSpeaker C: Right.\nSpeaker C: Right.\nSpeaker C: So that was number...\nSpeaker C: That was the second set of...\nSpeaker C: The second section.\nSpeaker C: And then the third thing was we looked at...\nSpeaker C: What we call interrupts, although that's maybe a misnomer.\nSpeaker C: Basically, we looked at cases where...\nSpeaker C: So we used the punctuation from the original transcripts.\nSpeaker C: And we inferred the beginnings and ends of sentences.\nSpeaker C: So...\nSpeaker H: Did you use Eberler case also or not?\nSpeaker H: Eberler case...\nSpeaker C: No, we only used, you know, appearance, question marks and...\nSpeaker C:...exclamation.\nSpeaker C: And we know that there's...\nSpeaker C: That's not a very...\nSpeaker C: I mean, we miss a lot of them.\nSpeaker C: Yes, okay.\nSpeaker H: How about also or not?\nSpeaker C: No, how about?\nSpeaker C: Okay.\nSpeaker C: And then we looked at locations where...\nSpeaker C: If you have overlapping speech and someone else starts a sentence.\nSpeaker C: You know, where do these...\nSpeaker C: Where do other people start their...\nSpeaker C:...turns, not turns really, but, you know, sentences?\nSpeaker C: So we only looked at cases where there was a foreground speaker...\nSpeaker C:...and then at the time...\nSpeaker C: So the foreground speaker started into their sentence...\nSpeaker C:...and then someone else started later.\nSpeaker C: Somewhere in between the...\nSpeaker C: And so...\nSpeaker C: What?\nSpeaker G: Sorry?\nSpeaker G: Somewhere in between the...\nSpeaker C: Yes.\nSpeaker C: So that there was overlap between the two sentences.\nSpeaker C: So the question was...\nSpeaker C: How can we...\nSpeaker C: What can we say about the places where the second...\nSpeaker C:...or actually several second speakers...\nSpeaker C:...start their interrupts, as we call them?\nSpeaker F: Three words from the end.\nSpeaker E: And pause the batteries.\nSpeaker C: And we looked at this in terms of...\nSpeaker C: On T closures, only.\nSpeaker C: So we had...\nSpeaker C: We had...\nSpeaker C: For the purpose of this analysis, we tagged the word sequences...\nSpeaker C:...and we timeline them.\nSpeaker C: And we considered an interrupt...\nSpeaker C:...if it occurred in the middle of a word...\nSpeaker C:...we basically, you know, considered that to be interrupt...\nSpeaker C:...as if it were at the beginning of the word.\nSpeaker C: So that if any part of the word was overlap, it was considered an interrupted word.\nSpeaker C: And then we looked at the location...\nSpeaker C:...the...\nSpeaker C:...you know, the features, the tags...\nSpeaker C:...because we had tagged these word strings.\nSpeaker C: Taked.\nSpeaker C: That occurred right before these interrupt locations.\nSpeaker C: And the tags we looked at are...\nSpeaker C:...the spurt tag, which basically says...\nSpeaker C:...are actually...\nSpeaker C:...sorry, end of spurt.\nSpeaker C: So whether there was a pause, essentially, here...\nSpeaker C:...because spurt's defined as being, you know, 500-minute seconds or longer pauses.\nSpeaker C: And then we had things like discourse markers, back channels, disloancies, field pauses.\nSpeaker C: So disloan...\nSpeaker C:...the ds are for...\nSpeaker C:...the interruption points of disloancies.\nSpeaker C: So where you hesitate or where you start them, repair.\nSpeaker C: What else do we have?\nSpeaker C: Repeated words, it's another kind of disloancies and so forth.\nSpeaker C: So we had both the beginnings and ends of these.\nSpeaker C: So the end of a field pause...\nSpeaker C:...and the end of a discourse marker.\nSpeaker C: And we just eyeballed.\nSpeaker C: I mean, we didn't really hand tag all of these things.\nSpeaker C: We just looked at the distribution of words.\nSpeaker C: And so every...\nSpeaker C:...so yeah, and okay.\nSpeaker C: And aha, where the word deemed to be back channels and well and so and...\nSpeaker C:...rights, where...\nSpeaker C:...not right is a back channel.\nSpeaker C: So we sort of just based on the lexical identity of the words we...\nSpeaker C:...tapped them as one of these things.\nSpeaker C: And of course, the interruption points we got from the original transcripts.\nSpeaker C: So...\nSpeaker C:...and then we looked at the distribution of these different kinds of tags...\nSpeaker C:...overall and in particularly at the interruption points.\nSpeaker C: And we found that there is a mark difference.\nSpeaker C: So that for instance, after...\nSpeaker C:...so at the end after a discourse marker or after back channel or after a field pause...\nSpeaker C:...you're much more likely to be interrupted than before.\nSpeaker C: Okay.\nSpeaker C: And also of course, after spurt ends, which means basically in point-side pauses.\nSpeaker C: So pauses are always an opportunity for...\nSpeaker C: So we have this little histogram that shows these distributions.\nSpeaker C: I wonder if it's not no big surprises, but this sort of...\nSpeaker C:...it's a nice actually measure.\nSpeaker F: I wonder about the cause and effect there.\nSpeaker F: In other words, if you weren't going to pause, you...\nSpeaker F:...will be because you're being interrupted.\nSpeaker F: Right, there's no statement about cause and effect.\nSpeaker F: Right.\nSpeaker F: It's just a statistic.\nSpeaker D: No, no, no.\nSpeaker D: Yeah, he's right to me.\nSpeaker D: Maybe we're intending to pause it all.\nSpeaker D: Right.\nSpeaker D: We're intending to stop for 57 milliseconds than chucking.\nSpeaker D: Yeah.\nSpeaker C: Pause for a second anyway.\nSpeaker C: And that was basically it.\nSpeaker C: And so we wrote this and then we found we were at six pages when we started cutting furiously.\nSpeaker C: And we thought half of the material again and played with the latex stuff.\nSpeaker C: And it was fun small, yeah?\nSpeaker C: No, no, well, you couldn't really make everything smaller, but we...\nSpeaker C: The abstract.\nSpeaker C: Oh, I...\nSpeaker C: You know, the gap between the two columns is like 10 millimeters, so I track it to eight millimeters.\nSpeaker C: That helps seven stuff like that.\nSpeaker F: Wasn't there some result under us?\nSpeaker F: I thought maybe Liz presented this at some conference a while ago about back channels.\nSpeaker F: And that they tend to happen when the pitch drops.\nSpeaker F: You know, you had a falling pitch.\nSpeaker F: Yeah.\nSpeaker F: And so that's when people tend to back channel.\nSpeaker C: We didn't talk about presenting.\nSpeaker C: Right.\nSpeaker C: Right.\nSpeaker C: So that's... I take it that's something that Don will look at.\nSpeaker C: Yeah, we're going to be looking at that.\nSpeaker C: So this is purely based on, you know, the words.\nSpeaker H: I have a reference for that though.\nSpeaker F: How you do?\nSpeaker F: So am I recalling correctly?\nSpeaker H: Well, I didn't know about Liz's finding on that, but I know of another paper that talks about something.\nSpeaker B: I think to see that reference too.\nSpeaker F: It made me think about a cool little device that could be built to handle those people that call you.\nSpeaker F: And just like to talk and talk and talk.\nSpeaker F: And you just have this little detector that listens for these drops and pitch and gives them the back channel.\nSpeaker F: So then you hook that to the phone and go off and do whatever you want to do while that thing keeps them busy.\nSpeaker C: There's actually... there's this former student of here from Berkeley, Nigel Ward.\nSpeaker C: He did a system in... he lives in Japan now and he did this back channeling, automatic back channeling system.\nSpeaker C: So very... so exactly what you described for Japanese.\nSpeaker C: And apparently for Japanese, it's really important that you back channel.\nSpeaker C: It's really impolite if you don't.\nSpeaker D: Actually, for a lot of these people, I think you just serve back channel continuously.\nSpeaker F: It wouldn't matter if it's when I ran the Minervals course.\nSpeaker E: There was of course a Monty Python sketch with that.\nSpeaker E: Or the barber who was afraid of scissors was playing a tape of clipping sounds.\nSpeaker E: And then they say, uh-huh, yeah, how about then Swordsteen?\nSpeaker C: Anyway, so the paper is online.\nSpeaker C: And I think I... I see a message to meeting recorded with the URL.\nSpeaker C: Or one more thing.\nSpeaker C: So I'm actually about to send Brian Kingsbury an email saying where he can find the material he wanted for this speech recognition experiment.\nSpeaker C: So, but I haven't sent it out yet because actually my desktop locked up.\nSpeaker C: I can't type anything.\nSpeaker C: So if there's any suggestions for that, I was just going to...\nSpeaker F: Is that the same directory that you had suggested?\nSpeaker F: I made a directory. I called it...\nSpeaker H: He still has his Unix account here, you know?\nSpeaker H: He does?\nSpeaker H: Yeah.\nSpeaker C: Yeah, but he has to...\nSpeaker C: He said he would prefer FTP.\nSpeaker C: And also, the other person that wants to do... There's one person that SRI who wants to look at the...\nSpeaker C: You know, the data we have so far.\nSpeaker C: And so I figured FTP is the best approach.\nSpeaker C: So what I did is I...\nSpeaker C: I made a new directory after a check set that was going to be a good thing.\nSpeaker C: So it's FTP, pub, real.\nSpeaker C: Real, exactly.\nSpeaker C: MTGC.\nSpeaker C: What is it again?\nSpeaker E: Has Danielis.\nSpeaker C: Yeah, right. The same as the main list.\nSpeaker C: And then under there... Actually, on this directory is not readable.\nSpeaker C: It's only accessible.\nSpeaker C: So in other words, to access anything under there, you have to be told what the name is.\nSpeaker C: So that's sort of a quick and dirty way of doing access control.\nSpeaker C: And the directory for this, I call it ISR0.1 because it's sort of meant for recognition.\nSpeaker C: And then there I have a file that lists all the other files so that someone can get that file and then know the file names.\nSpeaker C: And therefore download them.\nSpeaker C: If you don't know the file names, you can't...\nSpeaker B: Don't say.\nSpeaker C: Anyway, so all I was going to do there was stick the transcripts after the way that we munched them for scoring.\nSpeaker C: Because that's what he cares about.\nSpeaker C: And then the waveforms that Don segmented.\nSpeaker C: I mean, just basically tar them all up for each meeting and tar them all into one tar file and giz of them and stick them there.\nSpeaker E: So they put digits in my own home directory, home FTP directory, but I'll probably move them there as well.\nSpeaker F: So we could point Mari to this also for her March 01 request.\nSpeaker C: March 01.\nSpeaker C: Oh, remember, she was...\nSpeaker F: We wanted that also.\nSpeaker F: Well, she was saying that it would be nice if we had...\nSpeaker F: Or was she talking...\nSpeaker F: Yeah, she was saying it would be nice if they had the same set so that when they did experiments, they could compare.\nSpeaker C: But they don't have a recognition.\nSpeaker C: But yeah, I can see Mari on this one of that she knows.\nSpeaker F: So for the thing that we need to give Brian the Beab's file, so I was going to probably put it in the same place.\nSpeaker E: Yeah, I'll make another directory.\nSpeaker F: Yeah, exactly.\nSpeaker B: And Andreas, I think those files that I gave you are all down sampled.\nSpeaker B: They are?\nSpeaker B: I think so.\nSpeaker B: Yeah.\nSpeaker B: So either we should regenerate the original versions or we should just make a note of it.\nSpeaker C: Okay, because in one directory there's two versions.\nSpeaker B: Yeah, that's the first meeting I kept both versions.\nSpeaker B: Just to check which one.\nSpeaker B: There's significant difference.\nSpeaker B: Okay, so for the other meetings it's the down sampled version.\nSpeaker B: Yeah, down sampled, yeah.\nSpeaker C: Oh, okay.\nSpeaker C: Oh, that's important to know.\nSpeaker C: Okay, so we should probably give them the non-down sampled versions.\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker C: Okay.\nSpeaker C: All right, then I'll hold off on that and wait for you to...\nSpeaker C: Probably by tomorrow.\nSpeaker B: I'll send you an email.\nSpeaker C: All right.\nSpeaker C: Okay.\nSpeaker C: Yeah, definitely they should have the full bandwidth.\nSpeaker C: Yeah.\nSpeaker B: Yeah, because I mean, I think Liz decided to go ahead with the down sampled versions because we can...\nSpeaker B: Well, it's significant difference.\nSpeaker B: It takes a less this space from all over the world.\nSpeaker B: It does take a less this space.\nSpeaker B: And apparently it didn't even better than the original versions, which, you know, is just probably random.\nSpeaker B: It's a small difference, but they probably want the originals.\nSpeaker C: Okay.\nSpeaker C: Okay, good.\nSpeaker C: Good thing.\nSpeaker E: I think we're losing Dawn and Andreas at 330, right?\nSpeaker B: Yeah.\nSpeaker B: So, I'm going to put it in the description.\nSpeaker D: It's fine.\nSpeaker D: It's okay.\nSpeaker D: It's okay.\nSpeaker D: So, we're going to put the transcription in the next step.\nSpeaker D: Okay.\nSpeaker H: So, you know, Adam created a script to generate the beat file to then create something just into IBM.\nSpeaker H: And you should probably talk about that, but you were going to use the originally transcribed file because I tightened the time bins, and that's also the one that they had already been trying to debug the first stage of this.\nSpeaker H: And my understanding was that I haven't listened to it yet, but it sounded very good.\nSpeaker H: And I understand that you guys were going to have a meeting today before this meeting.\nSpeaker E: It was just to talk about how to generate it.\nSpeaker E: Just so that while I'm gone, you can regenerate it if you decide to do it a different way.\nSpeaker E: So Chuck and Tilo should now more or less know how to generate the file.\nSpeaker E: And the other thing Chuck pointed out is that since this one is hand-marked, there are discourse boundaries.\nSpeaker E: So when one person is speaking, there's breaks, whereas Tilo's won't have that.\nSpeaker E: So what we're supposed to do is just write a script that if two chunks are very close to each other on the same channel, we'll just merge them.\nSpeaker H: Oh, sure.\nSpeaker H: Yeah, sure.\nSpeaker E: Makes sense.\nSpeaker E: And that will get around the problem of the one word, deep one word, deep one word,\nSpeaker H: deep one word. Clever, yes.\nSpeaker H: Clever.\nSpeaker H: Yeah, excellent.\nSpeaker F: After this morning, Tilo came in and said that there could be other differences between the already transcribed meeting with the beeps in it and one that has just been run through his process.\nSpeaker F: So tomorrow, when we go to make the chunked file for IBM, we're going to actually compare the two.\nSpeaker F: So he's going to run his process on that same meeting.\nSpeaker F: And then we're going to do the beepify on both and listen to them and see if we notice any real differences.\nSpeaker H: Okay, one thing that prevented us from applying, you from applying exactly the training.\nSpeaker H: So that is a training meeting.\nSpeaker F: Yeah, and we know that.\nSpeaker F: We just want to see if there are any major differences between doing it on the hand.\nSpeaker H: Oh, interesting.\nSpeaker H: Ah, okay.\nSpeaker H: Interesting idea.\nSpeaker C: So this training meeting, is that some data where we have very accurate time marks for?\nSpeaker H: I went back and hand marked the buttons.\nSpeaker H: I mentioned that.\nSpeaker H: Okay, yeah.\nSpeaker F: But there's, yeah, but there is this one issue with them in that there are time boundaries in there that occur in the middle of speech.\nSpeaker F: So like when we went to, when I was listening to the original file that I had, it's like you hear word, then you hear beep, and then you hear the continuation of what is the\nSpeaker E: same sentence. That's because of channel overlap.\nSpeaker F: Well, and so there are these chunks that look like that have, I mean, that's not going\nSpeaker E: to be true of the foreground speaker that will only be if it's the background speaker.\nSpeaker F: Right. So you'll have a chunk of channel A, which starts at zero and ends at 10.\nSpeaker F: And then the same channel starting at 11, ending at 15, and then again starting at 16, ending at 20.\nSpeaker F: Right.\nSpeaker F: So there's three chunks where actually we can just make one chunk out of that, which is A, zero, 20.\nSpeaker E: That's what I just said.\nSpeaker F: Yeah.\nSpeaker F: So I just want to make sure that it was like, so if you were to use these, you have to be careful not to pull up these individual.\nSpeaker C: Right.\nSpeaker C: So I mean, what I would, I was interested in this having, having time marks for the beginnings and ends of speech by each speaker.\nSpeaker C: Oh, that's definitely probably because we could use that to find you in our alignment process to make it more accurate.\nSpeaker C: Battery.\nSpeaker C: So I don't care that, you know, there's actually a budding segments that we have to join together.\nSpeaker C: That's fine.\nSpeaker C: Okay.\nSpeaker C: So I think that's the reason why I think that the beginnings and ends are actually\nSpeaker F: close to the speech inside of that.\nSpeaker C: Yeah. I think Jane tightened these up by hand.\nSpeaker C: Okay.\nSpeaker H: So what is the sort of how tight are they?\nSpeaker H: Next good.\nSpeaker H: They were reasonably tight, but not excruciatingly tight.\nSpeaker H: That would have taken more time.\nSpeaker C: No, no, I don't actually have like, yeah, that's fine because we don't want to, that's perfectly fine.\nSpeaker C: In fact, it's good.\nSpeaker C: I always want to have a little bit of pause or non-speech around the speech, say, for recognition purposes.\nSpeaker C: But just, you know, again, I just want to have an idea of how much extra you allow so that I can interpret the numbers if I compare that with a forced alignment segmentation.\nSpeaker H: I can't answer that, but my main goal was in these areas where you have a three-way overlap and one of the overlaps involved, yeah.\nSpeaker H: And it's swimming in this huge bin.\nSpeaker H: I wanted to get it so that it was closely localized.\nSpeaker C: But are we talking about, I don't know, 10th of a second, you know, how much extra would you allow?\nSpeaker H: I wanted to be able to be heard normally so that if you play back that bin and have it in the mode where it stops at the boundary, it sounds like a normal word.\nSpeaker H: It doesn't sound like the person, it sounds normal.\nSpeaker H: It's as if the person could have stopped there.\nSpeaker H: Okay.\nSpeaker H: And it wouldn't have been an awkward place to stop.\nSpeaker H: Now sometimes, you know, these are involved in places where there was no time.\nSpeaker H: And so there wouldn't be a gap afterwards because, I mean, in some cases, there are some people who have very long segments of discourse where, you know, they'll breathe and then I put a break.\nSpeaker H: But other than that, it's really pretty continuous.\nSpeaker H: This includes things like going from one sentence into the next one sentence into the next without really stopping.\nSpeaker H: You know, in writing, you have this two spaces in a big gap, you know.\nSpeaker H: But some people are planning and, you know, we always are planning what we're going to say next.\nSpeaker H: But in which case, the gap between these two complete syntactic units, which of course spoken things are not always complete syntactically, but it would be a shorter break than maybe you might like.\nSpeaker H: But the goal there was to not have the text be so crudely parsed in a time bin.\nSpeaker H: I mean, because from a discourse purpose, it's more useful to be able to see.\nSpeaker H: And also, you know, from a speech recognition purpose, my impression is that if you have too long a unit, it doesn't help you very much either because of the memory.\nSpeaker H: That's why.\nSpeaker H: So, that means that the amount of time after something is variable depending partly on context, but my general goal, when there was sufficient space room pause after it to have it be kind of a natural feeling gap, which I don't know what it would be quantified as.\nSpeaker H: You know, well, the chase says that in producing narratives, the spurts that people use tend to be, what would be a pause might be something like two seconds.\nSpeaker H: And that would be one speaker.\nSpeaker H: The discourse, the people who look at turn-taking often do use, I was interested that you chose the, you know, the use, because I think that would be more consistent with sociolanguistics.\nSpeaker C: Or we chose, you know, half a second because if you go much larger, you have, you know, your statement about how much overlap there is becomes less precise because you include more of actual pause time into what you consider overlap speech.\nSpeaker C: So it's sort of a compromise.\nSpeaker C: And it's also based, I mean, Liz suggested that value based on the distribution of pause times that you see in switchboard and other corpora.\nSpeaker G: So yeah, I also used, I think, something around 0.5 seconds for the speech-man speech detector for minimum silence length.\nSpeaker G: I see.\nSpeaker G: So.\nSpeaker H: Okay.\nSpeaker H: In any case, this meeting that I hand, I handed just to two of them I mentioned before.\nSpeaker H: Okay.\nSpeaker C: And I sent email.\nSpeaker C: At some point we will try to find you in our first alignment, maybe using those as references because, you know, what you would do is you would play with different parameters and to get an objective, you need an objective measure of how closely you can align the models to the actual speech.\nSpeaker C: And that's where your data would be very important to have.\nSpeaker C: So.\nSpeaker G: And hopefully the new meetings which will start from the channelized version will have better time boundaries and alignments.\nSpeaker C: Right.\nSpeaker H: But I'd like this idea of, for our purposes, for the IBM preparation, having these joined together and it makes a lot of sense.\nSpeaker H: And in terms of transcription, it would be easy to do it that way, the way that they have, with the longer units, not having to fuss with adding these things.\nSpeaker G: Which could have one draw back if there is a back channel in between those three things, the back channel will occur at the end of those three and in the previous version which is used now, the back channel would be in between there somewhere.\nSpeaker G: So that would be more natural.\nSpeaker H: That's right.\nSpeaker H: But you know, this brings me to the other stage of this which I discussed with you earlier today, which is the second stage is what to do in terms of the transcribers adjustment of these data.\nSpeaker H: I discussed this with you.\nSpeaker H: So the idea initially was we would get for the new meetings.\nSpeaker H: So the E.D.U meetings that Tilo has now pre-segmented all of them for us on a channel by channel bases.\nSpeaker H: And so I've assigned them to our transcribers and so far I've discussed it with one.\nSpeaker H: And I had about an hour discussion with her about this yesterday.\nSpeaker H: We went through E.D.U.\nSpeaker H: One at some extent.\nSpeaker H: And it occurred to me that basically what we have in this kind of a format is you could consider it as a staggered mixed file.\nSpeaker H: We had some discussion over the weekend about at this other meeting that we were all at about whether the IBM transcribers should hear a single channel audio or a mixed channel audio.\nSpeaker H: And in a way, by having this chunk and then the back channel after it, it's like a staggered mixed channel.\nSpeaker H: The maximal gain from the IBM people may be in long stretches of connected speech.\nSpeaker H: So it's basically a whole bunch of words which they can really do because of the continuity within that person's turn.\nSpeaker H: So what I'm thinking, and it may be that not all meetings will be good for this, but what I'm thinking is that in the E.D.U meetings they tend to be driven by a couple of dominant speakers.\nSpeaker H: And if the chunked files focused on the dominant speakers, then when it got patched together when it comes back from IBM, we can add the back channels.\nSpeaker H: It seems to me that, you know, back channels per se wouldn't be so hard, but then there's this question of the time marking and whether the beeps would be, and I'm not exactly sure how that would work with the back channels.\nSpeaker H: And certainly things that are intrusions of multiple words taken out of context and displaced in time from where they occurred, that would be hard.\nSpeaker H: So my thought is, I'm having this transcriber go through the E.D.U. one meeting and indicate a start time for each dominant speaker and time for each dominant speaker and the idea that these units would be generated for the dominant speakers and maybe not for the other channels.\nSpeaker E: The only disadvantage of that is then it's hard to use an automatic method to do that.\nSpeaker E: The advantage is that it's probably faster to do that than it is to use the automated method and correct it.\nSpeaker H: Well, I think the original plan was that the transcriber would adjust to the boundaries and all that for all the channels, but you know, that is so time consuming.\nSpeaker H: And since we have a bottleneck here, we want to get IBM things that are usable as soon as possible, then the scene would be a way to get them a flood of data, which would be useful when it comes back to us.\nSpeaker H: And also at the same time, when she goes through this, she'll be, if there's anything that was encoded as a pause but really has something transcribable in it, then she's going to make a mark.\nSpeaker H: So that bin would be marked as double dots and she'll just add an S. And in the other case, if it's marked as speech and really there's nothing transcribable in it, then she's going to put a dash and I'll go through it and you know, with a substitution command, get it so that it's clear that those are the other category.\nSpeaker H: I'll just, you know, re-code them.\nSpeaker H: But the transcribable events that I'm considering in this continue to be left as well as speech and cough and things like that, so I'm not stripping at anything just, you know, being very lenient in what's considered speech.\nSpeaker F: So Jane, in terms of the new procedure suggesting, what is the, so I'm a little confused because how do we know where to put beeps?\nSpeaker H: Okay, so what it involves is really the original procedure, but only applied to a certain strategically chosen aspect of the data.\nSpeaker E: We pick the easy parts of the data basically and transcribable marks it by hand.\nSpeaker E: But after we've done T-Los thing.\nSpeaker E: No.\nSpeaker E: Yes.\nSpeaker E: Oh, after.\nSpeaker E: Okay, I didn't understand that.\nSpeaker G: Okay.\nSpeaker G: So I'm, no, I'm confused.\nSpeaker H: Okay, we start with your pre-segmented version.\nSpeaker E: Okay, leave the mics on and just put them on the table.\nSpeaker E: Okay.\nSpeaker B: Thanks.\nSpeaker H: We start with the pre-segmented version.\nSpeaker G: You start with the pre-segmentation.\nSpeaker H: And then the transcriber, instead of going painstakingly through all the channels and moving the boundaries around and deciding if it's speech or not, but not transcribing anything.\nSpeaker H: Okay, instead of doing that, which was our original plan, they just do that on the main channel.\nSpeaker H: Yeah, so what they do is they identify who's the dominant speaker and when the speaker starts.\nSpeaker G: Okay.\nSpeaker H: So, I mean, you're still going to, so based on your pre-segmentation, that's the basic\nSpeaker G: thing. And you just use the segments of the dominant speaker then for sending to IBM or.\nSpeaker F: Exactly.\nSpeaker F: So now Jane, my question is, when they're all done adjusting the time boundaries for the dominant speaker, have they then also erased the time boundaries for the other ones?\nSpeaker F: No, no, no.\nSpeaker F: So how will we know who?\nSpeaker H: That's why she's notating the start and end points of the dominant speakers.\nSpeaker H: So in EDU1, as far as I listen to it, you start off with a section by Jerry.\nSpeaker H: So Jerry starts at minute, so and so, and goes until minute, so and so.\nSpeaker H: And then Mark Pascon comes in and he starts at minute, such and such and goes until minute, so and so.\nSpeaker H: Okay.\nSpeaker H: And then meanwhile, she's listening to both of these guys' channels, determining if there are any cases of misclassification of speech as nothing and nothing in speech.\nSpeaker H: Okay.\nSpeaker H: Just the adjustments on those guys.\nSpeaker H: But you know, I wanted to say, his segmentation is so good that the part that I listened to with her yesterday didn't need any adjustments to the bins so far we haven't.\nSpeaker H: So this is not going to be a major part of the process.\nSpeaker F: So if you don't have to adjust the bins, why not just do it for all the channels?\nSpeaker F: Why not just throw all the channels to IBM?\nSpeaker H: Well, there's the question of whether, well, okay, it's a question of how much time we want our transcriber to invest here when she's going to have to invest that when it comes back from IBM anyway.\nSpeaker H: So if it's only inserting mums here and there, then wouldn't that be something that would be just as efficient to do at this end instead of having it go through IBM then be patched together then be double checked here.\nSpeaker G: But then we could just use the output of the detector and do the beeping on it and send\nSpeaker F: it to IBM without having to check anything.\nSpeaker H: Well, I guess for some meetings, I'm sure it wouldn't see how good they are. I'm open to that.\nSpeaker H: Yes, it's working well.\nSpeaker G: It's on some meetings, it's good.\nSpeaker H: I mean, we have to fix it when it comes back.\nSpeaker H: You were saying that they differ in how well they work depending on channels to systems.\nSpeaker G: We should perhaps just select meetings on which the speech non speech detection works well and just use those meetings to send to IBM and do the analysis.\nSpeaker D: So I forget, it's from the little town of the water.\nSpeaker G: It really depends.\nSpeaker G: My impression is that it's better for meetings with fewer speakers and it's better for meetings than nobody is breathing.\nSpeaker D: Yeah, that's it.\nSpeaker F: So in fact, this might suggest an alternative sort of a hybrid between these two things.\nSpeaker F: So the one suggestion is we run T-Lose thing and then we have somebody go and adjust all the time boundaries and we send it to IBM.\nSpeaker F: The other one is we just run his thing and send it to IBM.\nSpeaker F: There's another possibility if we find that there are some problems and that is we go ahead and we just run his and we generate the beeps file, then we have somebody listen to the beeps file and they write each section and say yes, no, whether that section is intelligible or not.\nSpeaker F: It's just a little interface which for all the yeses, then that will be the final.\nSpeaker H: It's interesting because that's directly related to the N-Task.\nSpeaker F: Stress test.\nSpeaker F: I mean, it wouldn't be that much fun for a transcriber to sit there, hear it, yes or no, but it would be quick.\nSpeaker D: It would be kind of quick but they're still listening to everything.\nSpeaker F: But there's no adjusting and that's what's slow.\nSpeaker F: There's no adjusting of time boundaries.\nSpeaker D: Well, this thing's just take time too.\nSpeaker D: Yeah.\nSpeaker D: I don't know.\nSpeaker A: I really have to answer your thoughts.\nSpeaker D: I mean, what's the worst that happens to transcribers?\nSpeaker D: I mean, as long as on the other end they can say there's something.\nSpeaker D: It's not a convention so they say, huh?\nSpeaker D: Right.\nSpeaker F: If I go I was later.\nSpeaker F: That's true.\nSpeaker F: We can just catch it, catch everything at this side.\nSpeaker F: Maybe that's the best way to go.\nSpeaker F: Just interesting.\nSpeaker F: I mean, it just depends on how.\nSpeaker F: Yeah.\nSpeaker H: So I was going to say E-DU1 is good enough.\nSpeaker H: Maybe we could include it in this set of this stuff we send.\nSpeaker G: Yeah, I think there are some meetings we're with.\nSpeaker E: I think it's possible like this.\nSpeaker E: Until we generate a bunch of beeps files automatically.\nSpeaker E: Yeah.\nSpeaker F: We won't be able to include it with this first thing because there's a part of the process of the beep file which requires knowing the normalization coefficients.\nSpeaker F: Oh, see.\nSpeaker E: It's hard to do.\nSpeaker E: It just takes five minutes rather than taking a second.\nSpeaker E: Right.\nSpeaker F: Except I don't think that the instructions for doing that was in that directory, right?\nSpeaker F: I didn't see where you had generous easy messages.\nSpeaker G: Doing the game that's no problem.\nSpeaker F: Adjusting the game?\nSpeaker F: Getting the coefficients for each channel.\nSpeaker E: Yeah, that's no problem.\nSpeaker E: Okay.\nSpeaker E: So we just run that one.\nSpeaker E: I have one program that I do it.\nSpeaker E: You can find it.\nSpeaker F: I do.\nSpeaker F: And that J sounds that.\nSpeaker D: Yep.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: I have a suggestion on that which is since really what this is is is trying to in the large send the right thing to them and there is going to be this post bus synced up.\nSpeaker D: Why don't we check through a bunch of things by sampling it?\nSpeaker D: Right.\nSpeaker D: And other than saying we're going to listen to everything.\nSpeaker E: I didn't mean listen to everything.\nSpeaker D: So you do much to see if there is a little bit here in there.\nSpeaker D: Yeah.\nSpeaker D: It sounds like it's almost always right.\nSpeaker D: There's not any big problem.\nSpeaker D: You send it to him.\nSpeaker D: Okay.\nSpeaker D: And then they'll send us back with what they send back to us and we'll fix things up.\nSpeaker D: Which is.\nSpeaker D: We should.\nSpeaker E: We should just double check with Brian on a few simple conventions on how they should mark things.\nSpeaker F: When they when there's either no speech in there or something they don't understand things like that.\nSpeaker E: Because what I had originally said to Brian was well they'll have to mark when they can't distinguish between the foreground and background.\nSpeaker E: Because I thought that was going to be the most prevalent.\nSpeaker E: But if we send them without editing then we're also going to have to have a notations forwards that are cut off.\nSpeaker E: Yeah.\nSpeaker E: And other sorts of acoustic problems.\nSpeaker F: And they may just guess at what those cut off words are.\nSpeaker F: But I mean we're going to adjust.\nSpeaker F: But everything.\nSpeaker E: But we would like them to do is be conservative so that they should only write down the transcript if they're sure.\nSpeaker E: And otherwise they should mark it to a check.\nSpeaker G: Yep.\nSpeaker H: Well we have the unintelligibility convention.\nSpeaker H: And actually they have one also.\nSpeaker A: Brian.\nSpeaker D: Maybe have an order of probably in the paper that I have not got my name.\nSpeaker D: An order magnitude notion of how on a good meeting how often do you get segments that come into work and so forth.\nSpeaker D: And then they're badly often.\nSpeaker D: And what is the good meeting?\nSpeaker D: The E.D.U meeting was a good meeting.\nSpeaker D: Yeah.\nSpeaker D: It was almost always doing the right thing.\nSpeaker D: So I wanted to get some sense of what almost always meant.\nSpeaker D: And then in a bad meeting or some meetings where he said he's had some problems.\nSpeaker D: What does that mean?\nSpeaker D: So does it mean one percent, ten percent?\nSpeaker D: Does it mean five percent, fifty percent?\nSpeaker G: So the number I gave in the paper is just some frame error rate so that's not really what will be effective for the transcribuses.\nSpeaker G: They have to ensure that that's a real support or something.\nSpeaker G: But the number is oops.\nSpeaker G: So the speech, the amount of speech that is missed by the detector for a good meeting is around or under one percent I would say.\nSpeaker G: But that can be more amount of speech, a more amount of the detector says there is speech but it's not.\nSpeaker G: So that can be a lot when it's really a breathy channel.\nSpeaker D: But I think that's less of a problem.\nSpeaker D: And that's for good meeting.\nSpeaker D: What about in a meeting that you said you had some more trouble with?\nSpeaker G: I can't really, I don't have really representative numbers I think.\nSpeaker G: I did this on four meetings and only five minutes of every meeting, of these meetings.\nSpeaker G: So it's not that representative but it's perhaps.\nSpeaker G: Yeah, it's perhaps then it's perhaps five percent or something which the speech frames which I missed but I can't really tell.\nSpeaker D: So sometimes when I want to go back and look at it more in terms of how many times is there a spurt that's interrupted?\nSpeaker H: The other problem is when it went on the breathy ones where you get breathing indicated a speech.\nSpeaker H: And I guess we could just indicate to the transgarbers not to encode that if they still do the B file.\nSpeaker D: That is probably less of a problem because if there's, if a word is split then they might have to listen to a few times to really understand it.\nSpeaker D: I think that's really, it doesn't happen very often that the various cut in the middle or something that's really not normal.\nSpeaker D: So what you're saying is that nearly always what happens when there's a problem is that there's some non-speech that there's marker speech.\nSpeaker D: Well then we really should just send this stuff.\nSpeaker D: That doesn't do any harm.\nSpeaker D: They say here dog bark and they say what was the word?\nSpeaker G: Yeah, I also thought of there are really some channels where it is almost only breathing in it and to reruns.\nSpeaker G: I've got a method that looks into the cross correlation with the PCM mic and then to reject everything which seems to be breath.\nSpeaker G: So I could run this on those pressy channels and perhaps throughout.\nSpeaker D: Yeah, I think that would be good.\nSpeaker D: I think none of this stuff is really something that these are.\nSpeaker H: I'd be delighted with that.\nSpeaker H: I was very impressed with the result.\nSpeaker D: Yeah, the thing I was concerned about was that seemed kind of specialized to the media meeting and that meeting like this or something.\nSpeaker D: Oh yeah, a bunch of different governments, speakers and how do you handle it?\nSpeaker D: Oh yeah.\nSpeaker H: I'm much prefer this. I was just trying to find a way because I don't think the staggered mix channel is awfully good as a way of handling overlaps.\nSpeaker F: Well good. That really simplifies saying that.\nSpeaker F: We can just get the meeting, process it, put the beeps file, send it off to IBM.\nSpeaker G: We have very little work on our processes here in Trit.\nSpeaker G: Listen to it and then sample it.\nSpeaker F: Yeah, that would be very good.\nSpeaker F: And then we can, you know, that'll be a good way to get the pipeline going.\nSpeaker G: And there's one point which I, which I recalculate when I listen to one of the new meetings and that somebody is playing sound from his laptop.\nSpeaker G: And the speech non-speech detector just assigns randomly the speech to one of the channels.\nSpeaker G: So I didn't think of this before, but what shall we do about things like this?\nSpeaker H: You suggested maybe just not sending that part of the meeting.\nSpeaker G: But sometimes the laptop is in the background and somebody is talking and that's really a little bit confusing.\nSpeaker G: It's a little bit confusing.\nSpeaker E: Even a hand transcription with a hand transcriber would have trouble.\nSpeaker G: Yeah, that's a second question. What will the transcribers do with the laptop sound?\nSpeaker D: What was the laptop sound?\nSpeaker G: Was it speech?\nSpeaker H: So my standard approach has been if it's not someone close-miked, then they don't end up on one of the close-mike channels.\nSpeaker H: They end up on a different channel.\nSpeaker H: And we have any number of channels available.\nSpeaker G: When this is sent to the IBM transcribers, I don't know if they can tell that's really...\nSpeaker E: Yeah, because there will be no channel on the sheet.\nSpeaker H: Well, they have a convention in their own procedures, which is for a background sound.\nSpeaker E: Right, but in general, I don't think we want them transcribing the background because that would be too much work.\nSpeaker E: Right, because in the overlap sections.\nSpeaker F: Well, I don't think Jane's saying they're going to transcribe it, but they'll just mark it as being...\nSpeaker E: There's some background stuff that's going to be right over the place.\nSpeaker E: How will they tell the difference between that sort of background and the normal background of two people talking at once?\nSpeaker H: Oh, I think it could be easy to say background laptop.\nSpeaker F: Why would they treat them differently?\nSpeaker E: Well, because otherwise it's going to be too much work for them to market.\nSpeaker H: They'll be marketing it all over the place.\nSpeaker H: Oh, background laptop or background LT wouldn't take any time.\nSpeaker E: Sure, but how are they going to tell the difference between that and two people just talking at once?\nSpeaker H: Oh, you can tell acoustically. Can't you tell?\nSpeaker G: It's really good sound, so...\nSpeaker H: Oh, is it?\nSpeaker D: Well, isn't there a category something like sounds for someone for whom there is no close-mike?\nSpeaker G: Yeah, that would be very important.\nSpeaker E: How do we do that for the IBM folks?\nSpeaker E: How can they tell that?\nSpeaker F: Well, we may just have to do it when it gets back here.\nSpeaker F: Yes, that's my opinion as well.\nSpeaker E: Okay, that sounds good.\nSpeaker E: And they'll just mark it however they mark it.\nSpeaker E: So it'll correct it when it gets back.\nSpeaker G: That is a category for...\nSpeaker G: Yes, that's a problem.\nSpeaker H: Well, as it comes back, when we have a...\nSpeaker H: When we can use the channelized interface for coding it, then it'll be easy for us to handle.\nSpeaker H: But if out of context they can't tell if it's a channeled speaker...\nSpeaker H: When a close-mike speaker or not, then that would be confusing to them.\nSpeaker H: Right.\nNone: Okay.\nSpeaker H: I don't know. I don't...\nSpeaker H: Either way, it'd be fine with me, I don't really care.\nSpeaker D: So, can we...\nSpeaker D: You did it, get out of here?\nSpeaker H: Yeah.\nSpeaker H: I have one question.\nSpeaker H: Do you think we should send that whole meeting to them and not worry about preprocessing it?\nSpeaker H: Or...\nSpeaker H: What I mean is, we should leave the part with the audio in the...\nSpeaker H: Beap file that we send to IBM for that one, or should we start after the...\nSpeaker H: That part of the meeting is over.\nSpeaker H: And what we send...\nSpeaker H: What?\nSpeaker H: So the part where they're using sounds from their...\nSpeaker H: From their laptop.\nSpeaker G: With the laptops on for...\nNone: Just...\nSpeaker H: Have speech from the laptop.\nSpeaker H: Should we just...\nSpeaker H: Exercise that from what we send to IBM, or should we...\nSpeaker H: Give it to them and let them do what they can.\nSpeaker F: I think we should just...\nSpeaker F: It's going to be too much work if we have to worry about that, I think.\nSpeaker F: Yeah, I think if we just send it all to them, you know.\nSpeaker F: Good.\nSpeaker F: Let's see how it works.\nSpeaker H: Yeah, and worry about it when we get back in.\nSpeaker H: And give them freedom to indicate if it's just not workable.\nSpeaker D: Yeah.\nSpeaker D: Okay, excellent.\nSpeaker D: Don't you lose mind having that.\nSpeaker D: I think that's right.\nSpeaker E: Yeah, we'll just have to listen to it and see how well it is.\nSpeaker E: Okay.\nSpeaker G: Sample it right there.\nSpeaker G: Yeah.\nSpeaker G: I think that will be a little bit of a problem, us.\nSpeaker G: It really switches around between two different channels, I think.\nSpeaker E: And they're very...\nSpeaker E: It's very audible on the closed talking channels.\nSpeaker G: Yeah.\nSpeaker E: Oh well.\nSpeaker E: Yeah, that's the same problem as the lapel mic.\nSpeaker G: Yeah.\nSpeaker H: Oh, interesting.\nSpeaker G: Comparable, real.\nSpeaker H: Okay, good.\nSpeaker H: Digits.\nSpeaker H: Okay, so we read the transcript number first.\nSpeaker E: So we're going to do it all together separately.\nSpeaker G: What time is it?\nSpeaker G: Ah, why do we go together?\nSpeaker D: Okay.\nSpeaker D: That's what we do.\nSpeaker D: One, two, three, go.\nSpeaker D: Transcript.\nSpeaker D: Transcript.\nSpeaker G: All 58.\nSpeaker F: 546.\nSpeaker F: 437.\nSpeaker F: 827.\nSpeaker F: 948.\nSpeaker F: 448.\nSpeaker F: 498.\nSpeaker F: 1, 2, 8.\nSpeaker F: 922.\nSpeaker F: 490.\nSpeaker F: 422.\nSpeaker F: 970.\nSpeaker G: 970.\nSpeaker A: 970.\nSpeaker G: 970.\nSpeaker G: 970.\nSpeaker G: 970.\nSpeaker G: 970.\nSpeaker G: 970.\nSpeaker G: 970.\nSpeaker F: 970.\nSpeaker F: 970.\nSpeaker F: 970.\nSpeaker F: 980.\nSpeaker G: 970.\nSpeaker A: 926.\nSpeaker A: 5886.\nSpeaker A: 6123.\nSpeaker A: 324.\nSpeaker A: 170.\nSpeaker A: 517.\nSpeaker A: 417.\nSpeaker H: Okay, it's going to be interesting if there are any more errors in these.\nSpeaker H: Yeah, they're really pretty low.\nSpeaker F: You guys plug your errors when you do it?\nSpeaker F: I do.\nSpeaker F: Nope.\nSpeaker F: I usually do it.\nSpeaker F: I don't.\nSpeaker F: You don't?\nSpeaker F: Nope.\nSpeaker F: How can you do that?\nSpeaker G: I have so many errors in it, but I- You hate to have your eyes plugged?\nSpeaker A: Really?\n"}, {"meeting_id": "IS1004b", "summary": "Project Manager started introducing meeting purposes on the functional design of the remote control. Group mates have agreed to name the project as 'Mando'. Next, Marketing presented user requirements and market reports on current remote improvements. User Interface presented the current trend on remote controls. User Interface compared scroll wheel and LCD screen. Industrial Interface gave a presentation on the working design of different interfaces. Lastly, Project Manager summarized the whole project meeting discussion.", "dialogue": "Speaker C: Okay everybody is ready.\nSpeaker C: Good morning again.\nSpeaker C: So today we are going to have a second meeting.\nSpeaker C: Oh Michael.\nSpeaker C: Hi, you're late.\nSpeaker C: You have a good reason for that?\nSpeaker C: Yes.\nSpeaker C: Very good.\nSpeaker C: Okay, let's have a look to the agenda today.\nSpeaker C: So we are going to have a meeting about the functional design.\nSpeaker C: So first before starting, I'm just going to go quickly through the minutes of previous meeting.\nSpeaker C: So basically we were not decided if we should go for universal or specific remote control but I have new inputs for about that topics.\nSpeaker C: I'm going to share with you.\nSpeaker C: And basically we decided to go to individual actions for each of you.\nSpeaker C: So industrial designer was supposed to work on the working design.\nSpeaker C: You prepare something for us.\nSpeaker C: The UE guy also worked on that.\nSpeaker C: And for our marketing expert should deliver some specs.\nSpeaker C: All right.\nSpeaker C: So we are going to go through three of your individual presentations.\nSpeaker C: But first I would like quickly to give a name to the project.\nSpeaker C: So I just put the quick clear remote.\nSpeaker C: But if you have any other names that we could decide for just to keep something fun for our project, we could discuss quickly.\nSpeaker C: Any ideas?\nSpeaker A: What about the power stick?\nSpeaker C: Yeah.\nSpeaker C: What else?\nSpeaker C: What else?\nNone: Maybe a Spanish name.\nSpeaker A: Especially if we sell it into the US market because there's a lot of Spanish speakers there.\nSpeaker A: Maybe something that sounds cool in English but sounds funny in Spanish.\nSpeaker A: Mando.\nSpeaker A: Mando.\nSpeaker C: Mando.\nSpeaker C: Mando.\nSpeaker C: Mando.\nNone: MA.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: D-O.\nSpeaker C: Okay.\nSpeaker D: It doesn't sound cool for me.\nSpeaker D: But maybe for Spanish, for I4.\nSpeaker A: What does it mean in Spanish?\nSpeaker A: Control.\nSpeaker A: Control.\nSpeaker A: Because it also, like in English it sounds like the man's tool, you know, because men like to have control of the remote.\nSpeaker D: Mando sounds Latino.\nSpeaker D: Okay.\nSpeaker C: So let's go for Mando.\nSpeaker C: Yeah.\nSpeaker C: Objection.\nSpeaker C: Yeah.\nSpeaker A: Great.\nSpeaker A: Maybe we could have some like, you could have the fonts, you know, special so you have man in like in one font and then the O is like.\nSpeaker A: Okay.\nSpeaker A: Well, you don't want to cut women out of the potential buyers.\nSpeaker A: Yeah.\nSpeaker B: They're the most TV watchers so we should be careful.\nSpeaker C: Okay.\nSpeaker C: I think this is more a question of, I think this is more a question of, of look and feel.\nSpeaker C: Something that should be addressed later.\nSpeaker C: We should go to the other top.\nSpeaker D: Yeah.\nSpeaker D: Because the product will be international.\nSpeaker A: Well, that's the thing.\nSpeaker A: We need to know who we're selling it to before we can really decide on it.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: So let's stick to Mando for the name and we'll see for the look and feel later.\nSpeaker C: So let's go for the three presentations right now.\nSpeaker C: So we want to start.\nSpeaker D: So maybe I should just start on the market here.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker C: Okay.\nSpeaker C: So I have your slides somewhere.\nSpeaker D: Yeah.\nSpeaker D: Should be in participant four.\nSpeaker C: participant four.\nNone: Yeah.\nSpeaker C: Yes, one.\nSpeaker C: Yeah.\nSpeaker C: That's coming.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker D: Great.\nSpeaker D: Okay.\nSpeaker D: So I will give a brief outline about what I prepared for this meeting for the functional requirements and especially for the user requirements.\nSpeaker D: I prepared a marketing report and just to find the weaknesses and the improvements we could do to the current remote controls.\nSpeaker D: And also I did a study with for the incorporation of new technologies.\nSpeaker D: It seems that the remote controls have remained the same for the last five, ten years.\nSpeaker D: There is no significant difference between the first new controls.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: Next slide.\nNone: Okay.\nSpeaker D: Well, most of the people think that remote controls are ugly, directly.\nSpeaker D: So and they admitted that they should spend more money in a fancier remote control, which is good.\nSpeaker D: Interesting point.\nSpeaker D: Also, the people are worried about the air as I disease, which is if you repeat the same movement, which is not a with a not very appropriate device, you will have problems when you will get all.\nSpeaker D: So people are worried about the shape of the remote control.\nSpeaker D: They are also they get angry very often because they lost the remote control very often.\nSpeaker D: So I think it would be a good point to find a solution to any beep, any alarm or something incorporated with the remote control every time it gets lost.\nSpeaker D: And also I found that the younger people are the more interested they are in incorporating new technologies in the remote control.\nSpeaker C: Okay.\nSpeaker D: So in my opinion, the Mando is Mando shouldn't be very small because the smaller it is, the more like the the likelier is to get lost likelier or more likely more likely like like.\nSpeaker D: Okay.\nSpeaker D: People are so complain because they they they all have the same size of the buttons for buttons which are not very used like memorizing channels or all these kind of actions which are not very often.\nSpeaker D: But they should they shouldn't have the same importance in the remote control.\nSpeaker D: Also the design should fit the hand shape.\nSpeaker D: So it may be interesting to to think in both prototypes for right and left handed people.\nSpeaker A: Well, the thing is the most remote controls they use by more than one person.\nSpeaker A: So unless you're kind of targeting single people, you know, you're gonna maybe cut out some a lot of your market.\nSpeaker D: I don't know if any way I think it could be interesting to to release some a small fraction of.\nSpeaker B: Oh, my face could be a universal design.\nSpeaker B: Sure.\nSpeaker B: A universal design which is which is good for both the hands.\nSpeaker A: Yeah, still straight for yeah for your hand but not for a particular.\nSpeaker B: That's right.\nSpeaker B: Yeah, whether it's left hand or right hand but but don't you think that the two points are clashing one thing you are saying design should fit the hand shape and it should not be very small.\nSpeaker B: Sorry.\nSpeaker B: The first and the third point they are clashing.\nSpeaker A: Well it can still be a bit you can still extend past the hand.\nSpeaker A: Yeah.\nSpeaker B: Okay.\nSpeaker B: Like so fitting the hand doesn't mean much then.\nSpeaker A: Well, it means like this right here is kind of is very thin and long.\nSpeaker A: So instead of having you know you might have it kind of yeah a bit bigger or you know with maybe some some finger molds or something.\nSpeaker B: So means design should be similar to that traditional ones.\nSpeaker B: Little sleek longer.\nSpeaker D: No, no.\nSpeaker D: I was thinking of like something.\nSpeaker C: Something with the shape of the pan.\nSpeaker A: Yeah.\nSpeaker A: Some finger grips maybe you could even have some buttons like you know.\nSpeaker D: Yeah, it should be symmetrical.\nSpeaker D: Not anymore.\nSpeaker D: That's what.\nSpeaker D: Yeah.\nSpeaker C: And then finally.\nSpeaker D: And finally the incorporation of LCD or a specific condition system in the remote control could also be interested but I don't know if the budget would be larger.\nSpeaker C: Yeah.\nSpeaker C: First I'm just wondering about the LCD stuff.\nSpeaker D: But most of the young people to 30 to 30 years old were really interested in this kind of technology.\nSpeaker C: Yeah.\nSpeaker C: So maybe it's a good time for me to to bring you to some new new information.\nSpeaker C: You had the new requirements from the head of thesis of the company and so they wanted so they want to they would like to be restricted to TV.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: I don't know if you had this information already.\nSpeaker C: No.\nSpeaker C: No.\nSpeaker C: So they want has to restrict the remote control to TV only because of time limitations.\nSpeaker C: They want also.\nSpeaker D: So this marketing report is restricted to TV remote control.\nSpeaker C: Excellent.\nSpeaker C: So we have also to focus more on the internet aspects because well well TLA text is how they did now.\nSpeaker C: And finally it should be clear that the corporate image that means color and logos of the car our company should be clearly identified in the product.\nSpeaker A: Are we still working on this 25 euro price point because I think actually having looked at some of the remote out there this is quite a low price.\nSpeaker A: If we're maybe I can get to this in my presentation though.\nSpeaker A: Yeah.\nSpeaker C: Yeah.\nSpeaker C: So maybe you can jump to your presentations right now.\nSpeaker C: Yep.\nSpeaker C: Okay.\nSpeaker C: So let's keep in mind about that this last point about LCD and speech.\nSpeaker A: I think even if it was within budget to speech record rec system it might be a bit difficult because if you think if you're watching TV you're going to have a lot of this background noise from the TV which might interfere with the.\nSpeaker C: Sorry.\nSpeaker A: What is your following?\nSpeaker A: Ah, participant three.\nSpeaker A: You might have some background noise from the TV which will make the speech recognition much.\nSpeaker D: Yeah but you should be able to activate or disactivate.\nSpeaker D: So yeah.\nSpeaker D: Right.\nSpeaker A: Press a button to talk and the TV.\nSpeaker A: 10 and 50.\nSpeaker B: 10 so. Yeah, no it could be command control kind of thing.\nSpeaker B: It require recognizes particular sequence and then it gets activated.\nSpeaker B: Means you say you should say like this that remote control being on or beyond kind of thing and then remote control comes into the picture for the speech recognition.\nSpeaker B: Because this kind of thing means speech is there from the TV also.\nSpeaker B: So there should be something command controlled you start and then you stop.\nSpeaker B: Yeah.\nSpeaker B: So you're going to be a editor you are having two modes similarly.\nSpeaker B: Otherwise it's just like idle.\nSpeaker C: Okay.\nSpeaker A: Okay.\nSpeaker A: So could I describe the mouse maybe a bit easier to use the mouse or.\nSpeaker A: Yeah.\nSpeaker A: Thanks.\nNone: Okay.\nSpeaker C: The wheel doesn't work.\nSpeaker A: Great.\nSpeaker A: Okay.\nSpeaker A: So while researching this topic I first of all just thought of a couple of things that I would like to see in a remote and just look to see if they're actually available in any current remotes and then also search for which are the top rated remote controls on opinions.com which is a you know a customer written basically review site.\nSpeaker A: So there's a pretty wide range of remote controls these days and this remote control on the right here is one of the more extravagant but it's not really, it's by no means you know on its own in being so expensive there are a lot of expensive remote control out there.\nSpeaker A: Yeah.\nSpeaker A: It looks like a PDA.\nSpeaker A: Yeah.\nSpeaker A: Well basically all the functions are controlled through the LCD screen except for the really really kind of main functions which have a couple of of their own buttons.\nSpeaker A: And if you look at a lot of the universal remotes out on the market I know we're working on television remote but a lot of the universal remotes out there have LCD screens which kind of helps when you're using multiple devices I suppose because you can have multiple kind of functions different functions on the screen at different times.\nSpeaker A: But the thing that I find most interesting about this remote control and it's kind of difficult to see in this light but it has a scroll wheel on it which is kind of like a mouse scroll wheel which I think is it's a really kind of important design aspect because the thing is what the presentation is about is what we want the remote control to actually do and obviously the simplest thing that a remote control does is it just changes the channel.\nSpeaker A: Yeah.\nSpeaker A: I think that a scroll wheel is actually a pretty handy way of changing the channel because I know when I use the remote to change the channel I rarely use the numbers on the pad I usually use the up and down because most channels are two digit numbers and you have to press a special button to enter a two digit number and then the number so it's annoying.\nSpeaker A: So I think a scroll wheel is quite handy.\nSpeaker A: Now the scroll wheel is much more useful if you have an LCD screen and this brings us to the point you were mentioning before about the internet capability.\nSpeaker A: One possibility if we, now we need to still talk about the price point because obviously a lot of this stuff can't be done for 25 euro but one possibility is to download program information into the LCD screen so that instead of actually saying I want to go to channel 37 because I know this program is on you know often you don't know what channel it's on or you don't know what's on.\nSpeaker A: If you have a list of programs on your LCD screen you just scroll to that program rather than to a channel.\nSpeaker A: So if you think about it's kind of like you know in mobile phones now you don't use, you don't remember people's phone number you remember the name and you go find that name and ring it.\nSpeaker A: So this would be pretty kind of anything to have but we really need to discuss the price.\nSpeaker A: So I mean there are cheaper, this is another multi-purpose remote control where it's very simple there's only a few buttons but each of those buttons does something different in a different context.\nSpeaker A: So this is something else we might want to consider is really kind of limiting the number of buttons because this is the top rating universal remote control on opinions.\nSpeaker A: It's really maybe worth thinking about limiting the number of buttons as much as possible because really I think people want to be able to find the button they're looking for without even looking at the remote control and give them a saying before about having different size buttons for different you know frequently used tasks.\nSpeaker A: But I think also you know the location and shape of the buttons is important but also the number of buttons so if you have too many buttons it increases the difficulty of finding.\nSpeaker B: But when this one problem means then the user has to understand each of that functionality.\nSpeaker B: Yeah well because the same button is doing to anything.\nSpeaker A: Yeah well we will have a bit of a simpler task in that we're only doing a television remote control.\nSpeaker A: I think maybe one option is to have you know a little flip open door that you have hidden most of the time but contains the extra buttons like say the number of buttons for instance.\nSpeaker A: I would if I had my perfect remote control I'd probably just have no numbers at all on it because they're just in the way they don't really do anything.\nSpeaker A: Maybe you know although I do also find flip open doors a bit of a pain because sometimes they can break off or whatever but maybe a door that you can permanently remove or permanently have on would be good.\nSpeaker A: But I think definitely you need to keep the buttons down to a minimum but not the let that kind of interfere with the functionality of the device.\nSpeaker D: I think that the transition to this new remote control shouldn't be very abrupt, very hard because if people see a remote control without numbers they will think it's very difficult to learn, very difficult to build.\nSpeaker A: It does something more important.\nSpeaker A: Very different too.\nSpeaker A: It depends on how you market it if you have the right advertisements showing how easy it is and how you can navigate to a program without the numbers then people might say that looks pretty easy.\nSpeaker A: Okay can you continue please mate?\nSpeaker A: Okay so I think one of the really kind of useful things you can do with internet connectivity would be to have this a program driven interface rather than the channel number.\nSpeaker A: So if we can have a higher priced remote control it would really be worth something that would be worth implementing.\nSpeaker A: There's the LCD screen which maybe is too expensive but I think also the scroll wheel I haven't mentioned it here, the scroll wheel could be used without an LCD screen just for changing channel numbers easily.\nSpeaker A: I think even that would be a fairly cheap thing compared to an LCD screen to implement but I think it would be quite useful as well.\nSpeaker A: And the other thing you said we need to keep it just television but I think maybe one option since this is supposed to be a kind of a fashionable device is you know there's a certain kind of cool or well factor that you can have with technology and maybe we want to make it something that's extensible to do other tasks.\nSpeaker A: Say you have like another little kind of base unit that can also receive signals as well the television where you can say change the lighting in the room.\nSpeaker A: That would be something maybe you could sell as an extra so that it doesn't have to be part of the initial development but you know later on you could, you know, you're also selling the potential of the device then you say potentially you can then do other cool stuff like change the lights, I don't know close the windows, whatever, turn the heating on.\nSpeaker A: And I think that's something we may need to have as is at least as an optional extra to kind of make out product cool since we say we're putting the fashion in electronics.\nSpeaker C: Okay, okay thanks.\nSpeaker C: Yeah, you want to go?\nSpeaker C: Yeah.\nSpeaker C: Give me.\nSpeaker B: So most of the things that you are discussing about a speech recognition are on my own end.\nSpeaker B: Yeah, it should be.\nNone: Great.\nSpeaker C: Nope, that's one.\nSpeaker C: That's your two.\nSpeaker B: Okay, so the working design is user interface could be of two types.\nSpeaker B: One is the usual press buttons which are there so that the user feels that he's doing some he's knowing about the technology.\nSpeaker B: So he's pretty comfortable if he wants to get this and on top of that there there could be a speech recognition technology also being sitting on the remote.\nSpeaker B: So the old kind of users who don't want to have any changes it can be useful for them.\nSpeaker B: And the new users as our marketing expert was saying they can use the new guest mode which is speech recognition kind of thing.\nSpeaker C: Okay, sorry to interrupt you but we have seen before that there is a new way of interacting that use wheel.\nSpeaker B: That's right.\nSpeaker B: So anyway that didn't come into my mind.\nSpeaker B: Okay.\nSpeaker B: So that is a possibility.\nSpeaker B: These could be other kind of interfaces.\nSpeaker B: It means we can have depending on the cost how much we can afford we can have different kind of interfaces.\nSpeaker B: So speed buttons are something which is very everybody's familiar with.\nSpeaker B: So if you go to the market and you say that buttons are there the people know what it is.\nSpeaker B: And on top of that if we are having extra functionality if I are willing to shell the 25 euros money which we are thinking otherwise we are just like others in the market.\nSpeaker B: So anyway that is the first user interface could be of more than one type and yeah that means we can do the online changes which cannot be done now actually.\nSpeaker B: So apart from speech we can have this crawl kind of thing with the buttons.\nSpeaker B: Now for buttons normal requirements like bit coding and all those things are required.\nSpeaker B: And for voice limited vocabulary automatic speech recognition system is required and we require a microphone also to be sitting there on the remote.\nSpeaker C: Yeah that increases the cost also.\nSpeaker B: That's right but means we have to see how much what kind of microphones and stuff like\nSpeaker C: that. Do you think that performance of such systems are enough to target of such technologies is enough?\nSpeaker B: Yes if it is limited vocabulary usually it's enough.\nSpeaker B: We can target means we can target 95% accuracy or somewhere.\nSpeaker C: I imagine also that the microphone will be an ambient microphone because you are not going to speak.\nSpeaker B: No it could be a little joke.\nSpeaker C: So it could be a few centimeters.\nSpeaker A: That's right.\nSpeaker A: One other thing that speech recognition could really blow out the price for is when you want to sell into other markets though because I'm not sure exactly where we are going to sell this but I presume it's not just going to be English speaking countries.\nSpeaker A: So then you have to train models for...\nSpeaker B: It's more like there are different speech technologies which are existing.\nSpeaker B: So DTW could be kind of which is the easiest.\nSpeaker B: So you have to store some templates on the chip itself and it's just dynamic time warping where you try to find out what it is.\nSpeaker B: Instead of having a model which is to be trained and being a microcontroller.\nSpeaker C: Okay we should discuss that later after this slide.\nSpeaker C: This is a very important issue in the discussion.\nSpeaker C: Okay next.\nSpeaker C: Yeah.\nSpeaker C: It's finished.\nSpeaker C: No no.\nSpeaker B: Components.\nSpeaker B: Components.\nSpeaker B: Yeah.\nSpeaker B: So we'll go to the next slide.\nSpeaker B: Yeah sure.\nSpeaker B: Yeah so this is the design which we are thinking.\nSpeaker B: So we are having the power button and the switch which is not much and then we are having the bulb with us which is to indicate whether the power is on or not.\nSpeaker B: And then there are two kind of things which can be...\nSpeaker B: So one is the button interface which has not been shown because of lack of time.\nSpeaker B: So we could not put that.\nSpeaker B: So now where the ASR decoder is sitting.\nSpeaker B: Similarly there are different kind of interfaces which could be there.\nSpeaker B: So there is ASR decoder which could be there and then there could be another scroll button, scroll, scroller and then there could be buttons.\nSpeaker B: And all of them they will just do the decoding and put it in the proper message format and then there is the chip which is sitting, the green one and it converts it into bitcodes and that bitcodes are sent by the infrared device to the receiver.\nSpeaker C: Okay.\nSpeaker B: So this is the easiest design that they could be.\nSpeaker B: So in ASR decoder we can have things in parallel to have different technologies.\nSpeaker B: So this was the...\nSpeaker B: My personal preference was that we can have ASR sitting there on the remote control.\nSpeaker A: Yeah I guess you could actually train the remote control as you are using it by saying you know, turn volume up and you press the button like people teach sign language to kids.\nSpeaker A: Well, by speaking and doing it.\nSpeaker B: As soon as you try to put a microchip kind of thing or something the price will go up.\nSpeaker B: So these are the slight problems.\nSpeaker C: So your opinion is that we should go for our subconscious.\nSpeaker B: Yeah, the reason is that if we go into the market means, though I do not have much idea, but as the marketing expert presentation was.\nSpeaker D: If you can sell a speech recognition remote control for 25 euros, everyone will buy it.\nSpeaker D: So if you go on the job show.\nSpeaker A: Actually, if I was using remote control to say turn the volume up because I can't hear it very well, I don't really want to drown out what people are saying by talking, you know, when I'm instead of pressing up on the remote control.\nSpeaker A: You know, there's some dialogue all of a sudden that I can't hear.\nSpeaker A: I'm trying to actually find out what's being said.\nSpeaker A: So maybe speech recognition gets in the way more than it helps.\nSpeaker D: Okay.\nSpeaker D: So yeah, but you know the average frequency of passion bottoms it's about.\nSpeaker A: Well, it depends if it's a remote control.\nSpeaker D: It's about 80, 80, 80 pashes per hour or something like that.\nSpeaker A: Maybe if the remote control is something that you don't actually have to pick up anymore, that would be a useful feature of the speech recognition.\nSpeaker A: If you can leave it sitting on the table and you don't actually have to find it then.\nSpeaker C: Okay, gentlemen, we have to take some decisions right now.\nSpeaker C: So if I kind of summarize everything we say, while targeting TV, we need to have remote control which is fancy, which is easy to handle, to spoil, not too big.\nSpeaker C: We have a good shape for that.\nSpeaker C: A good shape, yes.\nSpeaker C: We should bring new technologies for young people's.\nSpeaker C: And as we have also requirements to use to push through the internet, maybe this is something we can stick to it.\nSpeaker C: And also very interesting things I've seen on one of our competitors, this wheel that we can use to navigate.\nSpeaker C: So my feeling is that regarding costs, budgets we have and target price, it's not possible to go to LCD and also to go to automatic speech recognition technologies.\nSpeaker C: First, why not to go to LCD?\nSpeaker C: Because in fact, as we are targeting TV, in fact we can use TV screen as a screen to feedback, to give some feedback information about what we could have.\nSpeaker A: Well, it depends.\nSpeaker A: Unless we have some input, some video input to the TV, or we have control over the TV, then we can't actually display that.\nSpeaker A: Like if we produce the TV, then we can put menus up there.\nSpeaker A: But otherwise we need to actually have something sitting in between the video signal and the TV to superimpose those menus.\nSpeaker A: So that's an extra...\nSpeaker C: Yeah, that's right.\nSpeaker C: Don't we have contacts?\nSpeaker C: We have people on TV or systems that exist in use?\nSpeaker A: This is another...\nSpeaker A: We still haven't really defined the remote.\nSpeaker A: Are we still...\nSpeaker A: You said we're focusing on TV, but is it still kind of like a universal remote in that it's a replacement remote control?\nSpeaker A: Or is this something for our own line of televisions?\nSpeaker A: Because that really makes a big difference.\nSpeaker A: Because even if we have contacts, we can't really produce remote control that can bring up menus on other companies' TVs.\nSpeaker A: It's just there are too many TVs out there.\nSpeaker A: It's not really...\nSpeaker C: That's a good point.\nSpeaker C: What could be the cost of...\nSpeaker C: Well, could we fit the targets in terms of cost if we go to LCD and remote control?\nSpeaker A: For 25-year?\nSpeaker A: I think it's impossible.\nSpeaker A: It's impossible.\nSpeaker A: Yeah.\nSpeaker A: I think it would be good to know if there is any leverage in that Li-Wei in that 25-year?\nSpeaker A: Because for 25-year, I think all we can really do is provide a very basic remote control.\nSpeaker A: That seems to be kind of against the philosophy of our company, which is putting the fashion into electronics.\nSpeaker A: I would like to know if there's any chance of increasing the unit price.\nSpeaker C: What would be...\nSpeaker C: So, you mean we should target something maybe which would be more expensive, but really fancy in terms to have really an added value?\nSpeaker C: Yeah.\nSpeaker C: Because...\nSpeaker C: So, regarding the automatic speech recognition, I think this is...\nSpeaker D: Well, question.\nSpeaker D: What would be the goal of putting a LCD in remote control?\nSpeaker D: What kind of information?\nSpeaker A: Because you can have things like the program name instead of the channel number.\nSpeaker D: But it's like an interactive program.\nSpeaker D: Most of the TVs know where they are.\nSpeaker D: They have teletext.\nSpeaker D: Show the...\nSpeaker C: Wow.\nSpeaker C: Because they have teletext on it.\nSpeaker C: You have a teletext signal that you can...\nSpeaker C: Yeah.\nSpeaker D: But most of the TVs have teletext nowadays.\nSpeaker C: They have...\nSpeaker C: Most of them have teletext, but we want to get rid...\nSpeaker C: Well, one of our requirements is to move to teletext to the use of internet.\nSpeaker C: So, to...\nSpeaker A: You can get a lot more information.\nSpeaker C: It grows more easily in the teletext, for instance, through your remote control.\nSpeaker D: So, what would appear in the LCD?\nSpeaker A: So you could have the name of the program.\nSpeaker A: You could have the start time, you know, where it's up to...\nSpeaker A: To start all the programs you could have...\nSpeaker A: You could have even a little image of, you know, the main actors or something.\nSpeaker A: So you can quickly just kind of even without reading...\nSpeaker C: I don't know if this information is available from teletext also.\nSpeaker A: Well, no, but there are other electronic program guides.\nSpeaker A: They may not have pictures, but maybe they do.\nSpeaker C: It also depends on the country.\nSpeaker C: For the same reason that we cannot overlay information on the TV.\nSpeaker C: We couldn't grab information which is not there.\nSpeaker A: No, but I mean with the internet you have flexibility of where you get your information from.\nSpeaker A: It's impossible that there are people out there providing that.\nSpeaker C: So, I mean we need an extra internet connection to use the remote control if you're on the browser.\nSpeaker C: Yeah.\nSpeaker C: In addition to the TV.\nSpeaker C: Or it should be a special TV connection.\nSpeaker A: Well, I think we would definitely need the internet connection because even with...\nSpeaker A: I don't think you could even get teletext information from the TV onto the remote control.\nSpeaker A: Especially if we don't control the TV.\nSpeaker A: I don't know, we need to find that out.\nSpeaker C: Okay.\nSpeaker C: We need to close the meeting.\nSpeaker B: So just a small thing.\nSpeaker B: What kind of market we are targeting?\nSpeaker B: Is it that we are targeting the replacement remote market or what?\nSpeaker B: So the remote has gone bad and the person wants to buy a new remote or because the cost of LCD thing could be as high as the TV itself.\nSpeaker A: That is very important.\nSpeaker A: No.\nSpeaker A: If it's a really small TV.\nSpeaker C: Wow.\nSpeaker C: People go through buying another remote control when they broke.\nSpeaker B: Broke.\nSpeaker C: Broke they hold and they want to go for universal one and they take the fanciest they can have.\nSpeaker C: So this is that we should target.\nSpeaker C: So the commitment is following.\nSpeaker C: We don't go for speech recognition technology.\nSpeaker C: The LCD is still open to discussion.\nSpeaker C: It is up to you to go through this way and to report me back next meeting.\nSpeaker D: So...\nSpeaker D: I think that the speech recognition technology would be cheaper than the LCD.\nSpeaker C: Yeah, but not sure.\nSpeaker C: Maybe it's cheaper but we have no...\nSpeaker D: Because with the LCD you need more requirements.\nSpeaker D: You need internet connection.\nSpeaker D: You need more things.\nSpeaker D: But for the speech recognition you don't need anything.\nSpeaker D: You have to say, China 50 and that's it.\nSpeaker A: I think the type of people that are going to want to buy a very stylish remote control with lots of new technologies is the kind of people that are going to have a wireless internet connection maybe or...\nSpeaker D: But then we should move to another target because at $25.\nSpeaker A: This is for $25.\nSpeaker A: Can we increase the price point of this remote control?\nSpeaker A: Because otherwise we need...\nSpeaker C: Okay, this is not one question for you.\nSpeaker C: Yeah, this is up to you to tell us.\nSpeaker C: But I'm definitely not keen on...\nSpeaker C: To move to another target?\nSpeaker C: No, no, no.\nSpeaker C: I'm definitely not keen on going to speech recognition technologies.\nSpeaker C: I'm not confident enough.\nSpeaker C: I'm not sure that we have a product.\nSpeaker C: Really that work.\nSpeaker A: That work.\nSpeaker A: It's kind of hard to guarantee that you're going to...\nSpeaker C: Yeah, how to guarantee such performances is very hard.\nSpeaker D: The expert said 95%.\nSpeaker A: 95% is not good enough.\nSpeaker C: It's very bad.\nSpeaker C: So this is the end of this discussion.\nSpeaker C: Next meeting, you have the task you have to work on.\nSpeaker C: So you have to work on the component concept.\nSpeaker C: You have to work on user interface.\nSpeaker C: And you have to go through a trend watching.\nSpeaker C: So the question is still open about the LCD thing.\nSpeaker C: I hope that next meeting we will take some decisions that are there.\nSpeaker C: Thanks.\nSpeaker C: Bye!\nNone: this.\n"}, {"meeting_id": "ES2012b", "summary": "This is the second meeting of the design group. Project Manager went through the notes of the last meeting and informed the team of the management's requirements. They expected the remote control to cover television only and asked the team to incorporate the company logo and colours. As the team is looking at extreme simplicity, the buttons should be reduced to the absolute minimum. Voice recognition can also be added to reach this point. In addition, the remote control can be powered by a rechargeable or permanent battery.", "dialogue": "Speaker A: reminds me ofwater for dove year Okay, welcome to the second meeting of this design group.\nSpeaker A: Briefly go through the notes of the last meeting.\nSpeaker A: We've just done in note form and I haven't attributed anything to individuals because we're working strictly as a team here.\nSpeaker A: We saw that the problems with existing remote controls were the boring shape and boring color.\nSpeaker A: We saw that what we needed to do was to make sure that you guys control several items.\nSpeaker A: Switching was easy but you shouldn't need to point the thing at anything in particular.\nSpeaker A: That it needed to be controlled to make it interesting that the keys might be concave simply because that hasn't been done before that we know of.\nSpeaker A: Interchangeable, spacious so people can personalize it, illuminate it so that people can see it in dark rooms, that people might want it in addition to their existing remote controls.\nSpeaker A: It should always work whenever you use it but it should be too small but it gets lost.\nSpeaker A: Now I'll shortly ask for three presentations.\nSpeaker A: Before I do that however I will go through some new project requirements that the management has placed on us and will be challenging in terms of what we discussed at the first meeting.\nSpeaker A: The management has had its own thoughts on this and they didn't necessarily agree with what we thought.\nSpeaker A: As a result of that we will then talk through the functions that we see the device actually coming out and we have 40 minutes to do this in.\nSpeaker A: The new requirements are that the management team see that the teletext is no longer of any importance given the device on the internet and they wanted only to cover televisions.\nSpeaker A: What is not quite clear from their directive is whether they didn't want it to cover teletext or whether they didn't want it to cover videos, DVDs, satellite boxes which we saw as being fundamental to the exercise.\nSpeaker A: The actual wording of the directive is that it should cover television only on that basis.\nSpeaker A: I think we need to bear that in mind but possibly keep at the back of our minds.\nSpeaker A: The reality that people even with it no longer didn't look at teletext any more they certainly do look at other things.\nSpeaker A: The device has to incorporate the company logo and colors.\nSpeaker A: The logo being at the bottom of the screen there the two Rs in grey against the yellow background.\nSpeaker A: Now this doesn't necessarily mean that we have to give up some of our ideas about making it attractive to the market.\nSpeaker A: The other thing we have to do is to do some constraints as to how we might do that.\nSpeaker A: It also has to be simple which to some extent goes along with the first one and the way we've already said that it must be simple because that's what people want anyway.\nSpeaker A: We want it to be simple to get it to the market quickly which you see is their choice but we need to talk that through.\nSpeaker A: OK, so after meeting we'll be summarised and note sent out and etc.\nSpeaker A: OK, so we'll first of all have individual reports from everybody. Again there is no order of precedence here so I'll leave it up to you who thinks I'd like to go first.\nSpeaker B: I don't mind. I'm averse. I can steal the cable.\nSpeaker B: Cheers.\nSpeaker A: OK, so we'll start with the one that looks like a projection screen that I've got.\nSpeaker B: These are the functional requirements from there. Our guys don't have the research lab.\nSpeaker B: Take 100 people and cover all the aspects of what is needed by people and what they want to see.\nSpeaker B: Everything can from functionality and how individual functions are and how often they're used and which are necessary.\nSpeaker B: As we noticed, 5% of people find their remote controls ugly so some kind of a new style should be incorporated.\nSpeaker B: 8% of people said they'd spend more money on it which is a plus for us if we can make it a bit better or cost effective.\nSpeaker B: We can put price up.\nSpeaker B: The remote controls do not match the operate behavior of the user. I can take that to mean as the only one from the television or as a live live.\nSpeaker B: We've got six remote controls for our stereo system, a digital box, a DVD player, a video player and TV.\nSpeaker B: We've matched well to my behavior. Again, 75% use the sample.\nSpeaker B: We've changed channels of volume settings. I think a big issue is 50% usually use 10% of the buttons.\nSpeaker B: Some of the functions like audio settings aren't hardly used and are not considered relevant by the user.\nSpeaker B: Maybe fewer buttons which also make the design look sleager.\nSpeaker B: We've changed the illustrations of people losing the remote control.\nSpeaker B: We can use something else to find their remote control.\nSpeaker B: Time taking to learn new remote controls is to make it too complicated and easy to use for fresh time users.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker A: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker B: We've also asked if people pay more for speech records.\nSpeaker C: We've also asked if people pay more for speech records.\nSpeaker C: We've also asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: We've asked if people pay more for speech records.\nSpeaker A: If we wanted something different, truly different, then the button must remake control.\nSpeaker D: Well the only problem I can think of with that is if you've got a lot of people that don't want to be bothered learning how to use new remote controls, if you just kind of take away everything that they're used to knowing, that's going to be quite a change.\nSpeaker B: But if you just lift it up and say channel 1, or even BC, you could even just have it left on, just put it down once on top of your TV and never have to.\nSpeaker C: Yeah, have it be kind of like the satellite box or the cable box and have it just go on the TV and then it doesn't matter where in the room you are, you won't lose it.\nSpeaker A: I can see technical problems with that in terms of the sound from the television, because if somebody actually on the television says, it's actually like that, I TV, only watching BBC, then it might change itself, so it probably needs to be possibly actually need a button on it, or it's too activated, or something just to identify that you've lifted it up on its use, and then just say, I don't know if it's all, and then, I mean that would certainly be truly different.\nSpeaker A: Because their audio settings not 0.8%, and if they weren't there, would people miss them?\nSpeaker D: But look at the importance of them, the volume settings.\nSpeaker B: Relates of 2 out of 10, yeah.\nSpeaker A: Volume, yes.\nSpeaker D: They're not used often, but they are quite important when they use it.\nSpeaker A: We need to identify things that people actually need, and it's a functional frequency and relevance, and I would say ignoring power for the moment.\nSpeaker A: The Chamblum volume, given that we've been told to ignore teletext, the Chamblum volume are the only ones that would appear to be essential.\nSpeaker A: So if we can design something that looks interesting, it all looks different, it incorporates the logo and the colors, and we can still have our interchangeable face, even if it's the yellow and grey.\nSpeaker A: Are there no buttons or buttons? There's an option.\nSpeaker B: I just thought I actually saw interrupt.\nSpeaker B: We're thinking about the technical problems of someone on the television saying a channel number, and we can maybe have an activation word, because I've seen a lot of computers before, which is addressed, or more to address, the computer, and then give it a command.\nSpeaker A: It depends on whether if you want to make this so simple that anybody can walk into the room and lift it up and say, yeah, that's a PVC worm.\nSpeaker A: You could actually print it on the device itself.\nSpeaker B: I'm missing the point of view. You could still lose this.\nSpeaker A: That's always been a problem, I think.\nSpeaker A: I suppose one can make it so desirable that people lose it, they can buy another one.\nSpeaker A: Anyway, sorry, carry on. I'll just carry on with...\nSpeaker A: Oh, I have. I have.\nSpeaker A: What's up to you?\nSpeaker A: No, no, no, you have the option of the...\nSpeaker A: Oh, okay.\nSpeaker A: In the middle of your report, so...\nSpeaker B: Well, I was just kind of wrapping up there. Yeah, I was thinking, yeah, you're such things that are relevant.\nSpeaker B: You can make things much more. I think the 80% of people with them are more on...\nSpeaker B:...are more with it, looks better.\nSpeaker B: Combined with...\nSpeaker B:...decreach the... or take out the limited functions that we don't really use much.\nSpeaker B: Well, it's been... or ready to get to vertex, but...\nSpeaker B:...as my channel settings and stuff, it might turn people somewhere to build some people that want the whole functionality away.\nSpeaker B: Since if we're marketing under a kind of fashionable approach, then...\nSpeaker B:...it would be fashionable. Passionable of practicality.\nSpeaker A: Because we could make it dual-function, with our voice recognition and...\nSpeaker A:...we still have buttons on it.\nSpeaker A: Yeah, we could even have it as like...\nSpeaker B: Yeah, the buttons control this and the voice functions control the things that you would do all the time.\nSpeaker B: So...\nSpeaker A: Certainly could.\nSpeaker A: So, yeah, we could...\nSpeaker B:...power on and channel selection and...\nSpeaker B:...climps selection. We didn't have to, really.\nSpeaker A: Yeah, I mean, the advantage of doing the way the buttons all together is it makes them cheaper.\nSpeaker B: Yeah, it probably would look better, as well.\nSpeaker A: Yeah, it certainly...\nSpeaker A:...but it's a possibility for making it visually very distinctive.\nSpeaker A: Yeah.\nSpeaker A: Because it does not have to be...\nSpeaker A:...a oblong box...\nSpeaker B:...lamed with a number of buttons.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Who...\nSpeaker A: Yeah, yeah, that's definitely...\nSpeaker A:...given that we bought the...\nSpeaker A:...extensive discussion...\nSpeaker A:...of the senses.\nNone: Okay.\nSpeaker C: Well, I could do mine.\nSpeaker C: Don't like it.\nSpeaker B: Yeah.\nSpeaker C: Let's see if I can...\nSpeaker C:...make this work.\nSpeaker D: Oh, you have to take, like, function and...\nSpeaker D:...have something.\nSpeaker D: Oh, athlete.\nSpeaker C: Is it doing...\nSpeaker B: Yeah, give it a bit, 20 seconds.\nSpeaker D: Okay.\nSpeaker D: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: So this is just about the technical functions.\nSpeaker C: So the method I looked online for examples of other similar products and then just kind of...\nSpeaker C:...retrying to brainstorm some possible design ideas and...\nSpeaker C:...identify what the necessary things are, what people are...\nSpeaker C:...what you really want to have a remote control do.\nSpeaker C: And then there are two different kinds that I found.\nSpeaker C: There's a user-centered one and an engineering-centered one, which I will have pictures of.\nSpeaker C: And then we kind of have to decide which one this should be.\nSpeaker C: So these are the two...\nSpeaker C:...different ones.\nSpeaker C: This one.\nSpeaker C: This is the user-centered.\nSpeaker C: It has quite a few...\nSpeaker C:...fewer buttons.\nSpeaker C: And then this is the engineering-centered, which has a lot more buttons.\nSpeaker C: And probably this is one that people complain about about having too many buttons that you don't use.\nSpeaker C: So basically, what remote control is is you...\nSpeaker C:...it's to send messages to the television set, you know, turn on, off, switch the channels and the volume and things such as that.\nSpeaker C: And so for this product, it's going to be television only and then it has to have the logos...\nSpeaker C:...for the company and the colors.\nSpeaker C: And so for my personal preferences, I think this one is easier to use.\nSpeaker C: It has quite a, you know, fewer buttons.\nSpeaker C: We want something that sends messages easily to the television.\nSpeaker C: And I was kind of wondering about this example that they have...\nSpeaker C:...it looks kind of narrow at the top.\nSpeaker C: And I was thinking maybe if it were wider at the top, then that would be easier.\nSpeaker C: And so we have to decide what's going to make our product different.\nSpeaker C: The unique style, maybe have it light up.\nSpeaker C: It's visible in the dark, the changeable faceplates.\nSpeaker C: And the lighting up and visible, I was...\nSpeaker C:...when we were talking about losing it, maybe to have a button on the television that you press.\nSpeaker C: Maybe if it makes it in a ways or a life set for something like that.\nSpeaker C: So it's easier to find if someone has hidden under the couch or something like that.\nSpeaker C: So that's my presentation.\nSpeaker A: Can I...I'm actually going to use the...\nSpeaker A:...it's been of course great technical problems over here.\nSpeaker A: I'm actually going to use the...\nNone:...\nSpeaker C:...they probably clip.\nSpeaker D: They're not connected to anything on the table.\nSpeaker D: So I'll just leave one more ground.\nSpeaker A: It's rather than the traditional, in fact, I've got you to go that far.\nSpeaker A: Something like this shape.\nSpeaker A: So it's something that you can...a more vertical shape.\nSpeaker A: That you can sort of hold in your hand.\nSpeaker A: I think...\nSpeaker A:...such as, I mean, so we hold it like that, possibly with a couple of buttons like that.\nSpeaker A: But with the entire top, with the infrared or whatever source.\nSpeaker A: So that it's flying up in all directions.\nSpeaker A: So that, again, the neat look at the technicalities of...\nSpeaker A:...actually achieving that in terms of whether the power requirements of the...\nSpeaker A:...such a source, you know, compromise the...\nSpeaker A:...our need for, you know, it being...\nSpeaker A:...permanently...\nSpeaker A:...you know, available, whether different technology.\nSpeaker A: I mean, all these remakes are presumably...\nSpeaker A:...inforade, like they have been for a long time.\nSpeaker A: But we possibly need to be looking at something different.\nSpeaker A: You know, short range, not like the older radio remote controls.\nSpeaker A: We change next door to telly.\nSpeaker A: But I think basically if we're going to...\nSpeaker A:...if minimum number of buttons is our priority, then we should...\nSpeaker A:...it's going to say, we're going to really bite the bullets and reduce the buttons to...\nSpeaker A:...ampsit minimum, you know, possibly we'll back up channel and volume buttons and...\nSpeaker A:...on off, and nothing else.\nSpeaker A: So that it could...\nSpeaker A:...unform a same dot like that, because again, except with some...\nSpeaker A:...the wrist losing it.\nSpeaker A: Anyway, okay, so Kate, what's your...\nSpeaker A:...our help thoughts on this?\nSpeaker C: Which one's this plug into?\nSpeaker C: I think it's all...\nSpeaker C:...there.\nSpeaker D: I can't...\nSpeaker D:...did you see the most green?\nSpeaker D: Oh, yeah.\nSpeaker C: That's kind of...\nSpeaker C:...that's not cool to rain.\nSpeaker D: Oh, well, anyway.\nSpeaker D: Alright, yeah, so I'll just do my presentation on the working design.\nSpeaker D: There we go.\nSpeaker D: Okay, just at the...\nSpeaker D:...the whole method of how the remote control works.\nSpeaker D: The basic function of the remote control is to send messages to another system, the...\nSpeaker D:...the TV or the DVD player or whatever.\nSpeaker D: It does this by what we need to start off with in the energy source.\nSpeaker D: And this energy source will feed into an integrated circuit chip.\nSpeaker D: And the circuit chip is the part that actually composes the different messages...\nSpeaker D:...within the remote, which will then be sent to the...\nSpeaker D:...the television, the DVD to tell what to do.\nSpeaker D: And you need a user interface, which controls the chip and thus the messages.\nSpeaker D: And the user interface is basically just the design of the actual remote...\nSpeaker D:...what you'll hold in the hands and what buttons will be on it.\nSpeaker C: Shoot.\nSpeaker D: Okay.\nSpeaker D: Just general findings.\nSpeaker D: What we need, technically speaking, for remote control is some sort of energy source.\nSpeaker D: Some sort of user interface, which I think we've mostly been talking about the user interface and the design of that.\nSpeaker D: A circuit chip within that to control and send the messages and send her and receive her.\nNone: And...\nSpeaker D: Oops.\nSpeaker D: It's just sort of a little schematic diagram of what we're looking for.\nSpeaker D: This just kind of represents the energy source, which feeds into the circuit chip.\nSpeaker D: Which maybe then we could have that V into a switch, which would send signals to...\nSpeaker D:...a subcomponent and onto a light bulb to light up once we start pressing buttons.\nSpeaker D: Also send signals to the infrared bulb, which will be the part that actually...\nSpeaker D:...what?\nSpeaker D:...send signals to the television and then you've got your happy little TV watcher there.\nSpeaker D: So my personal preference is, I just think we need sort of a big energy source that won't die out.\nSpeaker D: Perhaps some sort of rechargeable battery or a battery dock you could place in so it would constantly be charged.\nSpeaker D: You wouldn't have to be worrying about it running out of batteries and not changing channels for you.\nSpeaker D: A wide range of sender receiver so that you can hit the buttons from basically anywhere in the room and the channel will still be changed.\nSpeaker D: Also definitely a user-friendly interface.\nSpeaker D: And I think we've also mentioned adding a locating device on it, so when it does get stuck out of the couch cushions, as they inevitably do, you can find them easily.\nSpeaker D: And that's pretty much it.\nSpeaker A: Okay.\nSpeaker A: It seems to me there are a number of fundamental decisions to make before we...\nSpeaker A:...you know, I think your point about the big energy source is a very valid one.\nSpeaker A: I don't suppose we've got any statistics on the die expectancy of the remote controls, particularly the independent ones.\nSpeaker A: Given the number of things you buy these days, we have a lithium whatever battery it that never needs to be placed in.\nSpeaker A: Perhaps we should have the disposable remote control of the battery relaxed, five, ten years, by which time...\nSpeaker A: I mean when all is set and done, digital television will be taken over in that time scale.\nSpeaker A: Perhaps we should have reduced the moving parts even more by not even having a battery compartment.\nSpeaker D: Just having one that's guaranteed to last five to ten years?\nSpeaker A: Yeah, if anybody manages to run it down, then you can run it well.\nSpeaker A: Thanks, Anna.\nSpeaker A: It's what it saves in cost and you know, there's a...\nSpeaker A: Well, so they marked it and it started to get impacted, so it's very practical.\nSpeaker A: I think the idea of a rechargeable one is...\nSpeaker A:...and I see a really high tech in it, it sort of just recharges itself if it's the buy, you know, magnetic waves or whatever.\nSpeaker B: It could have like a...\nSpeaker B:...like cordless floor in your house.\nSpeaker B: It's like a base.\nSpeaker A: Since they're old, a lot of people are really going to use it.\nSpeaker C: Yeah, people would think that people might forget...\nSpeaker C:...people forget to put their cordless phones back on there.\nSpeaker A: Yeah, it's... I mean, I know the promise, because my wife goes out in the morning and says I want you to put the phone numbers charged.\nSpeaker A: And then she's had those days so long when she doesn't look that pain by now.\nSpeaker D: I only remember to charge my cell phone when battery dies.\nSpeaker D: When it turns itself off, that's when I plug it in.\nSpeaker A: Yeah, right, so what we think about the...\nSpeaker A:...the department is the battery.\nSpeaker D: Yeah, I think that's a good idea.\nSpeaker A: That's pretty key to it.\nSpeaker A: We are really going for the ultimate in external simplicity here.\nSpeaker A: The cop costs in the manufacturing.\nSpeaker A: And, you know, they have a high tech in here, but then they will be cost effective.\nSpeaker C: Do they make batteries on that long?\nSpeaker A: I mean, certainly... I can't think of my head, but there are certainly things that you buy.\nSpeaker A: I mean, calculators, for example.\nSpeaker D: They usually have the little light source.\nSpeaker D: I don't think they're called...\nSpeaker A: They have the little solar.\nSpeaker A: Yeah, the little cells that...\nSpeaker A: Or something, I mean, the battery ones that are...\nSpeaker A:...firmarily...\nSpeaker A:...see, and in fact, I would...\nSpeaker D: Most of them don't, they have sort of a combination of the two.\nSpeaker D: Like, when there is light, they'll work off the light and then if they'll kick into this battery, so we can maybe do something like that, whereas there is a battery.\nSpeaker D: But if there's enough light, then it's using the light that's not actually draining the battery all the time.\nSpeaker D: But you will have the battery there for...\nSpeaker A: Yeah.\nSpeaker A: I mean, it's going into the technology a bit.\nSpeaker A: I mean, the actual time that a remote control is actually operating.\nSpeaker A: Our thing is...\nSpeaker A: It's probably...\nSpeaker A:...no more than minutes in its entire life.\nSpeaker D: Well, it depends who's using it, who's just sitting there clicking the phone.\nSpeaker A: Yeah, some people are getting the noise from it.\nSpeaker A: Yeah.\nSpeaker A: And they're clicking a lot, yeah.\nSpeaker A: That's what the wrong mouth is.\nSpeaker B: Like, this...\nSpeaker B:...my research thing says, number of times per hour, that's used, channel selection 168.\nSpeaker B: per hour.\nSpeaker C: Yeah, per hour.\nSpeaker C: Wow!\nSpeaker C: It's a lot.\nSpeaker C: Yeah.\nSpeaker A: I want to know, I can just say that.\nSpeaker A: That does sound excessive.\nSpeaker B: But then again, if you think it might have, you know, meant the use, it's like, you know, that's less than a second.\nSpeaker C: Yeah.\nSpeaker A: Well, that's right.\nSpeaker A: And I don't know what it is.\nSpeaker A: I don't know what it is.\nSpeaker A: I don't know what the signal is.\nSpeaker A: The signal is less as long as you keep the button pressed, or whether it's just a 10 to the second, no matter how impressive, or I don't know, I don't actually know.\nSpeaker C: But I think with digital TV, like I know on my cable box, you're not supposed to do that because the channel can't keep up with it.\nSpeaker C: If you just press it like that, so you're supposed to use the menu and go through the...\nSpeaker C:...different channels that way instead of...\nNone: Right.\nSpeaker A: So I've got my C-stage 5 minutes.\nSpeaker A: I don't know how long does that appear.\nSpeaker A: Oh.\nSpeaker A: Where are we getting?\nSpeaker A: Right.\nSpeaker A: So I'd need to sum up very quickly here.\nSpeaker A: We're looking at extreme simplicity.\nSpeaker A: We're looking at a radically different shape.\nSpeaker A: Possibly no buttons at all, but if you can incorporate channel change and volume buttons into the design, then that's fine.\nSpeaker A: Indeed, I mean, the role of the user interface, the designer becomes more involved here because, you know, shape is no longer a...\nSpeaker A:...series constraint.\nSpeaker A: But we clearly only need the main buttons, although we clearly only need the main functions.\nSpeaker A: I don't see why we shouldn't go for voice recognition.\nSpeaker A: And the only buttons that I think we need are channel control, volume control, and on-off.\nSpeaker A: It needs to incorporate the portrait logo, the grey and yellow colour scheme, and there's no reason why we can't introduce interchangeable covers.\nSpeaker A: Does that accurately summarise what we've discussed?\nSpeaker B: Yeah.\nSpeaker B: Yep.\nSpeaker B: Right.\nSpeaker A: So...\nSpeaker C: I just have one question.\nSpeaker C: So are we doing just the television?\nSpeaker C: Are we doing...\nSpeaker C: Yeah, I'm doing just the television.\nSpeaker C: So not DVD.\nSpeaker C: No.\nSpeaker A: But, okay.\nSpeaker A: I think that's quite clear.\nSpeaker A: But maybe the information would be giving me the...\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: That's the one.\nSpeaker B: And the constraints that you know what I got.\nSpeaker B: You know, I didn't even mention the text.\nSpeaker A: It's just a television.\nSpeaker A: That's one I sent to, which was mine too.\nSpeaker A: Oh, okay.\nSpeaker A: Oh, yeah.\nSpeaker A: Oh, yeah.\nSpeaker A: Okay.\nSpeaker A: I came there and sent head-off is...\nSpeaker A:...that's...\nSpeaker A:...that's there.\nSpeaker A: There we go.\nSpeaker A: Okay.\nSpeaker A: So we can all give some thought to that.\nSpeaker A: But what will it make me think about?\nSpeaker A: Okay.\n"}, {"meeting_id": "Bed008", "summary": "The group members mainly talked about the details of the belief-net framework. They first discussed the way to make the definition of user and situation clear and set up the standards, e.g., user thrift, based on user's budget. Then they turned to the discussion about endpoint decision. They mentioned approaches including setting features/modes of like selling, fixing and exhibiting, and this inspired the group members to design user compatibility nodes of which value would depend on the user status and situation. They further discussed Go-there decision, which is related to the previous idea using the similar set of the ontologies. Finally, they shifted the topic to different probability computation methods for belief-nets and thought of solutions to issues with limited data annotation.", "dialogue": "None: Yeah, yeah.\nSpeaker C: Hey, bus.\nSpeaker C: Yeah, it's a guy.\nNone: Yeah.\nNone: Yeah.\nNone: And just more to come.\nSpeaker C: And forever you get to.\nSpeaker C: I didn't look for air, but at the end.\nSpeaker D: And please turn off the microphone.\nSpeaker D: At the end, if you can remember, that's the most exciting light.\nNone: Great.\nNone: We'll move.\nNone: Make sure they're all on.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nNone: You know your record should be on, even if you're not talking like this.\nSpeaker C: Right.\nSpeaker D: Yeah.\nSpeaker D: Got anything good?\nSpeaker D: Good.\nSpeaker D: Well, Keaton, today we get to sort of do our stuff and drumming that.\nSpeaker F: Yeah.\nSpeaker F: That's all.\nSpeaker F: Yeah.\nSpeaker C: OK, got it.\nSpeaker C: Thank you putting it on backwards.\nSpeaker C: I think the hinge bent.\nSpeaker C: Anyway, there you go.\nSpeaker F: All right.\nSpeaker F: I should read all of these numbers.\nSpeaker F: 8551 14 8198 807 966 85 7691 9493 4771 246 00 7533 3226 3880 7491 9493 954896 097 347 163 56161 9731\nSpeaker D: Hey, transcript L87 864 664 171 803 256 803 225 202 311 611 7 424 058 8 4 2 1 8 3 6 9 1 7 5 2 1 4 3 5 9 0 9 7 6 1 6 9 6 6 0 281 209 8 1 3 0 4 5 or 7 8\nSpeaker B: Transcript L85 57367 1 348 4 545 8 5 9 9 6 5 4 7 3 2 0 1 8 3 7 8 2 2 1 9 0 9 2 0 6 6 7 7 2 6 4 6 1 6 4 4 4 3 9 7 5 4 1 1 0 5 3 3 2 6 4 3 8 4 8 5 1 6 7 5 5 5 7 8 4 4 4 4 4 4 6 7 0 9 0 4 5 9 8 4 1 1 6 8 0 1 1 1 1 4 1\nNone: 8 4 4 7 6 7 1 8 9 7 8 8 8 8 4 4 0 9 3 0 9 9 9 9 9 9 9 Transcript L83 9247 6 9 4 5 5 5 7 8 7 0 9 9 8 8 2 9 4 5 3 0 9 7 1 5 4 1 4 2 7 6 6 8 6 1 8 7 6 8 0 7 2 7 4 2 4 9 9 4 6 9 6 7 7 0 7 4 2 9 4 2 1 4 2 2 5 5 0 0 6 4 6 82 8 2 4 5 4 6\nSpeaker B: No 2 So my idea for today, and we can decide that that isn't the right thing to do, was to spend at least part of the time trying to build the influence links, which sets things are relevant to which decisions. And actually I had specific gestion to start first with the path ones, the database ones being in some sense less interesting to us although probably have to be done. And so to do that, so there's, and the idea was we were going to do two things.\nSpeaker B: Oh right. Well, we're going to do two things, one of which is just lay out the influence structure of what we think influences what. And then as a separate but related task, particularly buscrow and I were going to try to decide what kinds of belief nodes are needed in order to do what we need to do once. So we should sort of have all of the basic design of what influences what done before we decide exactly how to compute it. So did you get a chance to look at all? Yeah, look at some of that stuff. Great. Okay, so let's start with the belief that's general influence stuff and then we'll also, at some point,\nSpeaker D: break and talk about the techie stuff. Well, I think one could go, I think we can do this because everything. First of all, this I added, I knew from sort of basically this has to be there, right? Oh, are you going to go there or not? Yeah. So one, not transverse the castle, the decision is that's the first one to go there. Right. True. Does have to\nSpeaker B: be there. And I'm sure we'll find more as we go then. So go there in the first place\nSpeaker D: or not just definitely one of the basic ones we can start with that interesting effort. Is this basically true of false or maybe we'll get? Well, go there. So there is this question about we actually get just the entities, right? For each one here. Yeah, when we're done.\nSpeaker B: The reason it might not be true or false is that we did have this idea of when. So it's you know, current for next and so forth and so on or not at all, right? And so that's a decision would be do we want that? So you could do two different things you could do.\nSpeaker B: You could have all those values for go there or you could have go there be binary and given that you're going there. So let's say. I mean, it seems that you could. It seems\nSpeaker F: that those things would be logically independent like you would want to have them separate a binary go there and then the possibility is hard to go there because. Okay. Let's start that way because you know, it might be easy to figure out that this person is going to need more film eventually from their utterance, but it's much more complex to query when\nSpeaker D: would be the most appropriate time. Okay. And so I've tried to come up with some initial things one could serve. So who is the user? Everything that has user, it comes with the user model. Everything that has situation comes with the situation model. It should be here.\nSpeaker D: But when it comes to sort of writing down when you do these things, is it here? You sort of have to write the values this can take. Right? And here I was really, and sometimes I was really sort of standing in front of a wall feeling very stupid because in this case, it's pretty simple, but as we will see the other ones. For example, if it's a running budget, so what are the discrete values of a running budget? So maybe my understanding goes to a power. Right here that this is something a number that keeps on changing, but okay. This is understandable. Yes. So here. You've seen this before. I'll keep these\nSpeaker F: belief net things. No, but I think I'm following it. So here is the, that we had that the\nSpeaker D: user sponsored me influence the outcome of decisions. There we wanted to keep sort of a running\nSpeaker G: total of things. Is this like an humble represents how much money they have left to spend? Okay. Well, I mean, how is it different from user finance? The finance is sort of here\nSpeaker D: thought of as the financial policy person carries out in his life. He's actually cheap, average or spendy. Right. And I did become maybe a user. I don't know. I didn't want to write\nSpeaker B: that. But, or cheapness. Well, a user-thrift. Thrift is good. Right. There it is. So Keith, what's behind this is actually a program that will, once you fill all this in, actually solve your belief nets for you and stuff. So this is not just a display. This is actually a GUI to a simulator that will, if we tell it all the right things, will wind up with a functioning belief net at the other end. And it's so simple that I can use it. Wow,\nSpeaker D: that is simple. Okay. So here it was. Okay. I can think of people being cheap, average, or spendy, or we can even have a finer scale of the matter. It doesn't matter. It doesn't matter. Agreed there. But here, I wasn't sure what to write in. Let's go ahead. Well,\nSpeaker G: I mean, you've written in what seems to be required. Like what else do you want?\nSpeaker B: If that's permissible, then. Well, yeah. So here's what's permissible is that you can arrange so that the value of that is going to have to be updated. It's not a belief update, right? You took some actions, you spent money and stuff. So the update of that is going to have to be essentially external to the belief net. Right? And then what you're going to need is for the things that it influences. Well, let's see if it does influence anything.\nSpeaker B: And if it does influence anything, then you're going to need something that converts from the number here to something that's relevant to the decision there. So it could be right, there are quite different ranges that are relevant for different decisions or whatever, but for the moment, this is just a node that is conditioned externally and might influence\nSpeaker D: various things. Yeah, this is where, okay, anyways, let's get it. That's fine. Well, anyway.\nSpeaker G: Okay. And so the other thing is that every time that's updated, belief is up to be propagated, but then the question is, do you want to propagate, believe every single time it's updated\nSpeaker B: or when we need to? Yeah, that's a good question. And does it have a lazy mode? I don't remember.\nSpeaker G: Well, I mean, in Trini's thing, there was this thing, there was this option like propagate\nSpeaker B: differences, which suggests that it probably does. So what if the track hit down? And I think actually one of the items for the user home base should be essentially non-local. They're only there for the day and they don't have a place that they're staying.\nSpeaker D: I'll just accidentally erase this. I just had values here, such as, we had in our list, we had, is he staying in our hotel? Is he staying with friends? Yeah.\nSpeaker B: Okay. So it's clear where we are right now. So my suggestion is we just pick up.\nSpeaker B: Yeah, one of the, well, let's do the first one. Let's do the one that we already think we did.\nSpeaker B: So that would be endpoint.\nSpeaker G: So it's two or false? No, that's a BBA. So what's the same mode? Yeah.\nSpeaker G: And endpoint? Oh, that mode was, that's more of transportation? Yeah. Okay. Also true of false. Yeah, he hasn't filled the bin yet.\nSpeaker D: No, probably nothing done here. Oh, we're just doing the other ones. Okay. Make sense. Okay. So this was EVA.\nSpeaker D: Maybe we can think of more things across. Yeah. Okay. I'm Rob. I'm the merge.\nSpeaker C: Yeah. That's just the point. But it would be an endpoint if you were crossing over it.\nSpeaker B: Yeah, it would be for a given segment. You know, you first go to the town square.\nSpeaker F: Well, I mean, if you go to, you know, if you go to Prague or whatever, one of your key points that you have to do is cross the Charles Bridge.\nSpeaker F: It doesn't really matter which way you cross, which where you end up at the end, but the part that good part is walking over it.\nSpeaker B: That's subtle, but true. Anyway. So let's just leave it with three for now.\nSpeaker B: And let's see if we can get it linked up just to get ourselves started.\nSpeaker B: Okay. You'll see it. You'll see something comes up immediately. That reason I want to do this.\nSpeaker D: Well, the user was right. Definitely more likely to write for if he's a local, more likely to view if he's a tourist.\nSpeaker D: And then of course, we had the fact that given the fact that he's thrifty and there will be admission that we get all these cross.\nSpeaker B: We did, but the three things that contributed this, in fact, the other two aren't up there. So one was the ontology.\nSpeaker D: What type of building is it?\nSpeaker B: Yeah. And the third thing we talked about was something from the discourse.\nSpeaker D: What he has mentioned before.\nSpeaker B: Okay. So this is right. So what we seem to need here.\nSpeaker B: And this is why it starts getting into the technical stuff. The way we had been designing this, there were three intermediate nodes, which were the endpoint decision as seen from the user model as seen from the ontology and as seen from the discourse.\nSpeaker B: So each of those, the way we had it designed, now we can change the design. The design we had was there was a decision with the same three outcomes based on those three separate considerations.\nSpeaker B: So if we wanted to do that, would have to put in three intermediate nodes.\nSpeaker B: And then what you and I have to talk about is, okay, if we're doing that and they get combined somehow, how do they get combined? But they're not only going to be more things to worry about.\nSpeaker D: So this was just it for this one.\nSpeaker D: So let's, in our, in John was sort of pictogram everything that could contribute to whether person wants to interview or approach some.\nSpeaker B: Oh, it was called mode. So this is, this is mode here means the same as endpoint.\nSpeaker B: Okay, why don't we change that?\nSpeaker B: Yeah, but that was actually, yeah, unfortunately, that was a, kind of an intermediate version. That's, I don't think what we would currently do.\nSpeaker C: Can I ask about slurred and angry? Absolutely. Like, they're either true or false and the, well, I see if the, if the person talking is angry or slurs their speech, it might be tired or, you know,\nSpeaker B: they're strong. And therefore, and, you know, possibly less likely to enter some, yeah, that seems to be, yeah, so, somebody else must do is, is get this down to what we think is actually likely to be a strong influence. But yeah, that was what he had in mind.\nSpeaker B: Let's think about this, this question of how do we want to handle the two separate things. One is, at least two.\nSpeaker B: One is, how do we want to handle the notion of the ontology? Now, what we talked about, and this is another technical thing, mascara, is, can we arrange so that, I think we can, so that the belief that itself has properties.\nSpeaker B: And the properties are filled in from ontology items. So let's take the case of the, this endpoint thing. The notion was that if you had a few key properties like, is this a tourist site, you know, some kind of landmark, is it a place of business?\nSpeaker B: Is it something you physically could enter? Okay, et cetera. So that there be certain properties that would fit into the decision node.\nSpeaker B: And then, again, as part of the outer control and conditioning of this thing, those would be set so that somehow, someone would find this word, look it up in the ontology, pull out these properties, put it into the belief net, and then the decision would flow.\nSpeaker D: Now, it seems to me that we've sort of embodied a lot embedded a lot of these things we had previously, in some of the other final decisions done here, for example, if we would know that this thing is exhibiting something.\nSpeaker D: If it's exhibiting itself, it is a landmark, and it would be more likely to be viewed. If it is exhibiting pictures or sculptures and stuff like this, then it's more likely to be entered.\nSpeaker B: I think that's completely right, and I think that's good. So what that says is that we might be able to take and in particular, so the ones we talked about were exhibiting and selling accessibility.\nSpeaker D: If it's closed, one probably won't enter. Or if it's not accessible to a tourist ever. The likelihood of that person actually wanting to enter it, given that he knows it.\nSpeaker B: So let me suggest this. Could you move those up about halfway? The ones that youth in selling, I guess.\nSpeaker B: So here's what it looks like to me, is that you want an intermediate structure which is essentially the order of, for this purpose, of selling, fixing, or servicing.\nSpeaker B: So that is for certain purposes, it becomes important, but for this kind of purpose, one of these places is quite like the other. So it seems right?\nSpeaker B: So it's basically just merging those for just the take of endpoint decision. So the idea would be that you might want to merge those three.\nSpeaker B: Well, it's here's where it gets a little tricky. From the belief net point of view it is. From another point of view, of course, it's important to know what it's selling or servicing and so forth.\nSpeaker B: So for this decision, it's just a true or false. And in this is a case where the order seems just what you want. If any of those things is true, then it's the kind of place that you are more likely to enter.\nSpeaker B: So you just want to have them all pointing to a, yeah, so let's do that.\nSpeaker B: No, no, no, no, no, an intermediate node. That's the part of the idea.\nSpeaker D: Is that the object type node?\nSpeaker D: Is it the kind of object that sells fixes or servicing?\nSpeaker B: Open up object type and let's see what its values are.\nSpeaker B: So it's not so far. Oh, well, first of all, it's not objects we call them entities, right?\nSpeaker D: We have sort of the, let's say put commercial.\nSpeaker D: I was just going to commercial action inside.\nSpeaker B: Let's do commercial landmark and. What was the. Well, accessible. I think it's different because that's temporary. That varies temporarily. Whereas this is a.\nSpeaker C: What would a hotel fall under?\nSpeaker B: I would call that a service, but, but I don't know. Well, I mean in terms of entity type.\nSpeaker B: It's a, well, it's, I would again, and for this purpose, I think it's commercial.\nSpeaker B: Some place you want to go in to do some kind of business.\nSpeaker G: What is the underscore T at the end of each of those things?\nSpeaker D: Things. So places that service things.\nSpeaker D: So things. Okay. Or fix things. This is an exhibit things.\nSpeaker G: That also points to entity type, I guess.\nSpeaker F: So we're deriving. This, this feature of whether the main action at this place happens inside or outside or what we're deriving that from what kind of activity is done there.\nSpeaker F: Couldn't you have it as just a primitive feature of the. Well, you could. That's a choice.\nSpeaker F: So, I mean, it seems like that's much more reliable because you could have outdoor places that sell things and indoor places.\nSpeaker B: Yeah, the problem with it is that it's sort of putting in a feature just for one decision.\nSpeaker B: Now, we may wind up having to do that. That's an, at a metal level, that's what we're going to have to sort out.\nSpeaker B: So, you know, what does this look like? What are what are intermediate things that are worth computing?\nSpeaker B: So, what are the features we need in order to make all these decisions? And what's the best way to organize this so that it's clean and consistent and all that sort of stuff?\nSpeaker F: I'm just thinking about how people, human beings who know about places and places to go and so on would store this.\nSpeaker F: And it would probably, you wouldn't just sort of remember that they sell stuff and then deduce from that.\nSpeaker D: Well, it must be going on. I think entity maybe should be regard as a vector of several possible things that can either do sell things, fix things, service things, fix things.\nSpeaker D: It can be a landmark at the same time as doing these things. It's not either or.\nSpeaker D: It certainly best can be a hotel and a famous site.\nSpeaker D: Many people might think it can be generally a landmark and be accessible, i.e. a castle or can be a landmark and not accessible of some statue.\nSpeaker B: Anyway, so let me suggest to do something else, which is to get rid of that long link between the user and the endpoint.\nSpeaker B: No, no, I don't want the link there at all.\nSpeaker B: Because what we're going to want is an intermediate thing, which is the endpoint decision based on the user model.\nSpeaker B: So what we talked about is three separate endpoint decisions. So let's make a new node.\nSpeaker C: Just as a suggestion, maybe you could save as to keep your old one nice and clean and so you can mess with this one.\nSpeaker D: The old one was not there. Not that important, I think.\nSpeaker C: Okay, well, not a big deal then.\nSpeaker D: Let's do it then.\nSpeaker C: Isn't there a save as inside of Java page?\nSpeaker D: Okay.\nSpeaker D: This was user.\nSpeaker B: Let's put it this way. Let's do endpoint under bar U. End point. It's the endpoint. Let's say under bar U. So that's the endpoint decision as seen through the.\nSpeaker B: As related from the user model. Right. So let's actually. Yeah, so you can link that up to the.\nSpeaker B: Should I rename this? Yeah, so that I guess it's endpoint.\nSpeaker D: Under underscore e underscore e for entity and we may change all this, but. Right. And.\nSpeaker B: Actually, I guess the easiest thing would move move the endpoint. Well, go ahead. Just do whatever.\nSpeaker D: Isn't this possible?\nSpeaker C: Yeah, thank you. You have to be in move mode before.\nSpeaker A: Right.\nSpeaker B: So now we look for user related things that. Yeah. And maybe maybe it's just one who is the user. I don't know. Maybe maybe there's more.\nSpeaker D: Well, if he's using a car right now.\nSpeaker D: What was that people with Harry drove the car into the coffee? Never mind.\nSpeaker B: Anyway, this is crude. Now, but so the but then the question is.\nSpeaker B: So and we assume that some of these properties would come in directly through an ontology, but then we had this third idea of input from the discourse.\nSpeaker D: Well, that's finished. Sure. Okay. The user interests.\nSpeaker B: User thrift, the user budget. Well, maybe I again, I'd go, okay, put him in, but what we're going to want to do is actually.\nSpeaker D: This was one of my problems. We have the user interest is a vector of 500 values. So.\nSpeaker D: Oh, you mean this problem? No, not levels of interest, but things you can be interested in.\nSpeaker D: Somebody else has built this user model. So it's like a vector of 500 ones or zeros.\nSpeaker G: Like for each thing, I see.\nSpeaker B: Okay. So you could. So here, let me give you two ways to handle that. All right. One is.\nSpeaker B: You could ignore it. But the other thing you could do is have a, this is, you give me the flavor of the way it's going to.\nSpeaker B: You can have a node that's that was a measure of the match between the objects feature, you know, the match between the object, the entity.\nSpeaker B: I'm sorry. And the user. So you could have a fit node. And again, that would have to be computed by someone else.\nSpeaker B: But so that.\nSpeaker D: Just as a mental note.\nSpeaker B: Yeah, that's all.\nSpeaker D: And should we say that this interests affects the likelihood of entering?\nSpeaker E: Yeah. Which could.\nSpeaker D: And also, if it's an expensive place to enter, this may also.\nSpeaker F: Okay. User schedule.\nSpeaker F: Time to go in and climb all the way to the top of the kernel DOM. Or do I just have to.\nSpeaker C: It seems like everything in the user model affects.\nSpeaker B: Well, that's what we don't want to do. See that? Yeah. Because then we get into huge combinatorics and stuff like that.\nSpeaker C: Because if the user is tired, the user state, right, it would affect stuff.\nSpeaker C: But I can't see why anything, everything in the model wouldn't be.\nSpeaker B: Well, that's what we say we can't do that. So we're going to have to.\nSpeaker B: This is a good discussion. We're going to have to somehow figure out.\nSpeaker B: So if there is some general notion of, for example, the relation to the time to do this to the amount of time the guy has.\nSpeaker B: Or something like that is the compatibility with his current state.\nSpeaker B: So that's what you'd have to do. You'd have to get it down to something which.\nSpeaker B: It's the same thing. So it could be compatibility with his current state, which would include the money in this time and his energy.\nSpeaker C: Yeah, it just seems like it pushed the problem back a level.\nSpeaker G: It does. No, but it's more than that. The more sort of you break it up, the more that it.\nSpeaker G: Because if you have everything pointing to one node, it's like exponential.\nSpeaker G: Whereas if you like you break it up more and more, it's not exponential anymore.\nSpeaker B: So the two advantages. That's the one technical one.\nSpeaker C: It's a great solution. It gets you subgrouping. Subgrouping basically.\nSpeaker C: Yeah. So basically make it more tree like going backwards.\nSpeaker B: Right. Right. But there's two advantages. One is the technical one that you don't wind up with such big exponential.\nSpeaker B: Cpt's the others can be it presumably can be used for multiple decisions.\nSpeaker B: So if you have this idea of the compatibility with requirements of an action to the state of the user.\nSpeaker B: You can well imagine that that was not only is it is it cleaner to compute it separately, but it could be that it's used in multiple places.\nSpeaker B: So in general, this is the design. This is a really designed problem.\nSpeaker B: Okay. You've got to send it all into set of decisions.\nSpeaker B: How do we do this?\nSpeaker D: What do I have under user state? Because I named that already something. Oh, that's tired fresh.\nSpeaker D: Maybe should be renamed into physical state.\nSpeaker B: Or if it's user fatigue, even.\nSpeaker D: That's what the G.\nSpeaker D: Whatever that we can make a user state.\nSpeaker B: What's what we were talking about is compatibility.\nSpeaker B: Or something. I don't know.\nSpeaker C: I guess the question is it's hard for me to imagine how everything wouldn't just contribute to user state again.\nSpeaker B: Or user compatibility.\nSpeaker B: The thing is that we had some things that don't.\nSpeaker D: The user interest and the user who the user is are completely apart from the fact whether he is tired.\nSpeaker C: The node we're creating right now is user compatibility to the current action.\nSpeaker C: Everything in the user model would contribute to whether or not the user was compatible with something.\nSpeaker B: Maybe not. That's the issue.\nSpeaker B: Even if it was true in some abstract general sense, it might not be true in terms of the information we actually had and can make use of.\nSpeaker B: Anyway, we're going to have to find some way to get this sufficiently simple to make it feasible.\nSpeaker D: If we split it up again and if we look at the end point again, we said that for each of these things there are certain preconditions.\nSpeaker D: So you can only enter a place if you are not too tired to do so and also have the money to do so because of something.\nSpeaker D: So if you can afford it and perform it, preconditions.\nSpeaker D: Viewing usually is cheap.\nSpeaker B: With the way we're defining it, I think.\nSpeaker D: Viewing without definition of view is free.\nSpeaker F: What about the grand canyon?\nSpeaker B: I mean, never mind. No, you have to enter the park. Almost by definition, paying involves entering, going through some.\nSpeaker B: So let me suggest we switch to another one. I mean, clearly it's more work to be done on this, but I think it's going to be more instructive to think about other decisions that we need to make in path land.\nSpeaker B: And what they're going to look like.\nSpeaker C: So you can save this one as and open up the old one, right?\nSpeaker C: And then everything would be clean. So you could do it again.\nSpeaker B: I think we're saving this one, but I think I'd like to keep this one because I want to see if we're going to reuse any of this stuff.\nSpeaker D: What makes it?\nSpeaker B: Well, you tell me. So in terms of the planner, what's a good one to do?\nSpeaker D: Go there or not. I think it's a good one. It's a very basic one.\nSpeaker D: So what makes things more likely that?\nSpeaker B: Well, the first thing is getting back to the thing we left out of the other is the actual discourse.\nSpeaker B: So Keith, this is going to get into your world because we're going to want to know which constructions indicate various of these properties.\nSpeaker B: And so I don't yet know how to do this. I guess we're going to wind up pulling out discourse properties like we have object properties.\nSpeaker B: And we don't know what they are yet. So that the go there decision will have a node from discourse.\nSpeaker B: And I guess why don't we just stick a discourse thing up there to be as a placeholder for?\nSpeaker D: We also had discourse features, of course, for the endpoint.\nSpeaker D: Of course.\nSpeaker D: Identify that.\nSpeaker D: And so again, that's completely correct. We have the user model of the situation model here.\nSpeaker D: We don't have the discourse model here yet. We're just in the way. We don't have the ontology here.\nSpeaker B: Well, the ontology we sort of said we would pull these various kinds of properties from the ontology, like exhibiting, selling, and so forth.\nSpeaker B: So in some sense, it's there. But the discourse we don't have represented at all yet.\nSpeaker D: Let's be specific for a second here.\nSpeaker D: And we probably would have something like a discourse for endpoint.\nSpeaker B: But if we do, we don't have the three values. You don't have the EVA values.\nSpeaker D: If we have it, just for starters. And here discourse, for go there, probably is true and false, let's say.\nSpeaker B: That's what we talked about.\nSpeaker D: Well, I think, I mean, looking at the little data that we have, so people say, how do I get to the castle?\nSpeaker D: And this usually means they want to go there. So this should sort of push it.\nSpeaker D: Right. In one direction, however, people also sometimes say, how do I get there in order to find out how to get there without wanting to go there?\nSpeaker D: And sometimes people say, where is it? Because they want to know where it is. But in most cases, they're from...\nSpeaker B: Yeah, but that doesn't change the fact that you want these two values.\nSpeaker D: Yeah, true.\nSpeaker D: So this is sort of some external thing that takes all the discourse stuff and then says, here it's either, yep, A or A.\nSpeaker E: Okay.\nSpeaker B: And there'll be a user go there.\nSpeaker G: And maybe that's all I don't know. A situation go there, I mean, because whether it's open or not.\nSpeaker G: Okay, good. That definitely is.\nSpeaker G: But that... Now that kind of... What's the word...\nSpeaker G: That interacts with the EVA thing. If they just want to view it, then it's fine to go there when it's closed.\nSpeaker G: Whereas if they want to...\nSpeaker B: Right.\nSpeaker B: Right. So that's where it starts getting to be potentially more interesting.\nSpeaker B: So what Baskar says, which is completely right, is if you know that they're only going to view it, then it doesn't matter whether it's closed or not in terms of...\nSpeaker B: The time of day.\nSpeaker B: You know, whether you want to go there.\nSpeaker C: It does matter though if there's like a strike or a riot or something.\nSpeaker B: Absolutely. There are other situational things that do matter.\nSpeaker G: So, yeah, that's just that just having one situational node may not be enough because that node by itself wouldn't distinguish.\nSpeaker B: Well, it can have various values.\nSpeaker B: But you're right, it might not be enough.\nSpeaker G: Yeah, I mean, see, I'm thinking that any node that begins to go there is either going to be true or false.\nSpeaker F: Ah, I see. That could be.\nSpeaker F: So that node, I mean, the go there is S node would just be fed by separate ones for...\nSpeaker F: There's different things.\nSpeaker G: Next situation traffic and so on.\nSpeaker B: So now the other thing that Baskar pointed out is what this says is that there should be a link.\nSpeaker B: And this is where things are going to get very messy.\nSpeaker B: From the endpoint decision, maybe the final...\nSpeaker B: I guess the very bottom end point decision to the go there node.\nSpeaker B: And don't worry about layout. I mean, we'll go nuts.\nSpeaker G: But maybe we could have an intermediate node that just the endpoint and the go there S node sort of fed into.\nSpeaker G: Could be.\nSpeaker G: Right? Because that's what we... I mean, that's where this is.\nSpeaker B: Or actually the endpoint node could feed into the go there S.\nSpeaker B: Yeah, that's right. So the endpoint node, make that up to the go there there.\nSpeaker B: And again, we'll do layout at some point. But something like that.\nSpeaker B: Now it's going to be important not to have loops by the way.\nSpeaker B: It's really important to be in the belief world not to have loops.\nSpeaker D: How long does it take you to compete?\nSpeaker B: No, it's much worse than that.\nSpeaker B: Things don't converge. It's not well defined if they're loops.\nSpeaker B: You just... You have to...\nSpeaker B: There are all sorts of ways of breaking it up so that there isn't...\nSpeaker D: Okay.\nSpeaker D: This is not a loop yet. I'm just saying we...\nSpeaker G: But the good thing is we could have loopy belief propagation, which we all love.\nSpeaker B: Okay. So anyway, so that's another decision. What's another decision you like?\nSpeaker D: Okay. These have no parents yet, but I guess it sort of doesn't matter.\nSpeaker D: Right?\nSpeaker B: Well, the idea is that the go there you go comes from something about the user.\nSpeaker B: This comes from something about the situation. And the...\nSpeaker B: The discourse is a mystery.\nSpeaker B: Should I just make some?\nSpeaker D: Sure, if you want.\nSpeaker D: If there's parking, maybe...\nSpeaker E: Okay.\nSpeaker D: Okay. And if he has seen it already or not...\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: This discourse is something that sort of...\nSpeaker D: Should we make a key note here?\nSpeaker B: Sure.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker B: Then also the discourse endpoint.\nSpeaker B: I guess it points up to you if you want to make it consistent.\nSpeaker F: Actually, is this the right way to have it where go there from the user and go there from the situation?\nSpeaker F: Just sort of don't know about each other, but they both need to go their decision.\nSpeaker F: I think so.\nSpeaker F: I mean...\nSpeaker F: Okay.\nSpeaker F: And maybe not.\nSpeaker F: It still allows for the possibility of the user model affecting our decision about whether a strike is the sort of thing which is going to keep you going.\nSpeaker F: Right.\nSpeaker F: All that kind of decision making happens at the go there node.\nSpeaker B: If you needed to do that.\nSpeaker F: If you needed to do that.\nSpeaker F: Yeah.\nSpeaker F: But okay.\nSpeaker F: I was just thinking...\nSpeaker F: I guess maybe I'm conflating that user node with possible asking of the user.\nSpeaker F: Is this strike on?\nSpeaker F: Does that affect whether or not...\nSpeaker F: Good point.\nSpeaker F: I don't know how we're going to...\nSpeaker F: Yes, that might not come out of the user model, but directly out of direction.\nSpeaker B: I get the yes, my current...\nSpeaker B: Don't you know, yes, yes, enough.\nSpeaker B: My current idea on that would be that each of these decision nodes has questions associated with it.\nSpeaker B: And the question wouldn't itself be one of these conditional things.\nSpeaker B: Given that you know there's a strike do you still want to go.\nSpeaker B: But if you told him a bunch of stuff, then you would ask him do you want to go.\nSpeaker B: But I think trying to formulate the conditional question, that sounds too much.\nSpeaker B: Right, sure.\nSpeaker B: To me.\nSpeaker B: Let's stay with us a minute because I want to do a little bit of organization before we get more in the details.\nSpeaker B: The organization is going to be that the flavor of what's going on is going to be that...\nSpeaker B: As we sort of do in detail, indeed Keith is going to worry about the various constructions that people might use.\nSpeaker B: And John O. has committed himself to being the parser wizard.\nSpeaker B: So what's going to happen is that eventually, like by the time he graduates, okay, there'll be some sort of system which is able to take the discourse in context and have outputs that can feed the rest of the belief net.\nSpeaker B: I assume everybody knows that I just want to get closer that that'll be the game then.\nSpeaker B: So the semantics that you'll get out of the discourse will be values that go into the various discourse-based decision notes.\nSpeaker B: And now some of those will get fancier like mode of transportation and stuff.\nSpeaker B: So it isn't by any means necessarily a simple thing that you want out.\nSpeaker B: So if there is, and there is mode of transportation.\nSpeaker D: There's sort of also a split if you blow this up and look at it in more detail, there is something that comes from the discourse in terms of what was actually just set, what's the utterance giving us, and then what's the discourse history.\nSpeaker B: Yeah, well, we'll have to decide how much of where that goes.\nSpeaker B: And it's not clear yet.\nSpeaker B: It could be those are two separate things. It could be that the discourse gadget itself integrates which would be my guess.\nSpeaker B: You'd have to do, see in order to do reference and stuff like that, you've got to have both the current discourse and the context to say, I want to go back there. Why don't, what does that mean?\nSpeaker B: All right.\nSpeaker D: So, but is this picture that's emerging here?\nSpeaker D: Just my wish that you have noticed already for symmetry or is it that we get for each, each decision on the very bottom, we sort of get the sub e sub d sub u and maybe a sub o for autonomy.\nSpeaker D: Meta node.\nSpeaker B: I don't know.\nSpeaker B: But it might just, it could be, this is getting into the thing I want to talk about next, which is, if that's true, how do we want to combine those or when it's true.\nSpeaker D: But this would be nice though that we only have at most four at the moment.\nSpeaker D: Yeah.\nSpeaker D: And then, eros going to each of the bottom decisions. Yeah. And four, we can handle.\nSpeaker B: No.\nSpeaker B: Well, if it's four things and each of them has four values, it turns out to be a big CPT. It's not completely impossible.\nSpeaker B: It's not beyond what the system could solve, but it's probably beyond what we could actually write down or learn.\nSpeaker B: But, you know, it's four to the fourth.\nSpeaker A: It's pretty big.\nSpeaker C: 256, as I was saying.\nSpeaker B: Yeah, I mean, it's, and I don't think it's going to, I don't think it will get worse than that, by the way.\nSpeaker D: So that's a good, didn't we decide all of these are true or false?\nSpeaker B: Forgot there, but not for the other ones, three values for endpoint already.\nSpeaker G: Yeah, I mean, you'll need actually three to the five because, well, I mean, if it has four inputs and then it itself has three values.\nSpeaker G: I mean, it can get big fast.\nSpeaker D: For endpoint? No, it's the EVA.\nSpeaker D: Yeah, down here, but this one only has two.\nSpeaker G: No, it's still has three.\nSpeaker B: It will still have three.\nSpeaker B: From each point of view, you're making the same decision.\nSpeaker B: So from the point of view of the entity.\nSpeaker G: And also, I mean, there's other places where, like, for example, consider endpoint U.\nSpeaker G: It has inputs coming from user budget, user-driven.\nSpeaker B: So even, those are not necessarily binary.\nSpeaker B: We're going to have to use some care in the knowledge engineering to not have this explode.\nSpeaker B: And in fact, I think it doesn't, in the sense that, actually, with the underlying semantics and stuff, I think it isn't like you have 256 different ways of thinking about whether this user wants to go to some place.\nSpeaker B: So we just have to figure out what the regularities are and code them.\nSpeaker B: But what I was going to suggest next is maybe we want to work on this a little longer.\nSpeaker B: But I do want to also talk about the thing that we started into now of, well, it's all fine to say all these arrows come into the same place.\nSpeaker B: What rule of combination is used there?\nSpeaker B: So these things all affected. How do they affect it?\nSpeaker B: And belief nets have their own beliefs about what are good ways to do that.\nSpeaker B: So it's clear enough what the issue is.\nSpeaker B: So we want to switch to that. Now we want to do some more of this.\nSpeaker D: Basically, we just need to sort of get some closure on this.\nSpeaker D: Figure out how we're going to get this picture sort of completely messy.\nSpeaker B: Well, here's one of the things that I don't know how easy to just do this in the interface.\nSpeaker B: But it would be great if you could actually just display at a given time all the things that you pick up, you click on endpoint, okay, and everything else fades.\nSpeaker B: And you just see the links that are relevant to that.\nSpeaker B: And do anybody remember the GUI on this?\nSpeaker C: I would only say that the way to do that would be to open up or make, you know, and many belief nets and then open them every time you wanted to look at a different one versus that.\nSpeaker D: It's very easy to do it in HTML.\nSpeaker G: Yeah, but HTML.\nSpeaker D: Yeah, each of these things, each of the end leaflets be a patron and you click on the thing and then.\nSpeaker B: Yeah, anyway, so it's clear that even with this, if we put in all the arrows, nobody's going to be able to read the diagram.\nSpeaker B: Yeah.\nSpeaker B: All right, so we have to figure out some basically display hack or something to do is because, anyway, let me suggest that's not a first order consideration.\nSpeaker B: We have two first order considerations, which is what are the influences A and B, how do they get combined mathematically?\nSpeaker C: How do we display them as an issue, but?\nSpeaker C: I don't, yeah, I just don't think this has been designed to support something like that.\nSpeaker G: Yeah, I mean, it might soon, if this is going to be used in a serious way like JavaBase, then it might soon be necessary to start modifying it.\nSpeaker G: Right.\nSpeaker B: So, that's what we're going to do with our purposes.\nSpeaker B: Yeah, and that seems like a perfectly feasible thing to get into, but we have to know what we want first.\nSpeaker B: So, there's a little bit about decision nodes and what the choices might be for these.\nSpeaker C: You can technically wear that as you're talking.\nSpeaker C: Yeah.\nSpeaker E: I guess this board works fine.\nSpeaker G: So, recall the basic problem, which is that you have a belief net and you have a lot of different nodes, all contributing to one node.\nSpeaker G: Right. So, as you discussed, specifying this kind of thing is a big pain and it will take a long time to write down, because for example, if these have three possibilities each and this has three possibilities, then you have 243 possibilities, which is already a lot of numbers to write down.\nSpeaker G: So, what helps us in our situation is that these all have values in the same set.\nSpeaker G: Right. These are all like saying EV or A. Right.\nSpeaker G: So, it's not just a generalized situation.\nSpeaker G: I mean, basically, we want to just take a combination of, we want to view each of these as experts, each of them is making a decision based on some factors, and we want to sort of combine their decisions and create, you know.\nSpeaker G: Sort of a weighted combination.\nSpeaker D: The rover decision.\nSpeaker G: The what decision?\nSpeaker D: rover. All of their outputs combined.\nSpeaker G: Yeah. Yeah. So, the problem is to specify the, so the conditional probability of this given all those.\nSpeaker G: Right. That's the way belief nets are defined, like each node given its parent.\nSpeaker G: Right. So, that's what we want.\nSpeaker G: We want, for example, p of, let's call this guy y, and let's call these x1, x2, xn.\nSpeaker G: Right. So, we want probably that y equals, you know, for example, e, given that these guys are, as it's referred to this as like x, hat or something, like all of them, given that, for example, say this is, you know, a, v, a, e, or something.\nSpeaker G: Right.\nNone: Yep.\nSpeaker G: So, we would like to do this kind of combination.\nSpeaker B: All right. So, is that, I want to make sure everybody is with us before he goes on.\nSpeaker B: It's clear. Is it clear what he wants to compute?\nSpeaker G: So, right. So, basically, what we don't want to do is to, for every single combination of e and v and a, and every single letter e is given number, because that's obviously not desirable.\nSpeaker G: What we want to do is find some principle way of saying what each of these is, and we want it to be a valid probability distribution.\nSpeaker G: So, we want it to add up to one, right.\nSpeaker G: So, those are the two things that we need.\nSpeaker G: So, what, I guess what Jerry suggested earlier was basically that we, you know, view these guys' voting, and we just take the, we essentially take averages, right.\nSpeaker G: So, for example, here are two people of word for a.\nSpeaker G: One is word for v and one is word for e.\nSpeaker G: So, we could say that the probabilities are, you know, probably being e is 1 over 4, because one person word for e are 4.\nSpeaker G: And similarly, probably if, so this is probability of e, and then probably of a given all that is 2 out of 4.\nSpeaker G: And probably if v is 1 out of 4.\nSpeaker G: Right. So, that's, step, that's the, yeah, that's the basic thing.\nSpeaker D: Now, is that all the way?\nSpeaker D: So, step 1 outcome, that's, what?\nSpeaker D: If x, x 1 voted for a, x 2 voted for v, and so forth.\nSpeaker B: Right. Yeah.\nSpeaker B: So, this is sum symmetry and equal weights and all those sort of things, which may or may not be a good assumption.\nSpeaker G: So, that, yeah.\nSpeaker G: So, step 2 is, right.\nSpeaker G: So, we've assumed equal weights, whereas it might turn out that, you know, some, we, that, for example, what the, the actual, the verbal content of what the person said, like what, what might be, somehow more important than the, x 1 matters more in the next, sure.\nSpeaker G: Sure. So, we don't want to, like, give them all equal weights.\nSpeaker G: So, currently we've been giving them all weight 1, 4th.\nSpeaker G: So, we could replace this by, uh, w 1, w 2, w 3, and w 4.\nSpeaker G: Right. And in order for this to be a valid probability distribution for each, um, x hat, we just need that the w sum to 1.\nSpeaker G: So, they can be, for example, you know, you could have 0.1, 0.3, 0.2, and 0.4.\nSpeaker G: See.\nSpeaker E: That's 1.\nSpeaker G: And that would be 1.\nSpeaker G: So, that, um, also seems to work fine.\nSpeaker C: And, um, so it just, just to make sure I understand this, so in this case, um, we would still compute the average.\nSpeaker C: So, the probability of e would be, uh, so in this case, the probability that y equals a would be, uh, w 1 times, or a, or, let's say, 1, 4, quarter times 0.1.\nSpeaker G: Not 1 quarter. So, these numbers have been replaced with 0.1, 0.3, 0.2, and 0.4.\nSpeaker G: So, you can view this as gone.\nSpeaker G: Okay.\nSpeaker B: So, in this case, it would be, probably, a value of a would be 0.3.\nNone: Probably.\nNone: If v would be 0.3, and the probability of v would be 0.4.\nSpeaker A: Okay.\nSpeaker G: Yeah.\nSpeaker G: Okay.\nSpeaker G: So, all right.\nSpeaker G: So, this is a step two.\nSpeaker G: So, the next possibility is that, um, we've given just a single weight to each expert, right?\nSpeaker G: Whereas it might be the case that, um, in certain situations, one of the experts is more reliable, and in certain situations, the other expert is more reliable.\nSpeaker G: So, the way this is handled is by what's called a mixture of experts.\nSpeaker G: So, what you can have is, you augment this diagram slightly.\nSpeaker G: So, you have a new thing called H. Okay.\nSpeaker G: This is a hidden variable.\nSpeaker G: And what this is, is, it gets as input from x1, x2, x3, and x4.\nSpeaker G: And what it does is it decides which of the experts is to be trusted in this particular situation.\nSpeaker G: Right?\nSpeaker G: And then, these guys all come here.\nSpeaker G: So, this is slightly more complicated.\nSpeaker G: So, what's going on is that, um, this H node looks at these four values of those guys.\nSpeaker G: And it decides, in, given these values, which of these, isn't likely to be more reliable or most reliable.\nSpeaker G: So, H produces some, you know, it produces a number, either one, two, three, or four, in our situation, right?\nSpeaker G: Now, this guy, he looks at the value of H, say it's two, and then he just selects the, uh, thing.\nSpeaker G: Yeah. So, that's one thing you can do.\nSpeaker G: So, here we go on and say, you can say, you can say, you can say, you can say, I guess, about it.\nSpeaker B: What is it? Because the other thing you can do is you, so this, in this case, a mixture of x3, or just a mixture of x4.\nSpeaker G: Right. So, you can have a mixture of that.\nSpeaker G: So, what you can do, if you need it, is to say that, in here, there's a fourth frame.\nSpeaker B: There's a fifth frame, okay, which is the eighth frame.\nSpeaker B: So, let's suppose the H says, in such, let me do alpha.\nSpeaker B: Okay. In situation alpha, that's the waiting vector.\nSpeaker B: But in situation beta, is the other waiting vector.\nSpeaker B: So, you still, you might need to wait, but the weights aren't always the same.\nSpeaker B: That this guy, uh, detects for you, that this is a situation in which you should wait as more or less.\nSpeaker F: So, the function of the thing that comes out of H is very different from the function of the other input.\nSpeaker F: Right. It's actually, it's driving how the other four are interpreting.\nSpeaker B: It's, it's conditioning how the other four are.\nSpeaker B: Yeah. So, in what people do in another literature is learn, don't stop.\nNone: So, you learn the weights, you can learn the expert, etc.\nSpeaker B: If you have not dated, then make perfect good sense to do that.\nSpeaker C: So, H passes a vector on to the next node?\nSpeaker C: It could, it could.\nSpeaker C: No, I know, it could, could just pass a vector of the weights is, yeah, it could.\nSpeaker B: No, not normally, it would just pass to the selected.\nSpeaker B: So, it could pass either...\nSpeaker F: Well, a vector with three zeroes and one one.\nSpeaker B: Well, it could, yeah, it could, yeah, it could do that.\nNone: Okay. Uh, you know, uh, uh, a binary max.\nSpeaker C: Oh, it's basically to tell the bottom node which one of the situations that's in?\nSpeaker C: Or which one of the weighting systems?\nSpeaker C: Right. So, I would just, if you wanted to do it, you just...\nSpeaker C: I mean, the way you do it, or the one, you could pass a weighting system, though, too.\nSpeaker B: I think what...\nSpeaker B: That's not, that's harder than to just say there are discrete numbers,\nNone: in this situation, and this says, you know, in your situation alpha. Um, because each have to have another input to tell it alpha, beta, whatever.\nSpeaker F: Or is the, that's determined by what the experts are saying, like the type of...\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker B: Well, this definitely could have other, it absolutely could have other inputs from something\nNone: that has, how they can directly do with these experts. And it's actually often done.\nSpeaker F: So, I mean, it just seems that, like, without that, that outside input, you've got a situation where, you know, like, if, if, uh, X1 says no, you know, a low value coming out of X1, or if X1 says no, then ignore X1, you know, I mean, that seems like...\nSpeaker G: Yeah, well, it could be things like if X2 and X3C, yes, then ignore X1 also.\nSpeaker F: Oh, okay.\nSpeaker F: Or X2 is greater than X1.\nSpeaker B: It could have imagined it all towards the perfect.\nSpeaker B: All right.\nSpeaker B: So, these, so this is, this is one collection of technology that we could bring to the back of the app.\nSpeaker C: The situations that H has, are they built into the net or...\nSpeaker C: Okay.\nSpeaker C: So, they could either be hand coded or learned or...\nSpeaker C: Yeah.\nNone: Okay.\nSpeaker C: Based on training data, okay.\nSpeaker B: And there were some, some, some, actually, Michael Jordan did a lot of nice work in the earlier, on,\nNone: the situation which you could actually train and stuff and stuff. I don't think we're going to be in a situation where, realistic, we're going to have enough data to actually train these things.\nNone: Any time soon.\nNone: But, it still thinks a nice idea of the architecture is a force.\nNone: So, you know, you say here's the sign and, and, and, and, and, and, update, it comes along.\nNone: You could, you could, you could, you could, you could, you could do that.\nNone: I still have some of the things that I'm very, very, very, very, very, very, and, I don't know.\nSpeaker D: So, this is two for the case that X1 born at 4A, it's the word, but, yes.\nSpeaker D: Now, X1 puts the potentially awesome board key.\nSpeaker B: That would be another sort of thing.\nSpeaker B: Absolutely.\nSpeaker C: So, you specify one of these things for every one of those possible situations.\nSpeaker B: No, because, no, because you've got a general rule.\nSpeaker B: Oh, yeah.\nSpeaker B: You see, the point is rather especially on the combination, you could, you could, you could, you know, you've got a general rule that says, let's just take the average.\nSpeaker B: It doesn't matter who specifies one, we're just going to count the number of votes for each of the real three.\nSpeaker B: So, you know, like an actual combative.\nNone: So, that, that was the first thing about the first thing.\nNone: Forget who voting, let's just count the number of votes.\nNone: You know, you think about that much, much, much better.\nSpeaker E: Okay, so, the whole problem is, yeah.\nSpeaker B: And so, what he's saying here is, well, maybe that's right, but then you, you still, you can still get by without specifying all the combinations by saying, well, there are a few different situations in which you want to do a different one.\nSpeaker B: Or, you know, a different one.\nSpeaker B: Now, we might be able to learn, that was probably already possible that we could learn waiting on, you know, this, these, these gubby's.\nSpeaker G: Well, I mean, to learn them, we need data, where are we going to get data?\nSpeaker B: Well, so that's what's going on with this.\nSpeaker G: Well, I mean, we need data about people's intentions, right?\nSpeaker B: Which is slightly tricky.\nSpeaker B: I'm not so sure. So, we're having these people come in and have these dialogues.\nSpeaker B: Right.\nSpeaker B: Okay, that's what I just do it.\nNone: Okay, far here.\nSpeaker B: So, so we're going to get them. So they have some data.\nSpeaker B: They've taken before.\nNone: And, really able.\nSpeaker G: But what's the data about like, are we able to get these nodes from the data?\nNone: I think so.\nSpeaker F: Like how thrifty the user is, don't do we have access to that?\nSpeaker B: No, but you don't need that. This is just combining, this is simple. This is a very important point.\nSpeaker B: All you're saying here is give it these considerations from the user, the text, you know, what he says, and the kind of object, which is the best predictor of either he's going to use one of these nodes.\nSpeaker B: It doesn't say, how is it that you make that decision based on user or for that matter on the input line?\nSpeaker B: So there's a whole issue of how the hell do you do that and what constructions tell you from the input that you think he wants to handle?\nSpeaker B: All this says it is give it, we think these things are independent, which I think they are.\nSpeaker B: How do you wait?\nSpeaker B: Can we create an update for that?\nSpeaker G: Yeah, but that's my question. Like how do we, I mean, how do we have data about something like endpoint sub-e or endpoint sub?\nSpeaker C: Well, this is based on this dialogue that we have, which one of the things that they said, whether it was the entity relations or whatever, was the thing that determined what mode it was, right?\nSpeaker B: So it could be, for example, that you should really wait the intent, what you got to dialogue very strongly, because he said I want to enter it.\nSpeaker B: So, for example, it could be that our dialogue processing is not all that good, and that you shouldn't give it a high way. We just aren't good enough yet to get it in line.\nSpeaker B: And it may be that in this model, what it could be is that the autology information should be weighted quite highly.\nSpeaker B: The autology plus the situation information is basically what determines it.\nSpeaker B: Because in the kind of thing you could answer your US situation where it's possible to enter, and if those things are true and you've talked about it, it would be actually one ender.\nNone: I don't know.\nSpeaker G: So this is what we want to learn.\nNone: So, maybe we could do it over time that you have to.\nSpeaker B: The more important thing is to have an architecture in which, if you have more information as the designer you can add it, and if you get an update in so that you could turn a learning crank on it, the architecture supports that.\nNone: So, I think the language...\nSpeaker D: You're learning hard days, and you're going to walk around for this system, try to learn all the parking situations, and realize the trimester's approach.\nSpeaker D: And then they connect you down that and then you go.\nNone: So, there are, yeah, there's this geography, part of the entire version, has people doing busy on things like this.\nSpeaker B: Anyway, so this is one of these. Another one we talked about is the destruction. So this says these are all empty.\nSpeaker B: Oh, this is one over here about exhibiting, selling, fixing, et cetera. So this was not a situation where you would expect some weighted factors if it's a place that exhibits or sell, or repairs, or whatever, if they're commercial selling.\nSpeaker B: So it'll be some of which you want to disjunction. There will be some of which you probably have to actually lay out the product with the, you know, actions really lay out to do the hopefully will be those, you can think of it as small.\nNone: But that's, you know, the actual decision-making.\nNone: So, I'm going to do it.\nSpeaker E: So, and is there, and is there another one which we, if it turned out that there was something to do with this contract, you know, that you always think that we could be familiarized with sure that that's fine.\nSpeaker G: I don't think, well, you have a, can you bring up the function thing?\nSpeaker G: Where is the thing that allows you to sort of, that's only added variables, isn't it? Oh, a function property is decided. I guess not.\nSpeaker G: Yeah, that's. Right. Okay. And it, so either it'll allow us to do everything which I think is unlikely. I think more likely it allows to do very few of these things.\nSpeaker G: And in that case, we'll have to just write up little things that allow you to create such CPDs on your own in the Java-based format.\nNone: So, I think, yeah, I think the one with, what, here's what's tricky about it. It is that the, this is not only in Swayberg, this is actually a system that's all of these methods. And you can't just drop one of these in and expect all of its solution algorithms, because, oh, good.\nSpeaker B: And if it isn't, if it isn't, the problem of putting it is, it's not something you want to do. But you have to get into the guts of propagation algorithms.\nNone: And that, that, I would think it's possible. I mean, what Bofi was saying is, you can always convert into a CPD. You can have one macro that takes whatever role you have.\nSpeaker G: And then the system is going to have to go through its standard. I was assuming that's what we always do, because, yeah, I was assuming that's what we'd always do. It's.\nSpeaker B: So, and then the other question is, is this display question? Maybe you can check that out. What we can, you can look at this before.\nSpeaker C: How is, what are the options? Highly, is there some way to just see parts of things? How, how is it going to do it?\nSpeaker C: Well, in terms of Java based, I think it's basically what you see is what you get. And I don't, yeah, I would be surprised if it supports anything more than what we have right here.\nNone: And it doesn't matter. I mean, it does have enough space. It seems that you can stretch it out. Oh, you can make it very big. So that's not the problem. The problem I was interested in was, we all know that we can just argue and discuss it.\nSpeaker D: We should discuss from each detail and future control of free up new stuff. And in this, we would consider it as your final decision. And as we ended up at time with this, it was not simpler.\nSpeaker D: because you're just\nNone: You know, the other people are so much different. You're the one who really does have the understanding of the new.\nNone: And what the planning system is going to need is...\nSpeaker B: So we've done this guys coming in Germany next week.\nSpeaker B: So we'll probably be meeting you all in the last week as focus on the next guy ahead.\nSpeaker B: So we're really looking at how he manages the new.\nNone: So we're going to be doing probably...\nNone: We probably are talking about smart phones.\nNone: Parts in general.\nNone: So throughout the end of the smart phone operating, I understand that the media is promoted as a striped keyboard.\nNone: And that's why it's quite a lot of the picture of the system.\nNone: It's not too action-finding and the issue.\nNone: So the...\nNone: I think we're going to have him here on Monday.\nNone: And see what we do.\nSpeaker D: We really want to talk to them about the parsing issues.\nSpeaker D: And maybe we can expect out of the parsing issues.\nSpeaker D: We don't understand.\nNone: And maybe a lot of us are thinking about future work.\nNone: We're not in the business.\nSpeaker D: Wouldn't that be a goal to try to persuade them to abandon the way they're parsed\nNone: and some things for people and all these discussion-based work in the future? One of the questions is that.\nNone: What I think we want to do is get the first set of results and then show them.\nNone: Yeah.\nSpeaker D: So, in the missionary work only after you've done the system on a miracle.\nSpeaker F: Yeah.\nSpeaker B: That's my view of the...\nSpeaker B: The mirror of the smart mirror.\nSpeaker F: So...\nSpeaker F: By the way, I'm just talking about that general end of things.\nSpeaker F: Is there going to be data soon from what people say when they're interacting with the system and so on?\nSpeaker F: Like, I mean, what kind of questions are we going to be asking?\nSpeaker F: Okay.\nSpeaker F: Yeah, yeah.\nNone: The new data is on the website.\nNone: Okay.\nNone: You can just move by a lot.\nNone: And I hope the pay is working on this.\nNone: But I haven't heard from...\nNone: We have someone who's supposedly gathering us 50 or so.\nNone: Okay, you mean?\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker F: Okay.\nNone: Okay.\nSpeaker F: I'm just wondering because in terms of, you know, I mean, the figure...\nSpeaker F: I was thinking about this figure that we talked about 50 constructions or whatever.\nSpeaker F: That's a whole lot of constructions.\nSpeaker F: You bet.\nSpeaker F: And, you know, I mean, one might be fairly pleased with getting a really good analysis of five, maybe 10 in a summer.\nSpeaker F: So, I mean, I know we're going for sort of a...\nSpeaker B: Rough and rate is...\nNone: Including the little trigum of the appraisal.\nNone: It's not...\nNone: We're not going to talk about 50...\nNone: Okay.\nNone: So, you want the composition.\nSpeaker F: So, in order to do 10 potential constructions, I agree that we do 5.\nSpeaker F: You're going to need phrasal constructions or the VPs smaller.\nSpeaker F: Okay.\nSpeaker F: I mean, I was talking about the...\nSpeaker F: You know, if you wanted to do it really into tail, and we don't really need all the detail for what we're doing right now.\nSpeaker B: But anyway, in terms of just narrowing that task, you know, which 50 do I do?\nSpeaker B: I want to see what people are using, so.\nSpeaker B: And, you know, they inspire me.\nNone: Yeah, I agree.\nNone: I agree.\nSpeaker E: Do we only want some blend of what people use and what you actually do?\nSpeaker E: Right, sure.\nNone: We're going to be able to do this.\nSpeaker D: We're probably not going to have to take this because it can't be...\nSpeaker D: Right.\nSpeaker D: Yeah, sure.\nSpeaker D: Did you raise for sure?\nSpeaker D: Were they spontaneous people?\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: Right.\nSpeaker F: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: So, in terms of this thing, I believe that I've never seen this.\nNone: Let's see.\nNone: You\n"}, {"meeting_id": "IS1008c", "summary": "This meeting focused on the conception of the functional design of the new remote control. Industrial Designer evaluated several components in the technical design of the product and gave his personal preferences, especially on the chip and the material to use in the construction. Then User Interface presented a general layout of the functionality, the design of which should follow two principles: simpleness and customizability. Marketing, in his turn, explained the current market trend based on the results of his questionnaires on user requirements. The group also discussed where the remote control would be manufactured by taking into account the price of the local labour force.", "dialogue": "Speaker A: Welcome back.\nSpeaker A: I hope you had a good lunch together.\nSpeaker A: This meeting the main agenda to discuss about the conceptual design meeting.\nSpeaker A: And the agenda will be the opening and that's the product manager or secretary that's me and the presentations from the Christine and Agnes and Pernamishri Ed.\nSpeaker A: Finally in this meeting we have to decide and we have to take a decision on the remote control concepts and the functional design.\nSpeaker A: So we have 40 minutes.\nSpeaker A: I think it's a little bit low but I hope we can finish it up.\nSpeaker A: So I'll hand out to the functional team to the Christine to discuss about the components concept.\nSpeaker C: Okay so if you could open the PowerPoint presentation I'm number 2.\nSpeaker C: So, can we put it in slide show mode?\nSpeaker C: Yes, please.\nSpeaker C: Thank you.\nSpeaker C: So we were looking specifically at the components, the following components, the case, the power supply, the means of communications with the television set.\nSpeaker C: And then since we had talked about using some sort of speech recognition you have to have a microphone, well no you don't actually have to have a microphone in the device but maybe you do have a way, it has to hear the speaker and so it could be in the television set, it could be in the device but somewhere you have to put the microphone and a way of making beeps or sounds so you can find it when it gets lost.\nSpeaker C: So the other thing that we, our method for going about this is we've looked at the historical record, what's worked, what hasn't.\nSpeaker C: And then we also wanted to evaluate some new materials and we contacted manufacturing for their input because of course we might come up and choose the material that then manufacturing didn't have the technologies capabilities to offer us.\nSpeaker C: So this is the approach that we took during our research.\nSpeaker C: So for the case we told, we were making a specific assumption that it would be curved in design.\nSpeaker C: Of course you know I wanted it to be expandable and shrinkable but that doesn't seem to be one of the choice, not an option we can really seriously explore.\nSpeaker C: So then we were thinking about rubber but unfortunately that's been eliminated because of the heat factor and there might be some problems with how it goes with the board.\nSpeaker C: And then plastic also has this problem of melting and it's brittle, it gets brittle after a while so still had titanium and wood available but unfortunately titanium has also been eliminated.\nSpeaker C: The people in manufacturing said that it couldn't make curved cases out of titanium although how Apple did it with the power book I'm not quite sure but nevertheless they've eliminated all of our options except wood.\nSpeaker C: So this is my finding and as she said it's an environmentally friendly material.\nSpeaker C: So we're currently proposing, we'll get to all my personal preferences in just a second.\nSpeaker C: So then there's this other matter of the chips and well you could use a simple design on the board but these simple chips but that's only works for the buttons.\nSpeaker C: You don't get very much intelligence with this simple one.\nSpeaker C: And then there was the regular which I regret that I've forgotten exactly why I'm eliminating that one.\nSpeaker C: The other option was this advanced chip on print and we found that it includes this infrared sender which remember the beam was that was an important component of finding the right chip.\nSpeaker C: And manufacturing has told us that they've recently developed a sensor and a speaker that would be integrated into this advanced chip on print.\nSpeaker C: So we're now jumping right to our personal preferences.\nSpeaker C: I'd really think we should use some of some really exotic woods like well you guys come from tropical countries so you can kind of think of some trees and some nice woods.\nSpeaker C: I think that people might really want to design their own cases.\nSpeaker C: You see they could do sort of a three dimensional design on the internet and then they could submit their orders kind of like you submit a custom car order you know and you can choose the color and the size of the wheels and the colors of the leather and things like that.\nSpeaker C: And then I think we should go with the solar cells as well as the microphone and speaker on the advanced chip.\nSpeaker C: So this is the findings of our research and my recommendations for the new remote control would be to have to be made out of wood.\nSpeaker C: Do you have any problems with that?\nSpeaker A: Can you go back one slide?\nSpeaker C: I'm not sure.\nSpeaker C: Well I know let's see.\nSpeaker C: Let's go back up here.\nSpeaker A: Yes.\nSpeaker A: Question.\nSpeaker A: What's mean exactly it wants a chip on print.\nSpeaker A: What's the meaning of that?\nSpeaker C: I think it's a multiple chip design and it's maybe printed onto the circuit board.\nSpeaker C: I could find out more about that before the next meeting.\nSpeaker A: Is it on a microprocessor based or?\nSpeaker C: I don't know but I'll find out more on our next meeting.\nSpeaker A: Okay that would be great.\nSpeaker A: So if you find out from the technology background okay so that would be good.\nSpeaker C: Sounds good.\nSpeaker B: Well why was the plastic eliminated as a possible material?\nSpeaker C: Because it gets brittle, cracks, we expect these remote controls to be around for several hundred years.\nSpeaker C: So good expression.\nSpeaker C: Good expression.\nSpeaker D: Good expression.\nSpeaker D: World after us.\nSpeaker C: I don't know speak for yourself.\nSpeaker B: I think with the window you'd run into the same types of problems.\nSpeaker B: So I mean it chips it if you drop it.\nSpeaker B: I'm not sure.\nSpeaker A: So you're not convinced.\nSpeaker A: Actually I'm ready to sell it.\nSpeaker D: You're what?\nSpeaker D: I'm ready to sell it.\nSpeaker B: If you use really good quality wood then it might work but you can't just use.\nSpeaker B: No you can sell oils with it.\nSpeaker D: The only wood you can use are the ones that are hard to explain.\nSpeaker D: Yeah exactly.\nSpeaker D: But there's some very pretty woods out there.\nSpeaker D: That's actually a very innovative idea.\nSpeaker C: Okay good.\nSpeaker C: I'm sorry I'm having a hard time keeping with control over my face.\nSpeaker D: But it's actually a very innovative and different idea that you can choose your color of wood, your type of wood.\nSpeaker D: Mistain.\nSpeaker D: I mean each person is going to have their personalized individualized speech recognition remote control in wood.\nSpeaker D: That's not on the market.\nSpeaker A: Yeah so it looks good.\nSpeaker A: The design, the functional design, what about you?\nSpeaker B: In terms of comments on this or in terms of my own.\nSpeaker B: Yes.\nSpeaker B: In terms of wow.\nSpeaker C: In terms of comments first.\nSpeaker C: She works in the cubicle next to me so she was already a little bit prepared for this.\nSpeaker C: Yeah.\nSpeaker C: Luckily Ed was not.\nSpeaker B: I think if we can get the quality materials then it shouldn't influence the design principles too much which you'll see with my presentation.\nSpeaker B: One thing we'd have to check those with the users whether, how quickly the novelty wears off of having.\nSpeaker C: Yeah you would want to have to have splinters in your hand while you're using it.\nSpeaker B: Yeah for example.\nSpeaker B: So I have to see how kid friendly it is and all that.\nSpeaker C: But actually if your dog gets a hold of it they can use it for teething.\nSpeaker D: I did that anyway with rubber and plastic so.\nSpeaker C: Yeah I do it with other materials as well.\nSpeaker D: Yeah.\nSpeaker A: I'll show them up.\nSpeaker A: Okay then let's move to Agni.\nSpeaker B: Sure.\nSpeaker D: I think so.\nSpeaker B: Yeah that's the one.\nSpeaker B: Yeah it's a very short presentation because I'm actually going to draw you the layout on the board so if you want to just go straight to the second slide.\nSpeaker B: Which basically shows sort of I took the ideas that we were talking about last time and tried to put that into the remote control so the things that you can actually see on it are the on off switch volume and channel control the menu access button ergonomic shape which I completely agree with Christine's idea to have it sort of molded so it's slightly more ergonomic and comfortable to hold than the standard very straight remote controls.\nSpeaker B: And actually the other thing with the wood if we take your customizing idea is that people can actually do sort of quasi measurements on their hand size if someone has larger hands.\nSpeaker B: Right my hand you have a wider remote control.\nSpeaker C: And yours for example.\nSpeaker B: So that's actually a really good idea for custom customizability.\nSpeaker B: One thing I thought might be kind of interesting is to put a flip screen on it just like you have on flip phones so that you don't have this case where someone sits on the remote control or accidentally puts their hand on it especially if you have little kids around.\nSpeaker B: They're not pressing the buttons while you're trying to watch a TV show and accidentally change a channel or turn it off.\nSpeaker B: And also you had issues with batteries running at so I thought maybe we could put a little battery life light on it that kind of goes dimmer and dimmer and dimmer as your battery starts to die.\nSpeaker B: But in terms of invisible features, audio and tactile feedback on button presses and like you said speech recognition.\nSpeaker B: So in terms of what this thing would actually look like, despite working in interface design I'm not the greatest artist in the world so you'll have to forgive me.\nSpeaker B: You'd have something like this with an on off switch fairly big sort of in the corner and by itself so you don't accidentally turn your TV off while you're trying to maneuver other buttons.\nSpeaker B: And then you have sort of one of those toggle displays for channels and volume sort of for surfing channels and then volume.\nSpeaker B: So the volume would be the up and down because volume goes up and down and then channels left to right.\nSpeaker B: And then here you'd have your sort of standard telephone-ish number pad.\nSpeaker B: And then on one side you would have an access to the menu on your TV and on the other side a way to turn off the voice control so that if the user doesn't want to use their voice they can just turn it off and you don't have them out control accidentally changing things on you.\nSpeaker B: So again you can have a little LCD light somewhere, the flip thing and have I forgotten anything?\nSpeaker B: I don't think so.\nSpeaker B: So as you can see it's a very very simple design which is one of the things I really wanted to keep.\nSpeaker B: It's keep it simple not have too many buttons not have too many functionalities thrown into it.\nSpeaker B: I think the design can pretty much carry over to everything although with the wood the flip screen might have to do something slightly different.\nSpeaker C: It'd be like a copper hinge.\nSpeaker B: But you also have to start watching after the wait because depending on how much the flip screen will add to the wait of the remote control you don't want it to start getting too heavy.\nSpeaker B: But that's the general layout with the general functionalities if you come up with something else.\nSpeaker B: As you can see there's still lots of space on the actual remote control and if you do it customizably you can make this thing fairly small or fairly large depending on personal preferences.\nSpeaker B: So that's pretty much all I had to say.\nSpeaker B: I mean everything else in terms of design issues.\nSpeaker B: I mean the centering of the keypad and the channel is just depending on where your thumb is.\nSpeaker B: And you tend to use the volume control and the browsing more than the actual number pad so that would be sort of in direct line of where your thumb goes when you're holding the remote control.\nSpeaker B: The number pad a little bit lower because it's useless frequently.\nSpeaker B: So once we decide exactly what we want then we can figure out the exact positioning.\nSpeaker B: But more or less I think it should go along those lines.\nSpeaker A: So what's your comments or simple design?\nSpeaker D: So what consumers want?\nSpeaker D: It's almost like Houston we have a product here.\nSpeaker D: Problem is obviously going to be cost.\nSpeaker D: I also have a very simple presentation.\nSpeaker D: From the marketing point you have to see what the consumers want.\nSpeaker D: I also have a copy to a different type of remote.\nSpeaker D: This should only be one in here.\nSpeaker D: But try and watch.\nNone: Sure.\nSpeaker D: It doesn't be a modified.\nSpeaker D: They're stealing our product.\nSpeaker D: We've been giving simple questionnaires and different areas because obviously we have to see what the consumers are looking for today because trends change very very quickly.\nSpeaker D: In six months maybe this idea has already gone out the window.\nSpeaker D: So it's going to be a question of how fast we can act.\nSpeaker D: They already erased the rest of mine.\nSpeaker C: No.\nSpeaker C: Go to finance.\nSpeaker C: No, no, no.\nSpeaker D: Because I had another comment there.\nSpeaker D: The market trend, this is what we know from the last questionnaires and all the surveys we've done.\nSpeaker D: Fancy and feel good.\nSpeaker D: That's what we've been looking for.\nSpeaker D: Something that feels good in the hand that's easy to use.\nSpeaker D: Looking for a next generation of innovation because all the remotes out there now, they're all very similar.\nSpeaker D: They all do the same thing.\nSpeaker D: We have to have something completely different.\nSpeaker D: Easy to use has always become another major interest.\nSpeaker D: With the whiteboard we can see that it's a remote that's easy to use.\nSpeaker D: And I think this is another thing that's interesting is the consumers are actually willing to pay the price for exciting technology.\nSpeaker D: So even if we have a product that may be more expensive, if it comes out right, if they didn't look and it looks and feels good and has technology.\nSpeaker D: The second two, you can see the last one is a very easy, simple design.\nSpeaker D: The second one, there's about 45,000 different buttons on it, which makes it fairly hard to read, very hard to use.\nSpeaker D: The first one I see that they put in a display.\nSpeaker D: Now, this is something else with the little flip up now.\nSpeaker D: We're adding all kinds of things in, but with the little flip up, if you have a little display on the flip up and when you close it, everything is locked.\nSpeaker D: Maybe the display also makes it easier to use because sometimes when you're looking for buttons, maybe if you see a display.\nSpeaker C: Context sensitive instructions.\nSpeaker C: Depending on what the mode, the TV or the DVD or something else.\nSpeaker B: It was mostly the standard one.\nSpeaker D: Now you have one with a very simple also, the idea is simple, but with a display.\nSpeaker D: So you can see what you're doing.\nSpeaker D: So maybe if we can incorporate the easiness of use, trendy, fancy, feels good.\nSpeaker D: With a display, wood, designer wood, designer colors.\nSpeaker C: Maybe what you could do is if somebody orders the device, you could send them like some sort of foam rubber ball and then they would squeeze it and it would take the shape of their hands.\nSpeaker C: So it's really molded to your specific.\nSpeaker C: The geometry of their hands would be.\nSpeaker D: Hard they squeeze.\nSpeaker D: You know what kind of wood is there?\nSpeaker B: But for that you'd also have to do sort of an average across families.\nSpeaker B: Unless everyone has their own personal remote.\nSpeaker C: That actually would increase the revenue.\nSpeaker D: But incorporating the three obviously would be something totally new on the market.\nSpeaker D: Totally different.\nSpeaker B: Already the customizability is a really good sort of medium.\nSpeaker D: What it was, it was a Nokia that came out with the changeable colors.\nSpeaker D: Take apart and put it on another face, take it off and put it on another face.\nSpeaker B: And that took off.\nSpeaker D: They sold millions.\nSpeaker D: Millions.\nSpeaker D: So, so the findings were the research, easy to use, something totally new.\nSpeaker D: We have to come up with something totally new that is not on the market.\nSpeaker C: We'd also have to consider that who we were going to get to make these custom cases in terms of manufacturing processes.\nSpeaker C: We might want to learn about labor laws in different countries and stuff.\nSpeaker C: So we can do it cheap, but you don't want to exploit labor in third world countries.\nSpeaker C: So actually you could turn it around and say that the reason the cost is high for the device is because you're paying a working wage to the person who made the device.\nSpeaker A: But we can get a production in countries like India.\nSpeaker A: Cost of living is low.\nSpeaker A: Yes, if it is like India or China or Malaysia.\nSpeaker A: So you can go with better futures and better price and you can sell more.\nNone: Good.\nSpeaker C: Well, that would be something that manufacturing would have to explore more.\nSpeaker D: Where would you manufacture this is?\nSpeaker D: So another step.\nSpeaker D: Yes.\nSpeaker A: We're here to design.\nSpeaker A: Yes.\nSpeaker A: But that's a nice product.\nSpeaker A: We can talk about the production later, okay, depends on the quantity.\nSpeaker A: So we don't need to have our own fabric factory or something.\nSpeaker A: So we can have a tie up with who do the fabric with the different electronic items.\nSpeaker A: Then we can have a business tie up and to get to get the cost to sell more.\nSpeaker A: So let's decide first about the components concept and interface concept.\nSpeaker A: If it is acceptable for both of you, what Ed was talking.\nSpeaker A: And you will design with the display or without display or just simple.\nSpeaker B: I think it depends.\nSpeaker B: I mean, I think it's a good idea.\nSpeaker B: But we need to really think about how useful it's going to be because theoretically with a TV you already have a big display right in front of you.\nSpeaker B: So if we're trying to keep costs down, then maybe sacrificing the display is a way to go.\nSpeaker B: I mean, it depends on how much putting a display costs and what it would be used for.\nSpeaker B: Very specifically what it would be used for.\nSpeaker B: Because if it's only used for one little thing, then putting in a big display case or a big display that's probably expensive just to do the training on the chip for the speech recognition or whatever may not be the most cost efficient way to go.\nSpeaker B: But that's just sort of speculation.\nSpeaker C: I mean, what do you think Ed?\nSpeaker C: You like the display in one of the concepts that you showed.\nSpeaker C: Do you know how much it costs to add a little display like this?\nSpeaker D: No.\nSpeaker C: You want to take an action item to go find out?\nSpeaker D: It's because we have to find out a cost on it.\nSpeaker C: Sorry about that.\nSpeaker D: No, that's no problem.\nSpeaker D: I'm here for the pushing it after it's made.\nSpeaker D: I will market it.\nSpeaker D: Once we get a price on it, then we can market it.\nSpeaker C: So the advanced chip on print is what we've determined in the engineering industrial design is the recommendation.\nSpeaker C: And I think we've kind of come to some agreement regarding this concept of a wooden case.\nSpeaker C: A customizable and...\nSpeaker D: Nice, beautiful mahogany red wooden case.\nSpeaker B: Would the buttons be wood too?\nSpeaker C: I don't think so.\nSpeaker C: No, I think they could be rubber like they are now.\nSpeaker C: So you have that tactile experience.\nSpeaker A: Yeah, so what we'll do is we will stick with the simple design for time being until the ad find out about how much it's cost to the extra in case we go for the display.\nSpeaker A: So maybe what you can do is both of you can come up with the prototype, the model.\nSpeaker A: Okay.\nSpeaker B: Sure.\nSpeaker C: So are we done with this meeting?\nSpeaker A: Yeah, I hope.\nSpeaker A: Is it okay if they will come up with the prototype design?\nSpeaker A: Okay, then they can show you how it looks like and then we can submit to the...\nSpeaker A: I will submit to the management.\nSpeaker A: Okay.\nSpeaker A: Then mean that time you can come up with the price, how much it's cost as extra for the display.\nSpeaker A: And a marketing strategy.\nSpeaker A: And a marketing strategy.\nSpeaker A: That's very important.\nSpeaker A: Fire.\nSpeaker D: How much you can select?\nSpeaker A: Of course you'll make money too.\nSpeaker A: So not only fired, you make money too.\nSpeaker A: Your commission.\nSpeaker A: Okay, so any questions?\nSpeaker A: No.\nSpeaker A: So by next meeting, so please come up with the prototype.\nSpeaker A: And then we can press it from there.\nSpeaker A: Okay.\nSpeaker A: So thanks for all your efforts and coming for the meeting again and see you soon there.\nSpeaker A: Okay.\nSpeaker A: Okay.\nNone: Thank you.\nNone: Okay.\nNone: chime\n"}, {"meeting_id": "ES2005d", "summary": "In this meeting, the team was very satisfied with the three prototypes presented. However, they had to take the budget limits into consideration. Their current cost per remote control was fourteen point six Euros, while the budget required them to cut two more Euros. Therefore, a series of changes were done in order to meet the criteria, such as making the remotes flattened. In the end, the team was asked to make an evaluation about the project. All of them were very satisfied with the final product.", "dialogue": "Speaker C: Oh, is this how you are?\nSpeaker C: It's so fun.\nSpeaker B: You know?\nSpeaker B: You did something, you did something.\nSpeaker C: No?\nSpeaker C: It doesn't look as good as it is.\nSpeaker A: It's pretty cool.\nSpeaker A: You look kind of like Rambo.\nSpeaker A: I think...\nSpeaker C: Alright, hold on.\nSpeaker C: Let's play the game.\nSpeaker C: I think that's...\nNone: Come on, come on.\nNone: Dammit that.\nSpeaker A: Okay.\nSpeaker D: Okay, I'll use the power point.\nSpeaker D: How was that? Was that fun?\nSpeaker A: Very fun.\nSpeaker D: Okay.\nSpeaker D: Oh, I've forgotten to mail you the minute, but I will do.\nSpeaker D: Oops, it's easy.\nSpeaker D: We...\nSpeaker D: Alright, okay.\nSpeaker D: This one gets paid.\nSpeaker D: Alright, okay.\nSpeaker D: Were you guys first with your prototype?\nSpeaker D: I'll probably get to the good news.\nSpeaker C: Yeah, there's good news.\nSpeaker D: We have budget problems.\nSpeaker D: In the fridge, we'll sack.\nSpeaker D: Okay, have your presentation to make?\nSpeaker D: No, a North mic.\nSpeaker C: Okay, sure.\nSpeaker C: Okay, sure.\nSpeaker A: We made three for you.\nSpeaker A: One's based on the banana, one's based on the tomato, and the other one.\nSpeaker A: Oh, yeah, well, yeah, we had some red left over.\nSpeaker A: Okay, so this is the non-tough...\nSpeaker A: No buttons, one or as few buttons as possible.\nSpeaker A: Mainly speak recognition.\nSpeaker A: The yellow there is the slogan, yeah, that we need to incorporate.\nSpeaker A: Very simple.\nSpeaker A: If you do need buttons, you can flip it over and there's something there.\nSpeaker A: Okay.\nSpeaker A: But mainly it's speak tracking recognition.\nSpeaker D: Okay, so the buttons would be like individual users.\nSpeaker A: Yeah, yeah, they might project things onto the screen, which you can do on there.\nSpeaker A: I'm not sure about that.\nSpeaker A: And this one is the one, more like the one that we looked at earlier.\nSpeaker A: Right.\nSpeaker A: Yeah, you guys can...\nSpeaker C: Do you mind the feel of it?\nSpeaker A: Yeah, sure.\nSpeaker A: That one is... Oh, no, it's delicate.\nSpeaker A: It's already got it stand, that one, that's it stand.\nSpeaker A: It does also lie flat, but that yellow stand there represents the charging stand.\nSpeaker A: The black on the back is the slogan.\nSpeaker A: Okay, nice and obvious.\nSpeaker A: Yeah, that would have... Well, we did think of that.\nSpeaker A: It's on there, but also we're going to have the company name on the front, which is the little black kind of line.\nSpeaker A: Alright, okay, right.\nSpeaker A: And that's the transmittory thing.\nSpeaker A: These are the two scroll ones which we thought could be channel up and down and volume up and down.\nSpeaker A: We weren't sure about putting them there because it's... It kind of could get bashed.\nSpeaker D: Yeah, we're holding it kind of...\nSpeaker A: Well, if you hold it, you can... Well, hold it. It does actually feel quite ergonomic if you've got small hands.\nSpeaker A: Obviously, I don't think that's real sized thing, it would have to be a bit bigger.\nSpeaker A: Yeah, it's a model.\nSpeaker A: Okay, that's a speaker at the top, so you can speak into it like a little walkie-talkie as well for speaking recognition.\nSpeaker A: And then the buttons kind of self-explanatory, just buttons, whatever you need them, try to keep it simple.\nSpeaker A: Or that's the charging base prongs at the bottom.\nSpeaker A: We used those with tools and trade to that.\nSpeaker A: And then the big red button in the middle is the on and off one.\nSpeaker A: It's not in the traditional place, but it's quite an obvious place.\nSpeaker D: Excellent.\nSpeaker A: So there we go.\nSpeaker A: And we have the banana, the base one too.\nSpeaker C: This one is a... I suppose for the younger audiences.\nSpeaker C: Okay.\nSpeaker C: A more friendly type of...\nSpeaker C: So Barney the banana.\nSpeaker C: Right, right.\nSpeaker C: It's to induce more television watches.\nSpeaker C: Absolutely, just what you need.\nSpeaker A: Yeah.\nSpeaker D: Sort of Loch Ness banana.\nSpeaker D: Right.\nSpeaker D: Cool.\nSpeaker D: Well, yeah.\nSpeaker D: Nice to have options, at least.\nSpeaker D: Yeah.\nSpeaker A: So are there any improvements or issues?\nSpeaker A: Oh, there are issues.\nSpeaker D: Oh, there are issues.\nSpeaker C: It's way to lie down once.\nSpeaker D: Main problem that we have unfortunately being finance.\nSpeaker D: Ness is entering the evaluation criteria.\nSpeaker D: Unfortunately, the unit we are currently going to produce minus the extra scroll button.\nSpeaker D: And it's going to cost us 14.6 euros.\nSpeaker D: Oh.\nSpeaker D: So we have to re...\nSpeaker D: Sorry, I accidentally highlighted somehow.\nSpeaker D: There we go.\nSpeaker D: Oh, God.\nSpeaker D: Why is it doing that?\nSpeaker D: There we go.\nSpeaker D: So basically, in order to save our two euros, I was thinking that we could have essentially the same shape, but just have it flattened.\nSpeaker C: More likely to do it in all remote control.\nSpeaker D: Yeah.\nSpeaker D: I mean, it's already got a kind of cool shape.\nSpeaker D: So it wouldn't have to be curved in and out.\nSpeaker D: And by doing so...\nSpeaker D: Oh, no, hold on.\nSpeaker D: It doesn't save us quite as much.\nSpeaker D: I don't know what's going on, but let's again...\nSpeaker C: Why is there a double curve?\nSpeaker C: Two or four?\nSpeaker D: Oh, good point.\nSpeaker B: And double curve on both sides?\nSpeaker B: Curve on?\nSpeaker B: Yeah, this is double curve.\nSpeaker B: Sure.\nSpeaker B: Curve.\nSpeaker B: This is double curve.\nSpeaker B: This one is single curve.\nSpeaker B: This is single curve. This is one both sides.\nSpeaker B: Is that double curve?\nSpeaker D: No, I think it means double curve doesn't...\nSpeaker D: Like an S shape.\nSpeaker D: Like a single curve on that bottom half.\nSpeaker D: And the double curve would be if it was that similar curve.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker B: That might be wrong.\nSpeaker B: Like this one curve on this side, one curve on that side.\nSpeaker B: I don't like that.\nSpeaker B: It's a curve.\nSpeaker D: I think that's just a shape.\nSpeaker D: Okay.\nSpeaker D: The curve of a chair is like the...\nSpeaker D: It's a piece.\nSpeaker D: Okay.\nSpeaker C: That's the biggest expense.\nSpeaker C: Yeah, and why I've got it too.\nSpeaker D: I don't know.\nSpeaker D: I can't seem to select anymore, however.\nSpeaker D: Okay.\nSpeaker C: We can work around that.\nSpeaker C: Right.\nSpeaker C: No.\nSpeaker D: What do you think it should be one?\nSpeaker D: It's meant to be one.\nSpeaker D: Yeah, I don't know why I've been doing that.\nSpeaker D: But hold on until I find out that this shift button might be stuck again.\nSpeaker D: Yeah.\nSpeaker D: Yeah, this shift button is stuck.\nSpeaker D: Okay.\nSpeaker D: So take away three which would give us...\nSpeaker D: Oh, that's fine.\nSpeaker D: So 11 euro, 60.\nSpeaker D: Cool.\nSpeaker A: So we could even add something.\nSpeaker D: We could...\nSpeaker D: I don't know.\nSpeaker D: I don't know.\nSpeaker D: I don't know.\nSpeaker C: I don't know.\nSpeaker C: I'll just add a little bit more.\nSpeaker C: Fire the accountants.\nSpeaker C: Yeah, we could add things.\nSpeaker D: If you click back in that bottom right,\nSpeaker B: so where are you starting from? Yeah.\nSpeaker B: You can do one thing, right?\nSpeaker B: One of the buttons is stuck in the other one.\nSpeaker B: Okay.\nSpeaker B: Just to make it.\nSpeaker B: Okay.\nSpeaker B: Now, we'll wait like this.\nSpeaker B: It's not a shift button.\nSpeaker D: It's not a shift button.\nSpeaker D: It's not a shift button.\nSpeaker B: Yeah, it's not a shift button.\nSpeaker B: Should we ask about technical?\nSpeaker B: It's about Melissa.\nSpeaker D: That's fine.\nSpeaker D: We've worked out more.\nSpeaker D: It would be anyway.\nSpeaker C: Did you try both shift buttons?\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker C: That's it.\nSpeaker D: Oh, no, wait.\nSpeaker D: Right.\nSpeaker D: So that's fine.\nSpeaker D: I don't know what we could do.\nSpeaker D: What can we do?\nSpeaker D: We could add?\nSpeaker A: Well, maybe we could add something.\nSpeaker A: But maybe if...\nSpeaker A: I suppose that's our...\nSpeaker A: That's our design that we've got.\nSpeaker A: Yeah.\nSpeaker A: So...\nSpeaker A: If we're happy with designers, we'd have to put money if we don't have to.\nSpeaker A: But if there's anything we've missed out there, then feel free to add it.\nSpeaker A: Maybe...\nSpeaker A: I mean, obviously it would be bigger, so there might be more space for the slogan on the front.\nSpeaker A: Because it's not an ideal place right now.\nSpeaker D: That's...\nSpeaker D: Okay.\nSpeaker D: So, project variation.\nSpeaker D: We have under 12, that year is 50.\nSpeaker D: Project process.\nSpeaker D: Everything that went.\nSpeaker D: Are we happy?\nSpeaker C: We're happy.\nSpeaker C: I think we have a winning product.\nSpeaker D: Okay.\nSpeaker D: Project variation.\nSpeaker D: I don't really rate myself a bit more.\nSpeaker A: I think it went quite smoothly.\nSpeaker D: Room for creativity.\nSpeaker D: Were we happy with that?\nSpeaker A: I think we were very creative.\nSpeaker D: No, I mean, I think it means sort of individually.\nSpeaker A: Oh, right, okay.\nSpeaker D: Yes, maybe.\nSpeaker D: Okay.\nSpeaker D: So, we're just going to...\nSpeaker D: Yeah, okay.\nSpeaker D: Teamwork leadership, sorry.\nSpeaker A: It's a leadership.\nSpeaker D: You're okay.\nSpeaker D: You're all getting a raise.\nSpeaker D: Teamwork, I thought, out now?\nSpeaker A: Yeah.\nSpeaker A: Everyone got enough input, I think.\nSpeaker D: And, well, means, yeah.\nSpeaker D: Yeah.\nSpeaker D: We use technical stuff, which is brilliant.\nSpeaker D: That's why more.\nSpeaker D: Right.\nSpeaker D: I don't know what new ideas are firing to means, to be honest.\nSpeaker B: Yeah, these are new ideas, like, glow-in-the-dark or something like that.\nSpeaker B: We discussed all the new ideas, but, of course, we couldn't reach any proper goals.\nSpeaker B: We couldn't use these later, but we are using these scroll buttons like this.\nSpeaker B: These are new ideas, and new shapes, everything.\nSpeaker B: Yeah.\nSpeaker D: At least.\nSpeaker D: So, just a gen of thumbs up for all of us.\nSpeaker D: That, kind of, unfortunately, is too quick.\nSpeaker D: Okay.\nSpeaker A: Well...\nSpeaker A: Uh...\nSpeaker D: I don't have to do that.\nSpeaker D: I suppose, yeah.\nSpeaker C: So, let's talk about our bonuses and the raises we're getting.\nSpeaker D: I'm thinking another couple of days, holiday pay, maybe.\nSpeaker D: Well in order for all of you.\nSpeaker D: Yeah.\nSpeaker D: Let's see if I can get this.\nSpeaker D: It's not even good.\nSpeaker C: Oops.\nSpeaker C: Maybe start cleaning up the play.\nSpeaker B: Yeah, much.\nSpeaker D: Let's go back in.\nSpeaker D: Reusable.\nSpeaker A: I see.\nSpeaker A: Maybe.\nSpeaker A: Something.\nSpeaker D: I don't know what to say.\nSpeaker D: So, it's really, really annoying.\nSpeaker D: Oh.\nSpeaker B: Right?\nSpeaker B: Have you, have you finished?\nSpeaker B: Um, I have, yes.\nSpeaker B: I need also there.\nSpeaker B: I've just, my need also there.\nSpeaker B: Presentation.\nSpeaker B: All right.\nSpeaker B: Okay.\nSpeaker B: You've got more.\nSpeaker D: Oh, you got a potato star.\nSpeaker D: Yeah.\nSpeaker D: Oh, it didn't bother to tell me that on this thing.\nSpeaker D: Is it?\nSpeaker D: Okay.\nSpeaker D: It doesn't tell me.\nSpeaker D: Is the project evaluated that is my, man?\nSpeaker B: Oh, you're doing it.\nSpeaker D: Yeah.\nSpeaker D: We're going to evaluate ourselves.\nSpeaker D: We thought we were green.\nSpeaker D: Yeah.\nSpeaker D: Oh.\nSpeaker D: Love to be that now.\nSpeaker A: Kind of a green banana now.\nSpeaker A: I covered it.\nSpeaker A: It's, this is well, sorry.\nSpeaker A: We've got to mention the main thing.\nSpeaker A: We've got to mention we've made out of kind of a rubbery.\nSpeaker A: They text new material that we've got.\nSpeaker D: Okay.\nSpeaker D: Hold on.\nNone: Anybody?\nSpeaker C: I've got that.\nSpeaker D: One down, one would sell that one.\nSpeaker D: Yeah.\nSpeaker C: Yeah.\nSpeaker D: Okay.\nSpeaker B: So now is the final evaluation, final evaluation of the, of, of our product.\nSpeaker B: How we are going to, means at what standards, what standard, whether it means our standards or not, what rating we will give to these products.\nSpeaker B: So of course this will be a team work.\nSpeaker B: We together have to decide what rating we will give to this product and everything.\nSpeaker B: So, but methodology, I will tell you on what basis we are going to discuss all this.\nSpeaker B: We will give the rating to this product based on the user requirements, whether it meets the user requirements or not, this product then trends, whether it is as per the latest fashion trends or not.\nSpeaker B: That means because we have already studied that people do prefer fashionable things nowadays.\nSpeaker B: So this is also an important factor for evaluation also.\nSpeaker B: Then marketing strategy of the company.\nSpeaker B: As we have already discussed that our company is quite rapid in the market.\nSpeaker B: Not only in terms of providing quality products, not only in providing latest technologies, but also in terms of providing environmental.\nSpeaker B: So, I just say, but I also in terms of providing environmental safe products like keeping in mind all the safety issues.\nSpeaker B: So, now comes the criteria rating with seven point scale.\nSpeaker B: I am having this scale.\nSpeaker B: So, we have to do it on a board.\nSpeaker B: Okay.\nSpeaker C: I should be the user.\nSpeaker C: Thank you.\nSpeaker B: So,\nNone: thank you. Thank you.\nSpeaker B: Thank you.\nSpeaker B: So, these are the three criteria for evaluation of home of product.\nSpeaker B: First of all, comes user requirement.\nSpeaker B: So, we will see whether this product meets all user requirements or not.\nSpeaker B: First, I would like to have your views.\nSpeaker B: What do you think?\nSpeaker B: Whether it meets all user requirements or not?\nSpeaker A: I think, yeah, it did.\nSpeaker A: It had all the basic buttons that they needed as well as the new technology that people said they wanted.\nSpeaker B: So, what rating you will personally?\nSpeaker B: I would say seven.\nSpeaker A: Seven.\nSpeaker A: Seven is good, isn't it?\nSpeaker A: True or false?\nSpeaker A: No search.\nSpeaker B: One is true.\nSpeaker B: One is highest ranking.\nSpeaker B: But I think highest ranking is seven or one.\nSpeaker D: Okay.\nSpeaker B: So, it's one from your point of view.\nSpeaker B: And what do you say about industrial expert?\nSpeaker C: It's hard to know.\nSpeaker B: But you can tell only, I think she has given her views on the basis of design because she was our interface expert.\nSpeaker B: But you can give your views based on technology whether the technology meets the requirements of the customers or not.\nSpeaker C: Yeah, I think it might even exceed it.\nSpeaker C: But I guess there is kind of a shortage of buttons.\nSpeaker C: Say, I'm going to give it a two.\nSpeaker B: Two.\nSpeaker B: And what about you Brian?\nSpeaker B: I'll go for one.\nSpeaker D: You will go for one.\nSpeaker D: Basic requirements.\nSpeaker D: Okay.\nSpeaker B: Of the project.\nSpeaker B: For me personally, it is everything fine.\nSpeaker B: It may be having good design.\nSpeaker B: It may be meeting all the requirements of the customers like technology wise, price wise.\nSpeaker B: But there is one thing which limits the customers like we are having only two, three designs.\nSpeaker B: Like we are having one banana design.\nSpeaker B: And the other one is orange.\nSpeaker B: And yellow.\nSpeaker B: And the third one is what you get that is not of a fruit look.\nSpeaker B: But if a person doesn't like banana or orange, you are limiting him.\nSpeaker B: No, don't buy our product because we like this only.\nSpeaker B: Okay.\nSpeaker B: So we are showing our preference for particular fruits, two or three kinds and...\nSpeaker B: Fritest.\nSpeaker A: Yeah.\nSpeaker B: Is that not trends?\nSpeaker B: No.\nSpeaker B: Personally, as a marketing expert, I don't believe that because whatever companies they launch their products in the shape of fruits, they give a range of products, a range of shapes.\nSpeaker B: Like if we look at the smallest thing to office chocolate, they give a variety of different things.\nSpeaker B: Some children like to buy banana shapes, some apple shapes, some even pineapple shapes, some orange shape.\nSpeaker B: So you can't know means what shape a person will like.\nSpeaker B: So in this case, giving only one or two choices, we are limiting our customers.\nSpeaker B: And by limiting them, we are limiting our sales, limiting our profit also.\nSpeaker A: But in electronics, I think it's not always quite so.\nSpeaker A: You don't always have so many choices with chocolate.\nSpeaker A: I think, you know, if you go into buy a TV, maybe a company will have...\nSpeaker A: That you're going to choose from a company will have two or three choices, but their different designs will come up with one product.\nSpeaker B: Maybe.\nSpeaker B: Okay.\nSpeaker B: No, I will...\nSpeaker B: Obviously, you're potentially trying to...\nSpeaker B: Won't give it to you in beyond three.\nSpeaker B: Okay.\nSpeaker B: So we can...\nSpeaker B: Stop clicking.\nSpeaker B: Yeah.\nSpeaker B: On average, we can think of three people.\nSpeaker B: Seven.\nSpeaker B: Maybe...\nSpeaker B: Three or four.\nSpeaker B: No, sorry.\nSpeaker B: It should six.\nSpeaker B: Five or six.\nSpeaker C: What are we doing?\nSpeaker B: No, sorry, sorry, sorry, sorry.\nSpeaker B: We are doing very wrong things.\nSpeaker B: Yeah, yeah, we are taking average and that's...\nSpeaker A: I have taken it to I wrongly.\nSpeaker B: Yeah.\nSpeaker B: Three, four, four, six, seven, seven.\nSpeaker B: Yeah.\nSpeaker B: One point.\nSpeaker B: Exactly.\nSpeaker B: So we can say one or two.\nSpeaker B: Okay.\nSpeaker B: And because it is one point eight, just reaching two, so we will do two.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: So, however the trends...\nSpeaker C: Can you explain what you want us to write there?\nSpeaker C: Sorry?\nSpeaker C: How conforms to the current trends?\nSpeaker B: Yeah, again, the freshened trends, this also like...\nSpeaker B: Whether it will be fashionable to have these products in the...\nSpeaker B: As a fruit shape or something.\nSpeaker B: Okay.\nSpeaker A: Well, um...\nSpeaker A: Going on the specifications that we had, the fruit and vegetables are quite popular.\nSpeaker A: And that people like something that is good to look at, and not many buttons.\nSpeaker A: I would give it...\nSpeaker A: Well, because it's hard to make a fruit, good to look at.\nSpeaker A: That looks cool, you know?\nSpeaker A: So I would actually give it a three...\nSpeaker A: A three or four, I'm not sure.\nSpeaker A: Three.\nSpeaker A: Good for you.\nSpeaker C: As far as the technology, it's got the latest trends in speech technology, but it's missing the screen, as you said.\nSpeaker C: It does have the push buttons, or the scroll buttons, but doesn't have that fancy solar power, the vibrating energy mechanisms.\nSpeaker C: I give it a four, I give it kind of middle of the road for the technology.\nSpeaker D: And I'm sort of pretty...\nSpeaker D: Just the fruit, it does mean.\nSpeaker D: I mean...\nSpeaker D: Yeah.\nSpeaker D: It might be trendy to some, but I'm just not swallowing the fruit.\nSpeaker D: Yeah.\nSpeaker D: Also, I would like to see the LCD screen in it.\nSpeaker D: So yeah, I'd say about four as well.\nSpeaker B: Okay.\nSpeaker B: Personally, I mean...\nSpeaker B: I think that in terms of trends, these products are quite good, like...\nSpeaker B: These products are in fruit shape, because that people now...\nSpeaker B: Our fashion trends show that people like everything...\nSpeaker B: Everything is being advertised like clothes shoes, and everything is being advertised in the form of fruits and vegetables, or getting them some...\nSpeaker B: Or showing some association with them.\nSpeaker B: So...\nSpeaker B: And in this way, our product is good.\nSpeaker B: And the second thing now, people don't want any complicated or...\nSpeaker B: Bunkie project, and our system is quite simple and quite handy.\nSpeaker B: So that is also...\nSpeaker B: Our product meets the trends of the market.\nSpeaker B: And yes, it is Cuspongi also.\nSpeaker B: So they can play with it, it's quite good.\nSpeaker B: So then I think maybe I can give it to.\nSpeaker B: So...\nSpeaker B: Four, five, ten, thirteen, thirty, so we can...\nSpeaker B: Is it fine?\nSpeaker B: So what about company strategy?\nSpeaker B: Say...\nSpeaker C: Um...\nSpeaker C: Zialla...\nSpeaker A: Yeah, the company strategy, okay.\nSpeaker A: There was a lot of disgusting.\nSpeaker A: It was good.\nSpeaker A: I feel I got my say.\nSpeaker A: So I give the company strategy...\nSpeaker A: Uh...\nSpeaker A: Two.\nSpeaker C: Okay, well I think it's the remote control conforms to the company strategy.\nSpeaker C: Okay.\nSpeaker A: Is that the question?\nSpeaker A: Yeah.\nSpeaker A: Is it okay?\nSpeaker A: Okay, so...\nSpeaker A: One or two?\nSpeaker A: What...\nSpeaker A: Okay, just...\nSpeaker A: It's fine.\nSpeaker C: I'll go with two.\nSpeaker B: So what were you, Brian?\nSpeaker D: Um, yeah.\nSpeaker D: I agree...\nSpeaker D: Pretty much kept to the company strategy, so I would go for a...\nSpeaker D: Uh, one as we not only kept to it, but we were limited by it.\nSpeaker B: Um...\nSpeaker B: Yeah.\nSpeaker B: And...\nSpeaker B: Me also like this product...\nSpeaker B: Me...\nSpeaker B: This Me as all companies strategy, like...\nSpeaker B: Uh, product should be as per customer's requirement, as per latest technology, and it should be environmental safe.\nSpeaker B: So since our product meets all these requirements, so I would also prefer to give it rank one.\nSpeaker B: So...\nSpeaker B: Four, six, six.\nSpeaker B: Uh, so we can say two or one.\nSpeaker C: Two.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker B: Seven, seven.\nSpeaker C: And...\nNone: No.\nNone: Uh...\nSpeaker B: Overall, we are getting two pan-c's something, but we can round it as two.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: So I think overall, a evaluation of our product is quite good.\nSpeaker B: So we can launch it.\nSpeaker B: Yeah.\nSpeaker A: Mm-hmm.\nSpeaker A: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nNone: Mm-hmm.\nSpeaker D: In which case, we are done.\nSpeaker D: Because we have evaluated and we are within budget.\nSpeaker D: Okay.\nSpeaker D: Cool.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker D: Champing lunch?\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Good.\nSpeaker D: I've got a mic.\nSpeaker D: Mm-hmm.\nSpeaker D: Okay.\nSpeaker D: I'll wait with that.\nSpeaker C: So I'll have to write up a couple more...\nSpeaker C: More forms.\nSpeaker C: Yeah, yeah.\nSpeaker D: I've got a report to write, so...\nSpeaker D: Mm-hmm.\nSpeaker D: Mm-hmm.\nSpeaker C: So I like my job.\nSpeaker A: Yeah, with the claim.\nNone: Uh...\nNone: Yeah.\nSpeaker D: Can you hold up gender free?\nSpeaker B: My list is in...\nNone: Yours is in project?\nSpeaker D: Yeah.\nSpeaker D: That's good.\nSpeaker D: Okay.\nSpeaker D: Cool.\nSpeaker D: I'll get there eventually.\nNone: Okay.\nNone: Okay, oh...\nNone: terrible\n"}, {"meeting_id": "ES2008c", "summary": "This meeting was about the conceptual design of the remote control. Firstly, groupmates presented on the conceptual specification on components, properties and materials as well as a conceptual specification on user interface design. Also, they suggested conceptual designs including the buttons, the circuit board, the chip to print as well as the LCD screen with backlighting in the device. Then, the group had a discussion on the final remote control concepts. So, they decided to have changeable covers in fruity colours, rechargeable batteries and scroll wheels. And they would leave the spongy case and the standard of the chip to be discussed in the next meeting.", "dialogue": "None: I'm sorry, so...\nNone: I just didn't have to leave the end of the video.\nNone: Oh, I'm gonna leave it.\nSpeaker B: Mm-hmm.\nNone: You might be leaving the end of the back.\nNone: I'm just gonna leave it.\nSpeaker D: I'm sorry.\nNone: So...\nSpeaker B: I'm sorry, I'm sorry.\nSpeaker B: Okay.\nSpeaker B: I think I...\nNone: Yeah.\nSpeaker C: I'm sorry.\nSpeaker C: 14, 26.\nSpeaker B: Okay.\nSpeaker B: Lovely to see you all again.\nSpeaker B: It's our conceptual design meeting.\nSpeaker B: And it's starting at approximately 14, 25, and so we have 40 minutes for this one again.\nSpeaker B: So we'll go just after 3 o'clock.\nSpeaker B: Okay.\nSpeaker B: All right, Genda, we're gonna do an opening.\nSpeaker B: I'm gonna review the minutes of the last meeting.\nSpeaker B: Then we'll have your three presentations.\nSpeaker B: And then we'll have to make a decision on the remote control concepts.\nSpeaker B: Finally, we'll close.\nSpeaker B: So... opening.\nSpeaker B: These are our minutes from the functional design.\nSpeaker B: We decided our target group is the focus on who can afford it, because we have international appeal, and we said it's for all age groups, different functions of it.\nSpeaker B: Our main objectives were simplicity and fashion.\nSpeaker B: And in specific functions were something to keep the remote from getting lost.\nSpeaker B: Large buttons for the essential functions.\nSpeaker B: A possibility for extra functions like a sliding piece.\nSpeaker B: And a long-life battery or a charging station.\nSpeaker B: Okay.\nSpeaker B: Now three presentations.\nSpeaker B: I'd like to do it in this order.\nSpeaker B: First, do the conceptual specification of components, properties, and materials.\nSpeaker B: That would be me.\nSpeaker B: And then the conceptual specification of user interface.\nSpeaker B: And finally, trend watching.\nSpeaker D: Alright.\nNone: Well...\nSpeaker B: Okay.\nSpeaker B: Function of edit.\nSpeaker B: There we go.\nSpeaker D: Alright.\nSpeaker D: I'm very excited by this one, actually, guys.\nSpeaker D: I had a lot of fun doing it.\nSpeaker D: Components design.\nSpeaker D: This is where you look at what does it take to make a remote control, and what should we make our specific remote control out of?\nSpeaker D: So we need to examine each element separately, but we're designing a full thing, so you want to keep it integrated as a whole.\nSpeaker D: The main elements of remote controls, in general, and therefore ours as well, are the case, the buttons, the circuit board with the chip, and the battery.\nSpeaker D: These are all things that we had sort of addressed before, but we're going to take each one a little bit separately here as we figure out what they should be made of and what they should look like.\nSpeaker D: The case, the options that I've gotten from headquarters about what we can do, there's a shape of a case.\nSpeaker D: We could do a flat case, a curved case, or maybe even a double, double-curved case.\nSpeaker D: I haven't seen any pictures of what this exactly looks like yet.\nSpeaker D: Just keep that in mind, but these are the options that we have for manufacturing.\nSpeaker D: And we can make our case out of plastic.\nSpeaker D: The main base will be plastic, but we have all these sort of fashion and technological elements we can add in, wood, latex, titanium, rubber, or other colored types of plastic.\nSpeaker D: That would be our case.\nSpeaker D: For buttons, we have push buttons, which is what real reaction uses the most often, but we also have scroll wheels, which can have integrated push buttons, or we could go all high-tech and have an LCD screen.\nSpeaker D: Circuit board and chip, we can have a simple one, a regular one, or an advanced one, depending on what our other needs are.\nSpeaker D: And then battery, I think, is where things get most exciting.\nSpeaker D: We're talking about long-life batteries here.\nSpeaker D: We can have your basic AA batteries, but we also have these options of using a kinetic battery, like are used in high-tech watches, where you just have to move it a little bit to power up.\nSpeaker D: So simple movements like pushing buttons would recharge the battery, or solar battery, although there are slight complications with solar batteries, as in we can't use a latex case if we have a solar battery.\nSpeaker D: Or something they only described as the type of battery you find in torches from about 50 years ago, flashlights.\nSpeaker D: Not quite sure what that is, but that's the description that I received, so that's what I'm passing on to you.\nSpeaker D: So those are our options.\nSpeaker D: And personal preferences that I was thinking through.\nSpeaker D: Here's what we've been talking about all through fashion and simplicity.\nSpeaker D: So if we're going for fashion in our cases, I think that what we're going to want to look at is a curved or a double-curved case, probably with a variety of design elements, maybe titanium, maybe some wood.\nSpeaker D: We're going to have to investigate that better when I get specifics of the actual materials.\nSpeaker D: That's sort of what I have in mind.\nSpeaker D: And we want to go for simplicity.\nSpeaker D: Probably push buttons, but I'm sort of intrigued by the idea of a scroll wheel if anyone has anything.\nSpeaker D: Any ideas on that? I mean, I know the iPods and things right now have touch scroll buttons, which are exactly like what they're describing, so that might be something we want to look into.\nSpeaker D: And I'm really intrigued by the idea of a kinetic battery.\nSpeaker D: A solar, I don't think, would be such a good idea because how often are you sure that your remote control will get a certain amount of light?\nSpeaker D: But this idea of the kinetic that you don't have to replace and that a simple just shaking it around will make it work.\nSpeaker D: That would be a very interesting thing.\nSpeaker D: But I think we'd also want to go for a simple chip or regular chip to keep our costs down.\nSpeaker D: We really only need a regular or advanced chip if we're going to start using an LCD.\nSpeaker D: So I think we want to be aware of not making things overly technological if they don't need to be because that'll keep our manufacturing price way down.\nSpeaker D: That's what I have for options. I'd appreciate anyone's input, but that's what I'm seeing for the future of the look of this thing.\nSpeaker C: It's double curved, would that be like two hands kind of thing?\nSpeaker D: I'm not sure I haven't received any specific visuals.\nSpeaker C: Imagine like double curved is like, you know, like two sides, the curve and then one curve would just be like a single vertical.\nSpeaker D: Yeah, I'm not sure, but I'll let you know as soon as I get any pictures.\nSpeaker D: I wonder.\nSpeaker D: I know if we do have a double curved case, it can't have any titanium in it.\nSpeaker D: But the titanium they were quite they're marketing quite hard to us as being used in the space program.\nSpeaker D: So that could be quite interesting. Space age.\nSpeaker D: Just an interesting marketing kind of element.\nSpeaker D: That's been all I have to do guys.\nSpeaker D: I hope I didn't go through that too quickly.\nSpeaker B: Just a real quick question.\nSpeaker B: The weight of these different elements.\nSpeaker B: Have you? No idea.\nSpeaker D: I've no idea.\nSpeaker D: I'm assuming that a kinetic battery isn't going to take up that much weight.\nSpeaker D: And that titanium is very light.\nSpeaker B: I know.\nSpeaker D: But of the seasons, that's really basic.\nSpeaker D: I mean, that's all I have gotten so far.\nSpeaker D: Okay.\nSpeaker D: Any other questions?\nSpeaker D: No? Okay.\nSpeaker D: I'm kind of saving this in the shared projects.\nSpeaker D: If anyone wants to look, thank you.\nSpeaker D: And I have considerable notes on the topic as well.\nSpeaker D: If anyone needs any more information.\nSpeaker B: If you made notes yourself, you can put those on our underneath our book.\nSpeaker B: I mean, you're booked.\nSpeaker D: But if anyone has any specific questions, don't hesitate to email me or something.\nSpeaker D: All right.\nSpeaker D: I guess I can.\nSpeaker B: Okay.\nSpeaker B: Now we're concepts, concepts of user interface.\nSpeaker B: This one's so much tighter than the other.\nSpeaker B: Okay.\nSpeaker B: Nope.\nSpeaker B: There we go.\nSpeaker B: There you are.\nNone: Let's see.\nSpeaker D: Oh, no, Jason.\nSpeaker D: Oh, Jason.\nNone: Okay.\nSpeaker A: So I'm going to describe the concept of user interface.\nSpeaker A: This device.\nSpeaker A: We've talked about the two essential properties of the user interface.\nSpeaker A: We want it to be simple and we want it to be fashionable.\nSpeaker A: We've also got to remember that this is a device that serves as useful purpose.\nSpeaker A: We want people to be able to use it as a remote control.\nSpeaker A: So we need to determine what the essential functions of the device are and make sure that we include all of those and that we actually end up with a device that is going to be useful to people.\nSpeaker A: We have a number of different choices for a design concept.\nSpeaker A: And that's something that I'll show you some examples of.\nSpeaker A: But essentially we need to choose how is this device going to work?\nSpeaker A: How what kind of model is there going to be for user interaction with this device?\nSpeaker A: Once we've chosen a concept for it, we can then design the features of our concepts, making sure that we get all the essential functions in the device and the extra functions and the more advanced features.\nSpeaker A: And of course, we also have to make it look cool.\nSpeaker A: That's what we don't want.\nSpeaker A: We don't want lots of buttons complicated features.\nSpeaker A: We want something that looks nice and simple.\nSpeaker A: Here's a fairly simple device. This is an iPod for Mapple.\nSpeaker A: I think the main thing to notice about it is it just has four buttons.\nSpeaker A: It's very minimalistic in its design.\nSpeaker A: You use these four buttons to move around a range of settings on a small LCD screen.\nSpeaker A: The thing I like about this is that it's very, very quick to access the main features.\nSpeaker A: You can just about make up that the bottom three buttons are previous track, next track and playpours.\nSpeaker A: The main features of the iPod, the things you will use all the time.\nSpeaker A: Then if you want to do anything more advanced, you go into the little menu on the LCD screen and you use the buttons just to scroll around and find the more advanced features that you want.\nSpeaker A: I think that's a good model that we could have.\nSpeaker A: Here's another example.\nSpeaker A: It's an interesting idea. It looks pretty cool.\nSpeaker A: It's got novelty value. It's nice and colorful. It's nice and bright.\nSpeaker A: It's also something that you can feel your wear around.\nSpeaker A: The buttons are different shapes.\nSpeaker A: If you're watching TV in the dark or whatever, you can work out which buttons.\nSpeaker A: It's fun.\nSpeaker A: I like this idea of just having buttons for the features that you use most often.\nSpeaker A: You need a few buttons to select your favourite channels.\nSpeaker A: Most people when they watch TV, they have 250 channels on their TV and they watch them for them the most.\nSpeaker A: You have buttons for your favourite channel, changing the volume, which is something you do all the time when you're watching TV, and a button to switch off in case you get bored.\nSpeaker A: Other features, things like adjusting the brightness, tuning the TV, I don't know what else you do with the TV.\nSpeaker A: These are all necessary functions.\nSpeaker A: You can't have it. There's no point in having a television that you can't tune or that you can't set the contrast.\nSpeaker A: We need to find a way of including these somehow.\nSpeaker A: One of the suggestions I'd make is to include them in a menu system a bit like on the iPod.\nSpeaker A: We'd either have a small LCD display on the device itself or have a menu display that comes up on your television and can be controlled through the device.\nSpeaker A: I know it would allow you to access the advanced features whilst keeping a very small and simple set of buttons for the features you use most often.\nSpeaker D: Okay, so you'd be advocating on LCD then?\nSpeaker A: I think that's one way to go, yes.\nSpeaker A: The advantages and disadvantages, if you have an LCD display, it's nice because it lets you just sit there and control your television from your armchair.\nSpeaker A: There are disadvantages and LCD display would have to be quite small because we're...\nSpeaker A: I assume we're going to be making quite a small device.\nSpeaker A: It would also have to have a kind of backlighting in it because you tend to watch TV in the dark, but you need to be able to see your LCD display.\nSpeaker A: Your alternative is to have an on-screen display on your television that you control through your remote control.\nSpeaker A: I don't like how they have these digital boxes where you press the buttons and it comes up with your listing of what's on each channel.\nSpeaker A: That's also a good idea.\nSpeaker A: It does have its problems as well if you've got a small TV and you're sitting on the other side of a room.\nSpeaker A: It's hard to read the little text that comes up.\nSpeaker A: But that's the decision that we can make.\nSpeaker B: I do think that one of the important features for a remote is seeing a menu and seeing what's on.\nSpeaker B: And so, like favorite channels is applicable, but I think that you do need to have some kind of function where you can see the title of the show or possibly a description of it.\nSpeaker B: I know I use that often.\nSpeaker C: Is that separate from what he was saying?\nSpeaker B: Well, if it was an LCD on the remote, I don't know that you would be able to see it.\nSpeaker D: We're talking menu contrast and tuning the VCR or something.\nSpeaker D: If I've understood you rather than menu as what's on.\nSpeaker B: Okay.\nSpeaker A: That would be more specifically digital boxes.\nSpeaker A: If we put a display on the remote control, communication is one way.\nSpeaker A: So you can't have the televisions and information back to your mobile control.\nSpeaker A: Good point.\nSpeaker A: Okay.\nSpeaker B: Now we're moving on to market marketing.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker D: What F.G.\nSpeaker D: Press five eight.\nSpeaker C: I don't know how relevant all of this is going to be.\nSpeaker C: Anything.\nSpeaker C: Oh, snap plugged in.\nSpeaker B: All right.\nSpeaker D: There we go.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: So we're going to look a little bit at trend watching.\nSpeaker C: Basically, I was given an executive summary that was.\nSpeaker C: I was looking at the market investigation on road controls that was recently conducted.\nSpeaker C: And then also some fashion washers and Paris and Milan commented on some things that are going to be going on this year.\nSpeaker C: So first, they had people, they ranked the important aspects of remote controls.\nSpeaker C: So that's going to be a reference, but that really over the next year that's going to be switching to fancy look and feel remote.\nSpeaker C: So that just goes back to the whole desire of our real reaction company wanting to focus on fashion.\nSpeaker C: So even though we're stressing, when we're talking, we've all been talking about this simplicity and easy to use idea.\nSpeaker C: They're sort of wanting us to remember that the number one thing for everyone is that it's fancy looking feel.\nSpeaker C: And as these are ranked, the top one is doubly important to the second one, which is doubly important to the third one.\nSpeaker C: So just to take that weight into account.\nSpeaker C: The second thing that was mentioned as important was the technological innovations that would be like if we use something like the space material or the LCD screen, things like that.\nSpeaker C: And then ease of use was the third most important.\nSpeaker C: So really, no matter what we need to focus the most on fancy look of view, according to this, I don't know what you would get with that.\nSpeaker C: And then Fashion Watch talks about that this year's top trend for clothes, shoes and furniture is fruits and vegetables.\nSpeaker C: And that there's a preference for spungy type material.\nSpeaker C: And so that brings us to my personal preferences. Who wants this fun to your vote or want to their natural advantage?\nSpeaker C: Personally, I don't really think that I want one that's going to go out the style or go stale, excuse the pun, in a year.\nSpeaker C: So even though this is coming from us as, you know, trend watch market research, I don't know how much of a BNS is thoroughly want to take away.\nSpeaker C: Also, considering that the research has been carried out by real reaction, I'm a little hesitant as to like how these questions may have been worded.\nSpeaker C: And if necessarily this whole fashion to technology to ease of use is necessarily the right ranking, personally, like I might reverse it.\nSpeaker C: But if we're working for this company, then I guess no matter what we have to stress fashion the most.\nSpeaker C: But it doesn't necessarily need to be a spungy material.\nSpeaker D: There's all kinds of scope for imagination in that window.\nSpeaker D: Yeah.\nSpeaker C: I don't have a lot of notes to share if you want them that pretty much sums it up.\nSpeaker C: So yeah.\nSpeaker B: Okay, do we have any some questions for this?\nSpeaker B: Yeah.\nSpeaker C: What can I possibly might not?\nSpeaker B: Do you have any ideas how to possibly use these, how to use a fruitive vegetable or the spungy material at all?\nSpeaker B: Like, could we make a like, could we make a spungy remote?\nSpeaker B: It would be easier on the hands.\nSpeaker B: If it's latex, it's kind of, then it would have to find a way to protect like the chip and all that.\nSpeaker D: Yeah.\nSpeaker D: My understanding of latex cases that it's in fact hard to protect stuff inside, but it's covered with the latex, which is spun to your own softer on your hands.\nSpeaker D: There's something to be said.\nSpeaker D: I mean, we got that thing earlier from you about not wanting it to our repetitive stress injuries and things.\nSpeaker D: Yeah.\nSpeaker D: So something, something like that.\nSpeaker C: I mean, I think we don't want to go spungy.\nSpeaker C: I think I'm envisioning more like, you know, the material that you have when you sit on like a bicycle so that it doesn't hurt.\nSpeaker C: You're sitting on it for a long time.\nSpeaker C: I'm imagining that sort of thing.\nSpeaker D: I don't know what that feels like.\nSpeaker D: Given the list of materials I was forwarded, that seems doable.\nSpeaker D: Could we go in fruit and vegetable colors?\nSpeaker B: You could color corn either.\nSpeaker B: Or maybe like a couple main ones being like an 11s or strawberries.\nSpeaker B: The buttons could be finished.\nSpeaker A: Could they be smelly?\nSpeaker A: Oh, God.\nSpeaker A: Is it supposed to be shaped like a vegetable?\nSpeaker A: I don't know.\nSpeaker A: Like a carrots or something.\nSpeaker A: Oh, it's quite carrot shaped.\nSpeaker D: Like carrots.\nSpeaker D: Or maybe the buttons could be shaped like large, but that's what I was thinking of.\nSpeaker C: What about the idea like a round remote instead of like a vertical up and down one?\nSpeaker C: In terms of holding it.\nSpeaker C: That's a fruit.\nSpeaker C: Might have tied in a little.\nSpeaker B: I think you'd be able to hold it because I think the reason that long is.\nSpeaker D: I think the harder to press, the harder to push.\nSpeaker A: When you use a remote, you press the buttons with your thumb usually.\nSpeaker A: What are your fingers?\nSpeaker B: I usually hold it on one hand, maybe.\nSpeaker A: I maybe want something that's shaped like a mobile phone.\nSpeaker A: So you hold it in one hand and you press the button.\nSpeaker B: The buttons would have to be very small.\nSpeaker D: No, I tend to do it.\nSpeaker B: But you think they're just your fingers down around.\nSpeaker B: But I mean in order to get to all of them, you know, they would have to be within a certain amount of space.\nSpeaker A: But it's only got like four or five buttons anyway.\nSpeaker D: And it's not so much about it.\nSpeaker D: When I'm pressing buttons on my iPod, that's how I do it.\nSpeaker D: Hold it and press.\nSpeaker A: Hold it in one hand and you press the buttons.\nSpeaker D: Yeah, and use my thumb or my pointer finger on the touch scroll wheel.\nSpeaker C: I love the idea.\nSpeaker C: We all like the iPod.\nSpeaker A: The button on the iPod, is it just four buttons or is it?\nSpeaker C: It's like five.\nSpeaker D: There's one in the middle.\nSpeaker D: Yeah.\nSpeaker D: I mean each version of it has been a little bit different.\nSpeaker B: The one I have doesn't have the four on it.\nSpeaker D: Oh yeah, you have one of the in between ones when they weren't doing that anymore.\nNone: And you press the center button.\nSpeaker D: That's your all purpose select button right there.\nSpeaker D: Oh, okay.\nSpeaker C: I think why it would be good for us is because you could have the same wheel sort of effect for channel flipping and then the other one could be like for volume.\nSpeaker C: Like just the idea of like those so few buttons for main things.\nSpeaker C: But then how you go back to the menu and like I don't know if we would want it on the screen there on the actual TV.\nSpeaker C: I kind of am wanting to say on the TV because if you're changing the brightness, you want to see it happening kind of.\nSpeaker C: Yeah.\nSpeaker D: And it's still how that will be.\nSpeaker D: Yeah.\nSpeaker D: It might be good in theory but not as useful in practice.\nSpeaker A: Yeah.\nSpeaker A: Also, yeah, because you would be looking down at the LCD.\nSpeaker A: Right.\nSpeaker A: I'll put your TV.\nSpeaker A: Okay.\nSpeaker B: We have about 15 minutes left.\nSpeaker B: So I'm going to continue with my presentation.\nSpeaker B: I have one more slide before we close but in that slide is when we need to make decisions about these kinds of things.\nSpeaker B: So I'll just bring that up and show you all.\nSpeaker B: We'll move on.\nSpeaker C: I think I need more information on fruits and vegetables.\nSpeaker C: Can we let you know?\nSpeaker D: Can we have changeable covers like for your mobile?\nSpeaker D: In different fruit and vegetable colors?\nSpeaker D: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: And then the covers could be spongy latex, but the actual model could be tidy.\nSpeaker D: You could coordinate with your house or whatever.\nSpeaker D: It's option.\nSpeaker A: The packaging, it should be like a lemon and the packaging is like the peel.\nSpeaker A: So that's the opening of the box.\nSpeaker A: You just kind of peel it.\nSpeaker C: There you go.\nSpeaker C: But I've had packages.\nSpeaker C: So that was like half the fun.\nSpeaker C: It's like the way it all comes off.\nSpeaker B: Okay.\nSpeaker B: The components concept.\nSpeaker B: Energy.\nSpeaker D: Oh, yes.\nSpeaker D: Right.\nSpeaker D: I sort of skipped over that hoping it wouldn't be necessary.\nSpeaker D: But this is the agenda they gave me.\nSpeaker B: So what can you just explain with that as real quick?\nSpeaker D: Decisions.\nSpeaker D: Decisions on energy I'm thinking is based on the battery.\nSpeaker D: I don't know.\nSpeaker D: What do people think about this kinetic battery idea?\nSpeaker B: I think it's awesome.\nSpeaker B: I think it's really cool.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I mean, totally take care of our problem of not wanting to change batteries.\nSpeaker A: Yeah.\nSpeaker A: I think it's good.\nSpeaker A: As long as we consider the cost and how reliable it is.\nSpeaker A: Right.\nSpeaker D: I haven't gotten any more information on cost other than it's more expensive than a regular battery.\nSpeaker D: But over time.\nSpeaker D: But if we're using a cheaper chip and it even out, I think circuit boards.\nSpeaker D: Yeah.\nSpeaker D: I got a whole bunch of information on how circuit boards are produced.\nSpeaker D: They're thin fiberglass with copper wires etched onto them.\nSpeaker D: And I think they're quite easily printed on by machine, which is chip on print is where the machine prints on the wires and solders it all together for us.\nSpeaker D: I don't really know what to tell you as far as decisions.\nSpeaker D: I wasn't really given any options.\nSpeaker D: I was just given that this is how they're done.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: I can't really tell you.\nSpeaker D: I can tell you a whole lot about how it works.\nSpeaker D: But I don't know any decisions on.\nSpeaker B: If they're really options.\nSpeaker D: Yeah.\nSpeaker D: I'm sorry.\nSpeaker D: Okay.\nSpeaker A: I think we'll just get to the next one.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Then we'll move on to the case.\nSpeaker B: I guess maybe if we decided on a simple, a regular or an advanced chip.\nSpeaker D: Here's the thing on the chips that I got.\nSpeaker D: Simple regular advanced chip on print.\nSpeaker D: It includes an infrared sensor, so we don't have to worry about that.\nSpeaker D: Okay.\nSpeaker D: The push button, if we're going to have push buttons, they require a simple chip.\nSpeaker D: But a scroll wheel requires a regular chip and an LCD requires advanced.\nSpeaker D: Do we want a scroll wheel or do we just want push buttons?\nSpeaker A: I don't think we need a scroll wheel.\nSpeaker A: I mean, it might be nice for changing the volume.\nSpeaker B: But I don't know how useful it would be for changing the channel.\nSpeaker B: Because you don't have control over numbers or...\nSpeaker A: Yeah.\nSpeaker D: It'd be handy for going through if there was an on-screen menu of your channel choices.\nSpeaker D: But if you could scroll down on a scroll wheel.\nSpeaker C: If you could scroll through the channels.\nSpeaker C: And then the volume would just be...\nSpeaker B: We have five minutes left for the meeting.\nSpeaker C: And the volume would just be the same way.\nSpeaker C: Yeah.\nSpeaker C: Forward and backward.\nSpeaker C: And it would make it much sleeker sort of looking.\nSpeaker C: Yeah.\nSpeaker C: And otherwise, no matter how many buttons we have, we're going to have like, you know, black with red sticking out.\nSpeaker C: And it's going to inevitably sort of start looking like...\nSpeaker C: That was a group of sort of ugly ones that we saw stacked up.\nSpeaker B: So have a scroll for volume.\nSpeaker D: Or for all those secret functions.\nSpeaker D: When you get on the on-screen menu of all your functions that your remote...\nSpeaker D: Could do for you without the buttons.\nSpeaker D: And you could have a scroll wheel to go through those menus.\nSpeaker D: I think.\nSpeaker A: Yeah.\nSpeaker A: I think a scroll wheel would be nice, but it's not necessary.\nSpeaker A: Right.\nSpeaker D: So we could either go with a simple or a regular chip, depending on...\nSpeaker D: Maybe we could table that design for later.\nSpeaker D: I don't know.\nSpeaker B: I think, well, I think when we go into the user phase, we're going to have to decide...\nSpeaker B: The interface, we're going to have to decide whether we're going to have a scroll or not.\nSpeaker D: Well, let's think about that while we talk about the case.\nSpeaker D: Okay, let's do case.\nSpeaker D: I'm kind of liking the idea of latex.\nSpeaker D: If spongy is the end thing.\nSpeaker B: I'm a little hesitant about it because I'm worried about protecting the stuff on the inside.\nSpeaker B: Okay.\nSpeaker C: Well, could it be hard and then something around?\nSpeaker B: Yeah, I would be more okay with like a titanium actual thing and then maybe like a mobile phone.\nSpeaker D: What I've seen just not related to this.\nSpeaker D: But of latex cases before is that there's a lot like a hard plastic inside and it just covered with the latex.\nSpeaker D: Okay.\nSpeaker D: Not too thick a layer of latex, just enough to be grippable like bicandles or anything that you've seen like that.\nSpeaker D: The inside is hard.\nSpeaker D: I don't think we need to worry about protecting the circuit board.\nSpeaker D: I think that that's done for us.\nSpeaker B: Okay.\nSpeaker B: So we do want latex.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So, um, I'm probably in colors, maybe fruity vegetable colors.\nSpeaker B: Yeah.\nSpeaker B: Fruity colors.\nSpeaker B: Okay.\nSpeaker B: Um, let's go to the user interface then.\nSpeaker B: We'll come back to the chip, I suppose.\nSpeaker D: Oh, and we want a curved case.\nSpeaker D: Yeah.\nSpeaker D: Or a double curved.\nSpeaker B: Well, we don't really know what the difference is, right?\nSpeaker C: I think it's curved as some sort.\nSpeaker D: Yeah.\nNone: Yeah.\nSpeaker D: Okay.\nSpeaker D: So, um, okay.\nSpeaker B: Interface the type and the supplements.\nSpeaker B: So, push or scroll, right?\nSpeaker B: Yeah.\nSpeaker A: Or both.\nSpeaker A: Um, and I think if we want to keep our costs, um, we should just go for push buttons.\nSpeaker A: Because then we can have a simple chip and it's simpler.\nSpeaker A: It's cheaper to make push buttons than it is a scroll button.\nSpeaker A: Good point.\nSpeaker A: So, in terms of economics, it's probably better than push buttons.\nSpeaker B: And if we had an on screen, um, kind of thing that you could scroll through, like you could use your buttons to scroll through things.\nSpeaker B: Yes.\nSpeaker B: I think that, yeah, for channel surfing, I think a scroll, an actual, like an iPod's kind of scroll thing would be too fast.\nSpeaker B: Yeah.\nSpeaker B: I don't know.\nSpeaker D: I say push buttons, at least unless we get any information that I have no idea how much more expensive a scroll wheel is than a push button, but it's got to be some more expensive.\nSpeaker D: So, I think it might be better to put our money into the stuff like kinetic battery in the cool case.\nSpeaker D: Is that okay with you?\nSpeaker B: Yeah.\nSpeaker B: And let's like see if we get anything else.\nSpeaker C: I mean, I'm not like hard sold on the scroll wheel.\nSpeaker C: It's more just to give it a different kind of look.\nSpeaker C: But if it's going to be in a latex type thing, and that's going to look cool, then that's probably going to have a bigger impact than the scroll wheel.\nSpeaker C: Yeah.\nSpeaker D: Okay.\nSpeaker B: So we're going to go with type push buttons and then supplements.\nSpeaker B: How are we going to do that?\nSpeaker B: What do you mean by supplements?\nSpeaker B: I assume that's what else we're going to, like the additional buttons we can use.\nSpeaker B: So we're going to have like a menu button so that we can access on screen things.\nSpeaker B: Yeah.\nSpeaker A: Then, okay.\nSpeaker A: So we're doing an on screen menu.\nSpeaker A: So what are buttons going to be on off?\nSpeaker B: So like one through five or like a radio type sort of situation.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nNone: I feel like radio presets.\nSpeaker A: I think maybe numbers seems to come off.\nSpeaker B: Well, but in order to preset it, oh, you guess you can just hold it down when you get to one.\nSpeaker A: When you're scrolling through.\nSpeaker A: You need some kind of a nice up-down kind of button.\nSpeaker A: Yeah.\nSpeaker B: Up-down.\nSpeaker B: A volume control, a double, but for example.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Finishing the meeting now.\nSpeaker B: Our next meeting starts in 30 minutes.\nSpeaker B: You each have things to do, look and feel, design user interface design, product evaluation.\nSpeaker B: And you two are going to work together on a prototype using modeling clay.\nSpeaker B: You'll get instructions from your personal coach.\nSpeaker B: Wow.\nSpeaker B: Do we decide on a chip?\nSpeaker B: Let's go with a simple chip.\nSpeaker B: Okay.\nSpeaker B: We are done.\nSpeaker B: Thank you, everyone.\nSpeaker B: Oh, I think these are already in our shared folder.\nNone: Okay.\nSpeaker D: Cool.\nSpeaker D: Okay.\nSpeaker D: It's not too hot.\n"}, {"meeting_id": "ES2008b", "summary": "This was the second meeting with functional design discussion purpose. Firstly, the group gave three individual presentations on working design, technical functions and functional requirements. They mentioned the importance of the fancy appeal, the practicality and the quality of components. Then, the group had a discussion about general requirements on the remote control. The group decided to market the remote as a separate product and focused on the user-friendly as well as simplicity, instead of too much technological advancement. Also, they agreed to target on income group and would add extra functionalities like keep lost and rechargeable stations for the remote control.", "dialogue": "Speaker E: Good to see you all again.\nSpeaker D: This is our functional design meeting.\nSpeaker E: Okay, we put the fashion in electronics.\nSpeaker E: Okay, our agenda today, just took time.\nSpeaker E: I'm going to do an opening, talk about, did you all get the minutes?\nSpeaker E: I emailed them to you.\nSpeaker E: I'm also putting them in the shared folder.\nSpeaker E: We'll talk about our general objectives and have your three presentations.\nSpeaker E: I'll talk about the new project requirements I've just received.\nSpeaker E: And then we have to make a decision on remote control functions.\nSpeaker E: Finally, we'll just close.\nSpeaker E: Starting this meeting at approximately 12, 13, we have 40 minutes.\nSpeaker E: First of all, the functional design objectives.\nSpeaker E: We need to keep in mind the user requirements specification, what needs and desires are to be fulfilled, the textual functions design, what effects the apparatus should have, and the working design, how the apparatus actually works to fulfill its function.\nSpeaker E: We have three presentations.\nSpeaker D: Go in any order you choose.\nSpeaker C: Sure, please do.\nSpeaker A: Rest.\nSpeaker A: Put the spoolock in.\nSpeaker D: rise down again...\nNone: touch the blue moon, open?\nSpeaker E: Oh, it's like... is it plugged in all the way and you'd screwed it in and...\nSpeaker B: Oh, wait, it's screwed.\nSpeaker C: I think you just have to push it in really hard.\nSpeaker A: That's it.\nSpeaker C: Oh, got it.\nSpeaker D: Let's tighten it a little bit.\nSpeaker B: I'm not attached to anything.\nSpeaker B: There you go.\nSpeaker B: Alright.\nSpeaker B: So, I don't know if you guys are able to get access to the report that was online or if I'm the only one who is, but I don't even know how to play this.\nSpeaker E: Press the little presentation.\nSpeaker E: It's the... it looks like a Y kind of over there above draw.\nSpeaker E: There.\nSpeaker E: That one.\nSpeaker E: Yeah, right.\nSpeaker B: So, we're just going to talk a little bit about functional requirements that people specified when they were asked.\nSpeaker B: I guess real reaction did some market research.\nSpeaker B: They had 100 subjects in their usability lab and they watched them, watched TV and recorded the frequency that they used particular buttons and the relevance that those buttons had.\nSpeaker B: What they found was they analyzed people's desires and needs.\nSpeaker B: Focusing on their desires, people specifically said that they thought remotes were ugly.\nSpeaker B: 75% of the 100 subjects noted that and that they, more importantly though, 80% said that they would be willing to pay more for a fancier looking remote.\nSpeaker B: I don't know anything beyond what fancy means, but that's particularly of use to us, I think.\nSpeaker B: Also, they did some questions on voice recognition and found that the desire for voice recognition was inversely related to age.\nSpeaker B: So, younger people were more inclined to want something with voice recognition, whereas the older people in the 60 and above segment or so did not really think that they would pay more money for voice recognition.\nSpeaker B: People also had certain frustrations that I think that we could try to take into consideration with our design, that being people frustrated with losing their remotes.\nSpeaker B: I think over 50% of the people mentioned that that was their biggest frustration.\nSpeaker B: People were also frustrated with the difficulty it is to learn how to use a remote and I think that ties back to what you were saying before, just that there's too many buttons, it just needs to be easy to use.\nSpeaker B: It also mentioned something called RSI and I was hoping someone might be able to inform me as to what RSI is, because I don't know.\nSpeaker A: What?\nSpeaker A: There we go.\nSpeaker B: Well, people did not like that, so I guess sort of the carpal tunnel type thing.\nSpeaker B: People did not like that repetitive use, I guess, caused a strain.\nSpeaker B: Looking at the needs people specified, the problem right now is that people's remotes are not matching their operating behavioral.\nSpeaker B: People are only using 10% of the buttons that they have offered to them on their remote.\nSpeaker B: What people do most often is changing the channel and changing the volume.\nSpeaker B: People also zap to change the channel, about 65% during an hour of use.\nSpeaker B: We really just need to focus in on those volumes and channel changers, rather than things like the audio settings, the screen settings, and the channel settings, because they're used much more infrequently and probably just complicate what's going on.\nSpeaker B: So I think that some things that we might want to think about, the idea of an LCD screen was brought up, although they didn't have any details on what people's preferences on that were, so I don't know if that's coming to me later, something like that, but something for us to consider.\nSpeaker B: Also just the phenomenon that less is more when it comes to the buttons on their remote or what we want to make, easiest to use, make sure that something like an audio setting is given as much importance and visibility on the remote as something like channel changing that's used a lot more often.\nSpeaker B: And basically in order for us to win over to the consumer, we just need to focus on what it looks like that it has a fancy appeal and that it's not ugly, and that it feels like the way they're going to use it so it doesn't give them any hand injuries or things like that.\nSpeaker E: Thank you very much, that was great.\nSpeaker E: Let's move on to the next presentation on effects.\nSpeaker E: Is that you?\nSpeaker D: Great.\nSpeaker B: I think that's in shared if I did it right if anyone wants to look at it.\nSpeaker B: Okay, thank you.\nSpeaker D: Okay.\nSpeaker A: So I'm going to talk about the technical functions of the line of the remote control.\nSpeaker A: You need to start by considering what the remote control actually is.\nSpeaker A: It's a device that allows us to produce certain effects on our television.\nSpeaker A: So it's basically a communication device.\nSpeaker A: We tell the remote control what we want to be.\nSpeaker A: It sounds a message to the television saying change your channel or change your volume.\nSpeaker A: Adjusting the settings, adjust the brightness.\nSpeaker A: How do we actually go about designing any television remote control?\nSpeaker A: First thing to do is to come up with the design specifications.\nSpeaker A: We need to know what our final product is going to be like.\nSpeaker A: So we need a clear idea of exactly what this product does, how it works, and what the end user is going to want from this product.\nSpeaker A: A way I'd suggest that we could go about this is by designing several different prototypes of user interfaces for this product.\nSpeaker A: And then trying to get some feedback about how well these particular prototypes work.\nSpeaker A: So to find out what people think of them.\nSpeaker A: Using a remote control is quite a subjective experience.\nSpeaker A: And different people just refer different things.\nSpeaker A: We should remember that remote controls are a fairly standard piece of equipment when the user is using a remote control.\nSpeaker A: He or she expects the buttons to be in certain places.\nSpeaker A: So in some sense we're going to have to aim for the device which is fairly conventional in design so that we're not completely shocking people.\nSpeaker A: But I think within that there is also a room for us to introduce novel ideas and to make something that's perhaps a little bit different, something that stands out.\nSpeaker A: Also in designing the user interface we need to consider practicalities.\nSpeaker A: The first of these is technological.\nSpeaker A: What can we do with the current state of technology as it is?\nSpeaker A: The second is economic.\nSpeaker A: We need to find the balance between features and vice.\nSpeaker A: So as you mentioned things like voice recognition would add to the vice.\nSpeaker A: But it would also improve the design of the product.\nSpeaker A: So I had a look on the web to see if I could find a few examples of existing television in the remote controls.\nSpeaker A: In analyzing these we can consider what things go about them, what things do they get right, what's bad about them, what's wrong with them, how we can improve the designs that are found and what can we do to make our product stand out from the large majority of the remote controls.\nSpeaker A: Here's two examples, probably of the extreme ends of the spectrum.\nSpeaker A: On the left here we've got an engineering based design for the remote controls.\nSpeaker A: So it's one that's got lots of buttons, it's fully featured.\nSpeaker A: Everything you might possibly want to do is just go forward backwards, up, down, channel numbers, volume, these things.\nSpeaker A: It's fully featured and it might take a while to get to learn to use it, but once you've learned it, you can do whatever you want with your TV.\nSpeaker A: One on the right is a lot more basic.\nSpeaker A: It's just got the essential functions of the TV, changing the channel place, not volume.\nSpeaker A: It would be a lot quicker to learn to use it, but again it swings a lot about the design.\nSpeaker A: It's about the disadvantages you can't use it, so to please the television picture.\nSpeaker A: There's a lot of features that are missing from that remote control.\nSpeaker A: So we've got to find a way of striking a balance between the two.\nSpeaker A: As I said before, the very controls are subjective.\nSpeaker A: Different people want different things.\nSpeaker A: Personally, what I want remote control is a device that's simple. It's easy to use.\nSpeaker A: It's got big buttons for doing the things I do most often, changing the volume, changing the channel.\nSpeaker A: It does everything that I need it to.\nSpeaker A: As I said before, I'm quite lazy. I don't want to walk across the view just to adjust my television.\nSpeaker A: I also want something that looks cool and that feels economically designed.\nSpeaker E: Thank you very much. That was very useful.\nSpeaker E: It's funny to see the drastic difference between those two remotes.\nSpeaker E: Neither of them were very pretty.\nSpeaker C: I think that could be our selling point.\nSpeaker A: Fashion, fashion, the mode, right?\nSpeaker A: There's certainly a market technology that looks cool.\nSpeaker A: That's my company like Apple.\nSpeaker C: I really can't see what I'm doing, so does anyone have a...\nSpeaker C: There it is.\nSpeaker C: Ah, look at that.\nSpeaker C: I've been up already.\nSpeaker B: So, wait, did it let you go on the internet or was that just what you've seen?\nSpeaker A: That was just on the internet.\nSpeaker A: Okay.\nSpeaker B: Because I was like, yeah.\nSpeaker C: You're willing to go to my quiz.\nSpeaker C: Let me go.\nSpeaker C: All right.\nSpeaker E: No? How do I play again?\nSpeaker E: It's right above draw.\nSpeaker E: It's with bottom.\nSpeaker E: There are three icons and it's the one that looks like a desk.\nSpeaker C: So, this is our working design presentation.\nSpeaker C: I had a bit of some issues with this because I wasn't able to find everything I needed, but I guess that's...we're still in early stages.\nSpeaker C: So, yeah, this is this.\nSpeaker C: The thing about working design is what we're trying to do as a working designer is figure out how the apparatus can fulfill its function.\nSpeaker C: One of the examples that kept coming up for me is that a coffee grinder.\nSpeaker C: It works because it converts electrical energy to grinding the beans and then put the beans through a filter and that filters out.\nSpeaker C: And then you get coffee at the end.\nSpeaker C: That's nice and hot because of the combination of electrical energy and then other things that are brought in to make it work.\nSpeaker C: I don't know if I'm explaining that very well, but how do I get to the next...\nSpeaker C: So, the method as working designers figure out what you need to make it fulfill this practical function.\nSpeaker C: What needs to be done and how do we convert all the elements to make that done?\nSpeaker C: So, the easiest thing to do is to break down all the points at which you need something to happen.\nSpeaker C: So, you make a schematic of the way that the energy is converted towards this practical function.\nSpeaker C: And then I think the easiest thing to do is going to be work on each task separately.\nSpeaker C: So...\nSpeaker C: You just press the edge of the back.\nSpeaker C: So, the findings that I got very...just very briefly is that you have a choice of the way that the information is projected to the receiver.\nSpeaker C: And in my opinion, infrared is the best way to do that because you don't need a sight line.\nSpeaker C: So, that's one thing we're going to work on.\nSpeaker C: The user interface is critical here because a lot of the things that happen in a remote control happen through the chip that converts the electrical energy into data, which then goes through the infrared.\nSpeaker C: So, the chip that I think Ian is designing is going to be crucial.\nSpeaker C: And really it all comes down to the user because they're the one that's controlling most of the working design.\nSpeaker C: So, the components that we find here are the energy source, you know, the battery or whatever that's going to make it work.\nSpeaker C: Then the chip, which converts the data, the user that's controlling the chip and the infrared bulb that's going to let us move the data to the receiver.\nSpeaker C: So, you have four main components.\nSpeaker C: And they are designed sort of like this. You have your energy source right there, which then brings energy or information to the chip, which is also controlled by the user.\nSpeaker C: So, you have energy going to the user who's controlling the chip.\nSpeaker C: Ooh, that's not what I wanted to do.\nSpeaker E: Yeah, use that thing now that you can go back. Previous. Sorry about that guy.\nSpeaker C: Oh, no.\nSpeaker C: Okay, let's just get back to my skin out of here.\nSpeaker E: Double click on it with the right with the left hand one.\nSpeaker C: Yeah, I am.\nSpeaker C: I think it's frozen here. Don't show me that.\nSpeaker C: There we are. Sorry about this guys. I'm kind of pathetic with things like this.\nSpeaker C: All right, so you have your energy source, your user interface who's controlling the chip.\nSpeaker C: The chip also uses the energy and the chip through the use of the user interface is going to control the switch, which will work your infrared bulb, which will then bring the data to the receiver.\nSpeaker C: So, hopefully that makes sense for everyone in my kind of garbled way.\nSpeaker C: This is the parts of the working design that need to be figured out.\nSpeaker C: And personal references, it's a fact that I can't spell.\nSpeaker C: We need a long lasting energy source. People don't want to be changing it a lot.\nSpeaker C: We need a chip that works well with the user interface that isn't too complicated.\nSpeaker C: We get a straightforward interface like Ian was saying, simple, not overwhelming it with information.\nSpeaker C: And we need a reliable and clear infrared signal so that you're not waving your remote around it and having to get frustrated and go do it by hand.\nSpeaker C: That's pretty much it for the working design.\nSpeaker D: Excellent.\nSpeaker C: Let's do a second.\nSpeaker C: Do you think you can give me a similar? Yes, absolutely.\nSpeaker D: Ready, tidy, right? Lefty, loosey, ready, tidy.\nSpeaker E: I'm going to have to do.\nSpeaker E: I'm going to have to do a second.\nSpeaker E: We're not going to work with teletext because that's been taken over by the Internet so we're not going to worry about that.\nSpeaker A: No, no. Would you care to explain it?\nSpeaker A: I'm going to have to do a second.\nSpeaker C: I'm going to go to page 160 on C-fax now.\nSpeaker E: We're going to, like your question earlier, whether this is going to be for television, et cetera, just for television.\nSpeaker E: That's what we're focused on.\nSpeaker E: Otherwise, it becomes too complex. We want to get this out very quickly.\nSpeaker E: We only have a short amount of time.\nSpeaker E: And finally, this is more marketing, I think.\nSpeaker E: Our corporate image has to be recognizable.\nSpeaker E: While we're going to make it look pretty, we need to use our color and our slogan in the new design.\nSpeaker A: What's our corporate image like? It's kind of yellow color as well.\nSpeaker E: It's kind of a yellow and black and we have the emphasis on the R's in...\nSpeaker E: But we put the fashion and electronics, so we've got to keep that in mind.\nSpeaker C: So we want something that looks good.\nSpeaker E: Maybe some buttons could be yellow. We can play around with it a little bit.\nSpeaker E: We need to talk about our functions and our target group.\nSpeaker E: We decide who our target group is. You, in your analysis of different marketing, you identified that different groups wanted different things or had certain preferences.\nSpeaker E: For example, that older people didn't really care for voice recognition, but younger people did.\nSpeaker E: So, who are we ending this at?\nSpeaker C: If we're going to say that function, fashion, we put the fashion in electronics, then automatically I think, sort of younger group, that's who's going to be attracted to this.\nSpeaker B: We also think who's going to have the money to buy that's possible.\nSpeaker E: It's going to be 25 euro, remember. So, it has to be marketable to whoever it is.\nSpeaker A: Is it something that's going to be something that's going to be something that comes with a ton of money that's going to be marketable?\nSpeaker E: Well, at least right now, what we're doing is deciding on just the remote itself.\nSpeaker E: So, it would probably be sold separately.\nSpeaker E: 25 euro by itself.\nSpeaker B: The only breakdown that I was given in terms of age was the voice activation, and basically, there's a big jump after 35 people don't really care if it has voice.\nSpeaker B: So it's basically 15 to 35 that thinks that that's a good idea.\nSpeaker B: I don't know if I'm going to be given any other numbers broken down in terms of age later, but if that's sort of the only quality that we have that is divided into age, then we would want to stick between a 15 and 35 range.\nSpeaker C: That's probably a population of quite a little bit of disposable income for use on technology anyway, so that might be a fairly good target group for us.\nSpeaker E: Now that's all specific for speech recognition. Are we going to use that as one of our functions?\nSpeaker A: I would say no, because it's going to have too much interest, especially if we are marketing to the people who are going to be paying the 12 and a half year list.\nSpeaker A: To produce it? Yes.\nSpeaker A: I want to get place for permission to do that 12 and a half year list without money to make it a couple of months.\nSpeaker B: But what else are we going to put, I mean, not that I'm really going to tell about it, I don't know what else you can put in a remote to make it technologically advanced though, so like other than just making it look good, how is it, I mean, it can look really great, and still have the same up and down button, and why isn't I going to buy a new remote?\nSpeaker C: Why are we aiming for technological advancement?\nSpeaker C: Everything we're talking about is ease of use and simple, and that doesn't necessarily mean more technology.\nSpeaker C: That's a good thing to keep in mind.\nSpeaker C: If they might be overwhelmed with the remote that have too many buttons and too much technology,\nSpeaker B: if someone's looking to buy a new remote, don't they want an upgrade?\nSpeaker E: Well, we can look at upgrade or we can look at user friendly. Yeah, simplification.\nSpeaker C: It could have a crap remote that came with our TV that's just impossible to use, or maybe it broke, or maybe they're just missing it.\nSpeaker E: And we also need to talk about if we're only going to have the very simple ones, or also have the other ones just separate somehow, or...\nSpeaker B: Can you like, I mean, this might be too complicated, but I wish I had someone to explain it.\nSpeaker B: Like, if it was just a simple, either this way or this way, that had the mean buttons, and then you could pull something out kind of, and you got the rest of the buttons, but the rest of them went in. Do you know what I'm saying?\nSpeaker A: The remote can pull out of the side.\nSpeaker A: Yeah, like some TVs they have a slight experience on the remote control, but it's all the complicated buttons.\nSpeaker A: So if you want to do something complicated, like on some of it, it won't be too many.\nSpeaker A: You open up this little hat, and it slides me down in that little special buttons.\nSpeaker B: It sends people who don't want to ever look at them, never even have to see them.\nSpeaker B: And if you get the instruction manual that comes with it, and you just don't even read it, then you'll never even know that those things can pull out.\nSpeaker B: And you're a happy person, and everyone else doesn't have to have, like, two remotes, one that has the easy ones, and one that has more complicated ones.\nSpeaker D: That's a good idea.\nSpeaker E: We have to be careful that that doesn't impede the chip transmitting information.\nSpeaker E: Yeah.\nSpeaker E: But that's going to be mostly technological things.\nSpeaker E: Okay, so what are we emphasizing when in this project?\nSpeaker E: Simplicity and fashion.\nSpeaker E: Okay, those are very good goals, I think, that we have to keep in mind it with everything we do.\nSpeaker E: Simplicity and fashion, and we'll use abilities, however you want to say that, which includes an emphasis on making the infrared very functional, so that you don't have to travel around a lot.\nSpeaker B: What can you do to make the infrared more functional?\nSpeaker B: Like, why would it not be?\nSpeaker E: I think it's a lot to do with battery, but that's just my...\nSpeaker C: The battery, and that I think that the chip takes the data and presents it well without sort of scattering.\nSpeaker B: So it's just the quality of the chip?\nSpeaker C: Yeah, I think so.\nSpeaker C: The quality of all the components, really.\nSpeaker C: I mean, we can't really do anything.\nSpeaker C: Shotting work, because it's going to be visible down the line.\nSpeaker E: So our target group, we're saying 15 to 35?\nSpeaker B: Well, I don't know how useful that number is if we're not doing...\nSpeaker B: Yeah.\nSpeaker E: Which I kind of feel like voice recognition is not necessary in a remote control.\nSpeaker E: Like, it might be necessary for a TV, but not for the remote, you know?\nSpeaker A: Seems a little bit.\nSpeaker E: Yeah, and it might be too expensive.\nSpeaker B: And if the whole idea is you're using a remote, then why would you have a voice?\nSpeaker B: Like, you mean, then it's like you would need a remote, you would just talk to your TV.\nSpeaker A: Yeah.\nSpeaker A: Oh, I think we're going to be able to pick up the mic.\nSpeaker B: Yeah.\nSpeaker B: Maybe, I mean, if I get more numbers, I'll email you before the next meeting in terms of age-res.\nSpeaker B: But this doesn't divide up anything, and there was only 100 subjects also.\nSpeaker B: Which isn't... I mean, really isn't that representative, especially if it's 100 subjects so that they then can subdivide into age groups.\nSpeaker B: That means that very few in each age group, so...\nSpeaker C: I think regardless, we're aiming for the understanding.\nSpeaker C: The under 65 or something.\nSpeaker E: Under 65.\nSpeaker E: Okay, that's a good start.\nSpeaker E: I'd say we're...\nSpeaker E: Can we narrow it down to maybe teenagers and families?\nSpeaker E: Because that would go up to like 50...\nSpeaker C: Think of professionals or something.\nSpeaker C: I think so.\nSpeaker C: I think so.\nSpeaker C: Like 55.\nSpeaker D: Yeah.\nSpeaker E: It's hard to narrow it down.\nSpeaker C: It's hard to figure it out right now.\nSpeaker A: I think it's hard to figure it out right now.\nSpeaker A: But it feels impossible to write about age.\nSpeaker A: And we've said simplicity is one of the features, so it's going to appeal to people.\nSpeaker A: Maybe people who have problems with technology.\nSpeaker A: Okay.\nSpeaker A: And I think that's a good question.\nSpeaker A: That might be older people.\nSpeaker A: But then we've also got fashion, which is something that definitely...\nSpeaker A: Yeah.\nSpeaker B: Well, maybe we don't have to define this hard group by the demographic of age.\nSpeaker B: Maybe we can define it by the demographic of...\nSpeaker B: Like how much money they have to spend or something like that.\nSpeaker A: That's a good question.\nSpeaker B: Obviously, that's because someone who owns a television.\nSpeaker B: And like how recently have they bought their television?\nSpeaker B: Like that sort of thing.\nSpeaker E: So maybe it's more useful to define objectives.\nSpeaker E: Like fashion and simplicity than define a specific target group as far as age is.\nSpeaker E: Because yeah, things will different...\nSpeaker E: Will appeal to different people.\nSpeaker E: But...\nSpeaker E: Okay.\nSpeaker E: Oh, there are a couple of functions.\nSpeaker E: Do we want something so that...\nSpeaker E: Do we want some kind of thing to find it if it's lost?\nSpeaker E: Like a button on the TV, you can press and it'll ring or something?\nSpeaker E: I don't know.\nSpeaker E: Or beep?\nSpeaker B: I guess I have a more 50% of the frustration someone could have.\nSpeaker B: That was the biggest one.\nSpeaker B: And half people said that happened.\nSpeaker B: And we all mentioned it before we knew about it.\nSpeaker B: And if we're talking about making it make it easy, that sort of goes along with it so it wouldn't be like a random thing to sort of add in.\nSpeaker B: It would be relevant to like the overall goal.\nSpeaker C: Yeah.\nSpeaker E: Okay.\nSpeaker E: We have about four minutes left to define our functions.\nSpeaker E: So let's do that quickly.\nSpeaker E: So we want something to keep it from getting lost.\nSpeaker E: Yeah.\nSpeaker E: And we want...\nSpeaker E: We want large buttons for the essential things.\nSpeaker E: Yeah.\nSpeaker E: Large accessible buttons for the essentials.\nSpeaker E: We want a possibility to get the extra functions.\nSpeaker A: Yeah.\nSpeaker E: And we also want it to be fashionable.\nSpeaker E: Which I'm not sure if that's a function so much as a...\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker E: Okay.\nSpeaker E: Do any other essential functions that we need?\nSpeaker E: Battery.\nSpeaker E: Do we need a long life battery?\nSpeaker C: Battery use.\nSpeaker C: Yeah.\nSpeaker C: But I think that goes in with simplicity and ease of use really.\nSpeaker B: Both of them.\nSpeaker B: I think that's a good thing.\nSpeaker A: Yeah.\nSpeaker E: I would imagine just because it is an electronic device.\nSpeaker C: I think it does.\nSpeaker C: I don't think of a way you can operate a chip and convert that much data without one.\nSpeaker C: But you could maybe have it in a little charging station like a mobile phone.\nSpeaker C: Or like a little cradle for your iPod.\nSpeaker C: It's probably...\nSpeaker C: You could maybe do that instead.\nSpeaker C: So you don't have...\nSpeaker C: Or like a rechargeable battery.\nSpeaker C: I don't know.\nSpeaker C: That contributes to less people losing it too.\nSpeaker C: It stays with one place.\nSpeaker E: We have to think about space in living rooms too.\nSpeaker E: Like...\nSpeaker E: I mean, would you put it on top of the TV?\nSpeaker E: I don't know.\nSpeaker E: I just think...\nSpeaker E: Okay, that's a good idea.\nSpeaker C: We'll keep it.\nSpeaker C: Yeah.\nSpeaker E: Let's just soft it out my head.\nNone: Okay.\nSpeaker E: Okay, we're going to conclude now.\nSpeaker E: As everyone said, their functions.\nSpeaker E: Okay.\nSpeaker E: After the meeting, we're going to each complete questionnaire in a summer summary.\nSpeaker E: I don't know why it's summarization.\nSpeaker E: And then we'll have our lunch break.\nSpeaker E: After that, we have 30 minutes of individual work.\nSpeaker E: I'll make sure to put the minutes in the project documentation and including this PowerPoint slides.\nSpeaker E: If everyone could do that as well, that'd be great.\nSpeaker E: You'd have individual actions, ID, industrial design, if your components concept, user interface, the user interface concept, market trend watching.\nSpeaker E: And as per last time, you'll get specific instructions from your personal coach emailed to you.\nSpeaker E: And hopefully, I hope next time you'll be able to access more of the website that they seem to tell you that you could.\nSpeaker E: Yeah, it's kind of frustrating.\nSpeaker E: Be sure to note any frustrations or any issues that come up in your in your.\nSpeaker E: Summery.\nNone: Okay.\nNone: Okay.\nSpeaker C: That was good.\nSpeaker C: Great.\nSpeaker E: No.\nSpeaker C: That's you.\nSpeaker B: Did you, were you able to put yours in the group for?\nSpeaker B: Yes, I just did that.\nSpeaker C: Hopefully it is there for people.\nSpeaker C: Yeah.\nSpeaker E: It looks like there are.\nSpeaker E: Looks like there's a second one kind of mine.\nSpeaker E: Yeah.\nSpeaker E: That I didn't do.\nSpeaker E: It's from like an earlier project, I think so.\nSpeaker E: Okay.\nSpeaker B: Where is that?\nSpeaker E: Under the shared folder.\nSpeaker E: I don't know.\nSpeaker E: It might not even be under here as well.\nSpeaker C: So in there, we have technical functions, presentation, working design, presentation, and the functional requirements.\nSpeaker C: At least that's why I'm having three.\nSpeaker B: I just have like our three.\nSpeaker C: Yeah.\nSpeaker C: That's what I have as well.\nSpeaker E: You don't have mine.\nSpeaker B: No, but that's because I think yours is in the email separately.\nSpeaker B: It's not on the server.\nSpeaker B: Yeah.\nSpeaker B: But if I open it and then save it, probably will be there.\nSpeaker B: Oh, wait.\nSpeaker B: Never mind.\nSpeaker B: You can't say.\nNone: Okay.\nSpeaker E: Well, I'll figure that out in the meantime.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: Alright.\nNone: Okay.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: All right.\n"}, {"meeting_id": "ES2004a", "summary": "Project Manager introduced a new remote control project. The team started to know each other by drawing favourite animals. Project Manager proposed to price each remote control at 25 Euros, considering the 12.5-Euro production cost. The market range would be international and over all age groups. After the financial targets, the team discussed the remote design. Despite the \"one-for-all\" style, the remote was still supposed to be user-friendly. Standard features such as sorting through channels and adjusting volume would not lose their priority. The team, therefore, agreed on a style of basic buttons with a self-intuitive screen menu. Silver lightweight plastic was temporarily decided to be the material, for it is easily moulded and cost-saving.", "dialogue": "None: Are we not allowed to do the lights?\nSpeaker C: Are we not allowed to do the lights?\nSpeaker C: I might be standing up there.\nSpeaker B: So we've got both of these clipped on.\nSpeaker B: She can only see me.\nSpeaker B: Yeah, both of them.\nSpeaker B: Yeah.\nNone: God.\nSpeaker B: I'm just going to fall off.\nSpeaker C: Okay.\nSpeaker C: Hello everybody.\nSpeaker C: I'm Sarah, Project Manager.\nSpeaker C: And this is our first meeting.\nSpeaker C: It's a present meeting.\nSpeaker C: Okay, this is our agenda.\nSpeaker C: We will do some stuff, get to know each other a bit better, and then we'll go do chill training, talk about the Project Plan, discuss our own ideas and everything.\nSpeaker C: And we've got 25 minutes to do that as far as I can understand.\nSpeaker C: Now we're developing remote control, which you probably already know.\nSpeaker C: We want it to be original, something that people haven't thought of as it's not out in the shops.\nSpeaker C: It's a trendy, appealing to a wide market, but not a hunk of metal.\nSpeaker C: And user-friendly, granny to kids, maybe even coaches should be able to use it.\nSpeaker C: Okay.\nSpeaker C: First is the functional design.\nSpeaker C: This is where we all go off into our individual work.\nSpeaker C: What needs need to be fulfilled by the product, what effect the product has to have, and how it's actually going to do that.\nSpeaker C: Conceptual design, what we're thinking, how it's going to go, and then the detailed design, how we're actually going to put it into practice and make it work.\nSpeaker C: Okay, right.\nSpeaker C: We're going to practice with the pens and draw our favorite animal on the whiteboard, agrofers, and sum up the characteristics of that animal.\nSpeaker C: So, okay.\nSpeaker C: I'll just space for everyone else.\nNone: Okay.\nSpeaker C: We're running it, please.\nSpeaker C: Okay.\nSpeaker C: I'm not going to ask you to guess.\nSpeaker C: I'm going to tell you that it's supposed to be a tiger.\nSpeaker C: And I see them as majestic and independent and proud.\nSpeaker C: No.\nSpeaker C: We'd like to go next.\nSpeaker C: Okay.\nSpeaker D: Okay.\nSpeaker D: This is from.\nSpeaker D: Thank you.\nSpeaker D: Maybe you can guess what I'm trying to make.\nSpeaker B: I can't talk.\nSpeaker C: Yep.\nSpeaker D: It's actually sitting.\nSpeaker D: So, it's sitting.\nSpeaker D: It's not standing.\nSpeaker D: Okay.\nSpeaker D: I see it as one thing.\nSpeaker D: It's very supportive.\nSpeaker D: It's your best friend in your.\nSpeaker D: You can talk to a dog.\nSpeaker D: It can be a best friend.\nSpeaker D: It doesn't discriminate between you.\nSpeaker D: Second, it's loyal.\nSpeaker D: And third thing, it's got intuition.\nSpeaker C: It's a good dog.\nSpeaker B: Please leave me a space at the bottom, I'm little.\nSpeaker B: We're standing on a chair.\nSpeaker A: Since you guys have chosen the ones I wanted to do.\nSpeaker A: Okay.\nSpeaker A: I have to go to the same.\nSpeaker A: And also my drawing scale isn't that great.\nSpeaker C: Well, as you can see, the quality of the work today.\nSpeaker A: I think it's outstanding.\nSpeaker A: We're good.\nSpeaker A: Okay.\nNone: Now I'm going to have to change what it was originally going to be because that looks like a beak now.\nSpeaker B: Crocodile.\nSpeaker A: Yeah, it can be a crocodile.\nSpeaker A: It can be a crocodile.\nSpeaker A: Well, it was, it was a firstly, it was an attempt at a T-Rex.\nSpeaker A: And then it sort of changed into a pelican, but it can be a crocodile.\nNone: That's lovely.\nSpeaker C: Yeah.\nSpeaker A: And I have to think on the spot of things that it is.\nSpeaker A: Scary.\nSpeaker A: Strong.\nSpeaker A: Yeah, that's bad.\nSpeaker B: I'm very impressed with your artistic skills.\nSpeaker B: Minds are dreadful.\nSpeaker B: Oops, this is now coming apart.\nSpeaker B: I just put the top in.\nSpeaker B: I'll hold it on.\nSpeaker B: Oops.\nSpeaker B: Oh, dear.\nSpeaker B: What happened there?\nSpeaker C: Technical help.\nSpeaker B: Hopefully that'll stay on.\nSpeaker B: Two-handed version.\nSpeaker B: Okay.\nSpeaker B: Again, this is off the top of my head.\nSpeaker B: I was going to do a big cat too.\nSpeaker B: Oh, dear.\nSpeaker B: It's not a vampire bat, honestly.\nSpeaker B: And somewhere there's a body behind.\nSpeaker A: That's wonderful.\nSpeaker A: That's the worst.\nSpeaker B: Yeah.\nSpeaker B: It's meant to be an eagle.\nSpeaker B: Right.\nSpeaker A: You can tell it's a flying animal.\nSpeaker B: Could have been a seagull.\nSpeaker B: I never thought of a seagull.\nSpeaker B: I'm thinking on my feet, goodness.\nSpeaker B: I suppose they're also independent.\nSpeaker B: I'll put that one down again.\nSpeaker B: Dead end.\nSpeaker B: Independant, right?\nSpeaker B: Did you say they're good at golf?\nSpeaker A: Yeah.\nSpeaker A: Are they?\nSpeaker A: Oh, right.\nSpeaker B: I'm not good at golf.\nSpeaker B: I'll say they're quite free spirited.\nSpeaker B: Flying around everywhere.\nSpeaker B: Doing their own thing.\nSpeaker B: And birds of prey, aren't they?\nSpeaker B: Oh, dear.\nSpeaker B: Intrepid.\nSpeaker B: Intrepid.\nSpeaker B: There we go.\nSpeaker B: I hope that pen is going to be okay.\nSpeaker B: Whoops.\nSpeaker C: Okay.\nSpeaker C: Finance wise, we've got a selling price at 25 euros, which I don't actually know what that is sometimes at all.\nSpeaker C: Any ideas?\nSpeaker C: 17.\nSpeaker D: 1.4 euro would make a little bond or something.\nSpeaker A: Yeah, yeah, it's something like that.\nSpeaker A: Yeah, about 17.\nSpeaker A: 17.\nSpeaker A: 17 pounds.\nSpeaker A: Okay.\nSpeaker B: Should we be making notes with this?\nSpeaker B: We can just refer to this later.\nSpeaker C: I think so.\nSpeaker C: I'd be able to put that up.\nSpeaker A: Okay.\nSpeaker A: I haven't shared folders.\nSpeaker A: Having said that, though, if you want to get one of those, the ones on the market at the moment, they're about 20 pounds anyway.\nSpeaker A: Right.\nSpeaker A: Yeah.\nSpeaker A: We have to buy one.\nSpeaker B: So I suppose later it depends if we want to undercut the price.\nSpeaker B: Or is it going to make our product look a cheapy cheapy option?\nSpeaker C: Yeah.\nSpeaker C: So we have a production cost at 12.50.\nSpeaker C: So the huge margin.\nSpeaker C: Half of the selling prices.\nSpeaker C: Yeah.\nSpeaker C: Taking that by building it.\nSpeaker C: And profit aim is 15 million euros, which is a...\nSpeaker B: In our first year.\nSpeaker C: Yes.\nSpeaker C: Yeah.\nSpeaker C: I presume so.\nSpeaker B: You've got market range international, and you did to see earlier, it's got to be accessible and usable by sort of all age groups.\nSpeaker B: Just we're not focusing on business market, any particular thing.\nSpeaker B: It's everyone, user-friendly to everyone.\nSpeaker B: Okay.\nSpeaker B: Big target group.\nSpeaker B: Yes.\nSpeaker C: Yeah.\nSpeaker C: I don't think we have to...\nSpeaker C: I don't think it's a case of worrying about different languages and things like that.\nSpeaker C: Yeah.\nSpeaker C: Making that key point.\nSpeaker C: Just that it's going to be in the international market, like, Australia, America, things like that.\nSpeaker C: Okay.\nSpeaker C: What are your experiences with remote controls?\nSpeaker C: I mean, we got...\nSpeaker C: We had three videos, a TV, and a sort of AMP thing all set up.\nSpeaker C: So we got one of the universal remote controls.\nSpeaker C: Yeah.\nSpeaker C: And you program each of your things into, but that kept losing the signals, so we'd have to reprogram it every now and again.\nSpeaker C: I think it was quite a cheapy as well.\nSpeaker C: Yeah.\nSpeaker C: So that might have had fun to do with it, but that was quite good about the E-coup.\nSpeaker A: You saw the ones that didn't have them.\nSpeaker C: You didn't have six remote controls in front of you.\nSpeaker D: You want to include everything into one, like...\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker B: My experience has only been given the remote control with the object I buy, not doing any tampering with it and programming, using it to program TV videos and things.\nSpeaker B: But basically on, off, volume, up and down, channel one to that basic function, I don't think I could go any further with it than that.\nSpeaker B: So I suppose it's got to be something usable by someone like me as well.\nSpeaker C: Yeah, the main... that's the main stuff anyway.\nSpeaker C: I mean...\nSpeaker C: And you don't want to... I hate... I hate looking at a control and seeing a million tiny little buttons with tiny little words saying what they all do and just sitting there searching for the daily text button\nSpeaker B: and just symbols that you don't necessarily understand, symbols you're meant to understand that you don't.\nSpeaker C: So, when you've got the main things on the front of it, and a section opens up or something, to the other functions when you can do sound or options kind of recording, things like that inside it, because it doesn't make... when you pick it up, it doesn't make it really complicated to look at it.\nSpeaker C: Obviously, yes, on your day. Actually, that just raises a point.\nSpeaker B: I wonder what our design people think.\nSpeaker B: But you know on a mobile phone, you can press the key and it gives you a menu.\nSpeaker B: It's got a menu display. I wonder if incorporating that into the design of a remote control might be useful.\nSpeaker B: So you've got a little LCD display.\nSpeaker D: I was thinking on the same lines, you... instead of having too many buttons and make it complicated for the user, maybe have an LCD display or something like that, like a mobile?\nSpeaker D: With menus.\nSpeaker B: With menus.\nSpeaker D: Yeah.\nSpeaker D: And if it's somewhat similar to what you have on mobile phone, people might find it easier to browse and navigate elsewhere.\nSpeaker D: Yeah.\nSpeaker C: What about the older generation? What about granting granddads?\nSpeaker C: I don't understand.\nSpeaker C: My granddads can answer his mobile phone.\nSpeaker C: You can't even dream of texting or something like that.\nSpeaker B: Can he program his remote controller? Is he basic with that too?\nSpeaker C: I don't think they'd hate to.\nSpeaker C: Yeah.\nSpeaker A: My granddads are better than me using tablet tags, right?\nSpeaker B: Right.\nSpeaker B: So that's the problem regardless of any design modifications you come up with.\nSpeaker B: That's going to be a problem anyway with the older generation perhaps.\nSpeaker B: And that's another issue I would tackle that.\nSpeaker A: Why? It just needs to be... as long as it's sort of self-intuitive and you can work out why everything's doing.\nSpeaker A: Because I mean menus on sort of neat phones now, they've got all these pictures and stuff which makes it fairly obvious what you're trying to do.\nSpeaker B: I don't know how...\nSpeaker C: I don't know how.\nSpeaker C: I don't like the new phones that kind of got a Windows based running system.\nSpeaker C: I find it really confusing. I kept getting lost in the phone.\nSpeaker C: I have not got a new one but my friend got a new one and I was trying to do things with it.\nSpeaker C: And I just kept getting lost but that's just neat.\nSpeaker A: I don't know how for 25 or 12 euros 50 how much of an excellent screen you could get.\nSpeaker A: You'd have to sort of keep it down to a black and white LCD thing in your...\nSpeaker B: Is it possible that for the older generation you could have an extra button that you press for a large print like you do in large print books?\nSpeaker B: Obviously it displays less on the screen. It displays less on the screen but as long as they can read it that's the main thing.\nSpeaker C: Yeah. Or a broadband kind of a dual function in that you've got the basic buttons just for your play volume program things.\nSpeaker C: And also a menu to go into with obvious pictures, obvious symbols and that's where you control recording and things like that.\nSpeaker B: The other thing is just tucking into mobile phone design features again. You could have a flip top remote control so that when you flip over the top your screen is...\nSpeaker B: You can have a bigger screen in the flip over.\nSpeaker C: I think that's a cost thing. I don't know how much we're going to know about.\nSpeaker D: It might save a bit of space instead of looking bulky it might look small.\nSpeaker B: Yeah.\nSpeaker B: But it might have these cost implications.\nSpeaker B: There's no reason we need to make it look as fashionable and stylish as a mobile phone. It can still be lightweight plastic.\nSpeaker B: Something that's easily molded and produced. Sorry I'm treading on your territory guys.\nSpeaker C: Right. Okay. We've got half there for the next meeting so we're all going to go off and do our individual things.\nSpeaker C: I think that's probably about it and then we'll come back in the A's again and I get to do another fantastic print presentation.\nSpeaker A: Just a quick thing about the... What you're saying about the... Doesn't need to be fashionable.\nSpeaker A: So I had a quick look at the company website and it's like the... We put the fashion into electronics.\nSpeaker A: I really do that.\nSpeaker A: I think the whole design thing might be...\nSpeaker A: Sure.\nSpeaker A: You can still have plastic and it looks quite good.\nSpeaker B: Yeah. I mean it doesn't have to be... That was my main point. We don't have to use metal. I don't know if using plastic does make it cheaper.\nSpeaker B: I would.\nSpeaker B: Yeah.\nSpeaker C: I mean the sky remote controls have to be kind of molded and look a bit different than the telly west remote controls.\nSpeaker C: So silver plastic looks a bit lighter.\nSpeaker C: So yeah, I guess that's a second.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: So let's break it up there.\nSpeaker C: Okay.\nSpeaker C: So see you in half now.\nSpeaker B: Do we go back to our room?\nSpeaker C: Yeah.\nNone: Okay.\n"}, {"meeting_id": "IS1008b", "summary": "This is the second meet-up for the new product of television remote control, with a particular focus on its functional design. Industrial Designer, Marketing and User Interface were each invited to give a presentation on the subject that they are in charge of. On the basis of conducted user requirement survey, Industrial Designer and User Interface put forward their ideas on remote control features which are new to the existing market, along with the functionality and outlook of the product. Among all the features, speech recognition command takes up the majority of discussed subjects. The group also discussed the material appropriate to use in the construction. After the group determined the financial target of this project, as well as the production price and reasonable selling price, Marketing suggested taking a risk in price rise.", "dialogue": "None: It was supposed to change seats.\nNone: I didn't know this was mine.\nNone: It did.\nNone: It PowerPoint disappeared.\nNone: I did this no-no somewhere.\nNone: And it disappeared.\nSpeaker B: And whose thing and the jig is this?\nSpeaker B: There you go.\nSpeaker B: You need to.\nSpeaker D: There you go.\nSpeaker D: Agnes, you can help me for the smile when I...\nSpeaker D: Okay, welcome back.\nSpeaker D: I hope you have fresh hair and fresh time.\nSpeaker D: Now the meeting actually we gathering here to discuss about the functional design meeting.\nSpeaker D: Okay, and we received some information from all of you.\nSpeaker D: And it's in the, I think, in the sharing folder.\nSpeaker D: And I will invite the Christine and the aide and Agnes to discuss about on the various subjects.\nSpeaker D: So can you go to the next slide?\nSpeaker D: The agent of the meeting is opening.\nSpeaker D: Then I'm going to talk about the project management, what I'm going to do.\nSpeaker D: And of course I'm doing the project management and secretary both, okay, to take the minutes of the meeting.\nSpeaker D: And there are three presentations.\nSpeaker D: One is new project requirements and the second one is about decision on remote control functions.\nSpeaker D: And finally we are closing.\nSpeaker D: And the meeting time will be 40 minutes.\nSpeaker D: So we have to be very quick.\nSpeaker D: And I have come up with management come with the new proposal, okay, and I have to discuss a few points on this.\nSpeaker D: Both says new insights in the aim of your project.\nSpeaker D: The one is the teletext becomes outmodal, okay, because of the computer systems and new technology.\nSpeaker D: So we don't need to consider really about the teletext and all in our new project design.\nSpeaker D: The second one is about the remote control should be used only for the TV.\nSpeaker D: That's what our management says.\nSpeaker D: And the third point, it's very, very important to establish marketing or corporate image with this new project or new product.\nSpeaker D: So I will invite Agnes, can you go to the third slide?\nSpeaker D: No, this is the third slide.\nSpeaker D: So I'll invite Christine to discuss about the functional design.\nSpeaker C: Okay, do you want to open the ear participants?\nSpeaker B: Number two.\nSpeaker B: That's it.\nSpeaker C: Do you want the mouse or do you want me to?\nSpeaker B: I'll do it.\nSpeaker B: Thanks.\nSpeaker B: So, I figured we should identify some user requirements.\nSpeaker B: From my experience, I want to, and from research I did, the device has to turn the television on and off.\nSpeaker B: The first time you press on the big button, he can't have like waffling on this point.\nSpeaker B: And he needs to be able to find it because one of the biggest problems with remote controls is finding them.\nSpeaker B: So I also, since we have to establish our corporate image on the basis of this new product, that we better look at things that are popular and go beyond those and, as I said, in the first meeting.\nSpeaker B: And then we might want to talk eventually about the materials that are appropriate to use and the construction, especially on the outside of the product so that it gives the appearance and is reliable and so forth.\nSpeaker B: I did a little history on the remote controls and when they were invented and so forth.\nSpeaker B: So I guess this guy, Zenith, created the flashmatic, which I kind of like the idea because it made me think of maybe the remote control made a big flash when you turn the TV on and off.\nSpeaker B: That might be interesting.\nSpeaker B: And so it was a highly directional flashlight that you could turn the picture on and off and the sound on and off and change channels.\nSpeaker B: So I think those are still requirements we have today, 50 years later.\nSpeaker B: And it was really a pioneering innovation but it was sensitive to the sun so that it would start off by the, it would easily cause some problems.\nSpeaker B: So in addition to looking at the functional requirements, so all of these devices are examples of where they represent examples that are available today, which I think the one in the middle is really something to keep in mind.\nSpeaker B: And it would be easy to find and it would, you could throw it at things.\nSpeaker B: If the TV didn't turn on and off, you could use it for something else.\nSpeaker B: And since I'm not really an industrial designer, I didn't really know what to do with this slide.\nSpeaker B: But I just took some different schematics and I put them into this and I guess this is what a slide might look like if you were doing a circuit board.\nSpeaker B: I don't know why we were asked to do this.\nSpeaker B: So personal preferences, I think we could, I'm really thinking outside the box here and I think that we should consider perhaps having a size, a remote control that changes in size depending on the user preference.\nSpeaker B: So something that's very, very flexible and inflatable and then you could shrink it.\nSpeaker B: I think you could either be, you could go either one extreme, be very colorful or you could make it clear and kind of blend in with things so you didn't have to have a problem with the decoration of the user's home.\nSpeaker B: I think it needs to be waterproof because sometimes they fall into cups and you know, it might be out by the swimming pool or something like that.\nSpeaker B: If you, one of my requirements was about it needs to tell you when it's done as job or not because after time I keep pushing on the remote control and I don't know if it's actually understood in my message.\nSpeaker B: So I think it should give you some sort of an oral cue and of course I never want to replace the battery.\nSpeaker B: So those are my preferences and that's my presentation.\nSpeaker D: Yeah, let me interrupt you.\nSpeaker D: If you can add other facility, other future like unbreakable.\nSpeaker D: Yes.\nSpeaker D: Okay, because especially today you know, you have the family and the kids, okay and the kids throw it and they play with the remote sign.\nSpeaker B: Run over with the car.\nSpeaker D: Yes.\nSpeaker D: Okay, so if you can add the future, okay for your fabric, whatever in your outline design, okay, with unbreakable.\nSpeaker D: Okay, I think that will give a lot of advantage for our product.\nSpeaker D: Good idea.\nSpeaker B: Good idea.\nSpeaker D: I'll be, you can add it in that.\nSpeaker B: Yes, very good.\nSpeaker D: Okay, thank you Christian and any questions or clarifications or any discussion on the functional design.\nSpeaker C: Do you have any preconceived ideas in terms of materials?\nSpeaker C: Because for example, in the unbreakable thing doing something plastic would be harder.\nSpeaker C: We're having something like, I know, steal our time.\nSpeaker C: Titanium is an elite.\nSpeaker B: Titanium is economically viable.\nSpeaker B: Be heavy too.\nSpeaker B: Yeah.\nSpeaker B: No, I haven't really, I wanted feedback.\nSpeaker B: I think we need to rate rank these.\nSpeaker B: Yeah.\nSpeaker B: But we'll see what your personal preferences are in your class.\nSpeaker C: No, I'm just wondering whether that you had any.\nSpeaker B: I like titanium.\nSpeaker B: It's light.\nSpeaker B: Yeah.\nSpeaker C: Expensive.\nSpeaker C: Yeah, marketing comes out.\nSpeaker B: Who said, who said, you know, nobody told me how much what our financial objective is.\nSpeaker B: So, yeah.\nSpeaker B: It'd be hard to inflate something made out of titanium.\nSpeaker D: Yeah, I'm sorry because the last meeting we supposed to discuss about the financial thing, let me go quickly.\nSpeaker D: Maybe if I can go back and order project plan and the budget.\nSpeaker D: So, I can close this.\nSpeaker D: of course I can.\nSpeaker D: So let me see where is this file.\nSpeaker C: That's Christine's.\nSpeaker A: That's my mom.\nSpeaker D: I will say you may.\nSpeaker D: The project A, the end of the day.\nSpeaker D: The company wants to make at least the \u20ac50 million.\nSpeaker D: It will be a good idea for the market.\nSpeaker D: You can come back with your feedback.\nSpeaker D: I have one, maybe the suggestion or opinion.\nSpeaker D: It can be for a universal to use for any TV.\nSpeaker D: It will be slim.\nSpeaker D: It might be hard to find.\nSpeaker D: Any other questions?\nSpeaker A: I have a sales price of \u20ac25.\nSpeaker A: I think for what we are trying to find, I don't think that price is exactly in the market.\nSpeaker A: I explain myself here now in the sense that in recent surveys, from 15-35-80% are willing to spend more money for something as fancy as Trandy.\nSpeaker A: If we are going to take a risk and push this up a bit, make it more expensive, but give them added things they don't have now, it could possibly sell.\nSpeaker A: Obviously the risk is there.\nSpeaker A: There is one interesting thing that is interesting.\nSpeaker A: Speech recognition is requested.\nSpeaker A: We are talking between 75-90% of this group is willing to pay for a speech recognition on a remote.\nSpeaker A: Obviously we can't make a remote into a computer, but maybe simple commands, louder, softer, on, off.\nSpeaker A: That might be a possibility, even though it costs more, to be the first on the market to produce this.\nSpeaker A: 35% say they are too difficult to use.\nSpeaker A: We have to figure out a way of making it more user friendly.\nSpeaker A: 50% say they can't find the remote half the time.\nSpeaker A: Maybe one word speech recognition command say remote, and there is a BBB, and they can find it through 10 tons of newspapers, magazines, whatever you have at home.\nSpeaker A: The cost that the management is looking for is not going to be possible.\nSpeaker A: But if it is Trandy, if it is fancy, it has some color to it.\nSpeaker A: If it is very easy to use, if it has got simple speech remote control, like I said, louder, softer, change channel on, off, remote, it goes BBB.\nSpeaker A: I can find my remote without spending half a day looking for it and getting all upset because I can't turn the TV on.\nSpeaker A: We have to look at it in this global idea with the ideas of the industrial design.\nSpeaker A: But price obviously we have to talk about.\nSpeaker D: So what do you think about the design's characteristics?\nSpeaker D: Do you think you can make it or?\nSpeaker B: I'm sorry.\nSpeaker D: What do you think about the design, what you were talking about, speech recognition?\nSpeaker D: Speech recognition.\nSpeaker B: Well, training is always an issue with commands.\nSpeaker B: So we can perhaps do it if the user is willing to spend some time in the training process, it could reduce the overall cost.\nSpeaker B: I'm not sure how.\nSpeaker B: But anyway, I think also that this might impact the battery life.\nSpeaker B: And so maybe what we'll have to do is add something where you can recharge it wirelessly so that you know, send power to it.\nSpeaker B: Or maybe set it out in the sun and it gets from the light, a solar cell inside there so that you have enough juice to do all these fancy things.\nSpeaker C: It seems also like with the speech recognition, yeah, it's a great feature, but if you're watching TV, there's a lot of ambient sound and it's words.\nSpeaker C: It's not just, you know, noises like something hitting.\nSpeaker C: It's actual speech, so then you have to make sure that the speech recognizer is good enough to filter out the TV speech and the user's speech.\nSpeaker C: Otherwise you can say remote, but if someone on the screen is saying the same thing, all of a sudden you have someone in a movie saying off and your screen dies because they've triggered the remote control and it's turned off your TV.\nSpeaker C: So I think if we can find a speech recognizer that can handle those types of problems, then yeah, it'd be a really good marketing gimmick, but I think we seriously need to consider how that would impact the situation.\nSpeaker C: Very good point.\nSpeaker A: Because I would speech recognition, I'm not that good at that idea, but if it's a one word recognition because I know with telephones and cars and things, I've seen in the states, a friend of mine says, call mom and it calls up mom.\nSpeaker A: Yeah, the radio can be on and everything because I think with speech recognition, if the remote or like the telephone, it has an exact word that it has to hear, I don't think it would come through a sentence and a television.\nSpeaker A: If somebody's speaking on the television, they're not going to stop and say remote. Okay, so I think that something could be designed to recognize single word. Oh yeah. Yeah.\nSpeaker A: Like the telephone.\nSpeaker A: It's a great idea if we can design it.\nSpeaker A: This is years ago in the United States, we're driving down and he's called home on the telephone call immediately. So that's kind of cute.\nSpeaker D: What I can suggest to you, Christian, okay, if you need some technical feedback or some training, okay, about this facility, especially for the speech recognition, I can recommend you some companies like Intel or IBM, okay, because they're already in this speech recognition part.\nSpeaker D: Okay, and you can maybe have some technical backup from them, some kind of technical tyre. Okay, and if you want, I can coordinate to get some information.\nSpeaker D: And you can let me know, okay, so what kind of the details you required, okay, to add this future in this project. I don't think it's the difficult and we need to know how much is the timeframe you need to develop, apart from what you have today. Okay, we'll find that out.\nSpeaker A: From your side, you're gonna have to go back to management and be more precise. What do they want? A risk, take a risk on the market, something that's going to cost more, but could very easily make a boom in the market.\nSpeaker A: Because it has to be something totally different, it has to be totally new, something that nobody has right now. Yes, but it costs.\nSpeaker D: But at the end of the day, you're the sales guy, so I will come back and sit on your head because you're going to do your sales projection, okay.\nSpeaker D: Of course, it's good to tell the management how much it's cost and how much you're going to invent. Okay, and so I don't mind to conveyance, okay, the management to spend some more money on the project.\nSpeaker D: Okay, if you can make out of. Obviously. The money from the bottom line is positive. I don't mind to convince the management. Okay, the management says, okay, so they don't want certain facilities, which it's already worth. Okay, they want something new. Okay, I think like a speech recognition, definitely they will agree.\nSpeaker D: I don't think they say no for that. Okay, and I hope I can convince the management on that. Okay, so if you have any new ideas, okay, for you, you can always come up and you can tell me if you need any special, okay, coordination, okay, between any technical companies, which you can hide the technology backup.\nSpeaker D: Okay, for your functional design or technical design, okay, then I'm ready to do that.\nSpeaker D: And what's your comments about?\nSpeaker C: Well, I mean, maybe if I go through my presentation, you can sort of see what the user perspective is and how it ties into the other two comments. Yeah, so you'll finish at so I can.\nSpeaker D: Okay, I'll handle to technique.\nSpeaker D: Let's get it close this.\nSpeaker D: Where are you here?\nSpeaker C: It's been three.\nSpeaker C: Here, enjoy.\nSpeaker C: Okay, so basically, the method that we usually use in user interface design is that we need to look at what people like and what people don't like about existing products, so in our case existing remote controls.\nSpeaker C: And then what the good ideas are and what the bad ideas are and why they're bad and good, which isn't always as obvious, we seem to have intuitions about why things are good or things are bad, but when you look technically at how it works, sometimes that's not the case.\nSpeaker C: Then we need to decide what functionalities we really want to keep because that'll feed into both Ed's work and Christine's work.\nSpeaker C: And then what the remote control should look like, obviously, once we've got a good idea of what the functionalities are.\nSpeaker C: So in terms of the functionalities that we need, you obviously need to be able to turn the TV on and off.\nSpeaker C: You need to change channels, both by directly going to a specific channel or by channel surfing.\nSpeaker C: You need to be able to control the volume and then control any menus on the TV to regulate contrast or whatever.\nSpeaker C: So the problems that people have expressed is that there's too many buttons on remote controls in general.\nSpeaker C: The buttons, it's not clear what they're supposed to do.\nSpeaker C: Often you need to know specific button sequences to get certain functionalities done, which you don't necessarily always remember, especially if it's a functionality that you don't use very often, and that the buttons are too small.\nSpeaker C: So here we've got two examples where here on the left-hand side you can see a remote control that has lots and lots of buttons.\nSpeaker C: The buttons in a lot of cases are tiny.\nSpeaker C: They're hard to see and okay, they're labeled, but the labels don't necessarily tell you too much.\nSpeaker C: Whereas on the other side you have a much simpler remote control that I think basically has the minimum functionalities that are needed.\nSpeaker C: It sort of looks simpler and less imposing when you first look at it.\nSpeaker C: I would be inclined to go towards this in terms of design rather than this.\nSpeaker C: If there's specific functionalities that require more buttons, then we can figure out how to do it with existing buttons.\nSpeaker C: So my personal preferences are to keep the number of buttons to a minimum, sorry, make frequently used buttons bigger and more strategically placed.\nSpeaker C: So like the on button being really obvious when the channel changing and the volume.\nSpeaker C: And to keep the design basically sleek and simple, which I think ties into what Kristina and Ed have both said fairly reasonably.\nSpeaker C: So that's pretty much it.\nSpeaker C: I don't know if you guys have any questions or...\nSpeaker B: Oh, it seems very understandable. Clearly your research in hours heading in the same direction.\nSpeaker B: And the only thing that I saw missing from your research that we found was this ability to find the doggone thing when you need it.\nSpeaker B: Yes, that's true.\nSpeaker B: But that's okay. That's why we're all here at the table.\nSpeaker B: So that if we think of it in our research to indicate certain things and it's complementary.\nSpeaker B: I also think that the feel of it is when you hold it is something that was expressed more in my design.\nSpeaker B: And that's logical and normal because those are the parameters that an industrial designer is more thinking about the look in the feel.\nSpeaker C: And that's definitely a very important factor, especially to users who are going to be buying the thing and I'm using it almost on a daily basis in a lot of cases I think.\nSpeaker B: Okay, so I don't have any questions. Sounds good.\nSpeaker D: Okay, so anybody need any help for time being on this subject? Okay, so please come back to me.\nSpeaker D: And Christian, maybe I can try to help you to get some of the technical companies to help you for make a design slim.\nSpeaker D: Okay, and to add some futures like what we were talking about speech recognition at all.\nSpeaker C: Should we maybe make a decision about what features we actually want to include?\nSpeaker C: Because we've thrown a lot of features onto the table, but do we actually want to incorporate all of them or have we missed anything?\nSpeaker B: Do you want to go back and look at the closing slide? Maybe that would provide some guidance.\nSpeaker B: Sure.\nSpeaker B: Doesn't really tell us.\nSpeaker D: Not really, this is what we're talking about.\nSpeaker B: Well, it says individual actions.\nSpeaker B: It says user interface, so I'm supposed to do the components concept. I'm supposed to work on the user interface concept and you're supposed to keep watching the trends.\nSpeaker B: And specific instructions will be sent by our coach.\nSpeaker B: I think we should do as many features as start with all of them right now and eliminate them later in the process. That's my suggestion.\nSpeaker D: Okay, that would be great.\nSpeaker D: And I'll send you the minutes of me.\nSpeaker C: You can object if you want.\nSpeaker C: I'm just thinking in terms of time.\nSpeaker A: You're objecting.\nSpeaker C: Yes, now I'm objecting.\nSpeaker C: No, I mean, I was just thinking, is it really practical to start designing something with features that we're just going to end up throwing away?\nSpeaker C: I mean, it takes a lot of time and effort for everyone to consider different features.\nSpeaker C: And if we spend that time and effort on features that we're not going to use, maybe it's better to spend it on the thinking more about features that we actually do want.\nSpeaker B: I don't get to get a match item for after the meeting, because our meeting time is run out.\nSpeaker B: Somebody else just got to use this room and we can't hang out here and talk about this.\nSpeaker D: Okay, what we'll take for lunch break, then we can discuss for the more with our ideas and then we will come back again in the next meeting.\nSpeaker D: So thanks for coming and I'll send you the minutes of meeting and please put your all information in the sharing folder so everybody can share the information.\nSpeaker D: Okay, so let's go for lunch time.\nSpeaker C: I agree.\n"}, {"meeting_id": "Bmr023", "summary": "A pressing concern for the group is the DARPA meeting in July, which is only a short time away, and for which they would like to have some progress. PLP results for the front-end look good, with the group also reporting progress in segmentation: Thilo's segmenter will now be used and ways of improving performance investigated; The classifier segmentation is progressing well, especially in the use of prosody for identifying interruption. Work on the front end continues, with improvements of 3-5% being made.", "dialogue": "None: Hey, we're recording.\nNone: Oh, wait a minute, wait a minute.\nNone: Oh boy, I got the harness.\nSpeaker I: What's the channel?\nSpeaker A: Make sure to turn your microphone on.\nSpeaker A: There's a battery.\nSpeaker F: There we go.\nSpeaker A: Your channel number is already on this blank sheet.\nSpeaker D: Channel 5.\nSpeaker F: Channel 5.\nSpeaker F: Channel 5.\nSpeaker F: Channel whatever.\nSpeaker F: Camera 1.\nSpeaker F: Camera 2.\nSpeaker F: Channel 4.\nSpeaker D: Channel 5.\nSpeaker D: Channel 4.\nSpeaker A: The game's up at it what it usually is.\nSpeaker A: But if you think it's sort of a default.\nSpeaker A: But I can set it higher if you like.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Test test test test test test.\nSpeaker D: Okay, that seems better.\nSpeaker D: Yeah, okay, good.\nSpeaker D: That's good.\nSpeaker D: That's good.\nSpeaker E: Okay.\nSpeaker D: So I had a question for Adam.\nSpeaker D: Have we started to read?\nSpeaker A: Well, we started recording.\nSpeaker D: Yeah.\nSpeaker D: It's January.\nSpeaker D: I saw it earlier.\nSpeaker G: She can just walk in, I guess.\nSpeaker A: Yeah, right.\nSpeaker A: So we're starting late.\nSpeaker A: I figured we'd better just start.\nSpeaker D: I was going to ask Adam to say if he thought any more about the demo stuff.\nSpeaker D: It occurred to me that this is late May and the DARPA meetings in mid-July.\nSpeaker D: But I remember we...\nSpeaker D: I know we were going to do something with the transcriber interface.\nSpeaker D: There's one thing, but I thought there was a second thing.\nSpeaker D: Anybody remember?\nSpeaker A: Well, we were going to do a mock-up question answering or something I thought.\nSpeaker A: That was totally separate from the interface.\nSpeaker A: Do you remember?\nSpeaker A: Remember, like, asking questions and retrieving, but in a pre-stored fashion.\nSpeaker A: That was the thing we talked about, I think, before the transcriber.\nSpeaker A: Yeah.\nSpeaker D: All right.\nSpeaker D: So anyway, we have to sort of that out.\nSpeaker D: Somebody going on it.\nSpeaker D: Got a month left, basically.\nSpeaker A: You like these, right?\nSpeaker D: Okay, good.\nSpeaker D: Okay.\nSpeaker D: Okay, so what do we get else we got?\nSpeaker D: You just wrote about your stuff.\nSpeaker A: No, that was all previously here.\nSpeaker A: I was writing the digits and then I realized I could see Roxas.\nSpeaker E: Oh, okay.\nSpeaker A: Because I didn't want people to turn their heads from these microphones.\nSpeaker A: We have, by the way, have the same digit form for the record.\nSpeaker D: That's cool.\nSpeaker D: So the choice is which we want more the comparison.\nSpeaker D: Everybody is saying it at the same time or the comparison of people saying the same digits at different times.\nSpeaker A: It's just because I didn't have any more digit sheets.\nSpeaker A: I know that.\nSpeaker D: Which opportunity should we exploit?\nSpeaker A: It might be good to have them separately and have the same exact strings.\nSpeaker A: I mean, you could use them for normalizing or something.\nSpeaker A: But it, of course, goes more quickly doing a menu.\nSpeaker A: I guess we'll see.\nSpeaker A: I guess we'll see how long we go.\nSpeaker D: How long we go and how good this snack is out there.\nSpeaker H: Anyway, there's some pants talked to us.\nSpeaker H: Somebody is saying zero in some things.\nSpeaker H: Right.\nSpeaker H: Right.\nSpeaker H: It's really not identical.\nSpeaker D: Yeah, we'd have to train.\nSpeaker A: We'd be like a chorus.\nSpeaker D: We'd have to get some experience.\nSpeaker D: Yeah.\nSpeaker D: Really boring.\nSpeaker D: Chorus.\nSpeaker D: Do we have an agenda?\nSpeaker D: I had them usually tries to put that together.\nSpeaker D: I've got a couple of things to talk about.\nSpeaker D: Yeah.\nSpeaker C: What might those be?\nSpeaker C: IBM stuff and just getting meeting information organized.\nSpeaker D: Okay.\nSpeaker D: You're applying that it's currently disorganized.\nSpeaker I: In my mind.\nSpeaker D: Is there stuff that's happened about the SRI, recognizer, et cetera, those things that were happening before with?\nSpeaker D: You guys were doing bunch of experiments with different finance.\nSpeaker D: Is that still sort of where it was the other day we're improving?\nSpeaker C: Yeah.\nSpeaker C: Now, you saw the note that the PLP now is getting basically the same as the MOTC.\nSpeaker C: Yeah.\nSpeaker I: Actually, it looks like it's getting better.\nSpeaker I: Just with age.\nSpeaker I: With age.\nSpeaker I: Yeah.\nSpeaker I: But that's not tricky related to me.\nSpeaker I: So we can talk about it.\nSpeaker I: It looks like I haven't, but it's the experiment is still not complete.\nSpeaker I: But it looks like the Polotract-Actualization is working good.\nSpeaker I: Like, actually, using the warp factors that we computed for the SRI system, just applying them to the PC.\nSpeaker I: That's pretty funny.\nSpeaker A: So you just need to copy of it.\nSpeaker I: Just have to take the supercult to the number of the data.\nSpeaker I: They have different meanings in the system.\nSpeaker D: Yeah, all that's always good to do.\nSpeaker I: One issue, actually, that just came up in discussion with Liz and Darn, was, as far as meeting recognition is concerned, we would really like to move to doing the recognition on automatic segmentations.\nSpeaker I: Because in all our previous experiments, we were essentially cheating by having the hand segmentation system based on the recognition.\nSpeaker I: And so now, with Tilo's segment, we're working so well.\nSpeaker I: I think we should...\nSpeaker D: We think we should increase the error rate.\nSpeaker D: Yeah.\nSpeaker H: That's what I wouldn't do.\nSpeaker A: Yeah.\nSpeaker A: And even the good thing is that since you have high recall, even if you have low precision because you're over-generating, that's good, because we could train noise models and the recognizer for these kinds of transients and things that come from the microphones.\nSpeaker A: But I know that if we run recognition unconstrained on a whole waveform, we do very poorly because we're getting insertions in places that you may well be cutting out.\nSpeaker A: So we do need some kind of pre-sequentation.\nSpeaker I: Some extra things like retraining or adapting the models for background noise to the environment.\nSpeaker A: And using Tilo's, you know, pustyriers or some kind of...\nSpeaker A: Or right now they're discreet, yes or no, for a speaker to consider those particular speaker background models.\nSpeaker A: There's lots of interesting things that could be done.\nSpeaker D: Yeah, if you could do that.\nSpeaker D: So when we do the IBM stuff?\nSpeaker C: Yeah, so talked with Brian and gave him the alternatives to the single beep at the end of each utterance that we had generated before.\nSpeaker C: And so...\nSpeaker C: Truck, trunks.\nSpeaker C: Yeah, truck, trunks, right.\nSpeaker C: And so he talked over the transcriber and the transcriber thought that the easiest thing for them would be if there was a beep and then a number of digit and then a beep at the beginning of each one.\nSpeaker C: And that would help keep them from getting lost.\nSpeaker C: And so Adam wrote a little script to generate those style beeps.\nSpeaker C: And so where...\nSpeaker C: I think that the digits are...\nSpeaker C: I came up here and just recorded the numbers one through ten.\nSpeaker C: So...\nSpeaker C: That's a great idea.\nSpeaker C: So, yeah, we just used those.\nSpeaker I: A few splice to the right for four.\nSpeaker C: Yeah, I recorded, actually I recorded one through ten three times, three different speeds, and then he picked.\nSpeaker C: He liked the fastest one.\nSpeaker C: So he just cut those out and spliced them in between two beeps.\nSpeaker H: It will be funny.\nSpeaker H: It will be funny when you're really reading digits and then there are the trunks with...\nSpeaker H: Yeah, with my digits.\nSpeaker C: All right.\nSpeaker C: That'll throw.\nSpeaker D: Maybe we should have you record a bc for that one.\nSpeaker D: And she said it wasn't going to...\nSpeaker C: The transcriber said it wouldn't be a problem because they can actually make a template that has beep, number, beep.\nSpeaker C: So for them it will be very quick to put those in there and then transcribing.\nSpeaker C: So we're going to send in one more sample meeting and Tilo is running his segmentation.\nSpeaker C: Adam is going to generate the chunk file and then we'll give it to Brian and they can try that out.\nSpeaker C: When we get that back we'll see if that sort of fixes the problem we had with too many beeps in the last transcription.\nSpeaker D: Okay.\nSpeaker D: Okay. Do you have any idea to turn around on those steps you just said?\nSpeaker C: Are on our side or including IBMs?\nSpeaker C: Including IBMs.\nSpeaker C: Well I don't know. The last one seemed like it took a couple of weeks.\nSpeaker C: Maybe even three.\nSpeaker C: That's just the IBM side.\nSpeaker C: Our side is quick. I mean, how long does your...\nSpeaker C: Well I'm at the overall thing.\nSpeaker D: The reason I'm asking is because Jane and I have just been talking and Jesus has been doing a further hiring of transcribers.\nSpeaker D: We don't really know exactly what they'll be doing and how long they'll be doing it and so forth.\nSpeaker D: Because right now she has no choice but to operate in the mode that we already have working.\nSpeaker D: So it'd be good to sort of get that resolved.\nSpeaker C: I hope we can get a better estimate from this one that we send them.\nSpeaker C: So I don't know yet.\nSpeaker D: Yeah.\nSpeaker D: In particular, I would really hope that when we do the start meeting in July that we sort of have...\nSpeaker D: We're into production mode.\nSpeaker D: We actually have a stream going and we know how well it does and how it operates.\nSpeaker D: I think that would be a very good thing to be able to do.\nSpeaker E: Okay.\nSpeaker D: Maybe before we do the meeting info organized thing, maybe you could say a little bit stuff about where we are in transcriptions.\nSpeaker B: Okay. So the heat transfer was continued to our past one concept one, which was the second thing.\nSpeaker B: Talking about it at this point.\nSpeaker B: They got in five meetings down in that set right now they're in the positive end.\nSpeaker D: I hired two transcribers today and he hired another one, which was because we had a lot of nutrition.\nSpeaker C: They died off after they do this for a hell.\nSpeaker B: Burn out.\nSpeaker B: Well, that was a nice thing.\nSpeaker B: One of them had never planned a work past January.\nSpeaker B: It means that all these various things, because we represented it as possibly month project at 10.\nSpeaker B: And it's not an extensive nutrition way to be productive to two, but they're really solid.\nSpeaker B: We're really lucky to really have a lot of me.\nSpeaker A: I mean basically spending money on coffee to truck drives off\nSpeaker B: Computer safety sprints Backbie I'm just saying, the key point right now is to keep the staff on the leaner side, rather than hiring like eight to 10 right now, because if the IBM thing comes through really quickly, then we don't want to have to make people all the time so.\nSpeaker B: That's why I think. I got really a lot of response from my notice and I think I could hire additional people online.\nSpeaker D: Yeah, and the other thing is, I mean, in the unlikely event, since we're so far from this, it's a little hard to plan this way, in the unlikely event that we actually find that we have transcribers on staff who are twiddling their thumbs because there's all the stuff that was sitting there has been transcribed and they're faster, the pipeline is faster than the generation.\nSpeaker D: In the event that they actually don't, I bet we could find some other stuff for them to do. So I think that as we were talking, if we hired 12, then we could rent to a problem later. We also just couldn't sustain that for everything, but also it's a reason.\nSpeaker D: But if we hire, we have five on staff, five or six on staff, then we give time, then it's a small enough number so we can be flexible either way.\nSpeaker A: It'd be great too if we can, we might need some help again getting the tighter boundaries or some hand to experiment with, you know, to have a ground truth for this segmentation work, which I guess you have some already that was really helpful and could probably use more.\nSpeaker H: Yeah, that was the thing I planned working on is to use the transcriptions which are done by now and to use them as...\nSpeaker H: The new ones with the tighter boundaries, yeah.\nSpeaker H: And to use them for training or for whatever, yeah, to create some speech, non-speech labels out of them. But that's the thing was what I am just looking into.\nSpeaker H: Okay.\nSpeaker B: So, there you go, presentation to so much, I was so extremely helpful.\nSpeaker B: Now there was a couple weeks ago, I needed some new ones and it happened to be during the time that we was on the presentation.\nSpeaker B: I thought it was just very few days, but I happened to be during that time on humans, so I started them on the non-presegment, and then switched them on in two years and they always appreciate that.\nSpeaker B: And he's really, really appreciate it.\nSpeaker B: I was going to say that they do adjust once in a while, you know, once in a while, and they actually talk to them, didn't you?\nSpeaker H: Yeah, I talked to Helm.\nSpeaker B: And so I asked her, they're very perceptive.\nSpeaker B: I really want to have this meeting of trans farms.\nSpeaker B: I haven't done it yet, but I want to do that, and she's out there for a couple of weeks.\nSpeaker B: I'm going to do it as she returns.\nSpeaker B: Because she was saying in a span of very short-fared, it seems like the ones that need to be adjusted are these things.\nSpeaker B: As she was saying short-fared illnesses, you're here, you're aware of this.\nSpeaker B: But actually, it's so correct for so much of the time, it's not all this time, it's just this route to a long time.\nSpeaker A: That's great.\nSpeaker B: I think it would be interesting to take a moment.\nSpeaker A: Is there actually a record of where they change?\nSpeaker A: I mean, you can compare it, do a diff on the, just so that we knew...\nSpeaker B: It's complicated in that...\nSpeaker H: I feel when they create new segments or something, it will be not that easy, but...\nSpeaker A: I mean, if we keep an old copy of the old time marks, just so that if we run it, we know whether we're, which ones we're cheating and...\nSpeaker B: It would be great, which is a lot of good.\nSpeaker B: And then, when they start partly through, then when I do as I merge, I'll think down, they've been pre-signed into version.\nSpeaker B: So it's not a pure, it's not a pure condition.\nSpeaker B: And then, I think that they started with pre-signed methods and work with signal all the way through.\nSpeaker B: And I think it wasn't possible for that for many reasons, but it will be possible for future.\nSpeaker A: That's great.\nSpeaker A: As long as we have a record, I guess, of the original automatic one, we can always find out how well we would do from the recognition side by using those boundaries, you know, a completely non-shading version.\nSpeaker A: So if you need someone to record this meeting, I mean, I'd be happy to, for the transcribers, I could do it or chuck her.\nSpeaker D: So, you were saying something about organizing the meeting intro?\nSpeaker C: Yeah, so Jane and Adam and I had a meeting where we talked about the reorganization of the directory structure for all of the meeting record.\nSpeaker C: No.\nSpeaker C: For all of the meeting record today, we should have.\nSpeaker C: And so we've got a plan for what we're going to do there.\nSpeaker C: And then Jane also prepared a, started getting all of the meetings organized so she prepared a spreadsheet, which I spent the last couple of days adding to.\nSpeaker C: So I went through all of the data that we have collected so far and have been putting it into a spreadsheet with start time, the date, the old meeting name, the new meeting name, the number of speakers, the duration of the meeting, comments, you know, what it's transcription status is, all that kind of stuff.\nSpeaker C: And so the idea is that we can take this and then export it as HTML and put it on the meeting recorder.\nSpeaker C: Oh, great.\nSpeaker C: Keep people updated about what's going on.\nSpeaker C: I've got to get some more information from Jane because I have some gaps here that I need to get her to fill in.\nSpeaker C: But so far as of Monday, the 14th, we've had a total number of meeting, 62 hours of meetings that we've collected and some other interesting things.\nSpeaker C: Average number of speakers per meeting is six.\nSpeaker C: And I'm going to have on here the total amount that's been transcribed so far, but I've got a bunch of, that's what I have to talk to Jane about, figure out exactly which ones have been completed and so forth.\nSpeaker C: But this will be a nice thing that we can put up on the website and people can be informed of the status of various different ones.\nSpeaker C: And it'll also list, like under the status, if it's at IBM or if it's at ICSI or if it's completed or which ones were excluding, and there's a place for comments so we can say why we're excluding things and so forth.\nSpeaker D: Now with the ones that are already transcribed, we have enough there that, you know, we've already done some studies and so forth.\nSpeaker D: And shouldn't we go through and do the business of having the participants approve it for the transcriptions for distribution and so forth?\nSpeaker B: I would say as, although I still am doing some final pass and trying to convert it into a master file is being the channelized version of this.\nSpeaker B: It seems like I get into that a certain way that something else in the case is going to have to start cleaning up the things like the places where transcribed were in a certain way.\nSpeaker B: Doing a slouching here and there.\nSpeaker B: So I guess we may sense the weight of that's done.\nSpeaker D: Well, let me put another sort of milestone kind of as I get with the pipeline.\nSpeaker D: We are going to have this DARPA meeting in the whole July and I think it'll be given that we've been, we've given a couple public talks about already spaced by months and months.\nSpeaker D: I think it'll be pretty bad if we continue to say none of this is available.\nSpeaker D: We want to be able to say here is a subset that is available right now and that's been through the legal issues and so forth.\nSpeaker D: And they don't have to.\nSpeaker I: For July.\nSpeaker I: You know the netted version they can just give their approval to whatever version.\nSpeaker D: Well, in principle yes, but I mean if if if if somebody actually did get into some legal issue with it.\nSpeaker I: Yeah, but I mean the editing will continue presumably if there is a found they will be fixed but they won't change the content of the meetings.\nSpeaker A: Well, if Jane is clarifying question question, then you know how can they agree to it before they know her final version.\nSpeaker B: And the thing is, the subtleties where a person uses this word instead of that word which could have been transferred in the other way.\nSpeaker B: And not and they wouldn't have been signed or submitted in the other word.\nSpeaker D: You know there is a point at which I agree it becomes ridiculous because you know you could do this final thing and then a year from now somebody could say you know that should be a period not a question.\nSpeaker D: And you don't you there's no way that we're going to go back and ask everybody do you approve this document now.\nSpeaker D: So I think what it is is that the the the the thing that they sign I looked at it while but it has to be open enough that it sort of says, okay from now on.\nSpeaker D: Now that I've read this you can use do anything you want with these data.\nSpeaker D: And but I think we want to so assuming that it's that kind of wording I don't remember I think we just want to have enough confidence ourselves that it's so close to the final form is going to be in a year from now that they are.\nSpeaker B: It's just a question of if the person is using the transfer does the way of them judging what they said when it was signed as.\nSpeaker D: Well I forget how we end I figured how we ended up on this but I remember my taking the position of not making it so so easy for everybody to observe everything and Adam was taking the position of of having it be really straightforward for people to check every aspect of it including the audio.\nSpeaker D: I don't remember who won Adam.\nSpeaker D: That's really nice again because I can't remember how we ended up that it was the transcript he wanted to do a web interface that would make it.\nSpeaker D: That would give you access to the transcript and the audience that's what Adam wanted and I remember how we ended up.\nSpeaker A: I mean with the web interface it's interesting because you could allow the person who signs to be informed when their transcript changes or something like that.\nSpeaker A: I would say no like I don't want to know but some people might be really interested and then in other words they would be informed if there was some significant change other than typos and things like that.\nSpeaker A: I was like you were whispering satanic and continuing under your breath and you were like you know the small heads thing but I'm just saying that like you can sort of say that any things that are deemed...\nSpeaker A: Anyway I agree that at some point people probably won't care about typos but they would care about significant meaning changes and then they could be asked for their consent I guess if those change.\nSpeaker A: Because assuming we don't really distribute things that have any significant changes from what they sign anyway.\nSpeaker I: How about having an approval of the audio and of the transcripts?\nSpeaker A: Oh my god. But no one will listen to hours and hours of...\nSpeaker A: That's like...\nSpeaker I: Unfortunately, this is not a such transcript.\nSpeaker A: Really?\nSpeaker A: I think that's a lot to ask for people that have been in a lot of meeting.\nSpeaker D: We've gone down this path a number of times I know this can lead to extended conversations and not really getting anywhere so let me just suggest that offline that the people involved figured out and take care of it before it's still alive.\nSpeaker D: So that in July we can tell people yes we have this and you can use it.\nSpeaker D: So let's see, what else we got?\nSpeaker D: Don did the report about his project in class and we're all making a version so that was stuff he was doing with you.\nSpeaker A: I guess one thing we're learning is that we have eight meetings there because we couldn't use the non-native, all non-native meetings.\nSpeaker A: It's probably below threshold on enough data for us for the things we're looking at because the prosotic features are very noisy and so you need a lot of data in order to model them.\nSpeaker A: So we're starting to see some patterns and we're hoping that maybe with I don't know, double or triple the data with 20 meetings or so that we would start to get better results.\nSpeaker A: But we did find that some of the features that Jane would know about that are expressing sort of the distance of boundaries from peaks in the utterance and some local range pitch range effects like how close people are to their floor.\nSpeaker A: So we're also going to be able to see some of the different features that are showing up in these classifiers which are also being given some word features that are cheating because they're true words.\nSpeaker A: So these are based on force alignment. Word features like word frequency and whether or not something is a back channel and so forth.\nSpeaker D: So we're starting to see I think some interesting things including everything where those quasi cheating things.\nSpeaker F: I think depends what you're looking at actually sometimes positions and sentences obviously or in spurts was helpful.\nSpeaker A: I don't know if that's cheating too.\nSpeaker A: Spurts is not cheating except that of course you know the real words but roughly speaking the recognized words are going to give you a similar type of position.\nSpeaker A: Right.\nSpeaker F: Either earlier or maybe you gave a number of words.\nSpeaker A: Not exactly but it should be.\nSpeaker A: Well we don't know and actually that's one of the things we're interested in doing is I think time position like when the word starts.\nSpeaker I: I don't know if I can start.\nSpeaker I: Yeah.\nSpeaker F: I would have to know how these things to do.\nSpeaker F: Like there's a lot of different features you could just pull out.\nSpeaker I: Right.\nSpeaker I: Right.\nSpeaker A: And it depends on speaking right.\nSpeaker A: Yeah.\nSpeaker A: One of the interesting things was I guess you reported on some punctuation type finding sense boundaries finding disfluency boundaries and then I had done some work on finding from the foreground speech whether or not someone was likely to interrupt.\nSpeaker A: So where you know if I'm talking now and someone and Andreas is about to interrupt me is he going to choose a certain place in my speech either prosaacically or word based.\nSpeaker A: And there the prosaac features actually showed up and a neat thing even though the word features were available and a neat thing there too is I tried some putting the speaker so I gave everybody a short version of their name so the real names are in there which we couldn't use.\nSpeaker A: Should use IDs or something and those don't show up so that means that overall.\nSpeaker A: It wasn't just modeling Morgan or it wasn't just modeling single person.\nSpeaker A: But was sort of trying to get a general idea the tree classifier was trying to find general locations that were applicable to different speakers even though there are huge speaker effects so.\nSpeaker A: But the main limitation now is I because we're only looking at things that happen every 10 words or every 20 words we need more more data and more data per speaker.\nSpeaker A: So it also be interesting to look at the EDU meetings because we did include meeting type as a feature so whether you were in a meeting recorder meeting or robustness meeting did matter to interrupts because there are just fewer interrupts in the robustness meeting.\nSpeaker A: And so the classifier learns more about Morgan than it does about sort of the average person which is not bad.\nSpeaker A: It probably do better than but it wasn't generalizing.\nSpeaker A: So it's and I think Don what we have a long list of things he's starting to look at now over the summer where we can he'll be able to report on more things in the future but it was great that we could at least go from the you know James transcripts and the recognizer output and get it to this point and I think it's something Mara can probably use in her preliminary report like yeah we're at the point where we're training these classifiers and we just reporting very preliminary but suggestive results that some features both word and prozotic work.\nSpeaker A: The other thing that was interesting to me is that the pitch features are better than in switchboard and I think that really is from the close talking mics because the pitch processing that was done has much cleaner behavior than the switchboard telephone bandwidth.\nSpeaker A: First of all the pitch tracks are have less having the doubleings than switchboard and there's a lot less dropout so if you ask how many regions where you would normally expect some balls to be occurring are completely devoid of pitch information.\nSpeaker A: In other words a pitch tracker just didn't get a high enough probability of voicing for words for for you know five word they're much fewer than in switchboard so the missing we had a big missing data problem in switchboard and so the features weren't as reliable because they were often just not available.\nSpeaker C: So that's actually good with the lower frequency cutoff on the support.\nSpeaker A: Maybe I mean the telephone we had telephone bandwidth for switchboard and we had the annoying sort of telephone handset movement problem that I think may also affect it.\nSpeaker A: So we're just getting better signals in this data which is nice.\nSpeaker A: Anyway, Dawn's been doing a great job and we hope to continue with Andreas's help and also some of Tilo's help on this to try to get a non cheating version of how all this would work.\nSpeaker D: I think just talk about this the other day but has anybody had a chance to try changing insertion penalty sort of things with the using the tandem system.\nSpeaker I: Oh yeah I tried that. It didn't help dramatically.\nSpeaker C: Were they out of balance? I didn't notice.\nSpeaker I: We were a little relative number of, I think there were higher number of collisions actually.\nSpeaker I: Deletions?\nSpeaker I: So actually it preferred to have a positive, so negative insertion penalty which means that but you know it didn't change by adjusting that the, yeah the arrow changed by probably 1% or so.\nSpeaker I: I don't know if that's not the problem.\nSpeaker I: That's not the problem.\nSpeaker I: But we just, you know, Chuck and I talked and the next thing to do was probably to tune the size of the Gaussian system to this feature vector which we haven't done at all.\nSpeaker I: We just used the same configuration as we used for the standard system. And for instance, then they understand me a message saying that see me used something like 10 Gaussian per cluster.\nSpeaker I: Each mixture has 10 Gaussian.\nSpeaker I: We're using 64.\nSpeaker I: So that's obviously a big difference and it might be way off and give a very poorly trained, you know, Gaussian set way and poorly trained mixture.\nSpeaker I: So we have the turnaround time on the training when we train only the male system.\nSpeaker I: So our small training set is less than 24 hours. So we can run lots of, basically just brute force try a whole bunch of different settings with the new machines that will be better.\nSpeaker D: Yeah, we get 12 of those.\nSpeaker I: The PLP features work. You know, continue to improve the, as I said before, using dance, vocal track normalization option works very well.\nSpeaker I: So I ran one experiment where we just did the vocal track normalization only in the test data. So I did bother to retrain the models and all and the proof I 1% which is about what we get with, you know, just actually doing both training antists normalization with the standard system.\nSpeaker I: So in a few hours we'll have the numbers for the, for retrain everything with the tracking, so that might be the problem.\nSpeaker I: So it looks like the PLP features do very well after having triggered out all these little tricks to get it to work.\nSpeaker A: So you mean you improve 1% over a system that doesn't have any VTL and it already?\nSpeaker D: Okay, so then we'll have our baseline to compare the currently hideous new thing.\nSpeaker I: Right, and what that's just also is of course that the current switchboard, MOP, isn't trained on very good features.\nSpeaker I: Because it was trained on whatever, you know, was used the last time you did up five stuff, which didn't have any of the...\nSpeaker D: Right, but all of these effects were like a couple percent, right? I mean...\nSpeaker I: Well, but if you add them all up, you have almost 5% difference now.\nSpeaker D: And all of them, I thought one was 1.5% and one was 0.8%.\nSpeaker I: Yeah, and now we have another percent with the VTL.\nSpeaker D: That's 3.3%.\nSpeaker I: Actually, and it's...\nSpeaker I: What's actually interesting is that with...\nSpeaker I: Well, maybe another half percent if you do the VTL and training.\nSpeaker I: And then interestingly, if you optimize, you get more of a win out of restoring the...\nSpeaker I: the investments and optimizing the weights.\nSpeaker I: Then you do with the standard?\nSpeaker D: Yeah, but the part that's actually adjustment of the front end per se is opposed to doing putting VTN or something is...\nSpeaker D: It was a couple percent.\nSpeaker D: There was one thing that was 1.5% on the VTL.\nSpeaker D: And let's see if I remember what they were.\nSpeaker D: One of them was the change to...\nSpeaker D: because it did it all at once, from barc scale to male scale, which I really feel like saying in quotes, because...\nSpeaker D: Essentially the same...\nSpeaker D: Yeah, why did that change?\nSpeaker D: But any individual particular implementation of those things puts things in a particular place.\nSpeaker D: So that's why I wanted to look...\nSpeaker D: I wanted to look at exactly where the filters were in the two.\nSpeaker D: It's probably something like there's one fewer or one more filter in the sub...\nSpeaker D: when kill hurts band.\nSpeaker D: And for whatever reason, with this particular experiment, it was better.\nSpeaker D: It could be there's something more fundamental, but I don't know yet.\nSpeaker D: And the other...\nSpeaker D: That was like 1.5% or something, then there was 0.8% which was...\nSpeaker C: Well, that was combined with the triangular, right?\nSpeaker D: Yeah, those two were together.\nSpeaker D: We were able to separate them out as a system and one thing.\nSpeaker D: But then there was 0.8%, which was something else.\nSpeaker C: The low frequency cutoff.\nSpeaker D: Oh yeah, so that was...\nSpeaker D: That one I can claim credit for in terms of screwing it up in the first place.\nSpeaker D: So someone else fixed it, which is that I never put...\nSpeaker D: We had some problems before with offsets.\nSpeaker D: This one back to, I think, Wall Street Journal.\nSpeaker D: So we had...\nSpeaker D: Everybody else who was doing Wall Street Journal knew that there were big DC offsets in these data and those data and nobody had the mention into us.\nSpeaker D: And we were getting really terrible results, like two, three times the error everybody else was getting.\nSpeaker D: And then in casual conversation, someone mentioned, well, I guess, of course, you're taking care of the offsets.\nSpeaker D: I said, what offsets?\nSpeaker D: And at that point, we were pretty into the data and we never really looked at it on a screen.\nSpeaker D: And then we put it on the screen and we...\nSpeaker D: This big DC offset.\nSpeaker D: So...\nSpeaker A: There was that like a hammer?\nSpeaker A: Or when they recorded?\nSpeaker D: It's not at all uncommon for record electronics to have different DC offsets.\nSpeaker D: It's no big deal.\nSpeaker D: You could have 10, 20, 30 models, whatever.\nSpeaker D: And if it's consistently in there.\nSpeaker D: The thing is, most people, front ends, have pre-emphasis with zero-zero frequency so that it's irrelevant.\nSpeaker D: But with PLP, we didn't actually have that.\nSpeaker D: We had the equivalent of pre-emphasis in a pletcher-months-in-style waiting that occurs in the middle of PLP.\nSpeaker D: But it doesn't actually have a zero-zero frequency, like typical simple pre-emphasis does.\nSpeaker D: We had something more fancy.\nSpeaker D: It was later on, it didn't have that.\nSpeaker D: So at that point, I really...\nSpeaker D: Oh, we better have a high pass filter just, you know, just take care of the problem.\nSpeaker D: So I put in a high pass filter at, I think, 90 hertz or so for a 16-kiloward sampling rate.\nSpeaker D: And I never put anything in to adjust it for different sampling rates.\nSpeaker D: And so the code doesn't know anything about that.\nSpeaker D: So this is all at 8-kiloward, so it was at 45 hertz instead of 90.\nSpeaker D: So I don't know if Dan fixed it or...\nSpeaker I: Well, he made it the prime of her.\nSpeaker D: He made a parameter, so I guess if he did it right, he did fix it.\nSpeaker D: And it's taking care of sampling rate, which is great.\nSpeaker C: What is the parameters, just the lower cutoff that you want?\nSpeaker I: It's called H-P-F.\nSpeaker D: Yeah, that's H-P-F on its...\nSpeaker I: But H-P-F, you know, when you put a number after it, it's the set of the hertz-vide.\nSpeaker I: Yeah, cutoff.\nSpeaker D: I mean, frankly, we never did that with the Rasta filter either.\nSpeaker D: So the Rasta filter is actually doing a different thing in the modulations, where I could remain depending on what sampling rate you're doing, which is another old bugger pipe.\nSpeaker D: But...\nSpeaker D: So that was the problem there.\nSpeaker D: We had always intended to cut off below 100 hertz and we just wasn't doing this or now it is.\nSpeaker D: So that helped us by like a 10-3 percent, it still wasn't a big deal.\nSpeaker I: Well, but...\nSpeaker I: Well, again, after completing the current experiments, we can add up all the...\nSpeaker I: Oh, yeah.\nSpeaker D: I guess my point was that the hybrid system thing we did was primitive in many ways.\nSpeaker D: And I think I agree with you that if we fixed lots of different things and they're all that up, we probably have a competitive system. But I think not that much of it is due to the front end per se.\nSpeaker D: I think maybe a couple percent of it is as far as you can see from this.\nSpeaker D: Unless you call... Well, if you call VT, all the front end, that's a little more, but that's sort of more, both.\nSpeaker C: One experiment, we should...\nSpeaker C: We'll probably need to do, though, when...\nSpeaker C: At some point, since we're using that same...\nSpeaker C: The net that was trained on PLP without all these things in it for the tandem system, we may want to go back and retrain...\nSpeaker C: Well, that's what I meant.\nSpeaker C: Yeah, for the tandem, so we can see what effect it has on the tandem process.\nSpeaker I: So the thing is, do we expect, at this point, I'm wondering, can we expect the tandem system to do better than a properly trained...\nSpeaker I: A Gaussian system trained directly on the features with the right choice of parameters?\nSpeaker D: Well, that's what we're seeing in other areas, yes.\nSpeaker D: So it's...\nSpeaker C: So we may not...\nSpeaker C: I mean, if it doesn't perform as well, we may not know why, right?\nSpeaker C: Because we need to do the exact experiment.\nSpeaker D: I mean, the reason to think it should is because you're putting in the same information and you're transforming it to be more discriminative.\nSpeaker D: So... Now, the thing is, in some databases, I wouldn't expect it to necessarily give you much.\nSpeaker D: And part of what I view as the real power of it is that it gives you a transformation, an okay probability for taking all sorts of different wild things that we do, not just the standard front end, but other things like with multiple streams and so forth.\nSpeaker D: And it allows you to feed them to the other system through this funnel.\nSpeaker D: So I think that's the real power, but I wouldn't expect huge improvements.\nSpeaker D: But it should at least be roughly the same and maybe a little better if it's in way, way worse than...\nSpeaker C: So we're going to... Another thing that under S and I were talking about was...\nSpeaker C: In the first experiment that he did, we just took the whole 56 outputs and that's basically compared to a 39 input feature vector from either MFCC or PLP.\nSpeaker C: But one thing we could do is...\nSpeaker D: Let me just ask you, say take the 56 outputs, these are the pre-final nonlinearity outputs.\nSpeaker C: Yeah, through the regular tandem outputs.\nSpeaker C: Through the KLT.\nSpeaker C: Through the KLT, all that kind of stuff.\nSpeaker D: And so then do you use all 56 of the KLT or...?\nSpeaker C: That's what we did, right?\nSpeaker C: So one thing we were wondering is if we did principal components and say took out just 13 and then did Delta's and Double Delta's on that.\nSpeaker C: So we treated the first 13 as though they were standard features.\nSpeaker C: I mean, did Dandu experiments like that?\nSpeaker D: Talked with Stefan.\nSpeaker D: He did some things like that.\nSpeaker D: He was either him or Carmen, I forget.\nSpeaker D: I mean, all different databases and different HDK and all that.\nSpeaker D: So it may not apply, but my recollection of it was that it didn't make it better, but it didn't make it worse.\nSpeaker D: But again, given all these differences, maybe it's more important in your case that you now take a lot of these low variance components.\nSpeaker C: Because in a sense, the net's already got quite a bit of context in those features.\nSpeaker C: So if we did Delta's and Double Delta's on top of those, we're getting sort of even more...\nSpeaker C: Should be good, right?\nSpeaker D: Yeah.\nSpeaker I: But the main point is that, you know, it took us a while, but we have a procedure for coupling the two systems debug now.\nSpeaker I: I mean, there's still conceivably some bug somewhere in the way we're feeding the tenant features.\nSpeaker I: We're generating even more feeding them to the SRI system.\nSpeaker D: But there might be, because that's a pretty big difference.\nSpeaker I: Yeah.\nSpeaker I: I'm wondering how we can debug that.\nSpeaker I: I'm actually quite sure that feeding the features into the system and training it up.\nSpeaker I: I think that's essentially the same as we use with the POP features.\nSpeaker I: That's obviously working great.\nSpeaker C: Yeah, there could be a bug in the somewhere before that.\nSpeaker I: Another degree of freedom is how do you generate the POP transform?\nSpeaker E: That's right.\nSpeaker D: One other one is the normalization of the inputs to the net.\nSpeaker D: These nets are trained with particular normalization, and that could screwed up.\nSpeaker C: I'm doing what Eric coached me through them that part of it.\nSpeaker C: So I'm pretty confident in that.\nSpeaker C: I mean, the only slight differences that I use normalization values that underest calculated from the original PLP, which is right.\nSpeaker C: Yeah.\nSpeaker C: So I do, we actually don't do that normalization for the PLP doing for just the straight PLP features.\nSpeaker I: No, the SRI system.\nSpeaker C: SRI system does that, right?\nSpeaker I: So there's room for bugs that we might not have.\nSpeaker D: I would actually double check with Stefan at this point.\nSpeaker D: Because he's probably the one here and he and Dan are the ones who are at this point most experienced with the tandem.\nSpeaker D: There may be some little bit here and there that is not being handled right.\nSpeaker C: That's hard with features because you don't know what they should look like.\nSpeaker C: I mean, you can't just print the values out and ask you and you know, look at them and see if they're there.\nSpeaker A: And also, they're not, I mean, as I understand, you don't have a way to optimize the features for the final word error, right?\nSpeaker A: I mean, these are just discriminative, but they're not optimized for the final right.\nSpeaker A: So there's always this question of whether you might do better with those features if there was a way to train it for the word error metric that you're actually.\nSpeaker D: Well, you're actually, but in an indirect way.\nSpeaker D: Well, right. Just in direct so you don't know.\nSpeaker D: You may not be in this case come to think of it because you're just taking something was trained up elsewhere.\nSpeaker D: So what you do in the full procedure is you have an embedded training.\nSpeaker D: So in fact, the net is trained on a Vaterbi alignment of the training did it.\nSpeaker D: It comes from your full system.\nSpeaker D: And so that's where the feedback comes around so that it is actually discriminative.\nSpeaker D: You can prove that it's a, if you believe in the Vaterbi assumption that getting the best path is almost equivalent to getting the best total probability, then you actually do improve that by training up on local frames.\nSpeaker D: But we aren't actually doing that here because we did, we did that for a hybrid system.\nSpeaker D: And now we're plugging it into another system. So it isn't, it wouldn't quite apply here.\nSpeaker C: So another huge experiment we could do would be to take the tandem features, do SRI, forced alignments using those features.\nSpeaker A: Exactly. It's exactly.\nSpeaker A: And redo the net so that you can optimize it.\nSpeaker D: So since you're not using the net for recognition per se, but just for this transformation is probably bigger than it needs to be.\nSpeaker D: So that would save a lot of time.\nSpeaker I: And there's a mismatch in the phone sets.\nSpeaker I: So you're using a large of phone sets.\nSpeaker D: Yeah, actually all those things could could could could could could have acted as well.\nSpeaker D: The other thing, just to mention that Stefan, this was an innovation, I Stefan's, which was a pretty neat one. And my particularly apply here given all these things we're mentioning.\nSpeaker D: Stefan's idea was that discriminant approaches are great.\nSpeaker D: Even the local ones given, you know, these potential outer loops, which you can convince yourself, turn into the global ones.\nSpeaker D: However, this time is not good. When something about the test set is different enough from the training set that the discrimination that you're learning is not a good one.\nSpeaker D: So his idea was to take as the input feature vector to the Gaussian mixture system, a concatenation of the neural net outputs and the regular features.\nSpeaker A: Yeah, that didn't you do that already or oh, that makes a lot of sense.\nSpeaker I: When I first started corresponding with Dan about how to go about this, I think that was one of the things that we, I mean, I'm sure that Stefan was the first to think of it, but actually Stefan did it.\nSpeaker D: And it helped a lot.\nSpeaker D: So that's our current best system in the, yeah.\nSpeaker D: Yeah, that makes sense.\nSpeaker I: You should never do worse.\nSpeaker I: I'm on the combined feature vector. I miss what you said you do a KLT transform on the combined feature vector.\nSpeaker D: Yeah, well, actually, you should check with him because he tried several different combinations.\nSpeaker I: Because you ended up with this huge feature vector. So that might be a problem unless you do something from the first place.\nSpeaker D: Yeah, I, what I don't remember is which came out best. So he did one where he put a put the whole thing into one KLT.\nSpeaker D: And another one since the PLP things are already orthogonalized, he left them alone and, and just did a KLT on the, on the net outputs and again a that.\nSpeaker D: And I don't remember which was better.\nSpeaker C: Did he, did he try to, so he always ended up with a feature vector that was twice as long as either one of the, no, I don't know, I don't know.\nSpeaker D: You have to check with him.\nSpeaker D: I meant a big idea these days.\nSpeaker A: You need to close up because I need to save the data.\nSpeaker A: Not to mention that I have a few snacks.\nSpeaker A: Right. Did people want to do the digits or do them together?\nSpeaker D: I think given that we're doing for snacks, maybe we should do them together.\nSpeaker A: Okay. I mean, we're trying to do them in synchrony. That might be fun.\nSpeaker A: So he's not here to tell me no.\nSpeaker D: It's not going to work out, but we could, we could just see if we find the rhythm, you know.\nSpeaker D: Sure.\nSpeaker D: Oh, there's zero. So we want to agree on that.\nSpeaker A: Maybe just whatever people would naturally do. I don't know.\nSpeaker D: Well, but if we were singing group, we would want to decide.\nSpeaker A: Viet Harman, you know.\nSpeaker A: Sorry. So I set up and we didn't have enough digit form. So I zerox the same one seven times.\nSpeaker D: I'm going to have a problem with saying zero.\nSpeaker D: No.\nSpeaker D: Okay.\nSpeaker D: Okay. One and two and three.\nSpeaker E: Seven eight seven one five two zero three zero two eight one two zero two two six one four six zero three.\nSpeaker E: Two seven eight two six two two three four nine eight seven zero seven five two nine one zero nine one eight zero six zero five one five six two eight four three six five seven nine five one eight eight.\nSpeaker E: Eight eight four nine five three two five three eight one two zero zero one seven three three eight six zero zero.\nSpeaker D: What's more with feeling?\nSpeaker A: It's transcript L one three eight.\nSpeaker A: Oh, yeah.\nSpeaker A: It was.\nSpeaker F: It sounded like it.\nSpeaker E: It sounded very.\nSpeaker F: It was the lack of prasadica.\nSpeaker A: Exactly.\nSpeaker A: Everybody's sort of lowers their pitch.\nSpeaker A: And now we're going to go out and have some access.\nSpeaker F: Now we know.\nSpeaker F: No.\nSpeaker F: All right.\n"}, {"meeting_id": "Bed009", "summary": "The group discussed plans and concerns regarding the architecture of SmartKom, its proposed modules, and the types of interactions expected to take place between modules. The meeting was largely focused on SmartKom's decision making capacity and how to adapt this functionality to the tourist information domain. The group set a date for assessing SmartKom plans.", "dialogue": "None: Okay.\nSpeaker D: Testing channel 2.\nSpeaker D: Two.\nSpeaker D: Two.\nSpeaker C: Oh.\nSpeaker E: Hmm?\nSpeaker E: There.\nSpeaker E: Wow.\nSpeaker E: So, rough and terminate here.\nSpeaker E: Great.\nSpeaker E: Great.\nSpeaker D: Made it safety.\nSpeaker D: So, what we have been doing is they would like us all to read these digits.\nSpeaker D: But we don't already.\nSpeaker D: A couple people read it.\nSpeaker D: When it gives them all German accents.\nSpeaker E: Okay.\nSpeaker E: The way you do it is you just read the numbers not as each single.\nSpeaker E: So, just like I do it.\nSpeaker E: First you read the transcript number.\nSpeaker E: My transcript number is L95.\nSpeaker E: 3230977352214082115160087096440 64404562850593 5261 358105 624 468 48551 4229 05756 07 32\nSpeaker B: Okay. What's my transcript number is L91 141628613157 0395 8 06622 0316143416171340652105 9821408492 126586666 866161311102066654159\nSpeaker A: Okay. My transcript number is L92 85034213760632 4970 6068 636 564 9306 641 8706 807 3 1770 7556 445660 4510 845 841 1037 1072 6774 1700\nSpeaker C: Okay. Let's be done with this. Okay. This is Mommy.\nSpeaker C: Come on in.\nSpeaker C: Okay. So, we're going to try to finish by five.\nSpeaker D: So, people who want to go here, Nancy Chang's talk downstairs.\nSpeaker D: And you guys are giving talks on tomorrow and Wednesday much.\nSpeaker C: That's great. Okay. So, you know what we're going to do?\nSpeaker E: I thought two things we'll introduce ourselves.\nSpeaker E: What we do. And we already talked with Andreas Tilo and David and some lines of code were already written today and almost tested.\nSpeaker E: And just going to say we have again the recognizer to pass the thing where we're working on and that should be no problem.\nSpeaker E: And then that can be sort of developed as needed when we enter the tourism domain.\nSpeaker E: We have talked this morning with Tillman about the generator.\nSpeaker E: And they are one of our diligent workers has to sort of volunteer to look over Tillman's shoulder while he is changing the grammar to English because we have we face two ways.\nSpeaker E: Either we do a simple concatenating grammar for the English generation which is sort of starting from scratch and doing it the easy way.\nSpeaker E: Always simply adopt the more in-depth style that is implemented in the German system.\nSpeaker E: And are then able not only to produce strings but also the syntactic parts, not parts in the syntactic tree that is underneath the syntactic structure.\nSpeaker E: Which is the way we decided we were going to go because it's easier in the beginning.\nSpeaker E: And that's required some knowledge of those grammars and some linguistic background.\nSpeaker E: But it shouldn't be a problem for anyone.\nSpeaker C: Okay, so that's the answer.\nSpeaker C: You're going to have some time to do that with these guys.\nSpeaker E: Sure.\nSpeaker D: Because you're the grammar major.\nSpeaker D: Okay.\nSpeaker C: I mean it makes sense doesn't it?\nSpeaker D: Yeah.\nSpeaker D: Okay, so I think that's probably the right way to do that.\nSpeaker D: And yeah, so I actually want to find out about it too but I may not have time to...\nSpeaker E: The ultimate goal is that before they leave we can run through the entire system input through output on at least one or two simple things.\nSpeaker E: And by virtue of doing that, then in this case, John will have acquired the knowledge of how to extend it at infinitum.\nSpeaker E: When needed, if needed, when wanted.\nSpeaker E: So for...\nSpeaker C: Okay, that sounds great.\nSpeaker E: And also, Ralph has hooked up with David and you're going to continue either all through tonight or tomorrow and whatever to get the password interface working.\nSpeaker E: They're sending out and ticking out lattices and doing this kind of stuff to see what works best.\nSpeaker D: So you guys enjoy your weekend?\nSpeaker D: Yes, yeah, I'm sorry.\nSpeaker D: You ever got to put the work?\nSpeaker D: Okay, so that's sort of one branch is to get us caught up on what's going on.\nSpeaker D: Also of course would be really nice to know what the plans are in addition to what's already in code.\nNone: And we can...\nSpeaker D: I don't know what was or the time when we were set up to do that.\nSpeaker D: It probably will work better if we do it later in the week after we actually understand better what's going on.\nSpeaker D: So when do you guys leave?\nSpeaker A: We're here through Sunday.\nSpeaker A: Oh, okay, so...\nSpeaker D: So we'll find a time later in the week to get together and talk about your understanding of what smart comp plans are and how we can change it.\nSpeaker E: So we already said a day for that.\nSpeaker E: Might be a bit official while we're all here.\nSpeaker C: Okay.\nNone: What does not work for me is Thursday afternoon.\nSpeaker D: I can do earlier than day on Thursday or most of the time on Friday.\nSpeaker E: Thursday morning sounds fine?\nSpeaker E: What are your constraints?\nSpeaker E: Thursday afternoon doesn't work for me, but...\nSpeaker E: It's either Thursday morning, no.\nSpeaker C: Thursday morning should be fine.\nSpeaker C: 11?\nNone: 11 on Thursday.\nSpeaker C: I was just thinking I will 11 by 11.\nSpeaker D: This is a different to our morning people.\nSpeaker A: So he's there.\nSpeaker A: Third Sunday live.\nSpeaker D: And actually we could invite Andreas as well.\nSpeaker D: He will be in Washington.\nSpeaker D: That's true.\nSpeaker D: He's off the office trip already.\nSpeaker E: But David is here and he's actually...\nSpeaker E: Those everything about the smart comp plans.\nSpeaker C: Okay, we'll see if David can make it out.\nSpeaker E: Okay, so facing to what we've been doing here.\nSpeaker E: One thing we're also using this room to collect data.\nSpeaker E: Not this type of data, not meeting data, but sort of our version of a wizard experiment, not like the ones in Munich, but pretty close to it.\nSpeaker E: The major difference to the Munich ones is that we do it...\nSpeaker E: We add the telephone, even all the recording is done here.\nSpeaker E: And so it's sort of a computer call system that gives you tourist information to get places.\nSpeaker E: And it breaks halfway through the experiment and a human operator comes on.\nSpeaker E: And part of that is sort of try to find out whether people change their linguistic verbal behavior when first thinking they speak to a machine or then to a human.\nSpeaker E: And we're setting it up so that we can...\nSpeaker E: We hope to implant certain intentions in people, for example.\nSpeaker E: We have first looked at a simple sentence that...\nSpeaker E: How do I get to the powder tower?\nSpeaker E: Okay, so you have that castle of idle work and there is a tower and it's called powder tower.\nSpeaker E: And so what do you parse out of that sentence?\nSpeaker E: Probably something that we specified in M3L that is...\nSpeaker E: Action, go to whatever domain object, whatever, powder tower.\nSpeaker E: And maybe some model will tell us some GPS module in a mobile scenario where the person is at the moment.\nSpeaker E: And we've sort of gone through that once before in the DeepMap project.\nSpeaker E: And we noticed that first of all, what are...\nSpeaker E: I should have brought some slides.\nSpeaker E: But what are...\nSpeaker E: So here's the tower.\nSpeaker E: Think of this as a two-dimensional representation of the tower.\nSpeaker E: And our system that people here, to a point where they were facing a wall from the tower, there's no entrance here.\nSpeaker E: But it just happens to be the closest point of the road network to the geometric center.\nSpeaker E: And it just says how the algorithm works.\nSpeaker E: So we took out that part of the road network as a heck, and then it followed actually the way to the entrance, which was now the closest point of the road network to geometric center.\nSpeaker E: But what we actually observed in idle work is that most people, when they want to go, they actually don't want to enter.\nSpeaker E: Because it's not really interesting.\nSpeaker E: They want to go to a completely different point where they can look at it and take a picture.\nSpeaker E: So, let's say a simple parse from an utterance won't really give us, is what the person actually wants.\nSpeaker E: Does he want to go there to see it?\nSpeaker E: Does he want to go there now?\nSpeaker E: Later?\nSpeaker E: How does the person want to go there?\nSpeaker E: Is that person more likely to want to walk there, walk a scenic route, and so forth?\nSpeaker E: There are all kinds of decisions that we have identified in terms of getting to places and in terms of finding information about things.\nSpeaker E: And we are constructing, and then we've identified more or less the extra linguistic parameters.\nSpeaker E: It may play role, information related to the user and information related to the situation.\nSpeaker E: And we also want to look closely on the linguistic information, what we can get from the utterance.\nSpeaker E: That's part of why we implant these intentions in the data collection to see whether people actually are facing differently, whether they want to enter in order to buy something, or whether they just want to go there to look at it.\nSpeaker E: And so, the idea is to construct suitable interfaces and a belief net for a module that actually tries to guess what the underlying intention was.\nSpeaker E: And then, enrich or augment the M3L structures with what I thought, what more it sort of got out of that utterance.\nSpeaker E: So, if it can make a good suggestion, hey, that person doesn't want to enter.\nSpeaker E: That person just wants to take a picture, because he just bought a film, or that person wants to enter, because he discussed the admission fee before, or that person wants to enter, because he wants to buy something, and that you usually do inside of buildings and so forth.\nSpeaker E: These types of additional information are going to be embedded into the M3L structure, in a sort of subfield that we have reserved.\nSpeaker E: And if the action planner does something with that grade, if not, then that's also something that we can't really, at least we want to offer the extra information, we don't really, we're not too worried.\nSpeaker E: I mean, ultimately, if you have, if you can offer that information, somebody's going to do something with it sooner or later, that's sort of part of our belief.\nSpeaker E: For example, right now, I know the GIS from email is not able to calculate these viewpoints.\nSpeaker E: So, that's a functionality that doesn't exist yet, to do that dynamically.\nSpeaker E: But if we can offer that distinction, maybe somebody will go ahead and implement it.\nSpeaker E: Surely nobody is going to go ahead and implement it, if it's never going to be used.\nSpeaker E: What have I forgotten about?\nSpeaker D: Yeah, I'm happy to do it.\nSpeaker D: It's a good time to pause.\nSpeaker D: I see questions on people's faces.\nSpeaker A: What would the office want to be if you envision this as a module within SmartCom?\nSpeaker E: So far, I've sort of it sort of adding it on to the model and knowledge module.\nSpeaker E: So, this is one that already adds additional information to the, but it could sit anywhere in the attention recognition.\nSpeaker E: I mean, basically, this is what attention recognition literally sort of can.\nSpeaker E: That's why it should be.\nSpeaker A: Well, from my understanding of what the people at TIPS were originally trying to do, doesn't seem to quite fit into SmartCom currently.\nSpeaker A: So what they're really doing right now is only selecting among the alternatives, the hypothesis that they're given and enriched by the domain knowledge and the dismalarist modeler and so on.\nSpeaker A: So if this is additional information, that could be merged in with them.\nSpeaker A: And then it wouldn't be available to action planning and others.\nSpeaker D: Okay, that was one question.\nSpeaker D: Are there other things that, because we want to not pass over any questions or concerns that you have?\nSpeaker A: Whether they're two levels of giving an answer, I guess on both levels.\nSpeaker A: I don't have any further questions. The two levels of being, as far as I'm concerned, is standing here for the generation module and the other is my understanding of what SmartCom is supposed to be.\nSpeaker D: So, let me explain that a little bit from the point of view of the generation.\nSpeaker D: So the idea is that we've actually got this all laid out and we could show it to you.\nSpeaker D: I think Robert didn't bring it today, but there's a belief net, which is, there's a first cut of the belief net that doesn't, isn't fully instantiated in particular.\nSpeaker D: Some of the combination rules and ways of getting additional probabilities aren't there.\nSpeaker D: But we believe that we have laid out the fundamental decisions in this little space and the things that influence them.\nSpeaker D: So, one of the decisions is what we call this ABE.\nSpeaker D: You want to access view or enter.\nSpeaker D: So that's a discrete decision. There are only three possibilities.\nSpeaker D: And what we would like is for this knowledge modeling collection to add which of those it is and give it to the player.\nSpeaker D: But the current design suggests that if it seems to be an important decision and if the belief net is equivocal so that it doesn't say that one of these is much more probable than the other, then an option is to go back and ask for the information you want.\nSpeaker D: There are two ways one can imagine doing that. For the debugging, we'll probably just have a drop down menu and while you're debugging, you're just mucking.\nSpeaker D: But for a full system, then one might very well formulate a query, give it to the dialogue planner and say, are you planning to enter or whatever it might be. So that's under that model, then there would be a loop in which this thing would formulate a query, presumably give it to you that would get expressed and hopefully you get an answer back.\nSpeaker D: And that would of course, the answer would have a harsh two. You probably won't do this early on because the current focus is more decision making and stuff like that.\nSpeaker D: But while we're on the subject, I just wanted to give you a heads up.\nSpeaker D: It could be that some months from now, we said, okay, we're now ready to try to close that loop in terms of querying about some of those decisions.\nSpeaker A: So my suggestion then is that you look into the currently ongoing discussion about how the action plans are supposed to look like.\nSpeaker A: And they're currently agreeing or in the process of agreeing on an exemplification of something like a state transition network of how dialogues would proceed.\nSpeaker A: And these transition networks will be what the action plan of the next interprets in a sense.\nSpeaker C: You know this, honey?\nSpeaker A: And Mikhail is doing that right? Well, Markus is actually implementing that stuff and Markus and Michele together are leading the discussion there.\nSpeaker D: Okay. So we have to get it on that because partly those are like excements, the transition backgrounds.\nSpeaker D: And it may be that we should early on make sure that they have flexibility to be weak.\nSpeaker E: But they have understood this right? They govern more or less the dialogue behavior or the action.\nSpeaker E: It's not really what you do with the content of the dialogue, but it's.\nSpeaker D: I mean, there is this nice. So there's a. So the word action. Okay. Is what's ambiguous here. So one thing is there's an actual planner that tells the person first of me.\nSpeaker D: Where it tells the person how to go first go here first go there. Take a bus, whatever it is. So that's that form of planning and action and a round planner and GIS also to stuff.\nSpeaker A: But I think that isn't what you know. No, no, in smart home terminology that's called a function that's modeled by a function modeler.\nSpeaker A: And it's that that's completely encapsulated from the dialogue system. That's simply a functionality that you give data is in a query.\nSpeaker A: And then you get back from that functioning model, which might be a planner or a VCR or whatever. Some result. And that's then.\nSpeaker D: Okay. So that's what I thought. So action action here means that I have speech. Yeah. Dialogue.\nSpeaker D: Okay. I think that I think it's not going to. I think it's not going to be good enough. I don't know what I meant by that.\nSpeaker D: So I think the idea of having a transition diagram or the grammar of conversations is a good idea.\nSpeaker D: Okay. And I think we do have to get in on it. But I think that when so when you get to the tourist domain.\nSpeaker D: It's not just an information retrieval system. Right. So this is where I think people are going to have to think this to a bit more carefully.\nSpeaker D: So if it's only like in the film and TV thing, okay, you can do this. You just get information and give it to people.\nSpeaker D: But what happens when you actually get them moving and so forth and so on.\nSpeaker D: I think the notion of this is a self contained module. The functional module that interacts with where the tourist domain is.\nSpeaker D: Probably is too restrictive. I don't know how much people have thought ahead to the tourist domain.\nSpeaker A: Probably not another. Another more basic point there is that the current tasks and therefore the concepts of this.\nSpeaker A: What's called the action plan. It's really a dialogue manager is based on slots that have to be filled.\nSpeaker A: The kind of values in these slots would be fixed things like a time or a movie time setting like this.\nSpeaker A: Whereas in the tourist domain might be an entire route.\nSpeaker A: It's a very complex structure information in these slots. Not sure if complex slots of that type are really being taken into consideration.\nSpeaker A: So that's really something.\nSpeaker A: We need to be settled there. So this is really an ongoing discussion.\nSpeaker E: We have faced and implemented those problems once already.\nSpeaker E: Maybe we can even shovel some know-how from there to Markus and Micheal.\nSpeaker E: I'll talk to Micheal. How far is the M3L specification for the natural language input gone?\nSpeaker E: I haven't seen anything for the tourist path.\nSpeaker B: It's not defined yet.\nSpeaker E: You are probably also involved in that.\nSpeaker B: We'll meet next week.\nSpeaker E: Those are the two key issues. How does the input pipeline look like and what the action planner does with it?\nSpeaker E: I think of the internal working of the action planner and the function model as relevant.\nSpeaker E: That can be as detailed or as crude as you want it to be.\nSpeaker E: The internal workings of the action planner and the work with the state.\nSpeaker E: That shouldn't really matter too much.\nSpeaker E: It does have to keep track of your bare-on-part six of a route that consists of eight steps.\nSpeaker D: I think there are a lot of reasons why it matters.\nSpeaker D: The user says that the action planner told it if the parser and the language end doesn't know what the person has been told.\nSpeaker D: The person says that the planner says that the planner doesn't know that.\nSpeaker D: There are all sorts of dialogues that won't make any sense.\nSpeaker A: The point has been realized that it's not really defined yet.\nSpeaker A: There's going to be some kind of feedback from the action planner into all the analysis modules telling them what to expect, what the current state of the discourse is beyond what's currently being implemented.\nSpeaker D: This is not just the state of the discourse, this is actually the state of the plan.\nSpeaker D: It's great if people are already taking that into account.\nSpeaker D: The specifics are in this room.\nSpeaker D: The question is, can you put in this need a fair amount of feedback from planning it in these things which are much more continuous than the dialogue over movies and stuff?\nSpeaker A: The action planner needs to be able to have an expressive power that can deal with these structures.\nSpeaker D: The next question is, can you put in a fair amount of feedback from the action planner?\nSpeaker D: It ought to be called a dialogue manager.\nSpeaker D: What would happen if we said, we've talked about this and we've changed this.\nSpeaker A: Probably the most impossible.\nSpeaker A: Who you talk to, how we'll see.\nSpeaker A: I think this is just for historical reasons within the preparation phase of the project and not because somebody actually believes it ought to be action funded.\nSpeaker D: If that persists, then we're going to need another term for the thing that actually does the planning of the routes and whatever we're doing for the tourist.\nSpeaker E: That's external services.\nSpeaker D: That has all the wrong connotations.\nSpeaker D: It sounds like it's standalone, it does interact, it doesn't.\nSpeaker D: That's something I think you can't. It's fine for looking up when the show is on TV.\nSpeaker D: I think it's really wrong headed for something that has a lot of state it's going to interact in a complicated way with understanding the board.\nSpeaker E: I think just the spatial planner and the route planner, I showed you once the interaction between them among them in the deep map system.\nSpeaker E: A printout of the communication between those two fills up how many pages.\nSpeaker E: That's just part of how do I get to one place.\nSpeaker E: So this is definitely a good point to get into the discussion or to enter his discussion actually.\nSpeaker E: Is he new in the...\nSpeaker A: Yes.\nSpeaker A: He started, like, January.\nSpeaker A: He's going to be responsible for the interpretation of this action plan.\nSpeaker E: He's going to continue with the old thing.\nSpeaker E: Yes, I was just wondering the next question, we're going to stick to ProLoc or not.\nSpeaker E: But I do think the function modeling concept has a certain...\nSpeaker E: It makes sense in a certain light because the action planner should not be or the dialogue manager in that case should not...\nSpeaker E: We have to worry about whether it's interfacing with something that does route planning in this way or that way.\nSpeaker D: I agree. There is a logic to dialogue which is separable.\nSpeaker E: And it can sort of formulate what it wants in a rather abstract way.\nSpeaker E: Find me a good route for this. It doesn't really have to worry about how route planner AO or route planner B actually wants it.\nSpeaker E: So this seems like a good idea.\nSpeaker D: It's tricky because one could well imagine, I think it will turn out to be the case that this thing we're talking about, and the extended knowledge modeler will fill in some parameters about what the person wants.\nSpeaker D: One could well imagine that the next thing that's trying to fill out the detailed route planning, let's say, will also have questions that it would like to ask the user.\nSpeaker D: You could well imagine you get to a point where it's got a choice to make and it just doesn't know something.\nSpeaker D: So you would like it also be able to formulate a query and to run that back through the dialogue manager and to the output module and back around.\nSpeaker D: And a good design would allow that if you can't make it happen.\nSpeaker A: So that doesn't necessarily contradict an architecture where there really is a person that you will define the interface.\nSpeaker D: I totally agree. But what it needs, the point is, in that case the dialogue manager is sort of a vent driven.\nSpeaker D: So dialogue manager may think it's in a dialogue state of one sort.\nSpeaker D: And this one of these planning modules comes along and says, hey, right now we need to ask a question.\nSpeaker D: So that forces the dialogue manager to change state.\nSpeaker A: Okay, it can be true. Yeah, yeah, I think that's the concept of people.\nSpeaker A: And then the underlying idea, of course, is that there is something like kernel modules with kernel functionality that can pluck certain applications like tourist information or the whole scenario of controlling a VCR install.\nSpeaker A: And then extend it to arbitrary number of publications.\nSpeaker A: So that's an additional reason to have this well defined interface.\nSpeaker A: Keep these things like tourist information external.\nSpeaker E: Of course, there is another philosophical issue that I think you can debate.\nSpeaker E: But this makes sense to me that sooner or later a service is going to come and describe itself to you.\nSpeaker E: And that's sort of what Srini is working on in the Dumbled project where you find a GIS about that gives you information on Berkeley.\nSpeaker E: And it's going to be there and tell you what it can do and how it wants to do things.\nSpeaker E: And so you can actually interface to such a system without ever having met it before.\nSpeaker E: And the function modeler and the self description of the external service, handle it out.\nSpeaker E: And you can use the same language core understanding core to interface with planner a planner B planner C and so forth, which is, you know, a utopian completely utopian at the moment.\nSpeaker E: But slowly getting into the realm of the contingent.\nSpeaker E: But we are facing, of course, much more realistic problems and language input, for example, is of course crucial.\nSpeaker E: And also when you do the sort of deep understanding analysis that we envision, then of course, the, you know, what is the property of the stimulus, the last we get of that, the better.\nSpeaker E: And so we were thinking, for example, how much syntactic analysis actually happens already in the parser and whether one could interface to that potentially.\nSpeaker B: Currently, it's no syntactic analysis. But in the next release, and it's kind of, so we looked at the current pattern matching.\nSpeaker D: And as you say, it's just the surface pattern matching. So what are the plans roughly?\nSpeaker B: To integrate and syntactic analysis and some more features like segmentation. So then more than one utterance is there.\nSpeaker D: And this is all done, a pause between it, segmentation across. So the idea is to have a particular, particular parser in mind.\nSpeaker D: And if you thought through, is it an HBSG parser? Is it a, whatever? No, no, I think it's complicated for.\nSpeaker B: Okay. One person has to. Oh, you have to do it. Yeah. So things must be simple. I see. So.\nSpeaker D: But yeah, the syntactic analysis. People in finite state trans-susers. People at DFK, I have written a fair number of parsers. Other people over the years have written various parsers in DFK. None of them are suitable. I'm asking.\nSpeaker B: Yeah, the problem is that it has to be very fast because if you want to for more than one path and what's in the lattice from a speed track or not, so it's speed is crucial.\nSpeaker B: Not fast enough. It also has to be very robust cause of speed track recognition. I don't know. So there was a chunk parser in verbobile.\nSpeaker D: There was one of the branches. You know, I do, there were these various competing syntax modules. And I know one of them was a chunk parser. And I don't remember who did that.\nSpeaker B: I think that a tubing and I thought. I didn't know. Well, do you know something about it? Tubing was at least involved in putting the chunks together.\nSpeaker A: I can't quite recall whether they actually produced the chunks in the first place.\nSpeaker D: That's right. They had just done with a two-phase thing where the chunk parser itself was pretty stupid. And then there was a kind of trying to fit them together that used more context.\nSpeaker A: And especially you did some, was a learning based approach, which you learned from a big corpus of trees. And yes, the chunk parser was a financial machine that Mark Leidritch worked on and wasn't tubing them.\nSpeaker A: And somebody else was tubing that up, so it was done and tubing.\nSpeaker B: But is that the kind of thing you were thinking of? Yeah. It sounds like the star action. What? It's in this direction.\nSpeaker E: From Micheal Stubach, I've heard very good stuff about the chunk parser that is done by four-vice riches in embassy doing the parsing. So this sort of came as a surprise to me that embassy is featuring a nice parser.\nSpeaker E: But it's what I hear one could also look at that and see whether there is some synergy possible. And they're doing chunk parsing.\nSpeaker E: I can give you the names of the people who do it there. But then there's of course more ways of parsing things.\nSpeaker D: Of course, but given the constraints that you wanted to be small and fast and so forth, my guess is you're probably into some kind of chunk parsing.\nSpeaker D: And I'm not a big believer in this statistical cleaning up. That seems to be kind of a last resort if you can't do it any other way.\nSpeaker D: But I don't know. Maybe that's what you guys finally decided to do.\nSpeaker D: And if you looked just again for context, there's this one that they did at SRI some years ago, fastest pace.\nSpeaker B: Yeah, I've looked at that. But there's not much information available. But it's also finance Tetrance. It is. Yeah, it was pretty ambitious.\nSpeaker B: And of course it was English oriented. And fully finance Tetrance, I'm not so good for German.\nSpeaker D: Yeah, I guess that's the point is all the morphology and stuff. In English is all word order and it makes a lot more sense.\nSpeaker B: Yeah, okay. Good point. So in German, you've got most of this. So it's a choice between risk processing and set processing and template.\nSpeaker B: So what about like morphics? You've got stemmers or is that something? Yeah, but all in the in the lexicon.\nSpeaker B: But you have that. Yeah, information is a lot of that.\nSpeaker D: Okay, I see. So, but so you just connect to the lexicon. Yeah, at least for German, you have all of the stemming information.\nSpeaker B: Yeah, we can. We have knowledge passes from from Rappm\u00f6kels. Yeah.\nSpeaker D: But it doesn't look like it you're using it. I didn't see it being used in the current template parser. I didn't see any.\nSpeaker B: Which we actually only look at the English. But it's used for stem forms.\nSpeaker A: I think there's some misunderstanding. Morphics is not used online. So the lexicon might be derived by morphics, but what's happening online is just retrieving from the lexicon, which we call the stemming information.\nSpeaker A: So it will be a full form lexicon. That's what you have. Yeah.\nSpeaker E: We threw out all the forms. We threw out all the forms because English.\nSpeaker D: Oh, okay. So, yeah, so I thought I so in German, then you actually do case matching and things like that in the in the pattern matcher or not.\nSpeaker D: Not yet. I didn't. Okay.\nSpeaker D: I didn't think I saw it. Yeah. Getting it from the lexicon is just fine.\nSpeaker C: Yeah.\nSpeaker D: Yeah. Here's the case where the English and the German might really be significantly different in terms of if you're trying to build some fast parser and so forth. You really might want to do it in a significantly different way.\nSpeaker D: So you guys have looked at this also in terms of, you know, if you're doing this for English as well as Germans, you think now that it would be this similar way?\nSpeaker B: Yeah. I think it's possible to do list processing. And maybe it's more adequate for English and German set processing.\nSpeaker B: Maybe some extensions have to be made for English version.\nSpeaker E: I'm sure there's going to be more discussion on that after your talk. We're just going to foreshadow.\nSpeaker D: Now actually, are you guys three of five? Do you have to go somewhere at five o'clock tonight? No. I think I was just talking.\nSpeaker D: I'm just going to practice talk. Great. So you're going to. Yeah. That's good. Because that will tell you a fair amount about the form of semantic construction grammar that we're using.\nSpeaker D: So I think that's probably as good an introduction as you get to the form of conceptual grammar that will be having mind.\nSpeaker D: So I won't talk particularly about how that relates to what Robert was saying at the beginning. Let me give you a very short version of this. So we talked about the fact that they're going to be a certain number of decisions that you want the knowledge modeler to make that we then fed to the function module.\nSpeaker D: So they're these decisions. And then one half of this we talked about a little bit is how if you had the right information, if you knew something about what was said and about something about was the agent, a tourist or a native or a business person or a younger role, whatever.\nSpeaker D: So we're also about the what we're calling the entity is that a castle is it a bank is it a town square is a statue, whatever. So all that kind of information could be combined into decision networks and decisions.\nSpeaker D: So the other half of the problem is how would you get that kind of information from the parsed input. So what you might try to do is just build more templates saying we're trying to build a template, you know, build a template somehow would capture the fact that you want to take a picture.\nSpeaker D: And we could you could do this and it's a small enough domain that probably.\nSpeaker D: But from our point of view, this is also a research project and there are a couple of people not here. The various reasons we're doing Dr. of the citations on this. And the idea that we're really after is a very deep semantics based on cognitive linguistics and the notion that there are a relatively small number of primitive conceptual schemas that characterize a lot of activity.\nSpeaker D: So a typical one in this formulation is a container.\nSpeaker D: And the notion is that all sorts of physical situations are characterized in terms of container point in and out of portals.\nSpeaker D: But also importantly for lay coffin these guys is also a metaphorical picture also characters this way you get in trouble.\nSpeaker D: So what we're really trying to do is to map from the discourse to the conceptual semantics level and from there to the appropriate decisions.\nSpeaker D: So another one of these primitive what are called image schemas is goal seeking.\nSpeaker D: There's an ocean of the source, path goal, trajectory possibly obstacles. The idea is this is another conceptual primitive.\nSpeaker D: And that all sorts of things particularly in the tourist domain can be represented in terms of source, path goal. So the idea would be could we build and analyze it, we'd take an utterance and say, aha, this utterance is talking about an attempt to reach a goal.\nSpeaker D: The goal is this, the person, the traveler is that, the sort we are now is this, they've mentioned possible obstacles, et cetera.\nSpeaker D: And this is an attempt to get very wide coverage. So if you can do this then the notion would be that across a very large range of domains you could use this deep conceptual basis as the interface.\nSpeaker D: And then the processing of that both on the input end recognizing that certain words in a language talk about containers or goals, et cetera.\nSpeaker D: And on the output end given this kind of information you can then make decisions about what actions to take provides they claim a very powerful general notion of deep semantics.\nSpeaker D: And we're really, Nancy is going to her talk is going to be not about using this in applications, but about modeling how children might learn this kind of deep semantic grammar.\nSpeaker A: And how do you envision the deep semantic to be worked with would it be highly ambiguous. And then there would be another module that takes that highly underspecified deep semantic construction and map it onto the current context to find out what the person really was talking about in that context.\nSpeaker D: Well, that's that's where the belief that comes in. So the idea is let's take this business not going to the powder tower. So part of what you'll get out of this will be the fact that if it works right.\nSpeaker D: Okay, that this is an agent that wants to go to this place and that's their goal. And there'll be additional situational information.\nSpeaker D: Okay, part of comes to the ontology the tower is this kind of part of it comes with the user. And the idea of the belief that is it combines the information from the dialogue which comes across in this general way.\nSpeaker D: You know, this is this is a goal seeking behavior along with specific information from the ontology about the kinds of objects involved about the situation about is raining. I don't know, whatever it is. And so that's the belief that we've laid out.\nSpeaker D: And so the coupling to the situation comes in this model from at the at the belief net company evidence from the dialogue with the ontology with the situation.\nSpeaker D: Nancy isn't going to talk about that just about the first steps. Right, the construction grammar.\nSpeaker E: And she's going to start in a minute.\nNone: Yeah, I didn't want to give you a little.\nNone: Okay.\n"}, {"meeting_id": "ES2015b", "summary": "The meeting mainly discussed the design of functions and the appearances. On function wise, the group had a little argument whether to make the product multi-functional or single functional. In the end they went with single-functional because it was more focused and affordable. They also removed some of the functions being discussed such as LCD screen, teletext and voice recognition, because they wanted to stick with a simplistic design. On design wise, the group finalized on the colour yellow as their product appearance and they chose illuminated buttons as one of their highlights. They also included some humanitarian factors into the design.", "dialogue": "Speaker B: This is some great facial expressions, Matthew.\nSpeaker B: Remember it's not real.\nNone: Yeah, it's okay.\nSpeaker B: You're going to be the first one on PowerPoint.\nSpeaker B: I'll just like hook you up with that.\nSpeaker B: Am I starting now?\nSpeaker B: Welcome back to the second meeting.\nSpeaker B: I hope you had a productive last 30 minutes.\nSpeaker B: I'll be taking minutes on this one and being hooked up to the PowerPoint for this meeting isn't very necessary for myself because it'll be more about what you guys are bringing to the meeting today.\nSpeaker B: The first presentation we'll be looking at is Poppy's presentation.\nSpeaker B: So, take it away Poppy.\nSpeaker A: It's plugged in.\nSpeaker B: Sorry about this guy.\nSpeaker C: Right.\nSpeaker A: Okay.\nSpeaker A: I'll take this time just to apologize.\nSpeaker A: I'm only seeing my emails later on because I was too busy carrying away doing my own thing, which is not obviously not a very good part of a team working thing, but there we go from Shurds.\nSpeaker A: I was looking at how we're going to go about the working design and what we actually need to do and what the remote control needs to do.\nSpeaker A: It needs to allow a person to have a portable device so they can control the television from wherever they are.\nSpeaker A: They don't need to actually manually touch the television set.\nSpeaker A: So it gives them much more flexibility and allows them to view where they want to be.\nSpeaker A: From on the functional side of things, we found out from a previous meeting what we decided at certain points will make our project unique.\nSpeaker A: One is the visibility in the dark, which was Genevieve's idea.\nSpeaker A: So we need to think about how we could bring this in technically and we could use illuminated buttons, which we're all familiar with when we're using a mobile phone or something familiar.\nSpeaker A: It also massively lights up at first touch.\nSpeaker A: Or we could use fluorescent materials, which would just take in the light during the day and then as soon as they go off they would glow in the dark.\nSpeaker A: Also we could use an alarm.\nSpeaker A: So if we lost the remote control, perhaps there could be a button on the television set itself which you could press.\nSpeaker A: And then an alarm from the handset would sound where it was hopefully in the room.\nSpeaker A: Maybe behind a cushion or somewhere.\nSpeaker A: That would work.\nSpeaker A: Go back there.\nSpeaker A: Another thing I think we missed out on on the last meeting was the fact that we should consider the environmental impact of our design.\nSpeaker A: From previous researchers that have carried out another project.\nSpeaker A: We've learned about smart materials where specific alloys of metals have a shape memory so they can be heated and cooled and they change the shape of the metal.\nSpeaker A: So for example a screw that's holding something together could expand and it would force all the components apart.\nSpeaker A: So the benefits of this for our product would be that when it came to the end of its product life if it was heated everything would spring apart.\nSpeaker A: All the individual components could be easily separated and then some could be reused, some could be recycled and I think that would be very important for products now.\nSpeaker A: Especially because there's so much responsibility for all the companies who are coming up with new designs because we all know that our results have been limited and they have to be very environmentally conscious.\nSpeaker B: One question.\nSpeaker B: This self destructible metal it allows for recycling materials so that someone could have this product for as long as they felt that they wanted it and then once they contributed then that company can break down the parts better.\nSpeaker A: Yeah they would you would make the product as you normally would apart from the bits that hold it all together would be made out of the shape memory alloy and that's the part that would allow all the other parts to be separated at the end.\nSpeaker A: I mean the user would return the product to the company because it's a product's responsibility to get rid of what they've made and then the company could then make use of this shape memory alloys to split up the components and then either reuse some bits and add bits which are obviously going to wear out with time or not useable they might be able to put into scrap metal or something like the case if it scratched or something you wouldn't want to reuse it but you might be able to melt it down and use it again somewhere else.\nSpeaker B: Would we be the company that would break down these metals or would we...\nSpeaker A: We could probably employ a side company or something to do that for us but it would be our responsibility to get that done and to dispose of the products that we made for certain percentage at least.\nSpeaker A: No, not 100% of anything we could do.\nSpeaker B: This sounds like a really great idea.\nSpeaker B: One thing we have to consider is our 100% turnover goal that we have for our financial sector so we'll have to investigate how much that will cost us, cost the company because it sounds very labor intensive we have to hire a number of people and it might be more expensive.\nSpeaker A: Well the fact of the shape memory alloys is that they don't need to be manually constructed like you don't have to individually unscrew all the screws because of this property, the smart material all you need is just the heat so they self-destruct themselves.\nSpeaker A: All right, let's have to investigate the financial implications.\nSpeaker B: I like the environmental approach.\nSpeaker B: We'll have to see if that can meet our financial goals as well.\nSpeaker A: Also, there's some components that will be how it will actually work but I haven't put this plan together yet.\nSpeaker A: I'm sorry, can you, sorry should I go back?\nSpeaker A: This would actually show the circuit diagram although I haven't come back to the final circuit yet.\nSpeaker A: Okay, so I just don't would put all those components in.\nSpeaker B: So those are what will construct the remote?\nSpeaker A: Yeah, it just shows what sort of energy source it could be, a battery, like rechargeable probably.\nSpeaker A: So how the infrared will actually be sent to the city by the chip on television site.\nSpeaker A: All right, great, okay.\nSpeaker A: So now is it F8 again?\nSpeaker A: Okay, all right, thank you very much.\nSpeaker B: And the next presenter will be Tara.\nSpeaker C: Is it F8?\nSpeaker C: Yeah, yeah, yeah.\nSpeaker D: Okay.\nSpeaker C: I'm the user interface designer.\nSpeaker C: The technical functions design of the apparatus is the effect the apparatus you have.\nSpeaker C: In this case, it's the function of remote control, which is just a message to the television site.\nSpeaker C: By taking inspiration from other similar designs, you'll try and come up with an original trend of remote control, which is the elbow international.\nSpeaker C: There are two functional design options, a multifunctional remote control, which can be used for several entertainment devices, and a single function remote control used specifically for the television.\nSpeaker C: So it was a lot of multifunctional and site.\nSpeaker C: I think the function just the television, which is well.\nSpeaker C: Oh, I see.\nSpeaker C: Multifunction controls can be difficult to use as the notes and buttons can be confusing.\nSpeaker C: A single function remote control is simpler to use, but it means you have to have other remote controls for your other entertainment devices.\nSpeaker C: I think that a single function remote control would be preferable because it's easier to use.\nSpeaker C: It'd be more compatible with the range of television sets, making it more internationally available.\nSpeaker C: It would make an original design more attainable as the less functional necessities to include a new design, and it would be more profitable as it would be more simplistic and less functions would have to be included.\nSpeaker C: So it would be cheaper to make.\nSpeaker C: I'm probably more scalable just because it's more compatible with a wider range of devices.\nSpeaker C: Does anyone have any questions?\nSpeaker D: So as far as we know, a single function television remote control is usable internationally.\nSpeaker C: Well, it's just that when we're creating it, we have to make it compatible with different brands, devices.\nSpeaker C: And it would be easier to make it compatible with different brands, different devices rather than all you can do is use it.\nSpeaker C: Right.\nSpeaker B: Does everyone agree with this? Does anyone object and find the multifunctional might be a better way to go?\nSpeaker A: I was just thinking about what Jenna said before about having some hidden controls, having the outer casing.\nSpeaker A: What you said before about being a profitable simplistic design, if having that would complicate it a lot more.\nSpeaker A: The limit of design.\nSpeaker D: I think I agree with the single design thing now because we're trying to do so much that if we're trying to make a unique user friendly, that's also multifunctional.\nSpeaker D: We're going to go over budget.\nSpeaker C: Yeah.\nSpeaker C: And with this, we'll have more women in the budget probably to make a more legitimate design.\nSpeaker C: We'll have more money to go on to the design side of it.\nSpeaker B: Sounds great. All right. Well, are you ready for your presentation?\nSpeaker D: Jenna? Yes, I am.\nSpeaker B: Badness.\nSpeaker B: Except you're not hooked up to the.\nSpeaker D: Completely ready.\nSpeaker C: Great.\nSpeaker D: Oh, I just lost my microphone.\nSpeaker D: All right.\nSpeaker D: Okay, so I'll be discussing the functional requirements of this remote control.\nSpeaker D: And I'll give you a little briefing of what that means exactly.\nSpeaker D: I feel remember from the email we got before our very first kickoff meeting with the coffee machine.\nSpeaker D: The functional requirements of that was to produce hot coffee quickly.\nSpeaker D: So what I'll be talking about now is the equivalent for remote control.\nSpeaker D: So basically what needs and desires her to be fulfilled.\nSpeaker D: I've done some marketing research, a lot of interviews with remote control users and some internet research.\nSpeaker D: I'll show you my findings.\nSpeaker D: Oh, and firstly, I wanted to remind you about our company motto and purpose.\nSpeaker D: So we believe in providing international market with fashionable products.\nSpeaker D: Hence our motto, we put the fashion and electronics.\nSpeaker D: So I think that should be our priority here.\nSpeaker D: And we should also be looking to trends in clothing and interior design, not just in electronic fashion so that it's something that fits in the household.\nSpeaker B: I'm sorry.\nSpeaker B: What was that last thing that you just said?\nSpeaker D: We should be looking towards trends in both clothing and interior design.\nSpeaker D: Any trends that are going on in the public, even media, you know, who's famous, what TV shows are being watched to influence our remote control.\nSpeaker D: Okay, so the findings.\nSpeaker D: 75% of users of remote controls find them ugly, which is quite a significant number.\nSpeaker D: The other 25% didn't specify if they love them or found them, you know, neutral.\nSpeaker D: 80% of users would spend more money when a remote control would look fancy.\nSpeaker D: Current remote controls do not match well the operating behavior of the user.\nSpeaker B: I'm sorry, the 80% of users would spend more money when a remote control would look fancy, you mean that they would spend more money on a fancy looking remote control?\nSpeaker D: They're willing to spend money on a remote control with personality, as opposed to your basic, you know, oval, black, all same size button remote control.\nSpeaker D: So it is something that people care about.\nSpeaker D: It's not ignored in the household.\nSpeaker D: 75% of remote control users said that they zap a lot, zapping meaning they go through channels a lot, you know, thumbmasters.\nSpeaker D: And 50% of users say they only use 10% of the buttons.\nSpeaker D: That very small amount.\nSpeaker D: That was interesting.\nSpeaker B: So it might be very appealing if we have very concise buttons.\nSpeaker B: And another thing with lots of surfing, we probably have to work on something that could be a lot more durable because I find with channel changes that a lot of the numbers get rubbed down if they're printed on the button.\nSpeaker D: And actually to go with that, I'm going to give you some statistics on the relevancy of each of the buttons, how much they're used, and how important the users find them.\nSpeaker D: So the power button, obviously in an hour, is only used once.\nSpeaker D: Hopefully the person is not turning on and off the TV.\nSpeaker D: But the relevance of that button is 9 out of 10.\nSpeaker D: So people want to be able to turn on the TV with the remote control, as opposed to signing up and turning on the television set.\nSpeaker D: Channel selection is used 168 times on average per hour. That's a huge amount. This is the most important button.\nSpeaker D: So obviously when commercials come on, they're changing it.\nSpeaker D: So as you said, we want a durable button that's not going to run down.\nSpeaker D: Relevance of that button, our users found, was 10 out of 10.\nSpeaker D: Ditto for volume selection, it's a 10 out of 10.\nSpeaker D: And it's used on average four times an hour.\nSpeaker D: Not as much as channel selection, but still significant.\nSpeaker D: Audio settings is used on average 0.8 times an hour.\nSpeaker D: And relevance is 2. Screen settings, which means brightness, color, et cetera.\nSpeaker D: 0.5 times an hour. And relevance of 1.5. We're getting to specific statistics here.\nSpeaker D: Teletext. Now I'm not too clear on what that is. I don't know if you can help me flipping pages.\nSpeaker C: It has TV, has like information has information on holidays, news, entertainment.\nSpeaker B: So like a running banner? No, it's a button that you press and then you, like a menu pops up.\nSpeaker C: I haven't used it before, but... And you have page numbers like for the menu and you press the page numbers with your remote change.\nSpeaker A: Okay. Very basic internet. Very basic internet.\nSpeaker C: Okay. But you know... Like tells you the weather and... Yeah, but you have no interaction back with it.\nSpeaker C: You know, like internet content emails. Yeah.\nSpeaker A: Right. You know, information that I'm like television time data is what's on now from that from every time.\nSpeaker D: Yeah. Alright. Well, I guess I'm not with it because I wasn't, but it's being used 14 times an hour.\nSpeaker D: And has a high relevance in 6.5. So it looks like something that we're going to want to do some research on and include on our remote control.\nSpeaker D: Channel settings. 0.01 times an hour. Relevance of three. Channel settings.\nSpeaker C: Probably just tuning in the channels.\nSpeaker C: Are you sorry, changing channels?\nSpeaker C: Chilling in the menu. Sorry, you know, getting YouTube is that you tuning in the channels.\nSpeaker C: Oh, okay.\nSpeaker C: Yeah. Yeah.\nSpeaker D: Yeah. Okay.\nSpeaker D: So it's not used very often, but people still find it relevant.\nSpeaker D: Okay.\nSpeaker D: Biggest frustrations of the people that we interviewed, remote controls are often lost somewhere.\nSpeaker D: So that was already discussed by Poppy, how we can have an alarm system so that people can find it.\nSpeaker D: It takes too much time to learn how to use a remote control.\nSpeaker D: So it should be very user friendly, you know, people know what to do very quickly.\nSpeaker D: Remote controls are bad for RSI.\nSpeaker A: Yeah.\nSpeaker D: That's rather sad.\nSpeaker D: Maybe our designers can look into that.\nSpeaker D: Buttons that don't require, you know, very firm pushing if they respond.\nSpeaker D: But we'll have to also avoid buttons responding to the slightest touch as well.\nSpeaker D: That's a problem.\nSpeaker D: Okay. Did you guys get that one down?\nSpeaker D: Yep.\nSpeaker D: Okay. Here's some ideas for you.\nSpeaker D: A large percentage of the public would pay for voice recognition on the remote controls.\nSpeaker D: So I'll show you some numbers here.\nSpeaker D: So the youngest age group, 15 to 25, 91.2% said that they would pay extra money to have voice recognition included on the remote control.\nSpeaker D: You can see that number decreases a bit.\nSpeaker D: Interestingly enough, 25 to 35 is the lowest amount that we are willing to pay extra.\nSpeaker D: So I guess we're going to have to figure out what age group we're targeting and if voice recognition is something we want to look into.\nSpeaker D: And if we have the budget for it.\nSpeaker D: If we are targeting young adults, it looks like something that would pay off.\nSpeaker D: Seeing as 90% over 90% would pay for it.\nSpeaker B: I agree with if we're targeting young adults, then it would be something we should look into financially and functionally.\nSpeaker B: And especially if we are trying to be trendy, go with fashions, things like that.\nSpeaker B: And if we have a lot of people who are not interested in the same age as from 35 to 65, which show the lower numbers, probably won't be as concerned.\nSpeaker D: So that's a whole other field of research.\nSpeaker D: I don't know if you'd still have a remote or if you're talking to your television and saying change channel.\nSpeaker D: Depending on how many members you have in a household.\nSpeaker B: So it may be too complicated for us.\nSpeaker B: But it's something to keep in mind anyway.\nSpeaker B: And something that might further complicate it is that the TV makes noise itself.\nSpeaker A: And then we have the phone time.\nSpeaker A: Yeah.\nSpeaker A: And so it's panning to me when you're watching television.\nSpeaker A: But yeah.\nSpeaker A: It's very difficult to get specific.\nSpeaker B: Yeah.\nSpeaker B: If we're looking for simplistic design, if we need to decide if that is our intention is a simplistic design.\nSpeaker B: Because if it is, then I think voice activated.\nSpeaker B: It looks like.\nSpeaker B: And that would sort of negate the whole remote control thing because if people can activate the television with their voice and they won't be using it.\nSpeaker B: They won't be talking into remote.\nSpeaker D: I'm sure.\nSpeaker D: Okay.\nSpeaker D: And the last thing here was an LCD screen.\nSpeaker D: So I mean voice recognition might be a little too extreme for us.\nSpeaker D: Not practical.\nSpeaker D: An LCD screen though might be something that you know, you can shift through pages kind of the way this PowerPoint is working.\nSpeaker D: So that you don't have so many buttons to deal with.\nSpeaker B: I don't know what an LCD screen is.\nSpeaker D: Oh sorry.\nSpeaker D: Just a screen like a computer screen.\nSpeaker D: Or like a little bit.\nSpeaker D: Yeah.\nSpeaker D: Like an alarm clock.\nSpeaker D: You'd have an LCD versus just a.\nSpeaker A: I have no idea.\nSpeaker B: Sorry.\nSpeaker D: Oh just like an electronic screen as opposed to just buttons.\nSpeaker D: There'd be like a little.\nSpeaker B: Oh, on the remote.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Like on the top of the cell phone.\nSpeaker D: Yeah.\nSpeaker D: No, I don't know what exactly we put on there. I guess the channel that you're on volume setting.\nSpeaker C: Could it be good if it had the actual program that was on the what was next?\nSpeaker C: But that would probably be like LinkedIn.\nSpeaker A: Yeah.\nSpeaker C: That would be good.\nSpeaker C: Yeah.\nSpeaker A: That would be better.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Access that to you.\nSpeaker C: It might be quite expensive to do that.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker D: Right.\nSpeaker D: I guess that's something we can all take back to our respective research.\nSpeaker D: And finally, whoops.\nSpeaker D: My personal preferences and thoughts.\nSpeaker D: I think our priority really should be unique design.\nSpeaker D: We want something that people want in their home.\nSpeaker D: Every remote control looks the same.\nSpeaker D: So in my opinion, it should be user friendly and unique.\nSpeaker D: So the other stuff might be a little too gagedy for some people.\nSpeaker D: My self voice recognition kind of scares me off.\nSpeaker D: So if we're aiming to make this an international university universally accepted product.\nSpeaker D: And for all, the other thing is age market.\nSpeaker D: I mean, if we wanted to concentrate on 50 to 25 year olds, we could go with the fancy stuff.\nSpeaker D: But if we want to make 50 million and have everyone want this remote control, we should maybe stick to the basics.\nSpeaker C: I wish I could keep in mind that 15 to 25 year olds might not have 25 year olds to stand in a remote control.\nSpeaker C: Like their priorities might not be a fancy remote control, then they're just starting to like me.\nSpeaker C: Right.\nSpeaker D: And we had to keep in mind the reliability of our research.\nSpeaker D: I mean, you know, a 16 year old boy would say, yeah, I'd pay a temporary second next to them.\nSpeaker D: And they realized that's three months allowance.\nSpeaker D: Yeah.\nSpeaker D: So I think the older generations, we should be catering to a bit more early 20s.\nSpeaker C: Yeah.\nSpeaker B: And if one of the largest or most complained about thing is that it takes so long to get to know how to use remote control.\nSpeaker B: I'm sure that something like an LCD screen or remote control would be just furthering that problem.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: All right.\nSpeaker D: Okay.\nSpeaker D: That's it for the market research.\nSpeaker B: Okay.\nSpeaker B: Before we go into more discussion on we want this design to look like I've received some information from the management that will affect some of our decisions.\nSpeaker B: For one thing, because having controls with DVD, VCR, that sort of thing would really complicate the design of the remote control.\nSpeaker B: We've decided not to include them and make it a specific, just a specific television function, which is good as as we've sort of decided that we would like to go with that anyway.\nSpeaker B: For many reasons.\nSpeaker B: So we have that decision sort of made for us.\nSpeaker B: Another thing that might affect other decisions is that the management feels that teletext is outdated because more people are using the internet now.\nSpeaker B: And so we won't concern ourselves with navigating the teletext option.\nSpeaker A: Can I just enter?\nSpeaker A: Yep.\nSpeaker A: Maybe we can go on the PowerPoint?\nSpeaker A: Yeah, I do.\nSpeaker B: I'm looking at it right now.\nSpeaker B: There you go.\nSpeaker B: Thank you.\nSpeaker D: Oh, come back, Spring.\nSpeaker A: With the management suggesting use of the internet rather than teletext or just avoiding both altogether.\nSpeaker B: Well, I mean, we don't have the resources or possibility of using the internet with the remote control, but they were just pretty much saying that the teletext would not be used.\nSpeaker B: Right. And another thing this is for the design, the design of the product is that we want to create more of a sense that people know that this is from our company.\nSpeaker B: So all the remote controls must have our will incorporate our logo and color in some way.\nSpeaker B: Yeah.\nSpeaker B: So perhaps our logo on the bottom or wherever you feel like it would look good.\nSpeaker B: It would have to be the color of our company.\nSpeaker B: But another thing is that we probably would have to have that color and logo decided upon.\nSpeaker B: I'm assuming that we already have one, but for the purposes of this meeting, I wasn't offered a type of logo or color.\nSpeaker B: So if that could be somewhere on the design, so we can be recognizable.\nSpeaker C: Yeah.\nSpeaker B: Yeah.\nSpeaker B: The little are our yellow thing.\nSpeaker B: Okay.\nSpeaker B: Real reaction.\nSpeaker B: Okay.\nSpeaker B: Yes. Those are the changes.\nSpeaker B: So now we need to discuss.\nSpeaker B: And come to a decision on our remote control functions of how this is going to be.\nSpeaker B: I'm just going to look at my notes for a second.\nSpeaker B: We have to decide on a target group and the functions of the remote control.\nSpeaker B: So we already know that it'll just be for the television.\nSpeaker B: It won't have teletext, but we can discuss those other options that you brought up, Genevieve.\nSpeaker A: Okay.\nSpeaker D: So are we going to write off the LCD option?\nSpeaker B: Yeah.\nSpeaker B: Is that how most people feel about that?\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: No teletext and no voice recognition.\nSpeaker C: I think it would be annoying.\nSpeaker C: I don't use teletext that much, but if it was on your team, you'd want to be able to use it.\nSpeaker B: Yeah.\nSpeaker B: But another thing is that if we're reaching an international crowd, I know for one that in North America there is no such thing as teletext, so it would be really superfluous.\nSpeaker B: All right.\nSpeaker B: Yeah.\nSpeaker B: I don't know about other countries besides the UK.\nSpeaker B: Do you know if anywhere else?\nSpeaker B: I don't know.\nSpeaker A: How's it?\nSpeaker A: I think it's more recent.\nSpeaker A: All right.\nSpeaker A: It was a management decision that we have.\nSpeaker A: It was a management decision.\nSpeaker B: So it's pretty much out of our hands at this point.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: So I guess we're looking at something rather simple.\nSpeaker D: Well, I guess just from my findings, it looks like we want to minimize buttons.\nSpeaker A: Okay.\nSpeaker D: And what was the word that you used?\nSpeaker D: Findability is important.\nSpeaker D: I don't know what it is.\nSpeaker D: Yeah.\nSpeaker B: I think we should definitely go ahead with the alarm system idea that you had, because I'm sure that could be inexpensive because we could use the same kind of infrared.\nSpeaker A: Yeah.\nSpeaker B: The same signal.\nSpeaker B: The same signal through that nica just like a little beeping noise.\nSpeaker B: It's not that expensive to do.\nSpeaker A: Just the same as a mobile phone.\nSpeaker A: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker C: I like that idea.\nSpeaker C: I'm going to put the little device in here.\nSpeaker C: As an other not control for all two years.\nSpeaker A: If you're doing the link between the buttons.\nSpeaker A: Yeah.\nSpeaker A: The buttons are actually on.\nSpeaker B: So I know.\nSpeaker B: My computer there.\nSpeaker C: It's like an all-j\nSpeaker A: I'm not sure if I'm going to use the back of any set that would be not very obtrusive. Obviously something small.\nSpeaker A: Yeah.\nSpeaker A: That's good.\nSpeaker B: Yeah.\nSpeaker B: Then it wouldn't it probably wouldn't be able to use it would be able to use the same reception on the remote control, I guess, but the actual device would have to have its own infrared signal.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker B: Would it need a back to then?\nSpeaker B: Maybe.\nNone: Probably.\nSpeaker B: I mean, that's your department.\nSpeaker A: Sort that out.\nSpeaker A: Okay.\nSpeaker A: Unless some way it could have some universal connection to like the socket, the same socket that the TV's supplied from.\nSpeaker A: I mean, the power for the TV.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: You have to invest that one.\nSpeaker B: Do some research on that.\nSpeaker B: All right.\nSpeaker B: Great.\nSpeaker B: All right.\nSpeaker B: And I'm sure that the glow in the dark fluorescent, whatever system.\nSpeaker B: It's a go ahead.\nSpeaker B: Does everyone interested in that?\nSpeaker C: I like the light up suggestion.\nSpeaker C: Yeah.\nSpeaker C: Because you're like white.\nSpeaker C: Right.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker D: Yeah.\nSpeaker D: It doesn't, it could be a tactile thing as well.\nSpeaker D: Right.\nSpeaker D: If we're minimizing buttons, we might be able to make them actually larger and there's something on it.\nSpeaker D: Like a raised up arrow down arrow for volume.\nSpeaker D: And I don't know what we could do for channels.\nSpeaker C: Which is the numbers could be embossed.\nSpeaker C: Could be like raised.\nSpeaker C: The numbers themselves.\nSpeaker C: Yeah.\nSpeaker C: The 90s.\nSpeaker D: Like up button and down button for the channel.\nSpeaker D: Channel changing.\nSpeaker D: Just like a lot.\nSpeaker D: Yeah.\nSpeaker D: It just, it might be sucking more battery power if there, if it is a light up.\nSpeaker D: I'm not sure.\nSpeaker D: Yeah.\nSpeaker A: That's true.\nSpeaker A: And also, uh, having you mentioned before, um, like how it should be accessible to everybody.\nSpeaker A: Yeah.\nSpeaker A: And so like big, um, buttons made before people who are visually impaired, glow in the dark or light up, won't make any difference anyway.\nSpeaker A: So like you said, tactile might be better because it'd be more available to everybody.\nSpeaker B: Could we somehow, we could maybe possibly, sorry, incorporate them both so that.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: The buttons could maybe be in the shape of the numbers themselves and be made out of some sort of glow in the dark material.\nSpeaker B: Yeah.\nSpeaker B: So I don't think that little dark material, um, like the actual soft plastic, um, costs that much more than other colors.\nSpeaker A: Not these days and it's quite.\nSpeaker D: Yeah.\nSpeaker D: I guess the other option referring to the battery thing is, you know how cell phones will light up for 15 seconds.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And then it goes.\nSpeaker D: So if you're like changing the volume during a movie, I'm thinking of most of you watching a movie, turn all the lights off, right?\nSpeaker D: Yeah.\nSpeaker D: You don't turn on the lights to turn it down because of suddenly an explosion.\nSpeaker D: It's going to wake up the baby.\nSpeaker D: So if you touch the button, it kind of reactivates it.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker A: It's a self-tiler.\nSpeaker A: It's a self-tiler.\nSpeaker B: We have five minutes left, um, for the meeting, but I think we should discuss this light subject a little bit more before we close.\nSpeaker B: Um, what was, I missed the last moment reading that.\nSpeaker B: What were you talking about with the lighting up?\nSpeaker D: Oh, just if it was kind of the same way that a cell, you know, a cell phone will light up for 10, 15 seconds when you touch a button after having not touched it for a while.\nSpeaker D: Yeah.\nSpeaker D: Um, if instead of a constant light up on the, on the remote control, if it lights up for 10 seconds when it's touched again.\nSpeaker B: So it could be any button that would be pressed?\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: You just, and it just kind of lights up a bit and it gives a thing glow.\nSpeaker B: So if you have all the lights off in your living room, you'll, you'll temporarily see it because you're getting the glow in the dark thing and the shape of the numbers.\nSpeaker B: You have to kind of decide what we're going to do with this.\nSpeaker A: I think the shape of the numbers is really a good idea.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: And I think that's unique as well.\nSpeaker A: Yeah.\nSpeaker A: I mean, I haven't seen that.\nSpeaker A: And as you're saying, like numbers can wear off if they're just, yeah.\nSpeaker A: Yeah.\nSpeaker B: Maybe we have those.\nSpeaker B: If it's that softer rubber material, it'll be, um, uh, better for people with else.\nSpeaker B: No one's a cold RSI.\nSpeaker B: What was it though?\nSpeaker A: Oh, yeah.\nSpeaker A: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I think it's, it's, okay.\nSpeaker B: Um, did we want to go for the glow in the dark look or did we want to go for the lighting up instantly?\nSpeaker B: Should we do both or we can have one or the other because it might, for, for design purposes, I mean, the lighting up thing might be better because glow in the dark material has a funny kind of color and it might not go with different like face plates that we might come up with.\nSpeaker D: Exactly.\nSpeaker D: It might be perceived as tacky, glow in the dark.\nSpeaker D: Yeah.\nSpeaker D: It's kind of like 80s neon style.\nSpeaker A: Yeah, we could be trending fashion also.\nSpeaker A: Yeah, there are now loads of a huge range of different colours that you could light up in as well, which could glink in with the company colours, blue or green or yellow.\nSpeaker A: Right.\nSpeaker A: And like, we're just limited with the...\nSpeaker A: Just...\nSpeaker B: Alright, so we've just started unlighting up.\nSpeaker C: I was thinking of what was going on with the dark cupid, the real reaction symbol as going on the dark, and then it was a big compliment, right?\nSpeaker C: Avertise?\nSpeaker A: Yeah.\nSpeaker A: And I'm the...is it lit up?\nSpeaker A: That could light up as well.\nSpeaker A: Oh, oh.\nSpeaker B: But with the same thing, I mean, if you touch the button and then it could be lit up as well, is...\nSpeaker B: Are you okay with that?\nSpeaker B: Yeah.\nSpeaker B: Okay, cool.\nSpeaker B: Alright, so I think that completes most of our more practical decisions, and now it's up to designing, and making sure that this can be feasible.\nSpeaker B: And do you have any...\nSpeaker B: Oh, sorry.\nSpeaker B: Do you have anything to say?\nSpeaker D: Yeah, well, I'm just going to throw out there the thought about personalizing the remote control.\nSpeaker D: Because you mentioned faceplates.\nSpeaker D: So I don't know if there's something that...\nSpeaker D: You know, like five different faceplates.\nSpeaker D: I don't know if this would start making it more complicated, but it could increase the popularity of the remote.\nSpeaker A: Okay, like you can have changeable mobile Kevin's iPad or something.\nSpeaker B: Exactly.\nSpeaker B: Or like mobile.\nSpeaker B: Like a cell phone, yeah.\nSpeaker D: And I don't know if you'd want to go with TV show themes or something, like a Bart Simpson faceplate.\nSpeaker D: Yeah.\nSpeaker C: And then that would be more profitable, like as a side line to the remote control.\nSpeaker C: Yeah.\nSpeaker D: So, so you could start out with three, and if we hit it big, then we can add some more.\nSpeaker D: Yeah, last good.\nSpeaker B: Great.\nSpeaker B: Yeah, I think that we should incorporate that into changeable.\nSpeaker B: Because that wouldn't be very expensive at all.\nSpeaker B: You just get one mold.\nSpeaker A: Yeah.\nSpeaker B: Throw some plastic in it, you know.\nSpeaker A: And also possible, I mean, we could gain out of that by advertising certain TV shows.\nSpeaker A: Yeah.\nSpeaker B: That might be problematic with copyright issues.\nSpeaker A: So if we...\nSpeaker A: It takes off, then we'll try that out.\nSpeaker A: The environmental factor, we didn't bring that up again.\nSpeaker B: Right.\nSpeaker B: We'll have to do more research.\nSpeaker B: Like as of yet, that has nothing to do with the way to look.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Because we need to investigate the financial implications.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Does it need to be decided on now?\nSpeaker A: I think we can probably leave that till later.\nSpeaker A: Okay.\nSpeaker B: Good.\nSpeaker B: All right, then.\nSpeaker B: Anyone else have anything more to say before we close?\nSpeaker B: No.\nSpeaker B: All right.\nSpeaker B: Well, let's have lunch and we'll discuss this later.\nSpeaker A: Okay.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: All right.\nNone: All right.\n"}, {"meeting_id": "TS3007b", "summary": "Project Manager firstly gave some main points of the design of remote control. The Internet would be the main focus and the proposal of multi-purpose remote control was clearly rejected. Besides, the customer group expanded to those younger than 40. Then group members respectively gave presentations about the conception of the functions of remote control. Based on a questionnaire, Marketing concluded that the remote control should be simple, the buttons bigger and for minor functions. Industrial Designer rejected the idea of using Bluetooth and indicated that infrared would be better and within the budget. Industrial Designer believed that remote control should be simple and suggested sticking to basic things of remote control. User Interface also agreed to keep it simple and put forward the idea of using a touch screen and being user-friendly. Finally, Project Manager led the team to further discuss detailed questions like menus, parental control, update service.", "dialogue": "Speaker B: Hey guys.\nSpeaker A: Hi.\nSpeaker A: Hi.\nSpeaker A: I see my bunny is still standing.\nSpeaker A: Yeah.\nSpeaker A: All enjoying it.\nSpeaker A: Yeah.\nSpeaker A: I figured I had that much.\nSpeaker A: I'm going to wake up.\nSpeaker B: A minute please, my laptop is there.\nSpeaker B: There it is.\nNone: Thank you.\nSpeaker B: So, welcome back at the functional design meeting.\nSpeaker B: The plan is that each one of you, so not me, but only you, will present the things you worked on the last half hour.\nSpeaker B: I will take minutes and we'll put the minutes that I have at the end of the session in the shared folder.\nSpeaker B: Also, the minutes of the previous session are also in the shared folder now.\nSpeaker B: So, you can read that now or afterwards.\nSpeaker B: I had an email from the management board.\nSpeaker B: I don't know if you all also received it, but there were four points, which I think are very important.\nSpeaker B: The first one is they think that teletext, teletext, becomes outdated and internet will be the main focus.\nSpeaker B: The second one is also important because it's one of the discussion points of the previous session.\nSpeaker B: The remote controls should only be used for the television, so it's not going to be a multi-purpose remote control.\nSpeaker B: So, that's one thing to keep in mind.\nSpeaker B: Second, and I think that's important for the marketing expert.\nSpeaker B: The current customers are in the age group of 40 years and older, but with this new remote, they would like to reach a group younger than 40.\nSpeaker B: And a thing to keep in mind, but not really for now, is that they want the slogan and the logo to be recognizable in the remote.\nSpeaker B: So, we have 40 minutes, so I think not more than 10 minutes per presentation each.\nSpeaker B: And please use all the facilities so that you have the smart boards, the word files, whatever you want.\nSpeaker B: So, Tim, can you start?\nSpeaker D: Okay, yeah.\nSpeaker D: Okay, welcome.\nSpeaker D: I have some new findings on marketing expert level, which I will show you.\nSpeaker D: The method I used was giving orders to our usability lab to do a questionnaire.\nSpeaker D: 100 respondents were involved, and my marketing department generated a report with a lot of results.\nSpeaker D: These were a couple of findings, first page of three.\nSpeaker D: We have three audiences, of two audiences, I'm sorry.\nSpeaker D: The first one, this skill of 16 to 45 years in age.\nSpeaker D: The second one is from 46 to 65.\nSpeaker D: As you can see here, the market share for the first audience is about 60%, 65.\nSpeaker D: The second audience is 35%.\nSpeaker D: And some interest from the age groups.\nSpeaker D: It seems like the young users of remote controls really like the fancy new technology stuff, like an LCD screen on the remote control speech recognition.\nSpeaker D: I don't think that's really appropriate.\nSpeaker D: And when you see the audience, the age is going up.\nSpeaker D: Yeah, they don't really want it anymore.\nSpeaker D: At least the new technology.\nSpeaker D: The second findings out of the questionnaire are the opinions of the audience about current remote controls.\nSpeaker D: First point is 75% of the users find the most remote controls very ugly, and 80% of the users would spend more money when a remote control would look fancy.\nSpeaker D: So that's maybe something for the user interface.\nSpeaker D: Designing.\nSpeaker D: Third findings, according to a frequency of use versus importance investigation.\nSpeaker D: Following buttons are most important.\nSpeaker D: I will tell something about the way this test was done.\nSpeaker D: The persons were asked what the buttons were.\nSpeaker D: They used most, how much an hour.\nSpeaker D: And in the second table, the importance of those buttons.\nSpeaker D: When you multiply them, you get these three points.\nSpeaker D: Switching channels.\nNone: Yeah.\nSpeaker D: That's pretty normal.\nSpeaker D: That's what you do with a remote control.\nSpeaker D: The second, data text.\nSpeaker D: And the third, volume controls.\nSpeaker D: I think it's good that we know what the user wants.\nSpeaker D: At least these three points have to be very clear.\nSpeaker B: But it's strange that the management board said that the teletext will be outdated by the internet.\nSpeaker B: Yeah, okay.\nSpeaker D: At the moment, data text is the best thing you can get on TV.\nSpeaker D: Like getting information.\nSpeaker D: So when you ask people what do they use, they use data text and not the internet or remote control.\nSpeaker D: That's ridiculous.\nSpeaker D: It's a new technology, but it's not incorporated right now.\nSpeaker D: Okay.\nSpeaker D: My personal preferences.\nSpeaker D: I think we should aim at the audience from 16 to 45.\nSpeaker D: First of all, it's the biggest share.\nSpeaker D: The biggest audience.\nSpeaker D: 65%.\nSpeaker D: Second, I think you will get the most revenue from it.\nSpeaker D: Yeah, people from 16 to 45 watch a lot of TV more than people who are elder.\nSpeaker D: Second point.\nSpeaker D: We have to improve the most used functions.\nSpeaker D: As I said here, switching channels, data text and volume controls.\nSpeaker D: Third point that came out of the questionnaire.\nSpeaker D: People used to get lost of the remote control law.\nSpeaker D: So maybe it's an idea for us to design a kind of placeholder.\nSpeaker D: And sign off the TV where you can put the remote control in.\nSpeaker D: That's about it.\nSpeaker A: I think.\nSpeaker A: We mentioned improving functions.\nSpeaker A: What do you mean by that?\nSpeaker D: Not the functions, but it came out that a lot of buttons weren't even used on a remote control.\nSpeaker D: So you can have a remote control full of buttons, hundreds of buttons, but if you don't use them.\nSpeaker A: Oh, focusing more on the use buttons.\nSpeaker D: They have to be on it just to get it done if necessary.\nSpeaker D: But the most used buttons have to be bigger or.\nSpeaker A: Could you use perhaps one button for multiple functions?\nSpeaker A: For example, pressing it in longer makes it switch to a different function.\nSpeaker D: Yeah, perhaps.\nSpeaker D: Just for the minor functions perhaps.\nSpeaker A: Yeah, perhaps.\nSpeaker A: Just not yet.\nSpeaker D: Just to get less buttons on the remote control to make it easier and quicker to learn.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: Okay, that's it.\nSpeaker A: Thank you, Tim.\nNone: Yeah.\nSpeaker A: Right. I'll be explaining a bit about working design about the project. Well, what I did was I dissected current remote controls and I viewed how they worked, what kind of components are involved and how they are connected together. And after that I put up a scheme about how these things are organized and I'll show it to you in a few seconds. And I'll explain a bit about how it works and how we could build one and why I think several possibilities that we discussed in an earlier meeting falls off. Right. Well, what I did was I checked the remote controls and the remote controls of today are all infrared like you probably know. And the thing about that is the remote controls have to access a TV or a stereo or something and those have transmitted that's also focused on infrared. So if we want to build a remote control with Bluetooth for instance, then the TV should have Bluetooth too in order to communicate. So that would mean extra cost for the user and that wouldn't mean a cheap remote control for us. So that's probably why most controls are still infrared.\nSpeaker A: Firmware all have very simple structures that would probably mean lower cost and that could mean for us a good thing because we should be able to build a relatively cheap remote.\nSpeaker A: Well, as I mentioned, radio waves and Bluetooth, well, it may be possible but I figured it wouldn't be possible within our budget but that's not for me to decide but that's maybe something for marketing to look into because my personal opinion is not to do Bluetooth\nSpeaker D: or radio waves although. What do you think about incorporating Bluetooth or radio receiver in the place or next to the TV connected to the TV? Yeah, actually it's in our own product.\nSpeaker A: Yeah, yeah, I actually figured that would be rather nice but then you still have the infrared function so and very you actually just move the problem but what I did think about was when you mentioned about the couple is why not introduce a speech function like where is the remote? If somebody says where is the remote then it goes BPP for something.\nSpeaker A: I don't know maybe something to look into. I don't know what the cost for that something like that would be but it may be maybe something to explore. I'll just explain a bit of the components. First you have the energy source. The energy source would be a battery simple battery that you could find anywhere. I figured that would be best because when the battery stops functioning we could just use it, we could just go out and buy a new one so we don't have to do all to be too complicated about that. The energy source is connected to the infrared button but the infrared button works only via the chip and the sub-component to the switch. There is a switch between these. When the switch is pressed in, on this case the switch is a button. When the button is pushed in, the electric current goes through here and immediately a bulb lights up displaying to the user that something has happened. The user won't be thinking, well, did the button be pressed? What happened? I pressed the button but nothing is happening on the TV so it's something wrong or something. So that's just to explain, make it clearer to the user. While the signal goes via chip that's translated into electronic signals and then it's processed and then it's sent to the infrared bulb where it will be received on the receiving end and those interpreted by the device, well, in this case, television. Well, my personal preference is we have to keep it simple, not too many gadgets and functions. Just like you said, the most users, you have a lot of buttons and you usually don't use them. Why should we spend more time on those? I think we should stick with infrared transmitting and no receiving, so no input from the television. So I think we shouldn't be spending time on teletext and things like that because when you want teletext on infrared, you'd have to build a receiver too. So in order to receive the signals from what's on TV and such, so I figured that would be spending too much money and\nSpeaker D: time and maybe another problem. I think current TVs can even send. Yes, but should we, I agree\nSpeaker A: with you, but should we spend money and time on building a receiver into the remote control because that would be an extra component, extra design, larger remote control. These are all stuff that we have to take in account. So in my personal opinion, there's no receiver at all. Well, we should look into the design and the functionality. Like I said, use one button, for instance, for multiple functions or just hide a few buttons or switching it open or something. You use stuff. And don't overbuild. We shouldn't make big remote control for simple functions, but we should stick to the basics. So that was my personal opinion.\nSpeaker A: And that was my presentation. Okay, thank you, Janice. Yes, I can go ahead.\nSpeaker B: The last presentation, you have plenty of time. Tim and Janice, don't talk for 10 minutes.\nSpeaker B: So take your time. If you take your time too long, I will eventually warn you. Well, I\nSpeaker C: am going to give a presentation about some of the technical functions of the design and usability function. What's my opinion about? What's most important to combine the design technical possibilities and the use of friendliness in one. So if you go into design, remote, that looks good, it shouldn't weigh over the, if it's possible to make, of course, but also use friendliness. So that's some of the main points. And another one is the use of many functions. Well, make it more difficult. So use as little functions as possible, or at least don't display them all at once on the same remote if you have 50 functions. You don't want 50 buttons to be shown at the same time. When you visit an internet, the size, you don't want 50 links to see, but maybe use a hierarchy structure. And one of the ideas was maybe use touchscreen, but I don't know how far it is possible since we have taken to infrared and the remote cannot receive anything. But we might consider that. Well, of course, I hope this is all clear to you. You can use a remote like this with all the functions and many functions, but well, your thumb is a little bigger than this. You have to be very careful what you push. And if you're looking for teletext, you'll be searching for half an hour from, well, where is it? Well, here I guess, when you have to use something else. So just keep simple, make clear buttons easy to use. For example, if you want to use a play and back and stop, that's very important. Well, this was because of our last discussion.\nSpeaker C: If multiple machines are used, create easy switch between the machines, but it's no longer applying. Well, yeah, I prefer to use it only for TV. And not to give too many options.\nSpeaker C: And if possible, the buttons should give a direct action. So first select. You just\nSpeaker B: had, you wanted to combine more functions in one. So you want to keep it simple. But I think that if you want to do that, then you can't escape the fact that there will be buttons which gives more options than one. Yes. But thing you have to wait against yourself\nSpeaker C: do we want to use a few options and might not be so original or multipurpose as we thought. Or do we want to use many buttons? And wing those factors. It's maybe an option if you\nSpeaker D: use an LCD or touch screen that in the middle are the main keys like displayed on the, yeah, yeah, yeah, something like that. Okay, just in the middle the general functions like play channel switching. Yeah. And then at the top or at the bottom some menus like settings\nSpeaker C: or that you can drop down. Yeah, another question I had, do we want to use a menu display on the TV or have to everything be in remote? If you use a memory display on the TV, you can simply push a menu and then select the options you want to have and press OK. So that's my recommendation. If you use many options in one button, display the menu on the TV and use combination of two buttons at the same time or pressing buttons three times or five seconds is too complicated for most users. I think so too. And that's partly\nSpeaker D: because a lot of TVs have different menus and when you have a particular menu at your device, it could be that don't correspond to the menu what's actually on TV. Yeah, that will be a\nSpeaker A: problem. Well, you have to keep in mind that several TVs don't even have a menu structure or they have a very simple menu structure. So you have to keep in mind that not all our remote won't be able to work on all televisions. Yes. And that would be a considerable problem.\nSpeaker C: We have to stick with current technologies and well, the restrictions of what's on the market today. Yeah, you should keep it at this. Use big clear buttons, not too many. So maybe we'll lose a few options, but I think this is more important. Especially the important buttons. If you want to switch channel, change your volume, use teletext. It has to work at once and more advanced options may be put it somewhere away on the remote. It's behind something like a touchscreen or a good menu. Yeah, if you want to put the on standby or change the channel, that should always be possible to do not first change menu options or switch something. Well, yeah, as you already told, give some feedback. If the user is pushing a button, you should know if the television or at least remote is reacting and not just the batteries may be low. And well, my conclusion is less. It's more, keep it simple. So maybe we should just ease down on the functionality to keep it accessible because you all know if there are a lot of function on the television. Some you will never know and never use. And therefore, it's important if you want to change the volume or the channel that is always accessible and easy and other functions that are not so important. Well, we should consider just not using them or at least putting them somewhere on the remote where they're not in a way for\nSpeaker D: the most important function. Yeah. I think the idea about touchscreen is very good. I think it's a good idea because recently I saw a news item on TV about new telephones for elder people. They have like a touchscreen with really big pictures on it, like a call hang up. And that's a big advantage, I think, because on the one end, you make the remote control compatible for elder users just by scaling up the pictures or something. It's very visual intended. What was I to say more? Maybe the snorps, keep the primary buttons\nSpeaker C: visible and make a remote that fits easily in the hand and for some design issues, well, put the logo on it and maybe use it in some aesthetic form. But the important buttons make them always accessible and pushable and clear and maybe use a touchscreen or if that's will become too difficult. Just like some older telephones, maybe it's possible to flip them open and just expand the number of options that are normally visible. Yeah, okay, but\nSpeaker D: if you picked the left idea, then what's going to be displayed on a touchscreen? The extra\nSpeaker C: function. The extra function. Yeah, but like menu from system functions or teletext functions and you just choose one and then all the options will become available and you just go through\nSpeaker A: them. Wouldn't it be better to just make one big touchscreen? Yeah, I want one small touchscreen and I'll just make let's say 15 buttons on it and where three of those are actually just menus with sub menus or sub items or functions. I'd like to make a proposal. If you make one\nSpeaker C: big touchscreen, use the same concept as here. Keep the buttons always available and use the lower part of the touchscreen for the rest. Yeah, yeah, I like the iPod idea that we just saw.\nSpeaker A: You just have a few selected buttons and a few menus and with this idea you could actually make several. You can also improve later on. Yes. I think that would be great. Okay, so you think it\nSpeaker B: will be better to have a kind of total touchscreen. Yeah, I would actually go. Okay, I agree but I\nSpeaker C: think it's very important that always make the same buttons accessible. So use just for special options with part of the touchscreen. Yeah, and so an aloe designer picks up the of an elder parent or grandparent picks up the remote from the little child and who's all in the system functions.\nSpeaker C: You have to have the possibility to turn off the TV or the switch channel without well using all the menu structures to get back to the primary function. Yeah, I had another\nSpeaker D: idea about maybe parental control. Like building in some kind of pink code which allows parents to switch to all channels but children don't know the pink code. They can switch to violent\nSpeaker A: channels. That is possible. Well, that actually depends on the television but I think that would be just a simple login. Well, you see the thing is when you buy remote you set the channels. The channels are different on each television. They aren't certain preset order. So if you lock on a remote let's say channel 15, well channel 15 on this television is different than channel 15 on another thing. So that would be that would be actually the main concern. I think that he means that\nSpeaker C: maybe by some option make sure to remote control on a TV match and then after that you can use some inserts and passwords as being apparent that the children cannot use this. Change the\nSpeaker D: settings of the TV like color and volume and that kind of stuff but maybe if you log in first as a barrel you address the channels and like oh this channel 15 that's a violent channel.\nSpeaker D: My kids I don't want my kids to work. Yeah, they set the priority to only parents.\nSpeaker C: Yeah, for example but make it a separate option in the menu. Yeah, okay but the late\nSpeaker B: former yeah that's just late here so that's another future. Okay, but let's not go too wide about those things. It's a nice idea but I think that's we'll later in the stage. I have one little question about a total touchscreen or a partial. Yeah, a partial because I think elderly people may be not used to a touchscreen so they want the normal functions like tailored sex, volume changing to be kind of traditional and the other functions, the more difficult functions, to be maybe on a touchscreen but to keep this as normal as possible to keep it accessible.\nSpeaker D: But if you display it on LCD screen with really big numbers.\nSpeaker C: Yeah, you can display it on a new style.\nSpeaker A: I do agree because it's just not the same when you touch a touchscreen or when you touch a button.\nSpeaker A: That's different. We have to look at what target audience we are aiming for younger people.\nSpeaker D: Yeah, they chose below 40. Yeah, that's probably and those young people yeah you saw it in my marketing report. They like the new fancy stuff. Yeah, touch green like Microsoft already developed something like that for multimedia applications. I think we can do that too.\nSpeaker B: Yeah, okay as you can see the minutes from the second meeting, this one are done.\nSpeaker B: I've added this four things from the management board just to keep in mind.\nSpeaker B: Each time I had a sort of summary on what you told and what you personal think.\nSpeaker B: So that can be read out. A few things I noticed were...\nSpeaker B: The main points in this meeting is I think how it's going to look.\nSpeaker B: We must keep it simple but have the opportunity to have more options and have them hidden or something. So you don't have a big thing full of buttons or the point that you want to use one controller for, hypothetically, each television. So you must the functions like the menus or the parents or control must be all done by the remote control and not by the television. I think that's the point with... We discussed some of them.\nSpeaker D: The menus are not identical for all TVs so you have to display in a one-a-tv.\nSpeaker C: When you think of the remote and the TV, there's always possibilities to change the color and the brightness and volume. Maybe you can look out if there are options that are remote.\nSpeaker C: In this memory you can see what kind of TV it is.\nSpeaker C: For us, the feedback is in this and that and then give the options that are capable of.\nSpeaker B: That's an opportunity. You have an international market range.\nSpeaker C: So you have big range of personal remotes and they all have functionality for all the TVs.\nSpeaker C: So there wouldn't be an extra feature to incorporate the menus.\nSpeaker A: It's not too complex to do it. Well, they all have to be programmed to fit your TV.\nSpeaker A: And that is a bit of a tricky job. I actually use one of those.\nSpeaker A: They are kind of troublesome. But the thing is, when you start building something like this, you have to build a receiver into the remote. Because in order for the remote to process something from the TV, to synchronize, you have to send and receive.\nSpeaker C: No, no, no. You can just build it.\nSpeaker D: It means just one other thing. With the current remote controls, the universal ones, you have to press a code for a TV.\nSpeaker C: You look up, I have a Philips H55. It says press code 455 and you press code 455 and the remote and it displays all your menu options. Oh, yeah, sure.\nSpeaker D: Now we just connect the TV type to a set of options in just in a memory.\nSpeaker D: So if you like profile, if you touch in like 1410, kind of TV, the memory pops up the options.\nSpeaker D: Yeah, yeah, that would be possible. I don't think that takes a lot of storage space.\nSpeaker C: No, that wouldn't be just for arrival. If you look at the menus from universal remotes, there are maybe three, four hundred TVs at maximum. If you have all of them, all the old and new TVs summed up, so I think it is possible.\nSpeaker D: But on the other hand, five minutes to go, on the other hand, if you have a remote, a buy new TV that isn't incorporated in a remote.\nSpeaker C: Well, then you have to buy new ones, very good from what?\nSpeaker C: New remote, maybe, or an update software update.\nSpeaker D: Firmware, I would write. That's firmware. Ah, maybe.\nSpeaker A: That's maybe the cup holder.\nSpeaker D: Maybe we can incorporate some kind of USB or a firewire connection so that you can connect it to the PC and download newest firmware from the internet.\nSpeaker A: Well, not everybody has a PC at home.\nSpeaker A: Well, most people have, but not everybody.\nSpeaker C: You can go back to the shop and like a kind of service.\nSpeaker C: Yeah, maybe something like that.\nSpeaker A: You can download it for you.\nSpeaker A: Or you could actually look at the place, all you talked about earlier.\nSpeaker A: And you could probably make a connection to an telephone line or Internet connection.\nSpeaker C: Well, all the very digital information is sent to the standard TV connections.\nSpeaker C: You can see what's programmed on the new channels.\nSpeaker C: So maybe we can send that information along with standard TV.\nSpeaker A: Well, then you'd be back to the building and receiving.\nSpeaker A: Oh, yeah.\nSpeaker A: Yeah.\nSpeaker A: If it's actually worth it to build it in, we could actually look into it.\nSpeaker A: But I don't know. It would be bringing more cost for it with it.\nSpeaker A: Yeah.\nSpeaker D: I think it's most cheap or cheapest to just do the updates at the service center.\nSpeaker C: Yeah, that would be really best.\nNone: Yeah.\nNone: At the shop.\nNone: Yeah.\nSpeaker A: Like when you buy a TV, it just, I'll follow.\nSpeaker D: It's not a lot of work.\nSpeaker D: Just one docking station where you put it in, press start, bling, bling, updated.\nSpeaker A: Yeah, that would be best.\nSpeaker A: Yeah.\nSpeaker D: You don't buy a TV every week.\nSpeaker D: No, no.\nSpeaker C: It doesn't.\nSpeaker C: It doesn't.\nSpeaker C: It doesn't mean a while.\nSpeaker C: For which one are we going?\nSpeaker C: My mistake.\nSpeaker A: That's about that one or?\nSpeaker A: Yeah.\nSpeaker A: My vote goes out to the right here.\nSpeaker A: Your vote?\nSpeaker A: My vote.\nSpeaker A: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And your vote?\nSpeaker C: Well, I was doubting about which you want to take, but if convinced me that if you display buttons about the same as they would look on a normal remote or elderly, people will know what to do.\nSpeaker D: And also like a clapping, like if I said, almost not open.\nSpeaker D: Too difficult or not too difficult.\nSpeaker D: Maybe it's easier to break it.\nSpeaker C: Yeah.\nSpeaker D: Break it.\nSpeaker B: That is very sensitive.\nSpeaker B: Yeah.\nSpeaker D: Like my telephone.\nSpeaker D: It's sensitive.\nSpeaker D: Yeah.\nNone: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Well, it's almost at the end.\nSpeaker B: We have now a lunch break.\nSpeaker B: Oh.\nSpeaker A: Me?\nSpeaker B: Yeah.\nSpeaker B: After the lunch break, it's back to individual work.\nSpeaker B: Once again, 30 minutes.\nSpeaker B: I will put my minutes.\nSpeaker B: I have updated them.\nSpeaker B: So they're updated in the shared folder too.\nSpeaker B: 30 minutes.\nSpeaker B: 30 minutes.\nSpeaker B: Oh, minutes.\nSpeaker B: Fader.\nSpeaker B: The specific instructions for the next meeting you will all receive at the email.\nSpeaker B: I don't think I can say much about it.\nSpeaker B: So wait for the email and hopefully you get it done in the 30 minutes.\nSpeaker B: And I will see you after the lunch break in the 30 minutes.\nSpeaker B: One question.\nSpeaker D: Yeah.\nSpeaker D: How late do we have to get back?\nSpeaker B: Be back here.\nSpeaker B: Well, 30 minutes.\nSpeaker D: A quarter to one.\nSpeaker D: Yeah.\nSpeaker D: 30 minutes of lunch break.\nSpeaker B: 30 minutes of lunch break.\nSpeaker B: Okay.\nSpeaker D: About 45.\nSpeaker B: 45?\nSpeaker B: Yeah.\nSpeaker B: Well done.\nSpeaker B: Would it be one o'clock?\nSpeaker B: Yeah.\nSpeaker B: Oh, we ask our personal coach.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Thank you.\nSpeaker B: Thank you.\nSpeaker B: It was very good session I think.\nSpeaker C: We have possible distorted social documents.\nSpeaker B: Uh, well.\nSpeaker D: Save us.\nSpeaker B: Yeah, because all things are stored in Smartboard.xdk and that's in.\nSpeaker D: But you can open it from your, from your laptop.\nSpeaker C: You can save it as an image.\nSpeaker D: Yeah, maybe.\nSpeaker D: Save us.\nSpeaker A: I can move.\nSpeaker A: Maybe an export function.\nSpeaker A: Export.\nSpeaker B: Well, I can.\nSpeaker B: HTML.\nSpeaker C: Well, it usually makes you possible.\nSpeaker C: Huh?\nSpeaker C: Image?\nSpeaker C: Ypeg.\nSpeaker A: Yeah, a tree peg.\nSpeaker D: Ypeg.\nSpeaker D: Yeah, it's better.\nSpeaker D: Very precise.\nSpeaker D: A4.\nSpeaker D: One screen size.\nSpeaker D: And this.\nSpeaker D: Yeah.\nSpeaker D: Directory.\nSpeaker C: No.\nSpeaker C: Oh, yeah.\nSpeaker C: It's not connected to the...\nSpeaker C: You all have to have the...\nSpeaker B: The questionnaire again.\nSpeaker B: About the DFT work.\nSpeaker D: Yeah, it is connected.\nSpeaker D: It's connected.\nSpeaker D: Yeah.\nSpeaker D: I think so.\nSpeaker D: Do you?\nSpeaker B: Oh.\nSpeaker A: No.\nSpeaker A: I'll just save it in my documents.\nSpeaker A: Yeah, am I on messenger?\nSpeaker B: Yeah.\nSpeaker B: That's exactly what you're doing.\nSpeaker B: Yeah.\nSpeaker C: You give the name?\nSpeaker C: Oh.\nSpeaker C: Yes.\nSpeaker C: Okay, thank you.\nSpeaker C: The questionnaire filling with fill out after lunch.\nSpeaker B: Sure.\nSpeaker B: Well, it's simply filling...\nSpeaker B: Oh, no.\nSpeaker B: It's also filling out.\nSpeaker B: I do it after lunch.\nSpeaker B: I'm hungry, so I do it after lunch.\nSpeaker D: Yes.\nSpeaker D: Thank you all.\nSpeaker D: Thank you.\nSpeaker C: You're welcome.\nSpeaker C: You can leave the PC on, I think.\nSpeaker C: Yeah.\nSpeaker B: I'll bring it to my personal room.\nSpeaker B: Yeah.\nSpeaker B: My executive room.\nSpeaker B: My executive room with the painting.\nNone: I The\nSpeaker B: He went there.\nNone: you you you you you you you you you you you you you you you you you you you you you you\n"}, {"meeting_id": "IS1003b", "summary": "This was a functional design meeting. Based on the presentations of the user interface designer, the industrial designer and the marketing expert, the project manager concluded that the new remote control should be powerful, easy to use, fashionable, recognizable, only for television and Internet-connected. After discussing the industrial design of the product, the group reached the agreement to design an infrared-based controller with less buttons but whether the product would be wireless was still unsolved. As for the product appearance, the group decided to make the product look fancy and recognizable; nevertheless, they had different opinions on the colour of the product.", "dialogue": "None: P Resist he's fine\nNone: heh Alright, so...\nNone: Alrighty, sir.\nNone: Alrighty, sir.\nNone: Yeah, I'm a bit ready here.\nSpeaker A: So I will happen now, our functional design meeting.\nSpeaker A: So I will play the role of the secretary as an also program manager.\nSpeaker A: So we will have the three presentations from the industry designer, user interface designer, and...\nSpeaker A: I'm getting experts.\nSpeaker A: After that, we will have the new project requirements, the decision on the remote control functions, and we will close the meetings after.\nSpeaker A: So I suppose the first two-jute presentation will be the user interface designer.\nSpeaker A: Okay.\nSpeaker A: So you're a participant too?\nSpeaker A: No, you...\nSpeaker D: No, I'm a participant.\nSpeaker A: I'm a participant.\nSpeaker A: Okay, never mind.\nSpeaker A: Okay, did you save your presentation?\nSpeaker C: I'm fine.\nSpeaker C: In one, so...\nSpeaker A: Yeah?\nSpeaker A: This?\nSpeaker A: Yeah.\nSpeaker A: Technical functions?\nSpeaker D: I'm not sure.\nSpeaker C: So you didn't save it, maybe?\nSpeaker D: Yeah, it's mine.\nSpeaker D: It's David Jordan.\nSpeaker A: David Jordan?\nSpeaker A: Yeah.\nSpeaker A: So...\nSpeaker A: David Jordan?\nSpeaker D: Yeah.\nSpeaker A: So I'll let David Jordan do his presentation.\nSpeaker A: Okay.\nSpeaker B: No, no.\nSpeaker B: This one doesn't want to be moved.\nSpeaker B: Okay, so...\nSpeaker D: So first I will present the technical function design for user interface for our remote TV control.\nSpeaker D: I will focus on user interface design.\nSpeaker D: So move to the next slide.\nSpeaker D: As we know, removed TV controls has very sophisticated functions.\nSpeaker D: As we show from this picture, there's a lot of functions over, I think, over 12 or 20 functions of our remote TV control.\nSpeaker D: So how can we design a user interface with so many sophisticated functions?\nSpeaker D: Let's move to the next slide.\nSpeaker D: Yeah.\nSpeaker D: So we want to design an elegant, easy to use interface.\nSpeaker D: A very good example is Google.\nSpeaker D: As we know, Google is very successful because it's a powerful function, but it's very easy to use user interface.\nSpeaker D: So move to the next slide.\nSpeaker D: So my job is to design a Google TV control, which I want to have the face ticket functions, while it's very easy to use the interface.\nSpeaker D: So that's the end of the presentation.\nSpeaker A: So you proposed to have the remote control, which will be powerful?\nSpeaker D: It's a face ticket function.\nSpeaker D: So powerful.\nSpeaker A: So many functions and very easy to use.\nSpeaker B: So it's maybe difficult to move to system.\nSpeaker B: The one on the right doesn't look so simple.\nSpeaker A: You mean this one?\nSpeaker A: You have to learn the manual before using this remote control, I suppose.\nSpeaker D: But a very good user interface.\nSpeaker D: It takes the last time for user to learn how to use it.\nSpeaker C: Maybe you have some international standards where if you can use one, it's almost the same.\nSpeaker A: Are you mean for the pictograms or things like that?\nSpeaker C: For example, I don't know, here, escape.\nSpeaker C: You have escape in computers.\nSpeaker C: So if you see escape, you know, that it should be the same.\nSpeaker C: Yeah, it's for the user to know.\nSpeaker C: For example, the power of button, it is something really international.\nSpeaker C: You know that.\nSpeaker A: Yeah, such as maybe go on, go back.\nSpeaker C: The question is how to merge the Google system to the system.\nSpeaker B: That's my job.\nSpeaker D: That's not easy to say.\nSpeaker A: So you will propose something which is so powerful and easy to use.\nSpeaker A: Yeah.\nSpeaker D: The power of button is easy to use.\nSpeaker D: So that's the point.\nSpeaker A: So next I propose the industrial interface to present things.\nSpeaker A: So you're...\nSpeaker C: That's control.\nSpeaker A: Two?\nSpeaker C: Yeah.\nSpeaker E: Okay.\nNone: Okay.\nNone: Wait a minute.\nNone: Bye-bye.\nSpeaker A: So, bye-bye, yeah.\nSpeaker A: Industrial.\nSpeaker A: Design.\nSpeaker C: Oh.\nSpeaker C: Designer.\nSpeaker A: Okay.\nSpeaker C: So we can move to the next slide.\nSpeaker C: As you all know, you know, my job is to design, you know, to give an initial design of the remote control.\nSpeaker C: So basically the remote control will be, you know, infrared control.\nSpeaker C: So the problem is how to relate the remote control device like, for example, this one I'm holding in my hand and the TV.\nSpeaker C: So this one is laser based, but I propose an infrared based...\nSpeaker C: You know, so for me I think that it is better.\nSpeaker C: It's better to control TV.\nSpeaker C: So it will be cheaper and it is less technology than for the cheap price we have.\nSpeaker C: For the cheap price we want.\nSpeaker A: So you mean that infrared control is a cheap technology?\nSpeaker C: I think it's cheaper than laser.\nSpeaker C: Yeah.\nSpeaker A: Okay.\nSpeaker A: You know the requirements for the remote control.\nSpeaker A: Twelve, nearly thirty.\nSpeaker A: The cost.\nSpeaker C: Yeah.\nSpeaker C: I think for the cost you want, for the cost you want.\nSpeaker C: It's better to have...\nSpeaker C: Yeah.\nSpeaker C: So, yeah.\nSpeaker C: You can move to the next slide.\nSpeaker C: So the different, the findings, you know, what you will find inside the system here inside, it is just, you know, a bulb and an infrared bulb.\nSpeaker C: So here, for example, the infrared bulb will be here and the bulb, it will be somewhere inside.\nSpeaker C: That manages.\nSpeaker C: You can go to the next slide.\nSpeaker C: I have some kind of pictures, you know, here.\nSpeaker C: You have the bulb, it is the blue stuff here.\nSpeaker C: And the infrared bulb, it is the right stuff here.\nSpeaker C: And so, and you have the other sub-component that are not really here, so important.\nSpeaker A: What is this?\nSpeaker C: This is in fact some other components that are in between the electrical device.\nSpeaker C: But here, what I wanted to emphasize is just, you know, the bulb and...\nSpeaker C: Okay.\nSpeaker C: So, and the next slide is the last one.\nSpeaker C: So it's...\nSpeaker C: I think it should be why less than the remote control.\nSpeaker C: Because, you know, it's easier to manage and, you know, because if the battery problem can hold, you know, will use the battery or the wire to connect it to you.\nSpeaker C: Or maybe some kind of thing that we can have both, but, you know, I, myself, prefer a wireless.\nSpeaker A: Okay.\nSpeaker C: If you have some question, I didn't answer that.\nSpeaker B: What's the average price of this technology then?\nSpeaker C: The average prices, I think that this technology will cost, you know, around 8 euros or only, you know, the...\nSpeaker D: So what the wireless remote control?\nSpeaker D: There's the wire.\nSpeaker D: Wait for control.\nSpeaker C: Yes.\nSpeaker C: You can...\nSpeaker C: So, you have to decide how you will, you know, put some energy inside.\nSpeaker C: So you can think about wireless or, you know, it's a question that we can write.\nSpeaker C: So we can think that, you know, with a wire or without a wire.\nSpeaker C: We can have both also.\nSpeaker D: Okay.\nSpeaker A: So you think that...\nSpeaker A: Okay, you pulled between the remote control and the TV.\nSpeaker A: One secret idea.\nSpeaker D: Okay.\nSpeaker A: Now, it will be cheaper.\nSpeaker A: I don't know.\nSpeaker A: I just...\nSpeaker A: Maybe making...\nSpeaker B: This is what you would like to ask the user first.\nSpeaker B: I don't think the user would be ready to have a...\nSpeaker D: Yeah.\nSpeaker D: Wireless remote control.\nSpeaker C: Yeah, but...\nSpeaker B: I don't think...\nSpeaker C: Yeah, I don't think it would be a...\nSpeaker C: I always want to have, you know, sometimes I want to have wire because...\nSpeaker A: Yeah, but as industrial designer, do you think that it will be feasible to have a link or to have a link between the remote control and the TV?\nSpeaker A: No.\nSpeaker A: I'm just asking you, do you think it will be cheaper?\nSpeaker C: I don't think it will be.\nSpeaker E: Could you answer me?\nSpeaker E: I have to think about the question, you know.\nSpeaker C: I think it...\nSpeaker C: Okay.\nSpeaker C: You know, you can always, you know...\nSpeaker C: I think with the wire, it's cheaper, but the wireless, it's more practical.\nSpeaker A: Yeah, but I think the wireless problem is more the David Jordan's problem.\nSpeaker A: Okay.\nSpeaker A: So what?\nSpeaker A: The wire?\nSpeaker B: Yeah.\nSpeaker A: I think it's more your problem.\nSpeaker A: Maybe you have just to kind of research or try to ask people if they would like something wireless or not.\nSpeaker A: That's my job.\nSpeaker A: Okay, sorry.\nSpeaker C: No, but it should be an agreement, you know, because even if you can't think of the wireless, it is the job of the...\nSpeaker C: The graphical user design, but you know, if it is a lot cheaper and if you want to reach, for example, some...\nSpeaker C: The money you have if you want to.\nSpeaker C: So it can be good to have a wireless.\nSpeaker C: It is a question.\nSpeaker C: Okay.\nSpeaker A: So just think of the user...\nSpeaker A: The wireless remote problem.\nSpeaker A: Just think of the problem if one of you have wireless remote control and if you lose it, I think it would be easier to have that link between the remote control.\nSpeaker A: That's actually one of the points.\nSpeaker B: Yeah.\nSpeaker B: True.\nSpeaker B: You will see in my presentation then.\nSpeaker A: So I will let you...\nSpeaker A: Yeah.\nSpeaker A: To your presentation.\nSpeaker B: Which is part of the statement form.\nSpeaker B: I was trying to answer all the questions if the user would be happy to have something else.\nSpeaker B: We've made a study.\nSpeaker B: So could you go to next slide?\nSpeaker B: Sorry, for the functional requirements.\nSpeaker B: So that's the standard method for marketing.\nSpeaker B: Okay.\nSpeaker B: We had 100 subjects.\nSpeaker B: We observed their remote control use in the lab.\nSpeaker B: And then after all the experiments, they had to fill up a questionnaire and see what was okay or not for them.\nSpeaker B: Okay.\nSpeaker B: So then it's some statistics about what we have observed and what they answered.\nSpeaker B: Actually 75% of the users find the most remote controls ugly.\nSpeaker B: Okay.\nSpeaker B: So that's...\nSpeaker A: You mean the look is bad for them.\nSpeaker B: So it's your job, David, maybe.\nSpeaker B: Then 80% of users would spend more money.\nSpeaker B: Actually, if the remote controls were not so ugly.\nSpeaker B: So this is where we could have a good market, I guess.\nSpeaker B: If people are ready to pay more.\nSpeaker B: So it's interesting information, I think.\nSpeaker B: And then...\nSpeaker B: Yeah.\nSpeaker B: The remote controls, like the one you've shown previously.\nSpeaker B: It's not so...\nSpeaker B: Doesn't fit their user requirement sometime because it's too many buttons and so on.\nSpeaker B: So we should change this as well.\nSpeaker B: And...\nSpeaker B: Users are actually zapping a lot.\nSpeaker B: So they're using the device intensively.\nSpeaker B: That's something to take into account as well.\nSpeaker B: And...\nSpeaker B: You know, 10% of the buttons are actually used in our remote controls.\nSpeaker B: And...\nSpeaker B: This is one of the main points for me.\nSpeaker B: I'll come back to it later.\nSpeaker B: Okay.\nSpeaker B: Can you go to the next slide?\nSpeaker B: So as you said, remote controls are often lost somewhere in the room.\nSpeaker B: So this is from the experiments we've done.\nSpeaker B: So if we could solve this problem, this would be interesting for the user, I think.\nSpeaker B: It takes too much time to learn how to use the remote control for 35% of the people.\nSpeaker B: And the remote control are really bad for error as eye problems for 26% of the users.\nSpeaker A: What is RSI?\nSpeaker B: RSI is like when you're using the same...\nSpeaker B: Doing the same moment several times, then you get injured.\nSpeaker A: Okay.\nSpeaker B: Okay.\nSpeaker B: So those numbers are less important than the previous one, but still it has to be taken into account.\nSpeaker B: So last slide.\nSpeaker B: So my personal preference is for this problem is we have to meet the user needs.\nSpeaker B: If we can make it look fancy, then we might sell more than our concurrent.\nSpeaker B: Okay.\nSpeaker B: And definitely if you could have less buttons, still maybe the same number of functions, less buttons, this would definitely be a good way of selling more.\nSpeaker B: Okay.\nSpeaker B: So...\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: And just should you have an idea?\nSpeaker A: Do you think you as the user interface designer, it would be possible to have less buttons and still have the same functionality and to have powerful remote control?\nSpeaker A: Do you think it's possible?\nSpeaker A: Yeah, I think you're possible.\nSpeaker D: Because you can mix several functions in one button.\nSpeaker D: Yeah.\nSpeaker D: So that you have less buttons.\nSpeaker A: Yeah, but do you think it will be easy to use?\nSpeaker A: Because if you have many functions, just for one button, it would be quite difficult for the user to know.\nSpeaker B: Yeah.\nSpeaker B: Remember that the user is not happy to read the manual.\nSpeaker C: Yeah, I think it's...\nSpeaker D: You can have a switch menu, so you can...\nSpeaker A: You have to be intuitive?\nSpeaker D: Yeah, I think so.\nSpeaker D: For example, you can category the function into several classes.\nSpeaker D: Then you can have a switch menu.\nSpeaker D: Yeah.\nSpeaker D: You can put the switch menu to... it tends to this kind of... this category of functions.\nSpeaker D: Okay.\nSpeaker D: And then you put the switch button, then it... switch to another category of functions.\nSpeaker D: Yeah.\nSpeaker D: For example, if you have a remote control, you can control your TV.\nSpeaker D: And also you can control your recorder.\nSpeaker D: So there's different functions.\nSpeaker D: But if there's a button, you can switch between control TV and control recorder.\nSpeaker D: So you can have less buttons.\nSpeaker D: Yeah, but...\nSpeaker D: But I'm not sure it's about how to make the remote control look fancy.\nSpeaker D: Look fancy.\nSpeaker C: Not fancy.\nSpeaker C: The questions that should be asked to the...\nSpeaker D: Because if people have different opinion about fancy, you know...\nSpeaker C: If you ask the people, maybe the market...\nSpeaker C: Because...\nSpeaker D: Yeah, this is something...\nSpeaker D: Maybe a colorful, it's fancy for some people, but maybe simple and...\nSpeaker D: This was the first step, the...\nSpeaker D: Colorful, it's fancy for some other people.\nSpeaker C: So this was the solution is to have many colors of, you know, instead of having one grade.\nSpeaker A: Yeah, but I think it will increase the price of the...\nSpeaker A: Production, a remote control.\nSpeaker C: Yeah, maybe.\nSpeaker A: If you need to have special colors for remote controls, it will customize color.\nSpeaker A: Yeah, it will cost a little bit more.\nSpeaker D: Because maybe some people prefer a red, remote control, some people prefer black.\nSpeaker B: Yeah, but this is what we would ask to the users.\nSpeaker B: And also...\nSpeaker D: We can have several options, so users can select which color they prefer.\nSpeaker A: Yeah, but as soon as you speak about options, it means that the price increases.\nSpeaker A: And we don't really want the price to be too high, because we won't be able to produce it.\nSpeaker A: So we want something fancy, as previously said, florant.\nSpeaker A: Something very easy to use, powerful.\nSpeaker A: And also, as it's written here, 75% of users, they zap a lot.\nSpeaker A: So maybe just having many functions in one button is not that good, if you want to zap a lot.\nSpeaker A: They don't really want to zap between functions of their remote control.\nSpeaker A: They want to zap between channels on the TV.\nSpeaker A: So I think you have quite a lot of points to think about and discuss with the other members.\nSpeaker A: So is it okay for your presentation? Nothing else to add?\nSpeaker B: Yeah, it's done.\nSpeaker B: If we could remember, like, not too many buttons and make it look fancy, I think it would make it.\nSpeaker A: Okay.\nSpeaker A: So I had some new information about the project requirements.\nSpeaker A: So you will have to take care of it in your thinking, your designing of the remote control.\nSpeaker A: So the first one is that for the designing of the remote control, we don't really want to use the teletext anymore.\nSpeaker A: As it's something that is not.\nSpeaker A: Yeah, because now everybody has internet at home.\nSpeaker A: So it's better to use internet than did it text.\nSpeaker D: Yeah.\nSpeaker A: So you will have to think about it in the design of your remote control.\nSpeaker A: And also the remote control will only be used for television.\nSpeaker A: So for your designing, you won't be...\nSpeaker A: You won't have buttons to just manipulate, yet to control the recorder or maybe the garage door or things like that.\nSpeaker A: It's because if we want to do a remote control, what will be used for the television, for the recorder, for the camcorder and all the others, it will become too complex and it will increase the price and we can't do that.\nSpeaker A: We really want to focus on the remote control for the television.\nSpeaker A: Okay.\nSpeaker D: It's okay.\nSpeaker D: But that's the balance between the console and the cost.\nSpeaker A: It will be easier for you to design it to have very powerful and easy.\nSpeaker A: And also want the image of the real reaction to be recognizable in the product, such as the color and the slugging.\nSpeaker A: I mean, that's if you want to buy a remote control and if you go in a shop and if you see the remote control, you will recognize directly that it's our product.\nSpeaker A: So you will have to use the color of the product of them...\nSpeaker A: of the real reaction and also...\nSpeaker B: So it has to be yellow, yellow...\nSpeaker A: As we say, we put fashion in electronics, so it has to be a fashion remote control.\nSpeaker A: Fancy fashion, powerful, easy to use, quite a lot of requirements.\nSpeaker C: For cheap remote control, I'm cheap.\nSpeaker D: Yeah. A low cost.\nSpeaker A: Yeah.\nSpeaker A: But that's your job to find something...\nSpeaker C: What I'm thinking is how you will connect the TV to Internet.\nSpeaker B: I don't think it's part of the remote control design.\nSpeaker B: Maybe not, but maybe...\nSpeaker B: It will be more in browsing.\nSpeaker C: But this means that your TV will be able to connect to Internet, you know, save the wire.\nSpeaker D: Yeah.\nSpeaker D: There's a set box in the TV.\nSpeaker D: So with a set box, you can connect Internet with your TV.\nSpeaker C: Okay.\nSpeaker D: But it's not so popular now.\nSpeaker C: I don't think it's popular, so that's the problem.\nSpeaker C: If you are designing a remote control for the global usage, so if people don't have the technology, so...\nSpeaker A: So...\nSpeaker A: So everybody's okay with the new requirements?\nSpeaker D: For the color?\nSpeaker D: What do you think?\nSpeaker A: I think it has to be yellow.\nSpeaker C: Yellow?\nSpeaker A: Yellow?\nSpeaker C: Do you think that people like the color yellow?\nSpeaker D: The blue color?\nSpeaker A: Maybe you can change the color.\nSpeaker A: Okay.\nSpeaker A: But the image of the society has to be recognized.\nSpeaker A: Okay.\nSpeaker C: I think if you have, you know, like a yellow urban here, it has to be fashion.\nSpeaker A: Yeah.\nSpeaker A: So you have to...\nSpeaker C: It doesn't need to be completely yellow, but just...\nSpeaker A: Yeah.\nSpeaker A: It has to be a fashion and really be the image of the society.\nSpeaker A: Okay.\nSpeaker A: And just when you enter, I say, oh, oh, it's a real reaction.\nSpeaker A: I think it's a real reaction.\nSpeaker A: Okay.\nSpeaker A: Remote control.\nSpeaker A: And also it has to be attractive, of course, because if you want to sell the remote control...\nSpeaker A: Okay.\nSpeaker A: It's okay.\nSpeaker B: Regarding the first line, what?\nSpeaker B: So we're just expecting no more use of the teletext.\nSpeaker B: Yeah.\nSpeaker B: And that's it.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker C: I don't know what why no use of the teletext.\nSpeaker B: It's already changed for...\nSpeaker A: I think it would be simpler.\nSpeaker A: Yeah.\nNone: Yeah.\nSpeaker A: Okay.\nSpeaker A: So...\nSpeaker A: I will close the meeting.\nSpeaker A: Just after that, we'll have lunch break.\nSpeaker A: And you will have in 30 minutes of individual work.\nSpeaker A: You will have, as the idea, I don't remember, what is it?\nSpeaker A: In the shared designer.\nSpeaker A: In the shared designer.\nSpeaker A: Yeah.\nSpeaker A: To put...\nSpeaker A: Compoundant.\nSpeaker A: Yeah.\nSpeaker A: And the user interface designer to work on the user interface concept.\nSpeaker D: Okay.\nSpeaker A: And through home to work.\nSpeaker A: Work on a subject.\nSpeaker E: Yep.\nSpeaker A: And so specific instructions will be sent to you by your personal coach.\nSpeaker A: Okay.\nNone: It's okay.\nSpeaker A: Yeah.\nSpeaker A: I think the session is closed.\nSpeaker A: Okay.\nNone: Thank you.\nSpeaker A: And by the way, Mr. David Jordan, please record your presentations in your own folder.\nSpeaker A: Okay.\nSpeaker A: Not in mind.\nSpeaker C: It should be better.\nSpeaker C: Okay.\nSpeaker C: That's good.\nSpeaker A: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nNone: That was a funny one.\nNone: Check this out.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Now that's gonna be a bit...\nNone: It's very good.\n"}, {"meeting_id": "TS3011b", "summary": "This is the second meeting of the design group. Project Manager introduced new project requirements at first. The management required a remote control only for television and aimed at customers under 40. In terms of user interface design, User Interface focused on user-friendliness but still thought multi-function should be considered. Industrial Designer agreed and proposed to substitute voice recognition for the ten digits. He also suggested that they should use infrared so that the remote control could be connected with most TV sets. After that, in order to solve the problem of energy source, the team decided to include a cradle so that the remote control could be recharged.", "dialogue": "Speaker B: Welcome back. Hello. Let me see. There's one of mine.\nSpeaker B: The waiting is for Sebastian. There he is.\nSpeaker D: We have a flight problem. I opened the CD-ROM box.\nSpeaker D: You can't. Sorry.\nSpeaker D: During my work I have no time either.\nSpeaker B: Well, this is live. Sorry, Edel.\nSpeaker D: Yeah, I opened the CD-ROM box.\nSpeaker B: Okay. People, welcome back with the second meeting.\nSpeaker B: For now, on the schedule are a few points.\nSpeaker B: First of all, the opening, which we are doing now.\nSpeaker B: Second, I received some new project requirements.\nSpeaker B: I'm not sure if you received them as well, but I will tell you about it.\nSpeaker B: Then the three of you prepared a presentation, I think.\nSpeaker B: Sebastian? I think so too.\nSpeaker C: Ruh? Yeah.\nSpeaker C: Ruh? Almost.\nSpeaker B: Almost. Okay. Then we'll do your one last.\nSpeaker B: Top goal of this.\nSpeaker D: Your presentation from here, I believe.\nSpeaker B: We will figure that out.\nSpeaker C: If as well as I can just... I don't think it's why.\nSpeaker B: It is.\nSpeaker B: The top goal of this meeting is to reach a decision on the product, on the target group, and the functions of the remote control.\nSpeaker B: Keep that in mind.\nSpeaker B: We have 40 minutes.\nSpeaker B: The new project requirements.\nSpeaker B: First of all, we didn't speak about it, but we should not support data text in the remote, because our board feels that data text is out of date, and Internet is replacing data text.\nSpeaker B: So we are not even going to try to implement it in our product.\nSpeaker B: It's a board decision.\nSpeaker B: Remote control should only be used for television, because it's not feasible.\nSpeaker B: We cannot make it because of the time to market that we have to deal with.\nSpeaker B: The third requirement is that we should focus on customers that are younger than 40, which is important for you, Ruh, and as well for Ruh, because the product should be interesting and should be bought.\nSpeaker B: It should be bought by people younger than 40.\nSpeaker B: Then for Ruh, as well, important.\nSpeaker B: The corporate image should be recognizable in our product.\nSpeaker B: So the colors and the slogan do have to be in the product.\nSpeaker B: Is that clear?\nSpeaker B: Any questions on these requirements?\nSpeaker B: No?\nSpeaker B: The individual presentations.\nSpeaker B: Ruh, or Sebastian, who would you like to start?\nSpeaker D: Yeah, I'll start.\nSpeaker D: Jeez, this one.\nSpeaker D: Well, we discussed already in the previous discussion.\nSpeaker D: The function of remote control is basically to send messages to the television set.\nSpeaker D: The main important thing, what a remote control should do.\nSpeaker D: I found two different kind of remote controls.\nSpeaker D: The multi-function remote control with many possibilities, but the lack of the feeling I already mentioned in the previous discussion.\nSpeaker D: And the ease of use remote control with less possibilities, but a great feeling in touching the buttons.\nSpeaker D: Yeah.\nSpeaker D: My personal preference is the ease of use remote control, because the user-friendliness can be more trendy in user design.\nSpeaker D: But your new goal was for people less than 40 years old.\nSpeaker D: So maybe the multi-function can be implemented in our design.\nSpeaker D: But it should be a combination, but teletext buttons are not in our design.\nSpeaker D: So it should take out eight buttons or so.\nSpeaker D: But in my opinion, the ease to use RC is the best possibility for us.\nSpeaker B: Okay.\nSpeaker B: Rude, did you get that?\nSpeaker C: Yeah, most.\nSpeaker B: Okay, so the important thing here is-\nSpeaker D: And it's also indeed a good insight in the topic.\nSpeaker B: In the market.\nSpeaker D: Yeah. What does the market want? I don't know.\nSpeaker B: Yeah, okay.\nSpeaker D: Just for user-friendliness, I should choose for the ease of use remote control.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker D: Question.\nSpeaker C: Okay.\nSpeaker D: Excuse me.\nSpeaker D: Cool.\nSpeaker D: And still the right thing.\nSpeaker D: Okay.\nSpeaker D: Well, there are some changes in the design requirements.\nSpeaker D: So there are some changes in the method also.\nSpeaker D: Basically, all this device has to do is send messages to a TV set.\nSpeaker D: And the most easy way to do this is by sending pulses of infrared light to a TV set.\nSpeaker D: Well, I've tried to implement a picture here, but it's hardly readable.\nSpeaker D: Can you see it?\nSpeaker D: No, it's not visible.\nSpeaker D: Well, there's an energy source here.\nSpeaker D: And basically, it's connected to three things.\nSpeaker D: A user interface connected to a chip, which is connected to the sender, which generates messages using infrared light, which are sent to the receiver.\nSpeaker D: That's basically the idea.\nSpeaker D: And there's a little picture just for your imagination how a device like this should look or can look.\nSpeaker D: Okay.\nSpeaker D: What have I found?\nSpeaker D: Usually, these kind of things consist of battery, infrared, diode, buttons, chips, and circuit boards.\nSpeaker D: That's all.\nSpeaker D: Let's case together.\nSpeaker D: Nothing more than that.\nSpeaker D: It's almost every piece of equipment.\nSpeaker D: Every TV set is controlled infrared.\nSpeaker D: There are some exceptions, but most of all have infrared controls.\nSpeaker D: And the more luxury remote controls have little buttons.\nSpeaker D: And I think that's a little bit more fancy also.\nSpeaker D: So maybe we should consider that.\nSpeaker D: Okay.\nSpeaker D: I have a basic scheme of the things which are implemented.\nSpeaker D: Basically, this is all there is.\nSpeaker D: There are a few buttons connected.\nSpeaker D: The buttons are lit.\nSpeaker D: And the whole thing is transmitted by infrared light diode.\nSpeaker D: And there's not a power source here.\nSpeaker D: So that's basically the total design of this piece of equipment.\nSpeaker D: It's fairly easy.\nSpeaker D: It's fairly easy.\nSpeaker D: It's been done many times before.\nSpeaker D: I think we should succeed in our plan to do this.\nSpeaker D: Okay.\nSpeaker D: Personally, I think we should use infrared because otherwise our device cannot communicate with almost every TV set.\nSpeaker D: I think that should be clear.\nSpeaker D: Another important point is we really should use energy sparse friendly components.\nSpeaker D: There are specially designed energy friendly components which consume far less power than conventional components.\nSpeaker D: And if we want to use rechargeable design or energy safe design, we should really implement them.\nSpeaker D: For cost effectiveness, we should really use a very low cost circuit board because most of the production costs are in this part of the equipment.\nSpeaker D: Okay.\nSpeaker D: And the money we save when using this, we can use for elaborate fancy lighting techniques, blinking lights and all that kind of fancy stuff.\nSpeaker D: I think our users will really like that.\nSpeaker D: And if we use less, does it use much more energy?\nSpeaker D: No, the power friendly less also.\nSpeaker D: So we can use them.\nSpeaker D: So that's no problem.\nSpeaker D: For the same costs, they're a little bit more expensive, but by making low cost circuit boards, we can make it, I think.\nSpeaker C: You could only let the buttons that are used most.\nSpeaker C: So the channel switching.\nSpeaker B: Yeah.\nSpeaker B: But the question is whether two or four buttons makes a difference in the costs.\nSpeaker B: If you already plan to include fancy lighting techniques, I guess.\nSpeaker D: I think it's the same as in cell phone, just one light in the device that shines on all the buttons.\nSpeaker D: But one light, there are more lights in there.\nSpeaker B: Yeah, but not for each button.\nSpeaker D: No, no, no, no.\nSpeaker D: That's right.\nSpeaker D: Okay.\nSpeaker D: So, well, this should be it.\nSpeaker D: Okay.\nSpeaker A: Have a think about it.\nSpeaker C: Yeah.\nSpeaker C: Okay, dude.\nSpeaker C: Mine is already outdated.\nSpeaker B: Okay.\nSpeaker B: Well, we are very curious to what the market will feel about such a product.\nSpeaker B: Any income is welcome in PUT.\nSpeaker C: Well, then I'm sure the target market is 10 million units, of which we should sell about 40% to make the 5 million.\nSpeaker D: Could you just have a little bit more to the right?\nSpeaker C: Yeah, since the other part is 40% of the market, I thought earlier we might want to constrain that portion of the market.\nSpeaker C: But since the requirements changed, that's not a good idea.\nSpeaker C: Well, skip this.\nSpeaker C: But this is the truth, of course, that they only use temps and buttons.\nSpeaker C: The buttons to zap are used about 1500 times when the power button is only used one time.\nSpeaker C: And the volume button is only four times.\nSpeaker C: So they're obviously the most important button.\nSpeaker C: And lots of people complain that they can't find their remote control.\nSpeaker C: So we might want to build in a feature to support them.\nSpeaker C: Some audio signal like home phones.\nSpeaker B: Well, that's interesting.\nSpeaker B: Finding function.\nSpeaker B: Yeah, that's definitely interesting.\nSpeaker B: It separates our product from others as well.\nSpeaker B: Okay, go on.\nSpeaker C: Well, I just said that, and this is obvious, and he also said it.\nSpeaker B: Okay.\nSpeaker B: So that's what the market tells us.\nSpeaker C: That's about it.\nSpeaker B: Yeah.\nSpeaker D: Okay.\nSpeaker D: The volume and the zap buttons are mostly used.\nSpeaker C: The zap buttons most, they are used a lot, but more than the other buttons.\nSpeaker D: Just have a perfect design for those, only those buttons.\nSpeaker D: Yes.\nSpeaker D: First, we should focus on that, I guess.\nSpeaker D: Yeah.\nSpeaker D: Well, the technical, I think there are no technical difficulties.\nSpeaker D: There's no elaborate technique used in this kind of equipment.\nSpeaker D: So I don't think we have any hiccups there.\nSpeaker D: So we can fully concentrate on developing a product that is really what the market needs.\nSpeaker D: So maybe it's a good idea to think about these buttons.\nSpeaker D: Yeah.\nSpeaker D: And the sound signal.\nSpeaker D: And the sound signal.\nSpeaker D: That's one thing I'm just wondering.\nSpeaker D: The sound signal from where do you execute the sound?\nSpeaker D: Yes.\nSpeaker B: The sound.\nSpeaker B: None of the devices.\nSpeaker B: That's the reason.\nSpeaker B: Well, maybe you should be.\nSpeaker B: Maybe like clapping in your hands, like turning on and off the light.\nSpeaker D: Maybe there are some devices who incorporate this technique already.\nSpeaker D: There are video sets and there's a special button.\nSpeaker D: Find the remote control button.\nSpeaker D: You press it and your TV set starts to make a kind of weird sound.\nSpeaker D: And your remote controls then start beating.\nSpeaker B: Report.\nSpeaker D: And just like the phones.\nSpeaker D: Yes.\nSpeaker D: Same thing.\nSpeaker D: But the fees don't have all.\nSpeaker D: No.\nSpeaker D: So we must shoot you something else.\nSpeaker D: Yeah.\nSpeaker B: Because we do not have a home.\nSpeaker D: We do not control the TV set.\nSpeaker C: So even if the TV set would have such a button, you would have to walk to your TV.\nSpeaker D: And we want it to make sure it's as easy as possible for our customers.\nSpeaker D: Okay.\nSpeaker B: So what about the clapping technique?\nSpeaker D: I'm convinced Sebastian will find one solution for us.\nSpeaker D: It's quite complicated.\nSpeaker D: Because how can you separate the clapping sound from other sounds?\nSpeaker B: Well, you see it a lot in lightning.\nSpeaker D: Well, basically the characteristics, the characteristics of clapping is just an increase in the volume, the amplitude, the sound, which is generated a few times within a certain period of time.\nSpeaker D: But there are many other sounds, which are exactly the same from the point of view from a recording movie.\nSpeaker C: It would constantly beep.\nSpeaker C: Yeah, that's true.\nSpeaker D: So we don't want to have just a home station, a base station next to the TV.\nSpeaker D: Just a little antenna or something, something like that.\nSpeaker D: Well, if you love something more easily.\nSpeaker D: I don't think people would bother walking to the TV and press that button.\nSpeaker D: I don't think.\nSpeaker D: Because they lost their remote.\nSpeaker D: The most important thing for people is that there's a central point to which they can go and perform some kind of humbling.\nSpeaker D: And then the remote control reports itself.\nSpeaker D: So we should use something like that.\nSpeaker D: You do not want another device, which can be everywhere now, which you have to find first before you can find your remote control.\nSpeaker D: Just a base station next to the TV or something like that.\nSpeaker D: But that will be very costly, I think.\nSpeaker B: So that's not good.\nSpeaker B: Maybe we do agree on the thing that the remote should have such a function if it's possible within the costs and all that kind of things.\nSpeaker B: Maybe Sebastian should have a detailed look later on and come up with a solution.\nSpeaker B: Because that's his field of expertise.\nSpeaker D: Yes, but before I do it, I want to warn you that by implementing this kind of a function, the technical design will become more complicated.\nSpeaker D: And it will become more costly also.\nSpeaker D: Because there will be additional components which will be used.\nSpeaker D: And there are some people even prefer the sound above the lathes.\nSpeaker D: I think so, because when you have a newspaper over your remote control item, you cannot see it.\nSpeaker D: Yeah, it's a distinction in the market.\nSpeaker B: It's a different model.\nSpeaker B: Exactly.\nSpeaker B: It's in a unique feature.\nSpeaker B: And I think it's worth looking and probably more interesting than the light.\nSpeaker D: Just about the user interface, I come up with an easy remote control and advanced remote control.\nSpeaker D: What should we choose in design?\nSpeaker B: Yeah, well, according to Root, people do not use all the extra features very often.\nSpeaker C: I have the numbers right here.\nSpeaker C: Okay, well, we won't support teletext, which was the third most, the second most used function.\nSpeaker B: Okay, well, we do have a wise board, so I'm not questioning that.\nSpeaker C: Well, channel selection is obviously the most important.\nSpeaker B: Yeah.\nSpeaker B: Okay, so on the relevant scale, the channel selection, the volume selection and the teletext.\nSpeaker D: So just the basic functions.\nSpeaker D: And we don't have to use it for DVD players and all kinds of stuff.\nSpeaker D: So that's out of the question.\nSpeaker D: Yeah.\nSpeaker D: Just the easy.\nSpeaker B: I think we should go for the easy.\nSpeaker D: We can make a nice design when there are not much buttons in it.\nSpeaker D: We should save costs by not implementing a lot of functions and the money that we can save from that.\nSpeaker D: We should use for a recognized design and thinking about the user interface.\nSpeaker B: Okay, Root, how do you feel about that?\nSpeaker B: Do you agree that you think the market would respond to a simple?\nSpeaker C: Well, if we include other innovative functions, then they might.\nSpeaker C: And then we can use young people like new features.\nSpeaker D: Yeah, that's right.\nSpeaker B: Okay, so this is kind of a...\nSpeaker D: Just a few buttons, friendly design, nice, lightning effects.\nSpeaker A: Okay.\nSpeaker D: Well, maybe there's not a possibility.\nSpeaker D: You can make it look like an easy piece of equipment, but it's quite elaborated because it has many functions.\nSpeaker D: Yeah.\nSpeaker D: It's more important to make it look like a very friendly, easy to catch piece of equipment.\nSpeaker D: But are we not in the manual?\nSpeaker B: I mean, if you have few buttons, no display.\nSpeaker B: I mean, I guess it's working with shift functions and one button can...\nSpeaker D: Oh, like that.\nSpeaker D: I was more thinking about the more elaborate way of controlling by these kinds of sticks or something like that.\nSpeaker D: I don't know if it's user friendly, that's your field of expertise.\nSpeaker B: Yeah, but how does the remote report give back to the user in what state it is because we do not implement usage.\nSpeaker D: But there's no way to do that because we cannot implement that kind of the system.\nSpeaker B: I know, but if we use like a stick, for example, well, maybe we can use a light for that.\nSpeaker D: When you move the stick to a position, maybe a light next to it can lit up.\nSpeaker D: So you know, I've just pushed the button or changed the channel or tuned up the volume, something like that.\nSpeaker C: And if you use that stick for volume control and channel selection, you would add them to most important functions.\nSpeaker C: Yeah.\nSpeaker B: And we could have other buttons for the advanced functions.\nSpeaker D: Instead of the stick, like many cell phones, just around...\nSpeaker D: Yeah, draw them on board.\nSpeaker D: It's already...\nSpeaker D: Oh, we have a blank.\nSpeaker A: Oh.\nSpeaker D: Oh, yeah, something like that.\nSpeaker D: It's just an easy way of a round button which can be pushed in four directions.\nSpeaker D: Instead of a stick, a stick is vulnerable when it's running down.\nSpeaker D: It can break down.\nSpeaker D: Just around the button should be the trick, I think.\nSpeaker D: And there's also looks more fancy, I think.\nSpeaker D: I think it will attract more...\nSpeaker D: I'm a public, I think, but you're remarketing that.\nSpeaker C: And it's also quite easy to use, so it'll attract younger people because they're new.\nSpeaker C: And it might attract older people because it's easy to use.\nSpeaker C: Geez.\nSpeaker D: Well, volume and...\nSpeaker D: Something like that.\nSpeaker D: So, I'm up and down.\nSpeaker D: Okay, yeah.\nSpeaker D: The pen doesn't really do that's what I want.\nSpeaker B: Okay, that's good.\nSpeaker B: Okay, but we're still in the question of putting in advanced options.\nSpeaker B: So, the market is interested in some advanced new techniques.\nSpeaker B: However, keeping it simple is important for the costs and all those kind of things.\nSpeaker B: We need to find a balance between the advanced techniques and the user friendliness.\nSpeaker D: I think our next step to look at is just that.\nSpeaker D: So, I don't think it's...\nSpeaker D: We have something to do, we can discuss it right now because no one of us has the information to discuss about that.\nSpeaker B: Well, I think...\nSpeaker B: We do have to decide this meeting.\nSpeaker B: We need to have the user functions, the functions, decided, and our target audience.\nSpeaker D: Okay, the teletext is just...\nSpeaker D: Do you want a list of functions? Do you want an explicit list which incorporates all functions available on a device?\nSpeaker B: Well, I do not want a full...\nSpeaker B: It's not necessary to have a full list, but I want the kind of functions.\nSpeaker B: For example, the most important are volume and program, selecting...\nSpeaker B: The 10 digits?\nSpeaker B: Well, for example, that's what we should think about.\nSpeaker B: How is the remote going to look? Not in the user interface.\nSpeaker D: The program to program button, it switches to channels.\nSpeaker D: When you have something on channel 4 and something on channel 6, just one button which can...\nSpeaker D: A bit of a split note, like...\nSpeaker B: Yeah, yes.\nSpeaker B: Dual channel watch.\nSpeaker B: Okay, well, maybe I should write that down.\nSpeaker B: You're the secretary.\nSpeaker B: Ruth, what's your last name?\nSpeaker C: Mute.\nSpeaker B: Mute, because I was writing it down in the last...\nSpeaker B: Mute, right.\nSpeaker B: I put all the minutes I make into the shared folder, so if you want to have a look at it, you know where to find it.\nSpeaker D: Okay, we'll make a sub-folder for it because it's starting to fail up already.\nSpeaker B: Yeah, if you want to do it, it's almost the end.\nSpeaker D: Our BSCW hooroo.\nSpeaker B: Okay.\nSpeaker B: Okay, so...\nSpeaker B: Volume, program, dual channel.\nSpeaker D: That's important.\nSpeaker B: And do we want...\nSpeaker C: Ten digits?\nSpeaker C: Yeah, I believe so.\nSpeaker D: If you want to go to channel 90 and you only have that button.\nSpeaker D: That's complicated.\nSpeaker D: But is that so relevant?\nSpeaker D: Because I don't think...\nSpeaker D: I wouldn't buy it personally if it were more controlled without the ten digits.\nSpeaker D: I can imagine when all of you have a set-by-order and you have about 600 channels, I can imagine you want this.\nSpeaker D: But if you're a regular TV user and you just want to watch the Dutch television networks, well, you can use about ten buttons.\nSpeaker B: But we do have 13 different Dutch channels.\nSpeaker D: Yeah, a lot of people, how often do you watch all these channels?\nSpeaker C: They'll probably watch more channels than the older people.\nSpeaker D: Yeah, you're probably right.\nSpeaker D: Yeah, okay.\nSpeaker D: But I think these ten buttons look a bit dull, don't you think?\nSpeaker B: Well, it depends on the looks on the...\nSpeaker D: Okay, well, then there should be something specific.\nSpeaker B: You can make them very fancy by...\nSpeaker B: I mean, Nokia.\nSpeaker B: They have ten digits on their phones and it still looks very fancy.\nSpeaker B: So I do not agree.\nSpeaker D: Okay, so you can experiment with shape or with size or with color.\nSpeaker C: Exactly, exactly.\nSpeaker C: There are some numbers about speech recognition here that younger people would like it.\nSpeaker C: It might be expensive and hard to implement, but it would be a solution there.\nSpeaker D: Okay, speech.\nSpeaker D: Well, maybe it can be combined with the find-to-remote control.\nSpeaker D: So when you add speech recognition to your remote control, it's very easy to change the channel.\nSpeaker D: Yeah.\nSpeaker D: That's one thing.\nSpeaker D: And it's very easy to find your remote control.\nSpeaker D: Yeah.\nSpeaker D: So maybe that's a possibility, but I'm afraid it will be a bit costly.\nSpeaker B: Maybe when we... if we would drop the ten digits, but keep the program in the volume, because maybe people do not always want to use their voice.\nSpeaker C: Especially other people don't like voice recognition, so we should implement such a both.\nSpeaker B: Yeah, but we do focus on younger people.\nSpeaker B: It's a board decision.\nSpeaker D: I think it should work.\nSpeaker D: We should manage it yet.\nSpeaker D: Okay.\nSpeaker D: To make a decision between the ten digits or the voice recognition.\nSpeaker D: Yes.\nSpeaker D: Well, the voice recognition has the main advantage.\nSpeaker D: It makes it very easy for us to implement the find-to-remote control button, remote control function.\nNone: Okay.\nSpeaker D: That's a big advantage, I think.\nSpeaker B: Okay.\nNone: Good.\nNone: Okay.\nSpeaker D: There's only one disadvantage by implementing this.\nSpeaker D: The power saving, power saving will be a bit more difficult.\nSpeaker D: So we can expect that there will be a less longer life to the batteries.\nSpeaker D: But maybe we can think something smart about it.\nSpeaker D: There are some Hebrida hybrid devices, which incorporate the solar panel and rechargeable batteries.\nSpeaker D: Okay.\nSpeaker D: So when you just leave the device in a lightroom, it charges itself.\nSpeaker B: But can we manage it for the cause?\nSpeaker B: No, probably.\nSpeaker D: It's like a very...\nSpeaker C: Maybe, maybe not.\nSpeaker C: I'll have to find that.\nSpeaker C: And include a cradle in which it could recharge.\nSpeaker C: Yes.\nSpeaker C: So that wouldn't be a big problem.\nSpeaker D: No, that's very cheap.\nSpeaker D: Is the cradle very cheap?\nSpeaker D: Oh, it's very cheap.\nSpeaker D: There's no problem.\nSpeaker D: It's just a case with two metal contacts.\nSpeaker B: I know.\nSpeaker B: But there should be an adapter as well.\nSpeaker D: Yes, but their mass production, they're very cheap.\nSpeaker D: Okay.\nSpeaker D: It will cost us practically nothing.\nSpeaker B: We have about 10 minutes left this meeting.\nSpeaker B: I would like to hear maybe you all have things not spoken about yet, but that are important.\nSpeaker B: You came up with the voice recognition data.\nSpeaker B: Are there other things about the market we should know?\nSpeaker C: I think we dealt with the most important information.\nSpeaker C: Just that the younger part of the market isn't the 40% we want to sell to.\nSpeaker C: So we should, for at least a bit, look at all the people.\nSpeaker C: Okay.\nSpeaker C: But not like that.\nSpeaker C: We should use, we might attract them to.\nSpeaker C: Okay.\nSpeaker C: I think that's a big problem.\nSpeaker B: Okay.\nSpeaker B: So although we are focusing on younger people to make it profitable.\nSpeaker B: Yeah.\nSpeaker B: Profitable.\nSpeaker B: Profitable.\nSpeaker B: We still need to take into account the bit older people.\nSpeaker B: Yes.\nSpeaker B: Okay.\nNone: Ruh.\nSpeaker B: I have nothing to add.\nSpeaker D: Nothing to add.\nSpeaker B: I suppose, yeah.\nSpeaker D: I just want to make a summary of all things spoken.\nSpeaker D: Yeah.\nSpeaker D: Great.\nSpeaker D: The different possibilities.\nSpeaker D: So basically we've decided to implement a seemingly easy design.\nSpeaker D: So on the outside it looks easy, but we can make it easy on the inside or elaborate on the inside.\nSpeaker D: So those are the choices we have to make.\nSpeaker D: But I think there's a big advantage in making it more elaborate by implementing speech techniques and voice recognition, these kinds of things.\nSpeaker D: It makes it also more attractable, I think, to our audience.\nSpeaker D: There are some advantages and some disadvantages.\nSpeaker D: And the main advantages that we can implement fancy techniques, which I think our customers will like, the disadvantage is that there are some concerns about the costs and things like battery lifetime energy saving.\nSpeaker D: Yeah.\nSpeaker D: Another advantage is that a circuit board will become much more easier if we do not implement these buttons.\nSpeaker D: If we just add a chip, which does the voice recognition, our circuit board will become even more cheap.\nSpeaker D: Yeah.\nSpeaker B: And that was the main issue, right?\nSpeaker D: Well, it was an issue, but we also thought that we already thought we should do it as cheaply as possible.\nSpeaker D: Yeah.\nSpeaker D: And this even makes it more cheap.\nSpeaker D: But the cheaper the chip or the board, the fewer buttons you can use.\nSpeaker D: No, it's the fewer buttons you have.\nSpeaker B: Yeah.\nSpeaker B: The cheaper circuit board.\nSpeaker D: It's the other way around.\nSpeaker D: And it's very cheap to incorporate integrated circuits with chip.\nSpeaker D: So it's not an issue.\nSpeaker D: It shouldn't be a big issue.\nSpeaker D: Then we should just take a look at the costs.\nSpeaker D: And because I don't know.\nSpeaker D: Especially for the voice recognition.\nSpeaker D: I really don't know.\nSpeaker D: No.\nSpeaker B: It can be costly, maybe not.\nSpeaker B: I'm not sure how I will determine the costs.\nSpeaker B: I will have a look at it.\nSpeaker B: Maybe you will get some information on that.\nSpeaker B: I'm not sure how I will.\nSpeaker D: I hope my personal coach will have a look at it.\nSpeaker B: Okay, great.\nSpeaker B: Well, I think we are quite done.\nSpeaker B: For now we will have the lunch break.\nSpeaker D: How nice.\nSpeaker B: I don't know how long the break will be.\nSpeaker B: But we'll find out.\nSpeaker B: Then we will have 30 minutes of time to perform our individual work.\nSpeaker B: And I'm sure your personal coaches will assist you with it.\nSpeaker B: I will put the minutes I just made in the project.\nSpeaker B: Documents folder if you want to look at it.\nSpeaker B: Well, just do.\nSpeaker B: The interface designer.\nSpeaker B: I would like to, the next meeting, I would like to receive the components concept.\nSpeaker B: Ruh.\nSpeaker B: From you, I would like to see.\nSpeaker B: I'm sorry.\nSpeaker B: Yeah.\nSpeaker B: I'm sorry.\nSpeaker B: User interface designer.\nSpeaker B: Ruh.\nSpeaker B: I would like to see the user interface concept.\nSpeaker D: And the only problem is that it depends on the decisions the decisions Sebastian makes.\nSpeaker D: About the four is what we said.\nSpeaker B: Well, it should be easy.\nSpeaker B: That's what we concluded.\nSpeaker B: It should be an easy interface with not so much buttons.\nSpeaker B: Just a volume and a program one.\nSpeaker B: And yeah, some fancy lights.\nSpeaker B: I think that's what we decided.\nSpeaker D: So we dropped the first recognition?\nSpeaker B: No, no, no.\nSpeaker B: But it shouldn't be integrated.\nSpeaker B: Yeah, of course it's user interface.\nSpeaker B: But I was talking about really the design of the.\nSpeaker D: Well, do we really have to decide now?\nSpeaker D: No.\nSpeaker B: We can decide the next meeting.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: Because I'll have some updated info on that.\nSpeaker B: Yeah.\nSpeaker B: I guess all of you have updated info.\nSpeaker B: I'm sure the marketing expert will receive that you will do some trend watching.\nSpeaker B: And I'm sure the market will.\nSpeaker B: We'll change, adapt.\nSpeaker B: So we will see that.\nSpeaker B: Yeah.\nSpeaker B: That's it for now.\nSpeaker B: Okay.\nSpeaker B: Any other questions or can we have the lunch?\nSpeaker D: We can have the lunch.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Good.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "ES2003b", "summary": "After Project Manager's brief review of the last meeting, Marketing began a presentation of the user requirements. It frustrated users most when they failed to find the lost remote control. The market research also revealed a necessity to simplify the remote control interface, for most buttons were not used. User Interface and Industrial Designer agreed on the intuitive interface design with menu navigation on an LCD screen and very few buttons involved. The new remote control would only be for basic functions such as volume adjusting and channel flipping. A separate joystick would be for additional functionality. The new remote control would be the integration of original remote controls for different devices. Voice recognition would not be feasible due to the budget limit. A speaker and a transmitter were desirable, however, for the location function.", "dialogue": "Speaker B: Just put it on the desk.\nSpeaker B: On the desktop you should find that there's a project documents link.\nSpeaker B: Or actually just there.\nSpeaker B: Proge the documents, yeah.\nSpeaker B: That's it.\nSpeaker B: If you dump it in there.\nSpeaker A: What's the username?\nSpeaker A: What's the username I'm passing?\nSpeaker D: Sorry.\nSpeaker B: Okay.\nSpeaker B: There we go.\nSpeaker B: Excellent.\nSpeaker B: Right.\nSpeaker B: Hopefully that's us ready to go.\nSpeaker B: So, functional design meeting.\nSpeaker B: We'll have to flesh out some of the stuff.\nSpeaker B: We'll start with the minutes of the last meeting.\nSpeaker B: Not a lot, thankfully they say.\nSpeaker B: We'll introduce ourselves.\nSpeaker B: Discuss the possibility of a macro facility.\nSpeaker B: Interact, interact with the TV a bit more.\nSpeaker B: Mentioning of barcode.\nSpeaker B: Joystick for user manipulation.\nSpeaker B: And ergonomics of the remote control as well.\nSpeaker B: And it's come to my attention the following.\nSpeaker B: Teletext has come outdated due to the popular, popular TV internet.\nSpeaker B: Remote control should only be used for the TV.\nSpeaker B: Due for a time to market and possibly also cost issues.\nSpeaker B: Also key is the corporate image to stay recognizable.\nSpeaker B: Your color and slogan of course is down at the bottom there.\nSpeaker B: Now, just to say quickly, I would have thought that only being used for the television, the macro facility may now not be required.\nSpeaker B: Or its functionality would have been of limited use.\nSpeaker B: So, to the point in my opinion, anyway, that it might not be worth pursuing.\nSpeaker B: If anybody disagrees, we can definitely say so.\nSpeaker B: And hopefully we'll just crack on and we'll get everyone going.\nSpeaker B: I'd like to if it's possible here from our marketing expert first to help us gain an idea of where we're going to go.\nSpeaker B: So, I'll just load up here presentation from here if you want.\nSpeaker A: So, would you like to start from here?\nSpeaker B: Yeah, so, here's the, um, five second one maybe.\nSpeaker A: Right, yeah, maybe.\nSpeaker A: Yeah, okay.\nSpeaker A: Okay, right.\nSpeaker A: I'll put up with it in my last name, I guess not.\nSpeaker B: But, if he is, he, that's all right.\nSpeaker B: Okay.\nSpeaker B: Do you want me to just cycle through it for you?\nSpeaker B: Oh, yeah, that would be fine.\nSpeaker A: That would be great.\nSpeaker A: Okay.\nSpeaker A: So, I'm going to go ahead and do a little bit of this.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay, so, um, we did some research.\nSpeaker A: We asked 100 people.\nSpeaker A: The opinions on remote controls.\nSpeaker A: We had some open ended questions.\nSpeaker A: Just what are your opinions on the remote control?\nSpeaker A: Got a lot of responses.\nSpeaker A: And we asked some very specific questions and we got a lot of good feedback.\nSpeaker A: Please bear in mind, this is only a hundred people.\nSpeaker A: So, even when the groups are divided into 15 to 25, 25 to 35, there's only maybe 10 people, 15 people in each group.\nNone: Okay.\nSpeaker A: So, we got some of the best stuff we got.\nSpeaker A: Remotes are often lost.\nSpeaker A: I often lose my remote control, the back of the couch, someplace.\nSpeaker A: And even if it's not lost permanently, it takes me a few minutes to find it.\nSpeaker A: Most buttons are not used anymore.\nSpeaker A: Like you said, teletext is outdated now.\nSpeaker A: I remember trying to load a DVD player recently and there were so many buttons.\nSpeaker A: It took me, I don't know, maybe 10 minutes to go for each button because you have to press the shift button to access the yellow buttons.\nSpeaker A: You know, it's just a lot of stuff that no one really uses.\nSpeaker A: And if they do, not very often.\nSpeaker A: It takes too long to match the remote control.\nSpeaker A: I've seen some remote controls that are big.\nSpeaker A: They have a lot of buttons.\nSpeaker A: You have to hold down more than 1K at one time to do something.\nSpeaker A: They're just not great to use.\nSpeaker A: We just got a lot of bad complaints about remote controls.\nSpeaker A: People do not like remote controls.\nSpeaker A: Some of the good stuff we got.\nSpeaker A: Between the age of 15 and 25, most people will be willing to pay extra for voice recognition software.\nSpeaker A: I don't get excited yet.\nSpeaker A: I've got more to say on that.\nSpeaker A: Most people will be willing to pay for that.\nSpeaker A: Most people want remote controls to be pretty.\nSpeaker A: They want it to be fancy.\nSpeaker A: They want it to be different.\nSpeaker A: Everybody has a white remote control with black buttons and a red button and a green button.\nSpeaker A: Not everybody wants that.\nSpeaker A: Finally, my opinion.\nSpeaker A: The voice recognition thing is cool.\nSpeaker A: And voice recognition, the software opensource software exists already.\nSpeaker A: It's a bit sketchy sometimes.\nSpeaker A: You're not going to get good, always accurate results.\nSpeaker A: But for a very fixed number of words, you have how many different words can you have for remote control up, down, left, right, channel 5, channel 7.\nSpeaker A: You can't have that many words.\nSpeaker A: For a fixed vocabulary, it works quite well.\nSpeaker A: I'm pretty sure people will buy it.\nSpeaker A: But after a while, people may want to return it.\nSpeaker A: If you have to say, most people use a remote control to switch into channels.\nSpeaker A: They say they do that about 98% of the time.\nSpeaker A: Using a remote control, 98% of the time for changing channels.\nSpeaker A: And that's for flickering through channels.\nSpeaker A: So if you have to say up, up, up, up, up, up, if you have to do that all the time, then people might get a bit fed up with that and they may return it.\nSpeaker A: However, because the voice recognition software exists already, there's no need to spend money on research and development.\nSpeaker A: But this doesn't mean the need for microphones in the remote control, which is an unusual feature, in my opinion.\nSpeaker A: But if we do have the voice recognition thing, there's a lot of stuff that you can get rid of.\nSpeaker A: See, you get the two options.\nSpeaker A: Even you have voice recognition by itself, which I think is a bit impractical for like night time.\nSpeaker A: If you want to be watching television, you might be quiet.\nSpeaker A: Or I don't know, you have a visitor come around in the remote so only 20 per year.\nSpeaker A: It's been impractical to have just voice recognition by itself.\nSpeaker A: So you can have voice recognition and a regular remote.\nSpeaker A: But imagine you got rid of the regular remote part.\nSpeaker A: Then you can design the remote to look any way you want it to look, because there are no restrictions on physical size or shape.\nSpeaker A: It could be as fancy as you want it to be.\nSpeaker A: It could be like a lollipop, something like that, something weird like that.\nSpeaker A: As long as the voice recognition stuff works, that's fine.\nSpeaker A: So we have the free birds, we have the design bit.\nSpeaker A: We have the fancy bit, right?\nSpeaker A: The voice recognition is fancy. It's cool. It's different. It's radical.\nSpeaker A: And then we have an extra bit. I don't remember.\nSpeaker A: So I'm pretty sure people will buy the remote, but is it practical to have a voice recognition system in the remote control?\nSpeaker A: I think it's a big question.\nSpeaker A: Will people be willing to wait for the period that it will take to train the remote?\nSpeaker A: Because I think it will. The remote will get better over time with the same user.\nSpeaker A: But for the first week or for the first two weeks, are you willing to wait?\nSpeaker A: Are you willing to have a bad remote control?\nSpeaker A: And what if you have visitors come around?\nSpeaker A: They stay the night. They want to use the TV. They can't use the remote because they speak differently to you.\nSpeaker A: How do you account for regional accents and stuff like that?\nSpeaker A: Will people return the remote control? I think a lot of young people will buy the remote control if they have the money.\nSpeaker A: So do our audience have the money. But will they return it after a while because it's not as fast as pressing the button.\nSpeaker A: It's not practical. So these are things I think we should consider. I think it's cool.\nSpeaker B: He could just be up with it.\nSpeaker B: Sure. I'm about to end.\nSpeaker A: I think it's cool. But there are definitely some considerations.\nSpeaker B: Excellent.\nSpeaker B: Hear from the user interface designer now, I think.\nSpeaker B: Your presentation now is in the whole.\nSpeaker A: There was a website.\nSpeaker B: Technical functions.\nSpeaker A: I started making stuff up.\nSpeaker D: I read meme.\nSpeaker C: Okay.\nSpeaker C: This is a brief run through of the technical functions of the remote.\nSpeaker C: The presentation is already slightly obsolete because I had in mind something that would be aimed at controlling multiple devices.\nSpeaker C: But there's still stuff of relevance.\nSpeaker C: I've looked at a couple of other remote control models just for an idea of basic design principles.\nSpeaker C: And it wasn't really satisfied by what I have to say.\nSpeaker C: So it's more of a springboard for how these could be improved on.\nSpeaker C: Basically the job of the device is to send commands to an appliance, in this case the television set, to say you're getting off your backside.\nSpeaker C: And there's two main trends in the design practice.\nSpeaker C: On the one hand, this particularly relates to sort of earliest models where with so many buttons on it's about the size of the television set.\nSpeaker C: And every possible function that the device has the remote control controls and most of these functions are not going to be used.\nSpeaker C: It creates a rather user-unfriendly interface.\nSpeaker C: And there's a user-focused approach that pairs things down to just what is most likely to get used.\nSpeaker C: And of course there is a certain amount of loss of function here obviously, but mostly that's relatively peripheral functions.\nSpeaker C: And so you go for something that's fairly intuitive to use.\nSpeaker C: And for the most part, for TV remote, that would be channel control and volume.\nSpeaker C: And if you are one can cooperate to control a DVD or a VCR, put that into that little play-pause stop rewind.\nSpeaker C: Fast forward card so far.\nSpeaker C: My own user was definitely going for a user-focused design.\nSpeaker C: And I think that's something that is more programmable.\nSpeaker C: The full range of functions available to it, but you can then specify yourself what you're most likely to want.\nSpeaker C: So I like to have an user's idea, including a joystick in that, and possibly a fairly minimal number of function buttons.\nSpeaker C: And then you can do different modes for the device.\nSpeaker C: Again I was just going to write a principle from the\nSpeaker B: If we could hear from our industrial engineer.\nSpeaker D: Yeah, because I know. I'm still working as the fan got it finished.\nSpeaker D: What's it save?\nSpeaker D: What's it save?\nSpeaker B: If you look in the desktop, you'll find that there's a link to the project folder, or project documents.\nSpeaker B: If you save it now, we can open it up from here.\nNone: Right, hold on, what we're doing here is ended here.\nSpeaker D: And put it on here, damn sure.\nSpeaker D: I'm just gonna close it now.\nSpeaker D: I don't know where I'm saying, but.\nSpeaker A: I like if you go to one.\nSpeaker A: Oh whichever one it does today.\nSpeaker A: And just go ahead and say thanks.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Right.\nSpeaker D: I'm responsible for a working design.\nSpeaker D: This is basically the inside going on of the remote control.\nSpeaker D: So you have the energy source.\nSpeaker D: You have the user interface.\nSpeaker D: This is what I've seen.\nSpeaker D: The sender will push the button.\nSpeaker D: The chip will respond.\nSpeaker D: And then the digital signal sent to the TV.\nSpeaker D: So if you go to the next slide, you'll see what do we need on the user interface.\nSpeaker D: Do we need many buttons?\nSpeaker D: Or do we need many light bulbs that could be easier to realize which appliances are on or not?\nSpeaker D: Or would that take too much power?\nSpeaker D: Or would we need more components in there to supply the power?\nSpeaker D: The joystick is another thing if we're gonna add that.\nSpeaker D: And there'll be more components to deal with that.\nSpeaker D: So we need to.\nSpeaker D: I don't know exactly what.\nSpeaker D: That's the design of the the layout of the electronic design.\nSpeaker D: But obviously there'll be more details once we decide what we're putting on the user interface.\nSpeaker D: Okay, so that would be decided, I guess.\nSpeaker D: And the next slide.\nSpeaker D: If you go to the next slide then, I just used the.\nSpeaker D: It was a mess.\nSpeaker D: I was just putting, I think it's together at the end there.\nSpeaker B: Don't worry about it at all, mate.\nSpeaker D: So you have the energy source, obviously connected to the chip and the user interface.\nSpeaker D: Obviously it contains everything.\nSpeaker D: If the switch turned it on, the infrared bulb will contact the TV and have whatever device or TV player will turn on.\nSpeaker D: So do we need on the control different buttons to decide?\nSpeaker D: Or different light bulbs?\nSpeaker D: It'll be easier.\nSpeaker D: I don't know what we should decide on that.\nSpeaker B: Okay, well, I'll show you.\nSpeaker B: No, it's just, yeah.\nSpeaker B: Okay, right.\nSpeaker B: We can probably skip that.\nSpeaker B: So we've had some stuff put forward along with the new user requirements.\nSpeaker B: So I'm not sure what kind of input I'd say so far.\nSpeaker B: I hear what the marketing experts saying about voice activated control.\nSpeaker B: However, I've got a couple of worries about that.\nSpeaker B: The power required and the ability to cost.\nSpeaker B: It seems like for an embedded system, this could cause issues.\nSpeaker B: For example, you see that there's fairly robust services on computers via via voice IBM do drag and dictate.\nSpeaker B: But these require a lot of memory and usually state quite advanced processor requirements.\nSpeaker B: Voice activation could be an interesting idea, but I think that our industrial designer would probably, upon some research, say that it's maybe not feasible.\nSpeaker B: That's just my view right now.\nSpeaker B: However, the idea of a joystick and then maybe an LCD, which has been kind of put forward so far.\nSpeaker B: It's almost like having a small TV to say.\nSpeaker B: So we can control almost all of the functionality from display.\nSpeaker B: I don't know how much power now CD would take, but it might be quite low.\nSpeaker D: I see the under a mouth just telling you what's on.\nSpeaker B: If you can imagine a display, maybe slightly larger than both on a mobile phone.\nSpeaker B: If you can read a fair amount of information traversed, maybe quite a few menus.\nSpeaker B: If we maybe use mobile phones as a good example, they usually incorporate the keypad and then all the other functionality is usually associated inside of traversing around.\nSpeaker B: That could be one possibility.\nSpeaker C: Then you could have the common buttons as you say, volume control, changing channels.\nSpeaker D: Do you think that people who get mixed up, looking at this screen, not screens, trying to get everything working would be the confusing?\nSpeaker C: I think that size of display was about what I had in mind.\nSpeaker C: So you probably would require to, most of the mobile phone display you see these days is coloured, but you should probably just stick to black and white.\nSpeaker C: It's simply to keep the unit cast down.\nSpeaker C: Colour could be a subsequent development that costs a bit of extra money.\nSpeaker B: That would be my feeling as well. I think we could possibly create quite useful user interface.\nSpeaker A: The idea of a remote with a menu and a joystick, I think if I was in a habit of buying remote controls, then I would want one.\nSpeaker A: But I think we don't have a specific audience.\nSpeaker A: What does our target audience want?\nSpeaker A: Are we trying to mark it in the corner? For whom is this intended? Everybody?\nSpeaker B: I think it could probably be aimed at most people who have used a mobile.\nSpeaker B: That might be just another way of saying, trying to target most people.\nSpeaker B: If we were to follow that avenue, we might be almost relying on their experience with a mobile phone, I suppose.\nSpeaker B: But that does cover a very large section of the people out there.\nSpeaker B: I imagine as well that the actual LCD and maybe to a certain extent the joystick as well would be for the additional functionality, which maybe doesn't get used as often.\nSpeaker B: Maybe it makes it easy for them to figure out how to change the channels as in the frequencies and such for reprogramming it.\nSpeaker B: You can still have the main buttons that allow the people to turn the machine on and off.\nSpeaker B: It depends. Maybe we have the joystick as a separate one for the other functions and there's just a small number of buttons that have already been defined.\nSpeaker B: We go for the nice standard up and down and volume control so that literally anybody can come along and pick up the remote and still know what to do.\nSpeaker B: It can completely ignore most of the functionality, which would be through the actual owner of the key VN who would normally use it in it.\nSpeaker C: The thing I like about the joystick is that you can then basically put it for a TV remote alone.\nSpeaker C: You can put the main thing people do with the TV or the volume of the channel.\nSpeaker C: You can back all that onto a single control.\nSpeaker C: If you wanted to access further functions, you then get some menu navigation.\nSpeaker C: Things like tuning in channels, adjusting brightness and fiddly things that no one will be bothered with.\nSpeaker C: Most TVs these days are childy-have menus anyway, so this would be a fairly rational way of integrating the remote with the devices controlling.\nSpeaker C: I don't know if it would be a concern about a market.\nSpeaker C: Most of the replacement remote control rationalization is 20 different devices.\nSpeaker C: You want one that will do the job of all of them.\nSpeaker C: If all you're replacing is one of the many remote you have in your living room, you still need separate remotes for everything else.\nSpeaker B: This is a requirement that we have to stick to.\nSpeaker B: I can understand your point and I would agree with you, but this is our design spec for now.\nSpeaker B: I would say so, because then we have to be able to control the remote boxes and that is a separate unit.\nSpeaker B: I would say the design spec we've been laid with is for the television only for now.\nSpeaker B: The joystick is just for differentness.\nSpeaker C: Maybe the market should be TV manufacturers from the public.\nSpeaker A: In Hyperim money we waste the remote that no one buys, you know, like if we made a perfect if we made the very remote we made someone bought it. We have to set a lot of remote controls. Is this is\nSpeaker B: going to be enough to sell? Well something else that you brought up was the ability to lose a remote which as you said you've done that we've probably all done. Yeah. I don't know if it's a gimmick or not but if any of you ever seen those odd little key rings where if you whistle it will beep and\nSpeaker A: you tell you where it is. I had one of those and my brother or my dad could have beat me up because it went off all the time accidentally. Well the other option of course is that the clapping one.\nSpeaker B: The what's going to say clapping digital telephones for example one unit has of course you have to have that base unit somewhere where there's a button but maybe it's a button that you attach to the TV. It's very relevant. Something which when you press that it would be to give its location away on the remote unit. I think that could be something could separate us a bit and that way because we're attaching what would be a small button to the TV again say what it would be a small transmitter, watch battery type scenario I'd say or something that if it was problems enough you could actually even just I don't power and elsewhere so that might be something that we could look into.\nSpeaker B: Yeah. Yeah it's putting a lot on your shoulders there so I mean that could be something that separates us a bit from Mark as well. So to go on from here we have to decide exactly what we're going to do with the remote before we leave this meeting it'd be best to say this is what we're going this is the design we're going to try and get this is how we're going to make ourselves look unique.\nSpeaker B: Do we go for maybe a remote control sorry we're going to go for remote control obviously.\nSpeaker B: Do we go for a small joystick that would operate say changing channels up and down and then what another button that would say that it's now being used to maneuver around the LCD. Yeah.\nSpeaker D: Interesting volumes. Volumely important of the joystick do you think. We could use\nSpeaker B: say the left and right for changing channels and the up and down volume. Yeah but and we have a\nSpeaker A: frequency of how what people use a remote control for most. Okay. Power is used like once per hour channel selection is used 168 per hour that's by far the biggest one and then teletext is still\nSpeaker B: here that's 14. Go. I'm volume selection. Volume selection kit. Yep. The teletext we're gambling with we're going to say no it's dead. Okay. Wave the dodo. Yeah. So we well sorry we could maybe even go as far as saying power button small joystick LCD and then what maybe is it one more button to save it you're using the menu system and the possibility of pushing down on the joystick to say okay. So we're having very very few buttons involved navigation around a menu\nSpeaker C: for more things. I would say two function buttons the we're now using the menus button and an okay button. I mean something I found with the joystick on the joysticks on mobile phones is that a lot of the time the device mistakes you try to move it one direction or the other pushing down for okay.\nSpeaker C: Okay. That's really irritating. The other design constraint I think we should bear in mind is that people are very much used to using a remote control one-handedly so we need to make sure that even with the joystick something it can just sit comfortably in the palm of the hand and have the\nSpeaker B: joystick just control by the thumb. Okay. So I'd say that sounds like a rough idea. Do we incorporate the idea of trying to locate the remote control again via a beefing noise? Yeah. I think because\nSpeaker A: it's so small it might be and I mean if we only have like two or three buttons it might be essential\nSpeaker B: to have to be able to locate it again. Yeah. So that would require a transmitter maybe attached to the TV just so they says find me and what basically a small microphone on the actual unit can maybe hide it in the light bulb as well. So small speaker speaker. So yeah. Light bulb now to flash.\nSpeaker B: Now you'd see it anyway. We might be better with the sound possibly. Those little keyrings have both.\nSpeaker B: In true fact considering the cost of an LED we could just incorporate it anyway. The LEDs can be surprising on bright now. So by the sense of it what we're suggesting so far you design the user interface is still quite open. You could go for quite an interesting design because we're only going to have a very few and you've got you know a small LCD joystick. I think it leaves you open to a potential wide range of shapes something that can make it stand out slightly. If you could look into what we suggest to so far the feasibility of a small transmitter and such maybe if an LCD screen requires too much power or such. Every user. Mark and expert. Well I can give you the frequency.\nSpeaker A: What people what options people use most often I guess that's going to be important in defining the the software side of the interface right. You want to stuff that are more. If you could maybe as\nSpeaker B: well see what people think about the idea of this minimalist approach that we've taken where of a joystick to control the very basic functions and then the idea of manipulating it like you would a mobile. I don't know how easy that would be within the time frame but could be something we could maybe look into. Sure. Sure. Okay. Any last comments anybody would like to put forward anything that you think has been missed out. Bit of a wide open questionnaire of course. Feel free to email me if you think that we screwed something up and want to get it rectified as soon as possible.\nSpeaker D: Okay. So I should just look at the speaker about the speaker and the LED speaker and LED for locating.\nSpeaker C: Transmitter transmitter. One we thought about that. If we do end up trying to market this to television manufacturers then the transmitter could actually be built into the television.\nSpeaker B: Of course we're getting an external power source. Yeah that's quite true. And something that we can easily be adopted at the last minute as well. I'd say electronics can be either placed externally or internally. Big still difference to the final product of the actual remote control so that's good.\nSpeaker C: Maybe gives us a new potential market. If we get a major television manufacturer to say oh yeah we'll have that save us the bottle. Then that's vast amounts of sales. Oh one thing we've almost not\nSpeaker B: talked about at all my apologies for that. User interface we also need to maybe get the slogan in here. It's unpointing at my laptop. What in god? Real reaction and such. So as you put the fashion in electronics my apologies. No it could well be. I probably missed that. I think that's almost the last minute thing we can just incorporate into the actual plastic on top. So I'm not too concerned that we haven't gone into debt about that but it might be something you could consider when we're thinking about shapes. In fact we might like to put a slogan on and possibly two hours to\nSpeaker C: signify the company rather than real reaction. I thought the double arca perhaps be the end of\nSpeaker B: the video. Sounds good. And I'd say for that sauce for now. Okay.\n"}, {"meeting_id": "Bro016", "summary": "The meeting took place close to the finalization of the model, so participants were focused on further experimentation they could conduct. It began with a discussion about how the backend system could be further improved. The team thought that perhaps they could tweak language scaling and insertion penalties to see how the model changed its performance. Then, the team moved into discussing methods for detecting voice segments in their results. Finally, the team spent some time exploring normalization techniques and acoustic events, both of which could potentially improve model performance.", "dialogue": "Speaker E: Let's see. Test? Test? Yeah? Okay. Channel one. Hello.\nSpeaker A: Test.\nSpeaker E: I was saying he can be here next week, once day through Friday, through, I'm sorry, and I won't be here Thursday and Friday, but my suggestion is that, at least for this meeting, people should go ahead. I've seen him go be here.\nSpeaker E: You know, we don't have any check accent. Yeah, that's...\nSpeaker E: As far as I know, so. Okay.\nSpeaker E: There we go.\nSpeaker E: So other than reading digits, what's our agenda?\nSpeaker G: I don't really have anything new than working on meeting recorder stuff.\nSpeaker E: Okay. Do you think that would be the case for next week, also, or is...\nSpeaker E: what's your projection on?\nSpeaker E: Because the one thing that seems to me, we really should try, if you hadn't tried it before, because it hadn't occurred to me, it was sort of obvious thing, is adjusting the scaling and insertion.\nSpeaker E: L.T. sort of stuff.\nSpeaker G: I did play with that, actually, a little bit.\nSpeaker G: What happens is, when you get to the noisy stuff, you start getting lots of insertions.\nSpeaker G: Right.\nSpeaker G: And so I've tried playing around a little bit with the insertion penalties, things like that.\nSpeaker G: Yeah.\nSpeaker G: It didn't make a whole lot of difference.\nSpeaker G: Like, for the well-matched case, it seemed like it was pretty good.\nSpeaker G: I could do more playing with that, though.\nSpeaker E: But you were looking at Mel Capstrom.\nSpeaker E: Right.\nSpeaker G: Oh, you're talking about for our features.\nSpeaker E: Right. So, I mean, it's not the direction that you were working with, that we were saying, what's the best you can do with Mel Capstrom.\nSpeaker E: But they raised a very valid point, which I guess...\nSpeaker E: So, to first order, I mean, you have other things you're going to do.\nSpeaker E: But the first order, I would say, that the conclusion is, that if you do some munking around with the exact HTK training, and with how many states and so forth, that it doesn't particularly improve the performance.\nSpeaker E: In other words, that even though it sounds pretty dumb, just applying the same number of states to everything, or, no matter what language, isn't so bad.\nSpeaker E: Right.\nSpeaker E: And I guess you hadn't gotten to all the experiments you wanted to do with number of Gaussian's.\nSpeaker E: Right.\nSpeaker E: But let's just...\nSpeaker E: If we had to draw a conclusion on the information we have so far, we'd say something like that, right?\nSpeaker E: So, the next question to ask, which is, I think the one that Andreas was addressing himself to in the lunch meeting, is, we're not supposed to adjust the back end.\nSpeaker E: But anybody using the system would.\nSpeaker E: Yeah.\nSpeaker E: If you were just adjusting the back end, how much better would you do in noise?\nSpeaker E: Because the language scaling and search and penalty is over, they're probably set to be about right for milk capstrom.\nSpeaker E: But they're probably not at all set right for these things, particularly these things that look over larger time windows, in one way or another with LDA and KLT and neural nets, all these things.\nSpeaker E: In the past, we always found that we had to increase the insertion penalty to correspond to such things.\nSpeaker E: So, I think that's kind of a first order thing that we should try.\nSpeaker G: So, the experiment is to run our front end like normal with the default insertion penalties and so forth, and then tweak that a little bit and see how much of a difference it makes.\nSpeaker E: By our front end, I mean, take some version that Stefan has that is our current best version of something.\nSpeaker E: I mean, don't want to do this over 100 different things that they've tried, but for some version, he said he's a good one.\nSpeaker E: So, how much does it improve if you actually adjust that?\nSpeaker E: But it is interesting to say you have for the noise, how about for the mismatched or the medium mismatched conditions?\nSpeaker E: Have you, when you adjusted those numbers for milk capstrom, did it?\nSpeaker G: I don't remember off the top of my head.\nSpeaker G: Yeah, I didn't even write them down.\nSpeaker G: I don't remember I would hate to.\nSpeaker G: Well, I did write down.\nSpeaker G: So, when I was doing, I just wrote down some numbers for the well-matched case.\nSpeaker G: Yeah.\nSpeaker G: Looking at, I wrote down with the deletions, the substitutions, insertions, or four different numbers of states per phone.\nSpeaker G: Yeah.\nSpeaker G: But that's all I wrote down.\nSpeaker G: Okay.\nSpeaker G: I would need to do that.\nSpeaker E: Okay, so.\nSpeaker E: I can do that for next week.\nSpeaker E: Yeah.\nSpeaker E: And, yeah, also, sometimes if you run behind some of these things, maybe we can get someone else to do it, you can super-rise or something.\nSpeaker E: But I think it'd be good to know that.\nSpeaker G: Yeah.\nSpeaker G: I just need to get a friend and, uh, select for me, or you point me to some files.\nSpeaker G: Yeah.\nSpeaker G: You've already come to it.\nSpeaker G: All right.\nSpeaker B: Okay.\nSpeaker G: I probably will have time to do that and time to play a little bit with the silence.\nSpeaker B: Yeah.\nSpeaker G: Model.\nSpeaker G: Maybe I can have that for next week, maybe next week.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Because I mean, the other, that in fact, might have been part of what, uh, the difference was, at least part of it, that we were seeing, you know, we were seeing the SRI system was so much better than the tandem system.\nSpeaker E: Part of it just be that the SRI system, they always adjust these things to be so optimized.\nSpeaker G: I wonder if there's anything that we could do to the front end that would affect the insertion?\nSpeaker G: Yes.\nSpeaker E: I think you can.\nSpeaker E: Okay.\nSpeaker E: Well, um, uh, part of what's going on, um, is the, uh, the range of values.\nSpeaker E: So if you have something that has a much smaller range or a much larger range, and taking the appropriate root, you know, if something is kind of like the equivalent of a bunch of probabilities multiplied together, you can take a root of some sort of, like, seven probabilities together, you can take the seventh root, or something, or it's in the log domain, divided by seven, but, um, that has a similar effect because it changes the scale of the numbers, if the differences between different candidates from the acoustic model, as opposed to what's coming from the language model.\nSpeaker G: That's changing the value of your insertion.\nSpeaker E: Yeah. I mean, it's more directly like the, the language scaling or the, the model scaling or acoustic scaling, but you know that those things have kind of a similar effect to the insertion penalty anyway, a slightly different way of, of handling it.\nSpeaker G: So, um, so if we know what the insertion penalty is, we can get an idea about what range our numbers should be on.\nSpeaker G: I think so, yeah.\nSpeaker E: Yeah, so that's why I think that's another reason, another curiosity, as to why it would, in fact, be kind of neat to find out if we're way off.\nSpeaker E: I mean, the other things aren't, we're seeing, I'm sure you've already looked at this in these noisy cases.\nSpeaker E: We are seeing lots of insertions, right? The insertion number is quite high.\nSpeaker E: I know the VAD takes pretty care part of that.\nSpeaker C: I've seen that with the male capster. I don't know about the Aurora front end, but I think it's much more balanced with, uh, when the front end is more robust.\nSpeaker C: Yeah, I could look at it.\nSpeaker C: At this, yeah.\nSpeaker E: Yeah, what's the typical number?\nSpeaker E: I don't, I don't know.\nSpeaker E: Okay. I'm sure it's more balanced, but it wouldn't surprise me if there's still, I mean, in the, the old system is used to do.\nSpeaker E: I remember numbers kind of like insertions being half the number of deletions as being, and both numbers being, tend to be on the small side, comparing to, uh, substitutions.\nSpeaker G: Well, this, the whole problem with insertions was what I think, um, we talked about when the guy from OGI came down at one time, and, and that was when people were saying what we should have a, uh, voice activity detector.\nSpeaker G: Right. That, because all that stuff that we're getting, the silence that's getting through is causing insertions.\nSpeaker G: Right.\nSpeaker G: I've mentioned there's still a lot.\nSpeaker E: Yeah, and it may be less of a critical thing. I mean, the fact that some get by, maybe less of a critical thing if you, uh, get things in the right range.\nSpeaker E: So I mean, the insertions is, is a symptom.\nSpeaker E: It's a symptom that there's something wrong with the range, but there's a, your, your, your substitutions tend to go up as well.\nSpeaker E: So I, I think that, uh, the most obvious thing is just the insertions.\nSpeaker E: I don't know if it's, but, uh, um, if you're operating in the wrong range, I mean, that's why just in general, if you change what these, these penalties and scaling factors are, you reach some point that's, uh, that's a minimum.\nSpeaker E: So, um, we do have to do well over a range of different conditions, some of which are noisier than others.\nSpeaker E: But, um, I think we may get a better handle on that if we, if we see, um, I mean, it's, if we actually could pick a, a more stable value for the range of these features.\nSpeaker E: It, um, could, uh, even though it's, it's, it's true that in a real situation, you can, in fact, adjust the, these, these scaling factors and the back end.\nSpeaker E: And it's artificial here that we're not adjusting those.\nSpeaker E: You certainly don't want to be adjusting those all the time.\nSpeaker E: And if you have a nice front end that's roughly the right range, I remember after we got our stuff more or less together in the previous systems, we built that we tended to set those scaling factors at kind of a standard level.\nSpeaker E: And we would rarely adjust them again, even though you could get a, for an evaluation, you can get an extra point or something if you tweaked it a little bit.\nSpeaker E: But once we knew what, roughly the right operating range was, it was pretty stable.\nSpeaker E: And, uh, we might just not even be in the right operating range.\nSpeaker G: So with the, uh, what a good idea to try to map it into the same range that you get in the well matched case.\nSpeaker G: So if we computed what the range was and well matched, and then when we get our noisy conditions out, we try to make it have the same range.\nSpeaker E: No, you don't want to change it for different conditions.\nSpeaker E: No, no, I, what, what I'm saying.\nSpeaker G: Oh, I wasn't interested in changing it for different conditions. I was just saying that when we pick a range, we want to pick a range that we map our numbers into.\nSpeaker G: Yeah.\nSpeaker G: We should probably pick it based on the range that we get in the well matched case.\nSpeaker G: Otherwise, I mean, what range are we going to choose to map everything into?\nSpeaker E: Well, it depends how much we want to do, game ismanship and how much we want to do.\nSpeaker E: I mean, if it can be actually, even if you want to be played on the game ismanship side, it can be kind of tricky.\nSpeaker E: So I mean, what you would do is set the, set the scaling factors so that you got the best number for this 0.45 times the, you know, and so on.\nSpeaker E: But they might change that those weightings.\nSpeaker E: So I just sort of think we need to explore the space. Let's take a look at a little bit.\nSpeaker E: We may just find that that we're way off.\nSpeaker E: Maybe we're not.\nSpeaker E: You know, that's with these other things that may turn out the, it's kind of reasonable.\nSpeaker E: But then, I mean, Andreas gave very reasonable response and he's probably not going to be the only one who's going to say this in the future of, you know, people, people within this tight-knit community who are doing this evaluation are accepting.\nSpeaker E: More or less that these are the rules.\nSpeaker E: But people outside of it are looking at the broader picture are certainly going to say, well, wait a minute, you're doing all this standing on your head in the front end when all you could do is just adjust this in the back end with one knob.\nSpeaker E: And so we have to at least, I think, determine that that's not true, which would be okay.\nSpeaker E: Or determine that it is true, in which case we want to adjust that and then continue with what we're doing. And as you say, as you point out, finding ways to then compensate for that in the front end, also then becomes a priority for this particular test.\nSpeaker E: So you don't have to do that.\nSpeaker E: Okay.\nSpeaker E: So, what's new with you?\nSpeaker C: So there's nothing new.\nSpeaker E: What's old with you? It's developed.\nSpeaker E: I'm sorry.\nSpeaker E: Okay, what's old with you? That is developed over the last week.\nSpeaker C: So, if we're mainly working on the report, on the report of the work that was already done,\nSpeaker G: that's all. Anything new on the thing that you're working on with the... What was that?\nSpeaker G: The voicing detected.\nSpeaker A: What's going on now? What are you doing?\nSpeaker A: We try to use the variance, the difference between the effect spectrum and the male filter band spectrum. Also, the other parameters relates with the autocorrelation function.\nSpeaker A: Energy and the variance also helps the autocorrelation function.\nSpeaker E: So, that's what you were describing, I guess a week or two ago.\nSpeaker A: We don't have result of the aurora jet. We need to try and run network.\nSpeaker E: So, you're training neural networks now?\nSpeaker E: No, not yet.\nSpeaker E: So, what's going on?\nSpeaker A: Well, I work in the report too, because we have a lot of results, embedded in spares, and what necessary to look at the directory to give some structure.\nSpeaker E: So, yeah, if I can summarize, basically what's going on is that you're going over a lot of material that you've generated in a furious fashion of generating many results and doing experiments and trying to pull it together into some coherent form to be able to...\nSpeaker C: Yeah, basically we've stopped experimenting. We're just trying to think of some kind of technical report.\nSpeaker G: Is this a report that's for aurora?\nSpeaker G: No.\nSpeaker G: It's like a tech report for XC.\nSpeaker A: Yes.\nSpeaker E: So, my suggestion though is that you not necessarily finish that, but that you put it all together so that you've got a clearer structure to it, you know what things are, you have things documented, you've looked things up that you needed to look up so that such a thing can be written.\nSpeaker E: And when do you leave again?\nSpeaker A: July, first of July.\nSpeaker E: First of July, okay. And that you figure on actually finishing it in June, because you're going to have another much results to fit in there anyway.\nSpeaker E: And right now it's kind of important that we actually go forward with experiments. So I think it's good to pause and together everything together and make sure it's in good shape so that other people can get access to it so that it can go into a report in June.\nSpeaker E: But I think to really work on fine tuning the report at this point is probably a bad timing.\nSpeaker C: Yeah. Well, we didn't, we just planned to work on one week on this report, no more anyway.\nSpeaker E: I really want to add other things later anyway, because you're, this is more to go.\nSpeaker G: Yeah, well, so I don't know there are small things that we started to do, but maybe discovering anything that makes you scratch your head as you write this report.\nSpeaker G: Like, why did we do that? Why did we do this?\nSpeaker C: Yeah. Yeah.\nSpeaker C: Actually, there were some tables that were also with partial results. We just noticed that I get a ring of result that for some conditions we didn't have everything.\nSpeaker C: Yeah, yeah, we have extracted actually the noises from a speech that car.\nSpeaker C: So we can train neural network with speech and these noises.\nSpeaker C: It's difficult to say what it will give because when we look at the, or at the IDG experiments, there are these three conditions that have different noises and apparently the system performs as well on the scene noises, on the scene noises.\nSpeaker C: But I think there's something we have to try anyway. So adding the noises from, from the speech that car.\nSpeaker E: That's, that's, that's permitted.\nSpeaker C: Well, OGI did that.\nSpeaker C: At some point, they did that for the first activity.\nSpeaker G: Could you say it again? What exactly did they do?\nSpeaker C: They use some parts of the Italian database to train the voice activity sector, I think.\nSpeaker E: Yeah, I guess the thing is, yeah, I guess that's a matter of interpretation.\nSpeaker E: The rules, as I understand it, is that in principle the Italian and Spanish and the English, no Italian and the Finnish, and the English were development data. And what you could adjust things.\nSpeaker E: And the German and Danish were the evaluation data. And then when they finally actually evaluated things, they used everything.\nSpeaker E: Yeah, that's right.\nSpeaker E: And it is true that the performance on the German was, I mean, the improvement wasn't so good. The raw performance was really pretty good.\nSpeaker E: So, and it doesn't appear that there's strong evidence that even though things were somewhat tuned on those three or four languages, that going to a different language really hurt you. And the noises were not exactly the same, right, because it was taken from a different...\nSpeaker E: I mean, they were different drives. I mean, it was actual different cars and so on.\nSpeaker E: So, it's somewhat tuned. It's tuned more than, you know, you really like to have something that needed no particular noise at all, maybe just some white noise or something like that, at most.\nSpeaker E: But that's not really what this contest is. So, I guess it's okay.\nSpeaker E: That's something I'd like to understand before we actually use something from it, because it would...\nSpeaker G: It's probably something that, you know, the experiment designers didn't really think about, because I think most people aren't doing trained systems or, you know, systems that are like ours, but they actually use the data to build models, I mean, just doing things like all the processing.\nSpeaker E: Well, it's true, except that that's what we used in Aurora 1, and then they designed the things for Aurora 2, knowing that we were doing that.\nSpeaker E: That's true.\nSpeaker G: And they didn't forbid us, right, to build models on the data.\nSpeaker E: No, but I think that it probably would be the case that if, say, we trained on Italian data, and then we tested on Danish data, and it did terribly.\nSpeaker E: That it would look bad, and I think someone would notice, and would say, well, look, this is not generalizing. I would hope they would.\nSpeaker E: But it's true, you know, maybe there's parameters that other people have used, you know, that they have tuned in some way for other things.\nSpeaker E: So it's, we should, maybe that's maybe a topic, especially if you talk with him when I'm not here, that's a topic you should discuss with he may check it's okay.\nSpeaker G: The speakers or each of the training utterances.\nSpeaker C: What do you mean?\nSpeaker E: Social security number.\nSpeaker C: I think it.\nSpeaker C: Made a female just me up at least.\nSpeaker G: What kind of information do you mean?\nSpeaker G: Well, I was thinking about things like, you know, gender, you know, gender specific nets and both the track link on normalization.\nSpeaker G: Things like that.\nSpeaker G: I don't know what information we have about the speakers that we could try to take advantage of.\nSpeaker E: Right. I mean, again, if you had the whole system you were optimizing, that would be easy to see.\nSpeaker E: But if you're supposedly just using a fixed back end and you're just coming up with a feature vector, I'm not sure.\nSpeaker E: I mean, having the two nets, suppose you detected that was male, female, you can look at different both in as separate streams or something.\nSpeaker E: Maybe.\nSpeaker G: I don't know. I was just wondering if there was other information we could exploit.\nSpeaker E: Yeah, it's interesting thought maybe having something along, I mean, you can't really do vocal track normalization.\nSpeaker E: It's something that had some of that effect. Yeah.\nSpeaker G: No, I had no idea. I had thought it was too much about it really. It just something that popped into my head just now.\nSpeaker G: Normalization, you know, you have some sort of a general speech model, maybe just a mixture of galsians that you evaluate every utterance against.\nSpeaker G: And then you see where each utterance, like the likelihood of each other, and to divide the range of the likelihoods up into discrete bins.\nSpeaker G: And then each bins got some knob.\nSpeaker E: Yeah, but just listen to yourself. I mean, that really doesn't sound like a real time thing with less than 200 milliseconds latency that were you're not adjusting the statistical engine at all.\nSpeaker E: Yeah, well, not just expensive. I don't see how you could possibly do it. You can't look at the whole utterance and do anything.\nSpeaker E: Each frame comes in and it's got to go out the other end.\nSpeaker E: So whatever it was, it would have to be sort of on a per frame basis. Yeah. I mean, you can do fairly quickly. You can do male female stuff.\nSpeaker E: But as far as, I mean, like I thought, maybe I ended a thing with a vocal track normalization, ways back, maybe other people did too, with trying to identify third formant, average third formant, using that as an indicator of.\nSpeaker E: So, you know, third formant, if you imagine that the first order, what happens with changing vocal track is that the formants get moved out by some proportion.\nSpeaker E: So if you had a first formant that was 100 hertz before, if the 50, if the vocal track is 50% shorter, then it would be out at 750 hertz and so on.\nSpeaker E: So that's a move of 250 hertz, whereas the third formant, which might have started off at 2500 hertz, might be out to 3750, you know, so you frequently get less distinct higher formants.\nSpeaker E: It's still third formant is kind of a reasonable compromise.\nSpeaker E: So I think, if I recall correctly, they did something like that.\nSpeaker E: But that doesn't work for just having one frame or something. That's more like looking at third formant over a turn or something like that.\nSpeaker E: So, but on the other hand, male females is a much simpler categorization than figuring out a factor to squish or expand the spectrum.\nSpeaker E: You could imagine that, just like we're saying, voiced and voiced is good to know. Male female is good to know also.\nSpeaker E: But you have to figure out a way to incorporate it on the fly.\nSpeaker E: I mean, I guess, as you say, one thing you could do is simply have the male and female output vectors, you know, net strain only on males and drain only on females.\nSpeaker E: But I don't know if that would really help because you already have males and females. It's fitting into one net.\nSpeaker G: Is it balanced in terms of gender data?\nSpeaker C: Do you know? Almost.\nSpeaker E: Okay. You were saying before?\nSpeaker C: Yes. So this noise.\nSpeaker C: Yeah, the MSG.\nSpeaker C: There is something perhaps you could spend some days to look at this thing because it seems that when we train networks on, let's say on timid with MSG features, they look as good as network strain on BLP.\nSpeaker C: But when they are used on the speech data, it's not the case. The MSG features are much worse. And so maybe they are more sensitive to different recording conditions.\nSpeaker E: Yeah. But let me ask you this. What's the, do you know, recall of the insertions were higher with MSG?\nSpeaker C: I don't know. I cannot tell. But it's the error rate is higher.\nSpeaker E: Yeah, we should always look at insertions, solutions and substitutions. So MSG is very, very different. And BLP is very much like milk hamstring.\nSpeaker E: MSG is very different from both of them. So if it's very different, then this is the sort of thing. I mean, I'm really glad Andreas brought this point up. I sort of forgotten to discuss it.\nSpeaker E: We always have to look at how these adjustments affect things. And even though we're not allowed to do that, again, we maybe could reflect that back to our use of the features.\nSpeaker E: So if it, if in fact, the problem might be that the range of the MSG features is quite different, the range of the BLP or milk hamstring. And you might want to change that.\nSpeaker C: But yeah, but it's after, well, it's tandem features. So yeah.\nSpeaker C: Yeah, we have estimation of first, post-serials. Yeah, with BLP and with MSG as input. So why not? Well, that means they're between zero and one.\nSpeaker E: But it doesn't necessarily, you know, they could be, doesn't tell you what the variance of the things is.\nSpeaker E: So you're taking log of these things. We could be knowing what the sum of the probabilities are. It doesn't tell you what the sum of the logs are.\nSpeaker C: So, yeah, so we should look at the likelihood. Or what? Well, the log props. Yeah. Or what, you know, what you're the thing you're actually looking at.\nSpeaker G: So your, your values that are actually being fed into HTK. What do they look like? So the, for the tandem system, the values that come out of the net don't go through the sigmide, right? They're sort of the cream on linearity values.\nSpeaker G: Right. So they're kind of like log probabilities. So that's what goes into HTK.\nSpeaker E: Almost. Then you actually do a KLT. They are normalized after that. Are they?\nSpeaker E: No. No. Okay. So, right. So the question is, yeah, whatever they are at that point, are they something for which taking a square root or cube root or four-thread or something like that is going to be a good or a bad thing?\nSpeaker E: So, and that's something that nothing else after that is going to, things are going to scale it. You know, subtract things from it, scale it from it, but nothing will have that same effect.\nSpeaker G: So, anyway, if the log probs that are coming out, whether the MSG are really big, the standard insertion penalties are going to have very little effect compared to, you know, smaller set of log probs.\nSpeaker E: Now, again, you don't really look at that. It's something that, and then it's going through this transformation that's probably pretty close to, whatever the KLT is doing, but it's probably pretty close to what a discrete close-any transformation is doing. But still, it's not going to probably radically change the scale of things.\nSpeaker E: I would think, and yeah, maybe entirely off, and it may be, at least it may be quite different for MSG than it is for milk, up to MPLP. So that would be, so the first thing to look at without adjusting anything would just be to go back to the experiment, look at the substitutions, insertions, and relations.\nSpeaker E: And if there's a fairly large effect of the ratio between insertions and relations for the two cases, then that would be an indicator that might be in that direction.\nSpeaker C: Yeah, but my point was more that it works sometimes. Yeah, but sometimes it doesn't work.\nSpeaker E: And it works on the TID Jits and the speech that Gary doesn't work. Yeah, but some problems are harder than others. And sometimes there's enough evidence for something to work, and then it's harder to break.\nSpeaker E: But it could be that when you say it works, maybe we could be doing much better even if the TID Jits. Yeah, well, there is also the spectroscopy section, which I think maybe we should try to integrate it.\nSpeaker C: Yeah. Right. But I think that would involve to use a big bunch of the system of Ericsson.\nSpeaker C: Because the spectroscopy section then it's followed by other kind of processing that's dependent on the speech on silence.\nSpeaker C: And there is kind of spectral flattening after if it's silence. And I think it's important to reduce this musical noise and this increase of variance during silence portions.\nSpeaker C: So, this would involve to take almost everything from the this proposal and then just add some kind of fun like normalization and the neural network.\nSpeaker E: Okay, well, this would be I think something for discussion with Henik next week. Right. So, how are things going with what you're doing?\nSpeaker F: Well, it took a lot of time just getting my taxes out of the way. I'm starting to write code now from my work, but I don't have any results yet. It would be good for me to talk to Henik, I think when he's here. Do you know what his schedule will be like?\nSpeaker E: He'll be around for three days. Okay, so we'll have a lot of time. So, he'll be talking with everybody in this room.\nSpeaker G: But you said you won't be here next Thursday?\nSpeaker E: Not Thursday and Friday, so it would be a faculty retreat. So, I'll try to connect with him and people as I can on Wednesday.\nSpeaker E: How did taxes go? Next go, okay. Yeah.\nSpeaker E: Yeah, that's one of the big advantages of not making much money. The taxes are easier.\nSpeaker G: Unless you're getting money into countries. I think you're going to want to cut.\nSpeaker E: Can't do what? Can't do what's a cut? You have to do two returns. For 2000, I did. Yeah. Oh, yeah. That's right.\nSpeaker F: I'll still have a bit of Canadian income, but it'll be less complicated because I will not be considered a resident of Canada anymore, so I won't have to declare my American income on my Canadian return.\nSpeaker D: Very? Oh, right. Continuing looking at phonetic events. And it was Tuesday. I've been meeting with John and Halal and Chuck to talk some more about these neck events.\nSpeaker D: I came up with a plan of attack. Oh, well, I don't want to say something about what it is.\nSpeaker D: Okay, well, we're all gathered here together. I hope I can wave my hands. So once I'm thinking of getting a set of acoustic events to be able to distinguish between phones and words and stuff.\nSpeaker D: Once we figure out a set of these events that can be hand labeled or derived from hand labeled phone targets, we can take these events and do some cheating experiments where we feed these events into an SRI system and evaluate its performance on a switchboard task.\nSpeaker D: Can you give an example of an event? Sure. I can give you an example of 20 odd events. So in this paper, I'm talking about funding recognition using acoustic events.\nSpeaker D: So things like vacation or news. Who's paper?\nSpeaker E: From University of Hamburg and Bielfeld.\nSpeaker G: I think there's a difference between acoustic features and acoustic events. And I think of acoustic features as being things that linguists talk about.\nSpeaker G: So stuff that's not based on data. So they talk about features for phones like its height, its tenseness, laxness, things like that, which may or may not be all that easy to measure in the acoustic signal versus an acoustic event, which is just something in the acoustic signal that is fairly easy to measure.\nSpeaker G: So it's a little different.\nSpeaker E: When we did the spam work, we had this notion of an auditory event called an event with an A at the front.\nSpeaker E: And the idea was something that occurred that is important to a bunch of neurons somewhere. So a sudden change or a relatively rapid change in some spectral characteristic will do sort of this.\nSpeaker E: And there's certainly a bunch of places where you know that neurons are going to fire because something novel has happened. That was the main thing that we were focusing on there. But there's certainly other things beyond what we talked about there that aren't just sort of rapid changes.\nSpeaker G: It's kind of like the difference between top down and bottom up. I think of the acoustic, you know, phonetic features as being top down. And you look at the phone and you say this phone is supposed to be, you know, have this feature, this feature in this feature.\nSpeaker G: Whether that those features show up in the acoustic signal is sort of irrelevant. Whereas an acoustic event goes the other way. Here's the signal. Here's some event. And that, you know, that may map to this phone sometimes and sometimes it may not. It just depends on the context and things like that.\nSpeaker D: So, yeah. Okay. Using these events, we could perform these cheating experiments and how good they are in terms of phoneme recognition or work recognition.\nSpeaker D: And then from that point on, I would design robust event detectors in a similar spirit that Saul has done with his graphical models and this probabilistic and or model that he uses.\nSpeaker D: I tried to extend it to account for other phenomena like CMR, co-multulation, release. And maybe also the best of the ways to modify the structure of these models in a data driven way, similar to the way that Jeff, Jeff, builds his work.\nSpeaker D: And while I'm doing these event detectors, you know, I can measure my progress by comparing the error rates in clean and noisy conditions to something like neural nets.\nSpeaker D: And so once we have these event detectors, put them together and feed the outputs of the event detectors into the SRI, HMM system and test it on switchboard or maybe even Aurora stuff.\nSpeaker D: And that's pretty much the big picture of the plan.\nSpeaker E: By the way, there's a couple people who are going to be here. I forget I already told you this, but a couple people who are going to be here for six months. It's Professor Kolmeier from Germany who's quite big in the hearing aid signal processing area.\nSpeaker E: And Michael Klanchman who's worked with him who also looks at auditory properties inspired by various brain function things. So I think they'll be interesting to talk to in this sort of issue is these detectors are developing.\nSpeaker E: He looks at interesting things in different ways of looking at specter in order to get very speech properties out.\nSpeaker E: Okay, well, short meeting with it. Okay. And that's what we're doing. I encourage you to go ahead and meet next week with HINIK.\nSpeaker E: All right, I'll start. Okay, I'm doing transcript L76 032 36 5550 7058 592 4657 17 8034 6015 544 445 088 3666 601726 971 235 1588 4821 8042 3770 528 578 8674\nSpeaker C: Transcript L-77 5 845 474 163 130 287 452 121 161422 3891 4838 1740 651 8676 293 3132 6134 24\nSpeaker G: 0243 214 1337 001950 7956 transcript L-78 1543589276 449 746 646 787 337 618 2 689 033 313 375 175 536 1141 3007 510 782 461 64382 2503\nSpeaker D: transcript L-79 885 2526 17 445 191 288 4 13168 4 3004 668 3 9 8 6 4 8 6 6 6 7 5 0 3 1 4 1 7 0 8 5 9 5 0 3 7 6 0 2 6 0 8 2 3 0 7 1 9 8 7 5 7 4 2 8 8 3 7 1 6 8\nSpeaker F: transcript L-80 9 5 6 6 4 3 9 7 8 3 0 2 6 4 3 6 1 2 8 9 3 3 4 4 0 5 7 9 8 1 3 9 8 8 8 0 1 0 2 0 9 9 9 5 8 9 8 1 8 9 5 4 8 7 9 6 1 8 7 8 8 3 0 9 6 7 6 2 9 0 5 7 5 6 0 7 2 9 7\nSpeaker A: transcript L-81 7915 908 2 16168 6 4013 4 405 5 6 4 2 1 9 4 2 1 0 5 1 2 7 2 1 2 9 5 8 8 6 3 9 2 5 8 4 3 6 0 5 3 0 2 3 3 2 6 2 4 3 6 5 2 6 2 4 2 8 5 1 4 5\n"}, {"meeting_id": "Bro017", "summary": "The meeting focused on the Aurora project. The Professor began with some updates on how certain decisions made by those leading the project did not make sense. There was some confusion and debate about how to move forward. The team then moved onto a discussion about measuring the effect of a given feature on the model. The professor outlined the specific methodology. At the time, the team was focused on testing for C-zero and C-one only, but they decided to expand their testing. Finally, the professor told some stories to explain future directions and the team had a brief discussion about their hardware.", "dialogue": "Speaker D: is it starting now?\nSpeaker D: Yeah.\nSpeaker D: So from what whatever we say from now on, it can be heard against us, right?\nSpeaker D: That's right.\nSpeaker G: And you're right to remain silent.\nSpeaker D: So the problem is that I actually don't know how these heard mean things are heard.\nSpeaker D: They are very informal.\nSpeaker D: So there's just people that say what's going on and that's usually what we do.\nSpeaker E: We just sort of go around and people say what's going on.\nSpeaker D: What's going on?\nSpeaker D: Okay.\nSpeaker D: The reason it is if I make a report on what's happening in Aurora in general, at least from my perspective.\nSpeaker D: That would be great.\nSpeaker D: And so I think that Carmen and Stefan reported on Amsterdam meeting, which was kind of interesting because for the first time we realized we are not friends really, but we are competitors.\nSpeaker D: Until then it was sort of like everything was like wonderful.\nSpeaker D: Yeah.\nSpeaker E: It seemed like there were still some issues that they were trying to decide.\nSpeaker D: There's plenty of issues.\nSpeaker D: Well, what happened was that they realized that two leading proposals, which was French Telecom, Alcatel and us, both had voice activity that extra.\nSpeaker D: Right.\nSpeaker D: And I saw a big surprise.\nSpeaker D: I mean, we could have told you that four months ago, except we didn't because nobody else was bringing it up.\nSpeaker D: Obviously, French Telecom didn't volunteer this information either.\nSpeaker D: Because we were working mainly on voice activity that extra four past several months because that's why he got the most thing.\nSpeaker D: And everybody said, well, this is not fair.\nSpeaker D: We didn't know that.\nSpeaker D: And of course, it's not working on features, really.\nSpeaker D: And I agreed.\nSpeaker D: I said, well, yeah, you are absolutely right.\nSpeaker D: I mean, if you wish that you provided a better and pointed speech because, or at least that if you could modify the recognizer to account for these long silences, because otherwise, that wasn't the correct thing.\nSpeaker D: And so then everybody asked us, well, we need to do a new evaluation.\nSpeaker D: We don't want to have to do that.\nSpeaker D: Or we have to do something about it.\nSpeaker D: And in principle, we agreed.\nSpeaker D: We said, yeah, because about in that case, we would like to change the algorithm because if we are working on different data, we probably will use a different set of tricks.\nSpeaker D: That unfortunately, nobody ever officially can somehow acknowledge that this can be done.\nSpeaker D: Because French Telecom was saying, no, no, no, no.\nSpeaker D: Now everybody has access to our code.\nSpeaker D: So everybody is going to copy what we did.\nSpeaker D: Well, our argument was everybody has access to our code.\nSpeaker D: And everybody always had access to our code.\nSpeaker D: We never deny that we saw that people are honest that if you copy something and if it is protected by a patent, then you negotiate or something, right?\nSpeaker D: I mean, if you find our technique useful, we are very happy.\nSpeaker D: But French Telecom was saying, no, no, no, there is a lot of little tricks which sort of I cannot be protected and you guys will take them.\nSpeaker D: Which probably is also true.\nSpeaker D: I mean, it might be that people will take the algorithms apart and use the blocks from that.\nSpeaker D: But I somehow think that it wouldn't be so bad as long as people are happy about it.\nSpeaker D: Honest about it.\nSpeaker D: And I think they have to be honest in the long run because winning the proposal again, what will be available will be a code.\nSpeaker D: So the people can go to code and say, well, listen, this is what you stole from me.\nSpeaker D: Right.\nSpeaker D: So let's deal with that.\nSpeaker D: So I don't see the problem.\nSpeaker D: The biggest problem, of course, is that our country, the French Telecom claims what we fulfilled the conditions we are the best, the standard.\nSpeaker D: And other people don't feel that because they now decide that the whole thing will be done on a well-appointed data.\nSpeaker D: Maybe that somebody will point the data based on clean speech because most of this speech that car has also closed-picking in Mike and points will be provided.\nSpeaker D: And we will run again.\nSpeaker D: Still not clear if we are going to run, if we are allowed to run new algorithms, but I assume so because we would fight for that really.\nSpeaker D: But since at least our experience is that only end-pointing Melkeps Room gets you 21 percent improvement overall and 27 improvement on speech.car.\nSpeaker D: Then obviously database, that means the baseline will go up.\nSpeaker D: And nobody can achieve 50 percent improvement.\nSpeaker D: So they agreed that there will be a 25 percent improvement required on the bad-limit message.\nSpeaker E: I thought the end-pointing really only helped in the noisy cases.\nSpeaker E: But you still have that with the MFCC, okay?\nSpeaker D: Yeah, but you have the same problem.\nSpeaker D: MFCC basically has an enormous number of insertions.\nSpeaker D: So now they want to say we will require 50 percent improvement only for well-matched condition and only 25 percent for the severe cases.\nSpeaker D: And they almost agreed on that except that it wasn't 100 percent agreed.\nSpeaker D: And so last time during the meeting I just brought up the issue.\nSpeaker D: So well, you know, quite frankly I've surprised how likely you are making these decisions because this is a major decision.\nSpeaker D: For two years we are fighting for 50 percent improvement.\nSpeaker D: And suddenly you are saying, oh no, we will do something.\nSpeaker D: Maybe we should discuss that.\nSpeaker D: And everybody's all be discussed that and you were not there.\nSpeaker D: And I saw a lot of other people there because not everybody participates at this teleconferencing thing.\nSpeaker D: Then they said, oh no, no, no, because everybody is invited.\nSpeaker D: However, there is only 10 or 15 lines.\nSpeaker D: So people can't even participate.\nSpeaker D: So they agreed.\nSpeaker D: And I saw they said, okay, we will discuss that.\nSpeaker D: Immediately Nokia raised the question.\nSpeaker D: I said, oh yeah, we agreed.\nSpeaker D: This is not good to dissolve the criteria.\nSpeaker D: So now officially Nokia is complaining and said they are looking for support.\nSpeaker D: I think Qualcomm is saying too, we shouldn't abandon the 50 percent yet.\nSpeaker D: We should at least try once again, one more round.\nSpeaker D: So this is where we are.\nSpeaker D: I hope that this is going to be adopted next Wednesday.\nSpeaker D: We are going to have another teleconferencing call.\nSpeaker D: So we will see where it goes.\nSpeaker E: So what about the issue of the weights for the different systems?\nSpeaker E: The well matched and medium ones?\nSpeaker D: Yeah, that's a very good point.\nSpeaker D: David says, well, we can manipulate this number by choosing the right weights anyways.\nSpeaker D: So you are right.\nSpeaker D: But of course, if you put a weight zero on the mismatch condition, highly mismatch, then you are done.\nSpeaker D: Advates were also already decided half a year ago.\nSpeaker D: And they are saying the same.\nSpeaker D: Of course people will not like it.\nSpeaker D: What is happening now is that I think that people try to match the criterion to solution.\nSpeaker D: They have solution.\nSpeaker D: Now they want to make sure the criterion is.\nSpeaker D: And I think that this is not the right way.\nSpeaker D: It may be that eventually it may have to happen.\nSpeaker D: But it should happen at the point where everybody feels comfortable that we did all of what we could.\nSpeaker D: And I don't think we did it, basically.\nSpeaker D: I think that this test was a little bit bogus because of the data.\nSpeaker D: And essentially there were some arbitrary decisions made in everything.\nSpeaker D: So this is where it is.\nSpeaker D: So what we are doing at OGI now is working basically on the parts which we, I think, a little bit neglected.\nSpeaker D: Like a noise separation.\nSpeaker D: So we are looking in a way which we can provide the better initial estimate of the male spectrum, basically, which would be more robust to noise.\nSpeaker D: And so far not much success.\nSpeaker D: We tried things which long time ago Bill Burns suggested instead of using Fourier spectrum from Fourier transform, used the spectrum from LPC model.\nSpeaker D: The argument there was, the LPC model feels the peaks of the spectrum.\nSpeaker D: So it may be naturally more robust in noise.\nSpeaker D: And I saw that, that makes sense.\nSpeaker D: But so far we can't get much out of it.\nSpeaker D: We may try some standard techniques like spectral subtraction.\nSpeaker D: You haven't tried that.\nSpeaker D: No, not much.\nSpeaker D: Or even I was thinking about looking back into this totally ad hoc techniques.\nSpeaker D: For instance, Dennis Scott was suggesting the one way to deal with noise speeches, add noise to everything.\nSpeaker D: So I mean, add more than amount of noise to all data.\nSpeaker D: So that makes any additive noise less effective, right?\nSpeaker D: Because you already had the noise and it was working at the time.\nSpeaker D: It was kind of like one of these things.\nSpeaker D: But if you think about it, it's actually pretty ingenious.\nSpeaker D: So one, you know, just take a spectrum and add to constant C to every value.\nSpeaker D: Yeah.\nSpeaker E: So you're making all your training data more uniform.\nSpeaker D: Exactly.\nSpeaker D: And if the new test data becomes noisy, it becomes effective because less noisy, basically.\nSpeaker D: But of course you cannot add too much noise because then your green recognition goes down.\nSpeaker D: But I mean, it's yet to be seen how much it's very simple technique.\nSpeaker D: It's indeed.\nSpeaker D: It's very simple technique.\nSpeaker D: You just take your spectrum and use whatever is coming from FFT at constant, you know, to power spectrum.\nSpeaker D: Or the other thing is of course, if you have a spectrum where you can start doing, you can start leaving out the parts which are law in energy.\nSpeaker D: And then perhaps one could try to find a whole model to such a spectrum because our whole model will still try to put the continuation basically of the model into these parts where which you said to zero.\nSpeaker D: So that's what we want to try.\nSpeaker D: I have visited from Bernou, kind of like young faculty, really hardworking.\nSpeaker D: So he's looking into that.\nSpeaker D: And then most of the effort is now also aimed at this trap recognition.\nSpeaker D: This is this recognition from temporal patterns.\nSpeaker D: What is that?\nSpeaker D: Ah, you don't know about traps.\nSpeaker E: The traps sound familiar, right?\nSpeaker D: Yeah, I mean, this is familiar, because we gave it a name.\nSpeaker D: But what it is is that normally what you do is that you recognize speech based on short spectrum.\nSpeaker D: Essentially, LPC, Melcappes' room, everything starts with spectral slice.\nSpeaker D: So if you're given the spectrogram, you're essentially sliding the spectrogram along the frequency axis and you keep shifting the thing.\nSpeaker D: But you have a spectrogram.\nSpeaker D: So you can say, well, you can also take the time trajectory of the energy at a given frequency.\nSpeaker D: And what you get is then you get a vector.\nSpeaker D: And this vector can be assigned to some phoneme.\nSpeaker D: Namely, you can say, I will say that this vector will describe the phoneme which is in the center of the vector.\nSpeaker D: And you can try to classify based on that.\nSpeaker D: So you can say, I'm sorry, it's a very different vector, very different properties.\nSpeaker D: We don't know much about it.\nSpeaker D: But the truth is you have many of those vectors.\nSpeaker D: Wow.\nSpeaker D: So you get many decisions.\nSpeaker D: And then you can start thinking about how to combine these decisions.\nSpeaker D: So exactly that's what it is.\nSpeaker D: Because if you run this recognition, you still get about 20% error.\nSpeaker D: That's 20% correct.\nSpeaker D: You know, on like frame by frame basis.\nSpeaker D: So it's much better than chance.\nSpeaker D: How wide are the frequency?\nSpeaker D: That's another thing.\nSpeaker D: Well, currently we start always with critical vent spectrum for various reasons.\nSpeaker D: But the latest observation is that you can get quite a big advantage of using two critical vents at the same time.\nSpeaker G: Are they adjacent?\nSpeaker D: Adjacent.\nSpeaker D: Adjacent.\nSpeaker D: And there are some reasons for that.\nSpeaker D: Because there are some reasons I could talk about.\nSpeaker D: We have to tell you about things like masking experiments, which yield critical vents, and also experiments with the release of masking, which actually tell you that something is happening across critical vents, across bands.\nSpeaker E: And how do you convert this energy over time in a particular frequency band into a vector of numbers?\nSpeaker D: I mean, a time T0 is one number.\nSpeaker D: Yeah, but what time number is it just to see a spectacular energy?\nSpeaker D: A logarithmic spectrum of energy in that band.\nSpeaker D: Yes, in that time interval.\nSpeaker D: Yes.\nSpeaker D: And that's what I'm saying then.\nSpeaker D: This is a starting vector.\nSpeaker D: It's just like short-term spectrum or something.\nSpeaker D: But now we are trying to understand what this vector actually represents.\nSpeaker D: For instance, question is like how correlated are the elements of this vector?\nSpeaker D: And so they are quite correlated, because I mean, especially the neighboring ones, right?\nSpeaker D: They represent the same, almost the same configuration of the vocal tract.\nSpeaker D: So there is a very high correlation.\nSpeaker D: So the classifiers which use the diagonal covariance matrix don't like it.\nSpeaker D: So we think they are decorulating them.\nSpeaker D: And the question is, can you describe elements of this vector by Gaussian distributions?\nSpeaker D: To what extent?\nSpeaker D: Because and so on and so on.\nSpeaker D: So we are learning quite a lot about that.\nSpeaker D: And then another issue is how many vectors we should be using.\nSpeaker D: I mean, the minimum is one.\nSpeaker D: But I mean, it's the critical band, the right dimension.\nSpeaker D: So we somehow made arbitrary decision, yes.\nSpeaker D: And then now we are thinking a lot how to use at least a neighboring band, because that seems to be happening.\nSpeaker D: This I somehow start to believe that's what's happening in a cognition.\nSpeaker D: So a lot of experiments point to the fact that people can split the signal into critical bands.\nSpeaker D: But then so you can, you are quite capable of processing a signal independently in individual critical bands.\nSpeaker D: And the question is, what is the question?\nSpeaker D: What is the most important thing that you are making in a particular massaging experiment tell you?\nSpeaker D: But at the same time, you most likely pay attention to at least neighboring bands when you are making any decisions.\nSpeaker D: You compare what's happening in this band to what's happening to the band to the neighboring bands.\nSpeaker D: And that's how you make a decision.\nSpeaker D: That's why the articulatory bands which have pleasure talks about, they are about two critical bands.\nSpeaker D: You need at least two, basically.\nSpeaker D: You need some relative relation.\nSpeaker D: Absolute number doesn't tell you the right thing.\nSpeaker D: You need to compare it to something else.\nSpeaker D: But it's what's happening in a closed neighborhood.\nSpeaker D: So if you are making a decision what's happening at one kilohertz, you want to know what's happening at 900 hertz.\nSpeaker D: And maybe at 1100 hertz.\nSpeaker D: But you don't much care what's happening at three kilohertz.\nSpeaker E: So it's really, it's sort of like saying that what's happening at one kilohertz depends on what's happening around it.\nSpeaker E: It's sort of relative.\nSpeaker D: To some extent it's also true.\nSpeaker D: But for instance, what humans are very much capable of doing is that if they are exactly the same thing happening in two neighboring critical bands, the cognition can discard it.\nSpeaker D: This what's happening.\nSpeaker D: Hey, okay, we need another voice here.\nSpeaker D: Yeah, I think so.\nSpeaker D: So, so for instance, if you if you if you add the noise, then normally masks the signal.\nSpeaker D: And you can show that if you add the noise outside the critical band that doesn't affect the decisions you making about the signal within a critical band.\nSpeaker D: Unless this noise is modulated.\nSpeaker D: But the noise is modulated with the same modulation frequency as the noise in a critical band.\nSpeaker D: The amount of masking is less.\nSpeaker D: The moment you moment you provide the noise in neighboring critical bands.\nSpeaker D: So the masking can normally it looks like sort of I start from from here.\nSpeaker D: So you have no noise.\nSpeaker D: Then you you are expanding the critical band.\nSpeaker D: So the amount of marking is increasing.\nSpeaker D: And when you hit certain point which is critical band, then the amount of masking is the same.\nSpeaker D: So that's the famous experiment of lecture a long time ago.\nSpeaker D: That's where people start thinking, wow, this is interesting.\nSpeaker D: So but if you if you if you modulate the noise, the masking goes up and the moment you start hitting the another critical band, the masking goes down.\nSpeaker D: So essentially that's a very clear indication that that that cognition can take into consideration was happening in the neighboring bands.\nSpeaker D: But if you go too far, you know, if you if the noise is very broad, you are not increasing much more.\nSpeaker D: So if you if you are far away from the signal, the frequency which the signal is, then even when the noise is commodulated, it's not helping you much.\nSpeaker D: So so things like this we are kind of playing with and with with the hope that perhaps we could eventually use this in a in a real recognizer.\nSpeaker D: Like partially, of course, we promise to do this on the the roar.\nSpeaker E: Probably won't have anything before the next time we have to evaluate.\nSpeaker D: Probably not.\nSpeaker D: And the most likely we will not have anything which which would comply with the rules.\nSpeaker D: Like because latency currently traps require significant latency.\nSpeaker D: A amount of processing because we don't know any better yet than to use the neural and classifiers in the traps.\nSpeaker D: Though the work which by is looking at now aims at trying to find out what to do with this vector so that a simple Gaussian classifier would be happier with it.\nSpeaker D: Or to what extent the Gaussian classifier should be unhappy and how to gocheneize the vector center.\nSpeaker D: So this is so what's happening.\nSpeaker D: And Sunil asked me for one month's vacation and since he didn't take any vacation for two years I had no I didn't have a heart to tell him no.\nSpeaker D: So he is in India.\nSpeaker D: Wow.\nSpeaker D: And getting married or something.\nSpeaker D: Well, he may be looking for a guy.\nSpeaker D: I don't I don't ask.\nSpeaker D: I know that when last time Narend did that he came back engaged.\nSpeaker D: Right.\nSpeaker D: I mean I've known other friends who I know.\nSpeaker E: They go back home in India for a month to come back married.\nSpeaker D: I know.\nSpeaker D: I know.\nSpeaker D: And then of course then what happened with Narend was that he started pushing me that he needs to get a PhD because they wouldn't give him his wife.\nSpeaker D: And she's very pretty and he loves her.\nSpeaker E: So we had to really finally had some incentive.\nSpeaker D: Oh yeah.\nSpeaker D: Well I had an incentive because he always had this plan except he never told me.\nSpeaker D: Sort of figured that he told me the day when we did very well at this evaluation of speaker recognition technology and he was involved there.\nSpeaker D: We were after presentation, we were driving home and he told me.\nSpeaker D: When did you were happy?\nSpeaker D: Yeah so I said well okay so he took another three quarter of the year but he was out.\nSpeaker D: So I wouldn't surprise me if he has a plan like that though though probably but still needs to get out first.\nSpeaker D: Good advice there a year earlier and such he needs to get out very first because he already has a four years served.\nSpeaker D: One year he was getting masters.\nSpeaker E: So when is the next evaluation?\nSpeaker E: June or something?\nSpeaker D: Which speaker recognition?\nSpeaker D: No for Aurora.\nSpeaker D: There have been no evaluation.\nSpeaker D: Next meeting is in June.\nSpeaker D: But like getting together are people supposed to rerun their system?\nSpeaker D: Nobody said that yet I assume so but nobody even said up yet the date for delivering and pointed data.\nSpeaker D: Wow.\nSpeaker D: That sort of stuff.\nSpeaker D: But what I think would be of course extremely useful if we can come to an next meeting and say well you know we did get 50% improvement.\nSpeaker D: If you are interested we eventually can tell you how but we can get 50% improvement because people will be saying it's impossible.\nSpeaker E: Do you know what the new baseline is?\nSpeaker D: I guess 22% better than old baseline.\nSpeaker E: Using your voice acting?\nSpeaker D: Yes.\nSpeaker D: But I assume that it will be similar.\nSpeaker D: I don't see the reason why it shouldn't be.\nSpeaker D: I don't see reason why it should be worse.\nSpeaker D: If it is worse then we will raise the objection and say well you know how come because we just use our voice activity detector which we don't claim even that it's wonderful.\nSpeaker D: It's just like one of them.\nSpeaker D: We get this sort of improvement how come that we don't see it on the other end pointed.\nSpeaker C: I guess it could be even better because the voice activity detector that they choose is something that cheating.\nSpeaker C: It's using the alignment of the spiritual cognition system.\nSpeaker C: Only the alignment from the King channel.\nSpeaker D: David told me yesterday or Harry, he told Harry from Qualcomm and Harry brought up his suggestion we should still go for 50%.\nSpeaker D: He says are you aware that your system does only 30% comparing to the baseline.\nSpeaker D: So they must have run already something.\nSpeaker D: So and Harry said yeah but I mean we think that we didn't say the last word yet that we have other things which we can try.\nSpeaker D: So there's a lot of discussion now about this new criterion because Nokia was objecting with Qualcomm's we basically supported that.\nSpeaker D: He said yes.\nSpeaker D: Now everybody else is saying well you guys must be out of your mind.\nSpeaker D: The Ginta here who doesn't speak for Ericsson anymore because he is not with Ericsson and Ericsson may not maybe draw from horror activity because they have so many troubles now.\nSpeaker D: Ericsson is laying off 20% of people.\nSpeaker D: Where's Ginta?\nSpeaker D: Ginta is already he got the job or he was working only for past two years or three years.\nSpeaker D: He got the job some factual.\nSpeaker D: Technical college not too far from Archen.\nSpeaker D: So it's like university professor and not quite a university not quite a sort of Archen university but it's a good school and he's happy.\nSpeaker D: And he was hoping to work with Ericsson like on a consulting basis but right now he doesn't look like that.\nSpeaker D: Anybody is even thinking about speech recognition.\nSpeaker D: He's thinking about survival.\nSpeaker D: So this is being discussed right now and it's possible that it may get through that we will still stick to 50%.\nSpeaker D: That means that nobody will probably get this improvement with the current system.\nSpeaker D: Which essentially I think we should be happy with because they would mean that at least people may be forced to look into alternative solutions.\nSpeaker C: But maybe we're not too far from 50% from the new deadline.\nSpeaker C: But not 60% over the current baseline.\nSpeaker D: We're getting there.\nSpeaker F: We are on 50, 55.\nSpeaker D: Is it like how did you come up with this number if you improved by 20% of all the deadlines?\nSpeaker D: It's just a kind of quick competition.\nSpeaker C: I don't know exactly if it depends on the weightings.\nSpeaker E: How's your documentation or what was you guys working on last week?\nSpeaker C: Finally we've not finished with this.\nSpeaker B: Yes, I do read another more time to improve the English and maybe to finish something in a small detail.\nSpeaker B: Something like that but it's more or less ready.\nSpeaker C: Yes.\nSpeaker C: What do we have to do to explain?\nSpeaker C: To include the experiments.\nSpeaker E: Have you been running some new experiments?\nSpeaker E: I thought I saw some jobs of yours running on it.\nSpeaker C: Yes, right.\nSpeaker C: We've done some strange things like removing C0 or C1 from the factor of parameters.\nSpeaker C: We noticed that C1 is almost not useful at all.\nSpeaker C: You can remove it from the factor.\nSpeaker C: It doesn't hurt.\nSpeaker C: Is this in the baseline?\nSpeaker C: In the nozzle.\nSpeaker D: So we were just discussing, since you mentioned that, driving in the car with Morgan this morning, we were discussing a good experiment for beginning graduate student who wants to get a lot of numbers from something.\nSpeaker D: Which is, imagine that you will start putting any coefficient which you are using in your vector in some general power.\nSpeaker D: General power.\nSpeaker D: Sort of you take a square root of something.\nSpeaker D: So suppose that you are working with C1.\nSpeaker D: So if you put it in a square root that effectively makes your model half as efficient.\nSpeaker D: Because a Gaussian mixture model, right, computes the mean.\nSpeaker D: But the mean is an exponent of the, whatever, your compression range.\nSpeaker D: So you compress in the range of this coefficient.\nSpeaker D: So it's becoming less efficient.\nSpeaker D: So Morgan was initially saying, well, this might be the alternative way how to play with a fetch factor.\nSpeaker D: You know, just compress the whole vector.\nSpeaker D: And I said, well, in that case, why don't we just start compressing individual elements like when, because in all days, we were doing, when people still were doing template matching and, you know, the distances, we were doing this liftering of parameters, right?\nSpeaker D: Because we observed that higher parameters were more important than lower for recognition.\nSpeaker D: Right.\nSpeaker D: Basically, that C1 contributes mainly slope and it's highly affected by frequency response of the recording equipment and that sort of thing.\nSpeaker D: So we were coming with all these various lifters.\nSpeaker D: Bell ups had this raised cosine lifter which still I think is built into a HTK.\nSpeaker D: All of the reasons unknown to everybody, but we had exponentially if there are triangle or if there is a number of lifters.\nSpeaker D: And so they may be a way to, to fill it with the insertion, insertion, as well as the deletions or giving a relative, physically modifying, rather important of various parameters.\nSpeaker D: The only, of course, problem is that there is infinite number of combinations.\nSpeaker D: You need a lot of graduate students and a lot of computing power.\nSpeaker E: Genetic algorithm basically tries to write them.\nSpeaker D: Exactly.\nSpeaker D: Oh.\nSpeaker D: If you were Bell ups or I shouldn't be saying this on a mic, right?\nSpeaker D: Or I'd be, that's what maybe is for somebody who would be doing.\nSpeaker D: I mean, the places which have a lot of computing power.\nSpeaker D: Because it is really, it's, it will be reasonable search.\nSpeaker D: But I wonder if there is in some way of doing this.\nSpeaker D: Search, like when we are searching, say for best discriminants.\nSpeaker E: You know, actually, I don't know that this wouldn't be all that bad.\nSpeaker E: I mean, you compute the features once, right?\nSpeaker E: And then, these exponents are just applied.\nSpeaker D: And everything is fixed.\nSpeaker D: Everything is fixed.\nSpeaker E: Is something that you would adjust for training or only recognition?\nSpeaker E: For both.\nSpeaker E: You would have to do it on both.\nSpeaker D: Yes.\nSpeaker D: You have to do both, both.\nSpeaker D: Because essentially you are saying this feature is not important.\nSpeaker D: Or less important.\nSpeaker D: So that's a painful one, yeah.\nSpeaker E: So for each set of exponents that you would try to repair a training.\nSpeaker D: But wait a minute.\nSpeaker D: You may not need to retrain the model.\nSpeaker D: You just may need to give less weight to a component of the model, which represents this particular feature.\nSpeaker D: You don't have to retrain it.\nSpeaker E: So if you, instead of altering the feature vectors.\nSpeaker D: You just multiply.\nSpeaker E: You modify the Gaussian.\nSpeaker D: You modify the Gaussian in a model.\nSpeaker D: But in a test data, you would have to put it in the power.\nSpeaker D: But in a training, in a training model, all you would have to do is to multiply.\nSpeaker D: And I, a model, I appropriate a constant.\nSpeaker E: But why, if you're altering the model, why would, in the test data, why would you have to mark with the capture footage?\nSpeaker D: Because in the test data, you can't, don't have a model.\nSpeaker D: You have only data.\nSpeaker E: I think you're running your data through that same model.\nSpeaker D: That is true.\nSpeaker D: I mean, so what do you want to do?\nSpeaker D: You want to say if, if you observe something like Stefan observes, that C1 is not important.\nSpeaker D: You can do two things.\nSpeaker D: If you have a trained recognizer in the model, you know the component, which, I mean, the dimension.\nSpeaker E: All of the mean and variances that correspond to Stefan's.\nSpeaker D: You know it.\nSpeaker D: But what I'm proposing now, if it is important, but not as important, you multiply it by 0.1 in a model.\nSpeaker D: But what do you multiply?\nSpeaker E: Because those are means, right?\nSpeaker G: You multiply in a standard deviation.\nSpeaker D: I think that you multiplied, I would have to look in the math.\nSpeaker D: I mean, how does the, yeah.\nSpeaker E: I think you'd have to modify the standard deviation or something.\nSpeaker D: Yeah.\nSpeaker D: Effectively, that's what you do.\nSpeaker D: That's what you do.\nSpeaker D: You modify the standard deviation as it was trained.\nSpeaker D: Effectively, you know, in front of the model, you put a constant.\nSpeaker D: Yeah, effectively what you do is you are modifying the deviation.\nSpeaker D: Right?\nSpeaker D: Spread, right?\nSpeaker D: Yeah, spread.\nSpeaker G: So same, same mean.\nSpeaker E: And, and, and, and, so by making the standard deviation narrower, your scores get worse for, unless it's exactly right on the mean.\nSpeaker D: Yes, no, by making it narrower, you're allowing all as variance.\nSpeaker D: Yes, so you're making this particular dimension less important.\nSpeaker D: Because what you have to do, fitting is the multi-dimensional machine, right?\nSpeaker D: It's a, it has a set in nine dimensions.\nSpeaker D: Or, set in dimensions, if you ignore delts, doesn't double delts.\nSpeaker D: So in order, if you, in order to make dimension which, which step and sees less important, I mean, not useful, less important, what you do is that this particular component in a model, you can multiply by, you can, you can basically, deweight it in a model.\nSpeaker D: But you can't do it in a, in a test data, because you don't have a model for, I mean, the test comes, but what you can do is that you put this particular component in, in, you, you compress it, that becomes a, it gets less variance, subsequently becomes less important.\nSpeaker E: You just do that to the test data and not do anything with your training.\nSpeaker D: That would be very bad, because your model was trained expecting, that wouldn't work.\nSpeaker D: Because your model was trained expecting the certain variance, variance on C1.\nSpeaker D: And because some other things, C1 is important.\nSpeaker D: After you trained the model, you sort of, you could do, you could do still what I was proposing initially, that during the training, you, you compress C1.\nSpeaker D: That becomes, then it becomes less important in a training.\nSpeaker D: And if you have, if you want to run an extensive experiment without retraining the model, you don't have to retrain the model.\nSpeaker D: You train it on the original vector.\nSpeaker D: But after you, when you are doing this parametric study of importance of C1, you will deweight C1 component in the model, and you will put in, it will compress the, this component in a test data.\nSpeaker E: Could you also, if you wanted to, by the same amount?\nSpeaker E: If you wanted to try an experiment, by leaving out, say, C1, couldn't you, in your test data, modify the, all of the C1 values to be, um, way outside of the normal range of the Gaussian, or C1 that was trained in the model, so that effectively, the C1 never really contributes to the score.\nSpeaker D: No, that would be C1 mismatch, right?\nSpeaker D: What do you have?\nSpeaker D: Yeah.\nSpeaker D: No, you don't do one that, because that would, that your model would be unlikely, your likelihood would be low, right?\nSpeaker D: Because you would be providing C1 mismatch.\nSpeaker E: But what if you set it to the mean of the model then?\nSpeaker E: And it was a, you set all C1s coming in through your test data, you, you change whatever value was there to the mean that your model had.\nSpeaker D: No, that would be very good match, right?\nSpeaker D: Yeah.\nSpeaker D: But you have several means, so.\nSpeaker D: I see what you are saying, but, no, no, I don't think that it would be the same, let me, you set it to a mean that would, no, you can't do that.\nSpeaker D: That's true, right?\nSpeaker D: You can't, you can't, Chuck, you can't do that.\nSpeaker D: Yeah.\nSpeaker D: Because that would be a really, fiddling with the data, you can't do that.\nSpeaker D: But what you can do, I'm confident, well, I'm reasonably confident and I'm putting in on the record, right?\nSpeaker D: I mean, people will listen to it for centuries now.\nSpeaker D: Yes, what you can do is you train the model with the, with the original data.\nSpeaker D: Then you decide that you want to see how important C, C1 is.\nSpeaker D: So what you will do is that the component in the model for C1, you will divide it by two and you will compare your test data by square root.\nSpeaker D: Then you will still have a perfect match except that this component of C1 will be half as important in an overall score.\nSpeaker D: Then you divide it by four and you take a square root.\nSpeaker D: Then if you think that some component is more important than it is based on training, then you multiply this particular component in the model by, by, yeah.\nSpeaker D: You multiply this component by number larger than one and you put your data in the power higher than one.\nSpeaker D: Then it becomes more important in the overall score, I believe.\nSpeaker D: But you have to do something to the mean also?\nSpeaker C: No, no.\nSpeaker C: But I think it's the variance is on the denominator in the Gaussian equation.\nSpeaker C: So I think it's maybe it's the contrary.\nSpeaker C: If you want to decrease the importance of parameters, you have to increase its variance.\nSpeaker D: Exactly.\nSpeaker D: So you may want to do it the other way around.\nSpeaker E: But if your original data for C1 had a mean of two and now you're changing that by squaring it, now your mean of your C1 original data has is four.\nSpeaker E: But your model still has a mean of two.\nSpeaker E: So even though you've expanded the range, your mean doesn't match anymore.\nSpeaker C: You're going to see.\nSpeaker C: I think what I see, what could be done is you don't change your features, which are computed once for all.\nSpeaker C: But you just tune the model.\nSpeaker C: So you have your features, you train your model on these features.\nSpeaker C: And then if you want to decrease the importance of C1, you just take the variance of the C1 component in the model and increase it if you want to decrease the importance of C1 or\nSpeaker D: increase it. You would have to modify the mean in the model.\nSpeaker D: I agree with you.\nSpeaker D: But it's doable, right?\nSpeaker D: It's predictable.\nSpeaker D: It's predictable.\nSpeaker D: Yeah, yeah, yeah, it's predictable.\nSpeaker E: But there's a simple thing, you could just just adjust the variance to get the effect I think that you're talking about.\nSpeaker E: It might be.\nSpeaker E: Could increase the variance to decrease the importance?\nSpeaker E: Yeah, because if you had a huge variance, it becomes a large number of you in a very small contribution.\nSpeaker E: Yeah.\nSpeaker G: The sharper the variance, the more important.\nSpeaker E: Yeah, you know, actually this reminds me of something that happened.\nSpeaker E: When I was at VBN, we were playing with putting pitch into the Mandarin recognizer.\nSpeaker E: And this particular pitch algorithm, when it didn't think there was any voicing, was spitting out zeros.\nSpeaker E: So we were getting, when we did clustering, we were getting groups of new OLA, with the mean of zero and basically zero variance.\nSpeaker E: So when any time any one of those vectors came in the head of zero and we got a great score, I mean it was just, you know, incredibly high score.\nSpeaker E: And so that was thrown everything off.\nSpeaker E: So if you have very small variance, you get really good scores when you get something that matches.\nSpeaker E: So that's a way, yeah.\nSpeaker E: Yeah, that's a way to increase the variance.\nSpeaker E: Yeah, that's interesting.\nSpeaker E: So in fact, that would be, that doesn't require any retraining.\nSpeaker E: Yeah, that's right.\nSpeaker C: So that means it's just, it's using the mother's and this thing.\nSpeaker E: Yeah.\nSpeaker E: So that would be, you modify the models, you could copy of your models with whatever variance modifications you make and rerun.\nSpeaker E: And then do a whole bunch of those.\nSpeaker E: That could be set up fairly easily, I think.\nSpeaker E: And you have a whole bunch of, you know, Chuck is getting his head in trouble.\nSpeaker E: That's an interesting idea, actually.\nSpeaker E: For testing me.\nSpeaker E: Yeah.\nSpeaker G: Did you say you got these HTKs set up on the new Linux box?\nSpeaker E: That's right.\nSpeaker E: In fact, and they're just, right now they're installing, increasing the memory on that.\nSpeaker D: And Chuck is sort of really fishing for how to keep his computer busy.\nSpeaker D: Yeah.\nSpeaker D: Absent.\nSpeaker D: You know, that's five processors on that.\nSpeaker D: Yeah, that's a good thing because then you just write the do loops and you pretend like you are working while you are sort of, you can go fishing.\nSpeaker D: Yeah.\nSpeaker D: You have an encyclopedia?\nSpeaker D: Yeah.\nSpeaker D: You are sort of in this mode like all of the DARPA people.\nSpeaker D: I've seen cities on the record.\nSpeaker D: I can say which company it was.\nSpeaker D: But it was reported to me that somebody visited a company and doing a discussion.\nSpeaker D: There was this guy who was always hitting the college returns on the computer.\nSpeaker D: So after two hours, the visitors were, why are you hitting this college again?\nSpeaker D: So well, you know, we have been paid by a computer.\nSpeaker D: I mean, we have a government contract and they pay us by by amount of computer time of use.\nSpeaker D: It was in all days when the PDP8 and it sort of think.\nSpeaker D: So we had to make it look like.\nSpeaker D: Because so they had, they literally had a monitor at the time, at the time, on a computer.\nSpeaker D: How much time is being spent?\nSpeaker D: Or on this particular project.\nSpeaker D: Nobody was looking, you know, what was coming up?\nSpeaker E: Have you ever seen those little, it's this thing that's a shape of a bird and has a red ball and it's big dips into the water?\nSpeaker E: So if you can hook that up so that it hit the keyboard.\nSpeaker E: Yeah.\nSpeaker E: That's an interesting thing.\nSpeaker D: It would be similar to a new, some people who were, they were seen in old communities, the Czechoslovakia, right?\nSpeaker D: So we were watching for American airplanes coming to spy on us at the time.\nSpeaker D: So there were three guys stationed in the middle of the woods on the van, lonely, watching the hour, pretty much spending even a half there because they were the service, right?\nSpeaker D: And so they very quickly, they made a difference with local guys and local people in the village and here.\nSpeaker D: And so, but they, there was one plane flying over, always above.\nSpeaker D: And so they was the only work they had.\nSpeaker D: They like, for in the afternoon they had to report, they was a plane from Prague to Bernal, basically, flying there.\nSpeaker D: So the very first thing was that they would always run back and, and for a clock and quickly make a call that plane is passing.\nSpeaker D: Then the second thing was that they, they took the line from this post to a local pub.\nSpeaker D: And they were calling from the pub.\nSpeaker D: But third thing, which they made, when they screwed up, finally they had to pass.\nSpeaker D: The pub owner to make this phone call.\nSpeaker D: They didn't even bother to be there anymore.\nSpeaker D: And one day there was, there was no plane.\nSpeaker D: At least they were sort of smart enough that they looked if the plane is flying there, right?\nSpeaker D: And the pub owner just, oh my, four o'clock, okay, quickly pick up the phone call and it is a plane flying.\nSpeaker D: There was no plane for some reason.\nSpeaker D: It was downed.\nSpeaker D: So they got in trouble.\nSpeaker D: But, but, that's a lot.\nSpeaker E: Maybe I could set that up.\nSpeaker D: Well, at least go test the assumption about C1, I mean, to begin with.\nSpeaker D: But then of course, one can then think about some predictable ways how to change all of them here.\nSpeaker D: They just like be used to do this, this dance measures.\nSpeaker D: It might be this.\nSpeaker E: So the first set of variants, waiting vectors would be just, you know, one modifying, one leaving the others the same.\nSpeaker D: Yeah.\nSpeaker D: And you do that for each one.\nSpeaker D: Because you see, I mean, what is happening here in the, in the, in the such a model is that it's, tell us you, yeah, what has a low variance is, is, is, is more reliable, right?\nNone: Yeah.\nSpeaker F: When the data matches that.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: How do we know, especially when it comes to noise?\nSpeaker E: But there could just naturally be low variance.\nSpeaker E: Because I, I've noticed in the higher, capster coefficients, the numbers seem to get smaller, right?\nSpeaker E: So they just naturally, yeah.\nSpeaker E: Yeah.\nSpeaker C: They have smaller means also.\nSpeaker E: Yeah.\nSpeaker E: Exactly.\nSpeaker E: And so it seems like they're already sort of compressed.\nSpeaker E: The range.\nSpeaker D: Yeah.\nSpeaker D: That's why people use these lifters, so inverse variance waiting lifters basically, that makes, you can do in distance more like a, man, all this distance with a diagonal covariance and you use all the variances, where, over the all data, what they would do is that they would wait each coefficient by inverse of the variance.\nSpeaker D: Turns out that the variance decreases at least as fast, I believe, as an index of the capster coefficient.\nSpeaker D: I think you can show that analytically.\nSpeaker D: So typically what happens is that you need to wait the higher coefficient smaller than the lower coefficients.\nSpeaker D: So when we talked about Aurora still, I wanted to make a plea for, encourage for more communication between, between a different parts of the distributed center.\nSpeaker D: Even when there is absolutely nothing to say about it, whether it's good in work, in work, I'm sure that is being appreciated in Oregon and maybe it will generate a similar response here.\nSpeaker C: Can't set up a webcam maybe.\nSpeaker C: Yeah.\nSpeaker D: Well, you know, nowadays, yeah, it's up actually do ever almost.\nSpeaker E: Is the, if we mail to Aurora in house, does that go up to you guys also?\nSpeaker E: I don't think so.\nSpeaker E: No.\nSpeaker E: Okay.\nSpeaker E: So we should do that.\nSpeaker E: Do we have a mailing list that includes the OJIP?\nSpeaker C: No.\nSpeaker E: We don't do.\nSpeaker E: Oh.\nSpeaker E: Maybe we should set that up.\nSpeaker D: That would make it easier.\nSpeaker D: Yeah.\nSpeaker E: Yeah.\nSpeaker D: And then we also can send it to the same address, right?\nSpeaker D: And it goes to everybody.\nNone: Okay.\nSpeaker D: Because what's happening naturally in research, I know is that people essentially start working on something.\nSpeaker D: They don't want to be much bothered, right?\nSpeaker D: But what the, the, the, the danger is in a group like this is that two people are working on the same thing.\nSpeaker D: And of course, both of them come with a very good solution, but it could have been done somehow on a half of the effort.\nSpeaker D: Oh, that's another thing which I wanted to report.\nSpeaker D: Look-hash, I think, wrote a software for the Aurora 2 system.\nSpeaker D: Reasonably, the good one because he's doing it for Intel.\nSpeaker D: But I trust that we have a rights to use it or distribute it and everything.\nSpeaker D: Because Intel's intention originally was to distribute it free of charge anyways.\nSpeaker D: And so, so we make sure that at least you can see the software.\nSpeaker D: And if it is so-and-use, just that there might be a reasonable point for perhaps to start converging.\nSpeaker D: Because Morgan's point is that he's an experienced guy.\nSpeaker D: He says, well, you know, it's very difficult to collaborate if you are working with, supposedly, the same thing in quotes.\nSpeaker D: Except which is not the same, which one is using that set of filters and the other one is using another set of filters.\nSpeaker D: And then it's difficult to compare.\nSpeaker C: What about Harry?\nSpeaker C: We received a mid-nice tweak and he was starting to turn this on.\nSpeaker C: He got some experiments.\nSpeaker C: Yeah, they sent me.\nSpeaker C: And use this Intel.\nSpeaker C: Yeah, yeah.\nSpeaker F: Russian.\nNone: Yeah.\nSpeaker D: Because Intel paid us, should I say, on a microphone, some amount of money, not much.\nSpeaker D: Not much, I can say, on a microphone.\nSpeaker D: Much less than we should have gotten.\nSpeaker D: But it's a matter of work.\nSpeaker D: And they wanted to have a software so that they can also play with it.\nSpeaker D: Which means that it has to be in certain environment.\nSpeaker D: They use actually some Intel libraries.\nSpeaker D: But in the process, look at just re-road the whole thing.\nSpeaker D: Because he figured rather than trying to make sense of an including axis of the wear.\nSpeaker D: Not for training of the nets, but I think he re-roads the, or maybe somehow reused over the parts of the thing.\nSpeaker D: So that the whole thing, including MLP, trained MLP, is one piece of software.\nSpeaker D: Wow.\nSpeaker D: Is it useful?\nSpeaker D: Yeah.\nSpeaker G: I mean, I remember when we were trying to put together all the axis software for the submit.\nSpeaker D: That's what he was saying, right?\nSpeaker D: He said that it was like, it was like just so many libraries.\nSpeaker D: I'm not going to knew what was used when.\nSpeaker D: So that's where he started.\nSpeaker D: And that's where he realized that it needs to be.\nSpeaker D: It needs to be at least cleaned up.\nSpeaker D: And so I think it is available.\nSpeaker C: The only thing I would check is, does he use Intel?\nSpeaker C: That's the library.\nSpeaker D: If it's the case, maybe not in the first approximation.\nSpeaker D: Maybe not in the first approximation.\nSpeaker D: Because I think he started first with the plain C or C++ or something.\nSpeaker D: I can check on that.\nSpeaker D: And otherwise, Intel libraries, I think they are available freely.\nSpeaker D: They may be running only on Windows or on the Intel architecture.\nSpeaker D: Yeah, on Intel architecture, main or unknown sound.\nSpeaker D: That is possible.\nSpeaker D: That's why Intel, of course, is distributing.\nSpeaker C: Well, at least there are optimized versions for the architecture.\nSpeaker C: I don't know.\nSpeaker C: I never checked carefully this.\nSpeaker D: I know there were some issues that initially, of course, we do all the development on Linux.\nSpeaker D: But we have only three sounds.\nSpeaker D: And we have them only because they have a spare board.\nSpeaker D: Otherwise, almost exclusively, working with PCs now with Intel.\nSpeaker D: In that way, Intel succeeded in us because they gave us too many good machines for very little money or nothing.\nSpeaker D: So we run everything on Intel.\nSpeaker E: Does anybody have anything else?\nSpeaker E: So we read some digits?\nSpeaker E: Yeah.\nSpeaker E: Yes.\nSpeaker E: So, you never know.\nSpeaker E: We've ever done this way that it works is each person goes around in turn.\nSpeaker E: And you say the transcript number.\nSpeaker E: And then you read the digits, the strings of numbers as individual digits.\nSpeaker E: So you don't say 850, say 850.\nSpeaker D: OK.\nSpeaker D: OK.\nSpeaker D: So can I try maybe start there?\nSpeaker D: Sure.\nSpeaker D: This transcript L101, 850720538, 118, 528759, 961, 459500, 882758720, 9348, 36396545820, 8248, 36396545, 8248, 8208441810, 002586186551, 4374402890.\nSpeaker E: Transcript L102, 29564526, 768484913, 489025645223, 292901317, 80929366600, 567421337, 66833321, 032669137991.\nSpeaker G: Transcript L-103, 76377732, 21923460, 6314, 940853, 6862, 009187681809, 009434532, 711150319, 08323095, 4922986786.\nSpeaker A: Transcript L-104, 695007490.\nSpeaker A: 915005169, 9299151207, 853890321541, 2041138553775399817105082722691869352.\nSpeaker C: Transcript L-100014953649643373138811481203611078862765, 627114991, 6223682766914236458283, 09231750.\n"}, {"meeting_id": "ES2003c", "summary": "After a brief review of the last meeting, Industrial Designer, User Interface and Marketing gave their presentations respectively on components, interface design and user requirements of the new remote control. Considering Marketing's advice and the budget, the team temporarily decided on a double-curved rubber remote control with an ergonomic joystick and an LCD screen menu display. Users could program their remote controls on the LCD screen as well as using software on the computer. Power would be supplied to the remote control by standard batteries. The beeping sound of the location function could be customized. Project Manager informed the team of the plan for the next meeting. Industrial Designer and User Interface would present a prototype of the remote control. The team would also make a product evaluation.", "dialogue": "Speaker B: All right, yep, I just have a quick overview of the minutes.\nSpeaker B: I think to sum up the last meeting would be to say the requirements that we've set out.\nSpeaker B: We were going to go for what seemed to be a fairly minimal design based on a small joystick, LCD and a couple of other buttons for navigation with power being, let's suppose, one of the main single purpose buttons.\nSpeaker B: We were also going to use a novelty of being able to locate the remote control again via a small transmitter of any luck, the idea to try and separate us and also because of the minimal design, it looks like we'll be able to be fairly adventurous in the actual physical shape of our remote control of any luck.\nSpeaker B: Pretty much sums up the last one.\nSpeaker B: So we'll just crack on.\nSpeaker B: Like to maybe start with the industrial designer if it's possible.\nSpeaker B: As you can see today is the conceptual design phase and it's to cover things like what the parts might be made of, can we outsource these from elsewhere, what we have to construct the items ourselves.\nSpeaker B: I have a presentation area, I'll just say that in the, yep, okay, I'll just award the upman.\nSpeaker B: Which one?\nSpeaker B: Oh, interface concept.\nSpeaker B: That's you.\nSpeaker B: We've got trend watching.\nSpeaker D: Components design.\nSpeaker B: Okay.\nSpeaker D: So the case, that's for the role.\nSpeaker D: First of all, it could be plastic or plastic.\nSpeaker D: So later on, we found out that it can be rubber as well or titanium or even wood.\nSpeaker D: So it's probably plastic.\nSpeaker D: We need the infrared transmitter.\nSpeaker D: Get that off the shelf.\nSpeaker D: Joy-Sick will probably, if we're going to use it, could be plastic or rubber even as well.\nSpeaker D: If you go on to the next slide, if you go on to findings, it's two or three slides there.\nSpeaker D: So this is what a family can use.\nSpeaker D: Three different types of batteries can either use a hand dynamo or the kinetic type ones that they use in watches or else.\nSpeaker D: Solid power.\nSpeaker B: No, the kinetic one.\nSpeaker B: Of course.\nSpeaker B: Because that's the ones where like you, the movement causes it.\nSpeaker B: The battery for a watch wouldn't require a lot of power would be my one query.\nSpeaker B: Is a kinetic one going to be able to supply enough power?\nSpeaker C: Do you think watch moves around a great deal more?\nSpeaker D: I don't think it would.\nSpeaker D: And solar cells, I don't know about that.\nSpeaker A: Been inside and all, yeah.\nSpeaker D: You should probably just use conventional batteries.\nSpeaker D: Just like a new one can cause.\nSpeaker B: Which I suppose as well would allow us to go off the shelf again, you too?\nSpeaker D: Yeah.\nSpeaker D: These are two different types of shapes.\nSpeaker D: You can have one as a flat one.\nSpeaker D: And then more original ones are single curved one with a double curved.\nSpeaker D: And the materials are there as you can see.\nSpeaker D: But you can't have a titanium one for a double curved.\nSpeaker A: The titanium, they're really strong metal.\nSpeaker D: Is it also expensive?\nSpeaker D: It thinks so as well.\nSpeaker D: I'm curious.\nSpeaker B: I don't know if he'd be able to off the top of your head or not.\nSpeaker B: The single curved with double curved.\nSpeaker B: We could be able to give an example.\nSpeaker B: Could you maybe draw something?\nSpeaker B: It doesn't have to be perfect.\nSpeaker B: It's just because I'm not quite sure if I understand the difference between the two.\nSpeaker D: For curved, well I was thinking to sit in the palm of your hand.\nSpeaker D: Maybe like this with the joy pod here.\nSpeaker D: Do I stick here?\nSpeaker D: Maybe an okay button around here so that the thumb can use it quite easily.\nSpeaker D: But don't exactly double curves.\nSpeaker D: The problem means this is probably double curves.\nSpeaker D: Whereas a single curve would be like that.\nSpeaker D: I guess.\nNone: Or not.\nSpeaker B: So it might literally just be two curves.\nSpeaker B: Okay.\nSpeaker D: Whereas this is two curves.\nSpeaker D: So I guess that's what they mean by double curve.\nSpeaker D: It looks better than a single curve.\nSpeaker D: You can't have it in titanium.\nSpeaker D: A nice material.\nSpeaker D: For the buttons, you can have the scroll wheel which they use in mouse.\nSpeaker D: It's a little bit more expensive.\nSpeaker D: If you want to use LCD, it's a little more expensive.\nSpeaker D: So you can decide to trade off there.\nSpeaker D: If you want the buttons to be, if you have a double curve, control and it's rubber, then you have to use rubber buttons as well.\nSpeaker D: You're going to have to have a number keypad for the amount of channels these days.\nSpeaker D: You wouldn't want to just have to scroll through all the channels together when you want to.\nSpeaker D: You want to enter just a number if you know.\nSpeaker D: So I think we're going to have to have a number keypad anyway.\nSpeaker D: Do you think?\nSpeaker B: Okay.\nSpeaker B: I'm going to have to make sure that we can talk about it.\nSpeaker B: So you've got a little bit about the chip that we might require as well.\nSpeaker D: Yeah.\nSpeaker D: So it depends where you're going to spend the money.\nSpeaker D: If you want to fancy LCD display.\nSpeaker B: Do you have any idea so far when we're saying that we need an advanced chip for an LCD, does that shoot the cost up by a drastic amount or?\nSpeaker A: You need an advanced chip for the LCD.\nSpeaker B: I think compared to say with pressing of buttons.\nSpeaker B: Advanced light.\nSpeaker B: If you press a button and it sends a certain transmission through the infrared, whereas if we're controlling the LCD, we definitely require a much more powerful chip.\nSpeaker B: Just compared to the chip you would use for pushing buttons.\nSpeaker B: I think see.\nSpeaker B: Okay.\nSpeaker B: If I've not overstepped.\nSpeaker B: Okay.\nSpeaker B: Should I go on or go for that?\nSpeaker A: If we only have 12 pounds 50.\nSpeaker A: 12 euros, not even 12 pounds.\nSpeaker A: 12 euros was that like 8 pounds or something like that.\nSpeaker B: Well, we'd also be relying on the bulk buying and produce and such.\nSpeaker B: I assume we have to look into the cost of those.\nSpeaker B: So sorry.\nSpeaker D: The previous slides just explain what the internal components of the control.\nSpeaker D: If you don't want to be further.\nSpeaker D: So it just says what it does.\nSpeaker D: Translate the key press into the infrared signal and this is received by the TV.\nSpeaker D: The chip just needs to detect the key press and then it's added to the amplifier.\nSpeaker B: Okay.\nSpeaker D: And then the next slide shows the copper wires interact with the buttons to get sent to the chip.\nSpeaker D: So that's just how the power works inside.\nSpeaker D: It's a side on what buttons are they use.\nSpeaker B: Okay.\nSpeaker B: You just got a section of the press.\nSpeaker B: So in the information that you've been supplied, how feasible would you say that the idea of using an LCD looks?\nSpeaker D: I think we can do it if we use conventional buttons and not have solar cells, they're connected.\nSpeaker D: And then maybe use a single curved case because you might need to be curved for the film to use the joystick easily.\nSpeaker D: And then you'll need the advanced chip for the LCD.\nSpeaker B: I mean that sounds like a good requirement to me.\nSpeaker B: Conventional battery would seem to make sense.\nSpeaker B: I'm not sure if there's any cost differences between single and double curved.\nSpeaker B: And I don't know about anybody else but plastic or rubber is a traditional format casing would seem to be a good way forward.\nSpeaker C: I'm actually going to be a golden go street for rubber.\nSpeaker A: Okay.\nSpeaker A: I also have a preference for rubber.\nSpeaker B: First of my reasons.\nSpeaker B: Yeah.\nSpeaker B: Well, we move on to user interface and...\nSpeaker B: Yeah.\nSpeaker B: Sorry, as long as we use...\nSpeaker B: Yep.\nSpeaker B: Okay.\nSpeaker B: And I'll...\nSpeaker B: And interface concept.\nSpeaker C: I'm going to have to work between the slides and the whiteboard because I actually have some fairly concrete things.\nSpeaker C: It's time.\nSpeaker C: I was given an HTML file giving various ideas from other previous remote control designs and pretty much decided to just dump them off.\nSpeaker C: I wasn't very impressed by them and I didn't seem to...\nSpeaker C: Nothing I saw seemed to meet the sorts of design specifications that were going for today.\nSpeaker C: So rather than looking at other remote, it would be better to simply look at the human hand and try and figure out a way of laying out the elements we've already decided on.\nSpeaker C: So the finger is...\nSpeaker C: Each finger or thumb is wherever it needs to be already.\nSpeaker C: So...\nSpeaker C: Next slide, please.\nSpeaker C: And what we've basically decided on was the joystick two function buttons and the LCD.\nSpeaker C: Just keep it paired down to the absolute minimum.\nSpeaker C: I don't actually think we need the numeric keypad because one of the menus that we could have available via the LCD is one where you scroll through channels.\nSpeaker C: So if there's something...\nSpeaker C: Bear in mind, since this isn't meant to work for...\nSpeaker C: For digital or for cable or whatever, you're basically looking at four or five terrestrial channels and then the channel for your VCR and R DVD player and R set top box.\nSpeaker C: So it's not really excessive to the inside of that.\nSpeaker C: Because I have two modes, basic mode where the joysticks left right for channels up down for volume and the menu mode for further functions.\nSpeaker C: The reason I was particularly interested in using rubber for this is if we're going to have a highly ergonomic design, it needs to be ergonomic for left or right handed people.\nSpeaker B: Can I jump in slightly there?\nSpeaker B: That would suggest that double curve design is probably going to be most appropriate then.\nSpeaker C: Yes, absolutely.\nSpeaker C: Basically what I...\nSpeaker C: Basically what this would be having, I would say the whole thing articulated at two points.\nSpeaker C: So if you're handing it from a left or right handed, you can adjust it so that the LCD and the joystick could be in the right place.\nSpeaker C: Also this is a rather nicer design gimmick that the whole thing should have sort of organic feel to it because it should be soft to touch and can be moved around.\nSpeaker C: Onto the next slide.\nSpeaker B: Just to let you know, we'll probably be quite tight for time as well.\nSpeaker B: Because I think you've probably got a lot you'd like to say I guess.\nSpeaker C: Basically, I can add pretty pictures to this.\nSpeaker C: Assuming the hand to be in a sort of disposition, not holding the remote, the joystick unit should rest up with the joint of the forefinger.\nSpeaker C: So it's directly visible for the thumb.\nSpeaker C: And it would need to be articulated just below that so that it could be switched around for left or right handed users.\nSpeaker C: So...\nSpeaker C: You then have a grip section that can be more or less the same irrespective of handedness.\nSpeaker C: You just have two big buttons that cover most of the area so it can end the upper part, one for the forefinger, one for the middle finger.\nSpeaker C: This part here is the joystick.\nSpeaker C: This would be the actual grip, probably where you don't have the battery as well.\nSpeaker C: That would be probably the bulkiest part.\nSpeaker C: And you then have at the bottom the LCD and this would need to be articulated as well.\nSpeaker C: And basically I'd want this to rest here right at the base of the wrist so it would fit just nicely in the hand.\nSpeaker C: And again, this part could be rotated so it can...\nSpeaker C: So that it can be adjusted to either left or right handed user.\nSpeaker C: So the top function button in basic mode would be the on off switch and menu mode would be the enter button.\nSpeaker C: And then the bottom function button switches between modes.\nSpeaker C: Now, programming it...\nSpeaker C: Actually, this is one thing I found with the replacement remote control.\nSpeaker C: Programming them can be a right pain.\nSpeaker C: So I thought that the simplest way around that would be to have a cable to connect it to the computer.\nSpeaker C: So I'm fairly simple software on the computer just so that you could program it in a rather more comfortable interface.\nSpeaker C: And you could download programs for TVs from all the main manufacturers.\nSpeaker C: Though it would be necessary to have a mode for programming it without the computer.\nSpeaker C: Just in case there are still people left out there who don't have them yet.\nSpeaker B: Okay, that's my idea.\nSpeaker B: Excellent, right?\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker A: So to gather my research to be basic methods, we compared whether people want the remote control to do a lot of stuff, or they want it to look cool.\nSpeaker A: And then we researched our fashion trends in Europe.\nSpeaker A: What's the new black, you know, as it goes?\nSpeaker A: Next slide, okay.\nSpeaker A: We found in order of importance, people want the remote control to look cool more than they wanted to be cool.\nSpeaker A: And they wanted to do a lot of good stuff.\nSpeaker A: They wanted to look like it does a lot.\nSpeaker A: And if it does do a lot, that's a bonus, but they don't care so much, you know.\nSpeaker A: They want it to be...\nSpeaker A: Sounds a bit like a contradiction.\nSpeaker A: Technically, technically innovative.\nSpeaker A: People want it to be that, but still they care more about the way it looks than what it does.\nSpeaker A: So like the interface is really important.\nSpeaker A: And easy to use.\nSpeaker A: It just so happens that from the second point to the third point is twice as important.\nSpeaker A: I mean, the second point is twice as important as the third point.\nSpeaker A: People want it...\nSpeaker A: It has to be cooler than easy to use, you know, if it has the newer speeches, even if it's difficult to use, to prefer it to have the newer speeches.\nSpeaker A: And if it's easy to use, that's a bonus.\nSpeaker A: The fashion now, this seems better to me, but fruit and bears is the new black for furniture, for clothes, for shoes.\nSpeaker A: I had that relates to a remote control, I don't know.\nSpeaker A: But I see, I come onto that in the next slide.\nSpeaker A: Spongy.\nSpeaker A: I've always been saying everything is the new black.\nSpeaker A: Well, spongy is the new black as well.\nSpeaker A: So we had the choice between rubber and plastic.\nSpeaker A: If it's a type of rubber that you can squeeze, you know, it's spongy, then that's good for it.\nSpeaker D: I mentioned that the rubber material is the type of stress ball material.\nSpeaker A: Not just normal rubber.\nSpeaker A: Okay, so kind of spongy material.\nSpeaker A: So my personal opinion, we want some advanced internet technologically innovative, obviously.\nSpeaker A: What we need is something that looks like it's from the future that looks cool.\nSpeaker A: That's different, you know, that's everyone has a white remote control, black remote control.\nSpeaker A: You need something cool.\nSpeaker A: Like titanium is cool, but it's expensive.\nSpeaker A: And maybe it's a bit of overkill for a remote control.\nSpeaker A: Now the fruit and veg options.\nSpeaker A: Even we go in that direction or we stay totally away from it.\nSpeaker A: I mean, the research did come up with fruit and veg, so maybe it is important.\nSpeaker A: It's up to the interface guy.\nSpeaker A: So if we stay away from it, you know, stay away from it.\nSpeaker A: But if we're going to go along with it, then it doesn't necessarily have to be like an apple or something like that or a kiwi fruit.\nSpeaker A: There can be something like, I say potato peeler, but I'm sure you guys have a much cooler idea than I do.\nSpeaker A: So I think cool is the key.\nSpeaker A: A few questions about a spongy remote control. I've never seen one before. I've seen plastic remote controls.\nSpeaker A: I think maybe they were.\nSpeaker A: I don't know, back in the day when they first come up with remote controls, they had a reason for it being.\nSpeaker A: Sturdy, you know, for being stronger, sturdy.\nSpeaker A: So if we want something stronger than sturdy, I say stay with plastic or titanium.\nSpeaker A: But if we go with spongy, we can stress that you can drop this as many times as you want.\nSpeaker A: It doesn't matter. Spongy material is not going to break.\nSpeaker A: I just don't know how the LED and the lights are going to fit into a spongy material because it's not going to be completely squeezable.\nSpeaker A: So how do things fit in? And if we are going to use spongy, we can say it's long lasting.\nSpeaker A: You know, it's damage resistant and stuff like that.\nSpeaker A: So just to summarize, people want stuff that's cool, that looks like it's cool.\nSpeaker A: And if it is cool, it's a bonus where there doesn't have to be people like Fruit and Veg.\nSpeaker A: We can either go down that alley or stay totally away from it.\nSpeaker A: People like spongy material, if we're going to use spongy material, we have to think of how practical it is and how we can further promote that idea.\nSpeaker A: So this was this year. So things change all the time, every year, you know, they always talk about this year.\nSpeaker A: This is a new black. Well, next year, something else is going to be the new black and we're stuck with last year's Fruit and Veg type stuff.\nSpeaker A: Okay. That's me.\nSpeaker B: Well, I would say that the most you could probably hope for is gaining a year's lead than most people anyway. You always have to bring out new designs. So if we can get next year, then that's possibly a good place to start anyway.\nSpeaker B: Seems like a lot of the components could be off the shelf. So I don't know exactly what cost would be incurred. I can see your point about the numbered key back.\nSpeaker B: But I would say that we can probably incorporate it into the menu system.\nSpeaker B: If you need to do traversal of a large number of channels, my feeling would be that even if at a later date this was to be taking control of set boxes as well.\nSpeaker B: And you should be LCD and a joystick would probably allow you to manipulate enough channel numbers for you to be.\nSpeaker C: Actually, if you've got a lot of channels, the numbered key back can be quite annoying as well because if you're trying to remember what numbers the discovery child is.\nSpeaker A: That's a good point. You can incorporate names into different.\nSpeaker C: But if you have a menu structure, then you can subgroup them.\nSpeaker B: So what it looks like to me is we'll use a large number of standard components.\nSpeaker B: Say something like lithium ion battery, the kind that you find in most small hand-held devices now.\nSpeaker B: It looks like we're going for a double curve design. I don't know what cost implications there are in that.\nSpeaker B: It looks like we're pretty much decided on some kind of flexible rubber.\nSpeaker B: So I'd have to say, depending on how flexible it is, we might need to have some kind of inner frame.\nSpeaker C: Yeah, I would say definitely.\nSpeaker C: I mean, the idea of having it articulated, there would be basically two points of articulation.\nSpeaker C: One below and one above, the main sort of grip.\nSpeaker B: Do you think there's any way we can maybe remove the articulation?\nSpeaker B: I can see why it looks appealing, but it could be a weak point in the structure, do you think?\nSpeaker B: That would be a worry of mine.\nSpeaker D: If you're going with the fruit and vegetables, it looks like a banana.\nSpeaker D: I don't want to sign it that way.\nSpeaker C: I don't think it would be a structural weakness.\nSpeaker C: I mean, if you have a firm substructure, you can then incorporate articulation into that.\nSpeaker C: If you then have other outer skin of flexible, spongy rubber, then you have something very much organic.\nSpeaker C: I think it would look rather cool.\nSpeaker C: I mean, drawing there doesn't do it just as it makes it.\nSpeaker C: It makes it look a little bit more like a vibrator than a remote control, but we won't add that functionality.\nSpeaker B: One interesting point is I don't know how serious you were there, but if we take some of the ideas of a banana, why don't we make the damping yellow to incorporate the color of the logo?\nSpeaker B: I don't know.\nSpeaker B: It's certainly a different color from your average...\nSpeaker C: You make it harder to lose, as well.\nSpeaker C: Yeah, that's true.\nSpeaker B: Was there anything in your research?\nSpeaker A: The noise for when you lose the banana.\nSpeaker A: Yeah, but when you lose the remote control, it could be like a monkey noise or something like that rather than a standard B.\nSpeaker A: You lose the monkey, the banana.\nSpeaker A: You lose the banana, you press the button, and you hear like monkeys scream to someone like that to direct you towards the banana.\nSpeaker C: I think that's something that could perhaps be programmable, though we have monkeys default.\nSpeaker A: Oh, I was going to talk about the programmable remote control.\nSpeaker A: I think it would be cool if you had one remote control that can work with all TVs, you know, you click a button and it finds the frequency that works with your particular television.\nSpeaker A: But why you'd need to program it like as a standard feature?\nSpeaker C: I don't know.\nSpeaker C: Well, basically, the...\nSpeaker C: For...\nSpeaker C: I mean, each manufacturer will have a particular command set that the TV responds to.\nSpeaker C: It's not simply a matter of frequency.\nSpeaker C: So, usually what's done is you have a big booklet full of different possibilities.\nSpeaker C: You'll have all of them sort of programmed into the...\nSpeaker C:...remote and you'll have this little booklet of codes.\nSpeaker C: You then find your...\nSpeaker C:...find your manufacturer and try the different codes that come under that manufacturer's name until you find the right one.\nSpeaker A: That's because televisions, they don't give the remote control any type of feedback when you send it.\nSpeaker A: Yeah, that's right.\nSpeaker C: Okay.\nSpeaker C: So, that's quite annoying and we probably would still have to do that.\nSpeaker C: But if we had some sort of hookup to your computer as an option, then if you've got a computer, you can avoid that rather irritating.\nSpeaker B: An interesting point is that if the person doesn't have the computer, we can still make the process easier because instead of having to look up codes, maybe we can have it for one of the options they have, is it the lookup, the name of the company on the LCD, and then maybe look up different names of different actual units that have been produced.\nSpeaker B: Or we have the little control, just go through them until it's like, does a Stufa correct function and such?\nSpeaker A: Is it actually a book of names to digits or is it like a few pages?\nSpeaker C: booklet.\nSpeaker C: Some pages.\nSpeaker A: I just think if we were to store this information, some type of mapping, this person probably need to use this feature like once.\nSpeaker A: When they first buy the remote control, whenever they buy a new television.\nSpeaker A: So, once every-\nSpeaker B: It's not to be used very often, that's right. Yeah.\nSpeaker C: But it's a nuisance.\nSpeaker C: And it's a nuisance very close to the, to actually purchasing it.\nSpeaker C: So, it's something you're likely to be thinking about.\nSpeaker C: If there's one in the shop that says it can avoid much of that nuisance, you might be favorably inclined towards it.\nSpeaker B: Okay, this is just to give us a rough idea of what we're meant to be doing for the next stage.\nSpeaker B: I'm pretty sure that you've got that anyway.\nSpeaker B: This was a conceptual one.\nSpeaker B: I think we've covered everything we need to here.\nSpeaker B: I think we've decided on standard items for most of them, rubber and such.\nSpeaker B: So, we're now looking at the detail design and what we need to be doing for the next meeting.\nSpeaker B: So, for example, we'll just start the talk.\nSpeaker B: But yourself and the industrial and designer are going to be working quite closely at this stage because it'll be hard not to, obviously.\nSpeaker B: Looking for a prototype thing of some of the description using clay.\nSpeaker B: I suppose you'll be told a lot more about that as progress.\nSpeaker B: The user interface design.\nSpeaker B: It looks like the idea of push forward so far is maybe more of a physical one now, or possibly be more interested in maybe how the LCD is going to incorporate, you think?\nSpeaker B: Or do you perceive that most of the design decisions still need to be based on the physical layout?\nSpeaker C: Well, I think other than getting a sort of more aesthetically pleasing form for it, most of the sort of layout design decisions are made, I would say.\nSpeaker C: But then again, the menu structure to a considerable extent is going to vary according to the model of television and also any customizations that the user might wish to incorporate.\nSpeaker C: So, one of the nice things about having an LCD and a menu structure is that there is that flexibility to it.\nSpeaker B: That's very true.\nSpeaker B: OK.\nSpeaker B: I've got product evaluation as well.\nSpeaker A: Yeah, see, some of these things kind of logically follow the others.\nSpeaker A: How can product evaluation doesn't that come after?\nSpeaker A: Have you seen any product prototype?\nSpeaker B: I think it would be more case of how do you envisage us performing the product evaluation once we have a model to test with?\nSpeaker B: Oh, that's true.\nSpeaker B: OK.\nSpeaker B: So, it would be a case of do you think that just bringing users in, do you have them test out?\nSpeaker B: Maybe putting a certain number of products into the field in certain places, which I suppose is quite similar, or just kind of hitting the streets and saying, this is a new remote control, what do you think it will look at it?\nSpeaker B: OK.\nSpeaker A: So, at this stage, we still have an old target audience.\nSpeaker B: I think the target audience is pretty much anybody under the age of, say, maybe 60 or something, with the address.\nSpeaker B: The reason I'm saying that is that we're just looking at a replacement remote control, something that's stylish.\nSpeaker B: So, maybe you're even just, maybe you're narrowing down your target audience simply by saying, it's nervous.\nSpeaker B: If they're buying it that often, or they're maybe looking to replace something, then they've got a bit of free cash, so that puts them potentially in the younger age bracket.\nSpeaker B: Maybe even single, just for the reasons of excess cash flow.\nSpeaker B: Of course, I mean, at 25 euros, I don't think we're looking at charging the air for the device.\nSpeaker B: Yeah.\nSpeaker B: I think that's a lot of the best way to do that, as well as the normal bracket.\nSpeaker B: Your idea of a USB, I think, would largely depend on the cost.\nSpeaker B: The USB is definitely one of the cheapest interfaces out there, but it might push the cost of the overall unit up, because we require not just the connection, but the chip for communicating with the rest of the system.\nSpeaker B: That one would have to be based on the USB approach for the remote control.\nSpeaker B: The idea was that maybe it could connect tops for allowing a software interface in the PC for a larger programming, due to the fact that we've gone for such a nice, easy, minimal design normally.\nSpeaker A: Definitely broken some type of...\nSpeaker D: But didn't they just say it was just for TV?\nSpeaker B: It's just for TV, but for different programming it to use your TV, you might hook it up to the PC.\nSpeaker B: I'm not sure, but I'm thinking maybe that the additional cost of the USB might be prohibitive, but we don't know unless it would make sense to.\nSpeaker A: But you're going to need some type of flash memory or something.\nSpeaker A: Well, something that doesn't...\nSpeaker A: You wouldn't have to redo the whole thing once the batteries are changed.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I think it would need to have...\nSpeaker C: It would need to have some sort of onboard memory in any way.\nSpeaker C: For one thing, I do think the menu system should be...\nSpeaker C: Although it's not going to be terribly complicated.\nSpeaker C: It's just controlling one device, the menu system ought to be customizable.\nSpeaker A: Different languages and different schemes and stuff like that.\nSpeaker C: How you want to say the menu structure for looking through channels, if you've got lots of channels.\nSpeaker A: If it was customizable, it's in different languages and stuff.\nSpeaker A: It is supposed to be international, right?\nSpeaker B: It would make sense to you.\nSpeaker B: It would make sense to you.\nSpeaker B: And we'd better be careful at the time as well.\nSpeaker B: Everybody had a five minute warning, so...\nSpeaker B: I would say international would make sense.\nSpeaker B: You're going to look at product evaluation.\nSpeaker B: I'd better start writing up the hell of a lot of crap.\nSpeaker B: You're going to look a bit more at the physical makeup.\nSpeaker B: It looks like we're going to need to maybe prototype and play.\nSpeaker B: We'll see how that goes.\nSpeaker B: We'll be supplied with the clay as well.\nSpeaker C: Where is the clay?\nSpeaker B: I think we've got an idea of where we need to go for the next meeting.\nSpeaker D: We'll go with the fashion thing.\nSpeaker D: The fruit and veg.\nSpeaker D: It's a spongy, it's over.\nSpeaker B: Yeah, that would be my idea.\nSpeaker C: I think I would largely not have to ask you to clear the whole fruit and veg.\nSpeaker C: But the spongy, I like it a lot.\nSpeaker B: I have quite a general consensus that the idea of the rubber could be quite...\nSpeaker C: I don't know if you're wondering about having bright yellow.\nSpeaker C: Maybe we could have options for colors as well.\nSpeaker C: And again, it has the advantages of being harder to lose.\nSpeaker A: Sure.\nSpeaker A: We are trying to promote a remote control, but we want to keep the company brand as well.\nSpeaker B: So we always need to remember that somewhere we're getting the slogan on.\nSpeaker B: Possibly, I'm not sure, but I'm seeing a little bit of space around the joystick area, which could be used.\nSpeaker C: Like I said before, I think we should have the RRR on the top function.\nSpeaker B: Okay.\nSpeaker B: And I think that says it all really.\nSpeaker B: I'm going to throw it to it.\nSpeaker B: Right.\nSpeaker B: We'll see everybody in a half hour.\n"}, {"meeting_id": "Bro003", "summary": "The meeting included an introduction to the project, the mechanics of training the data, standardizing the data for the models, the time involved, and planning efficient use of computational resources . The team members shared and discussed the existing model as well as the tests they intended to run. There was a problem standardizing labels as well as converting HTK labels to the format that the team wanted. Adding more dimensions to the current model was also causing concern because of the upper bound on computational resources.", "dialogue": "None: Challenge?\nSpeaker B: Challenge.\nSpeaker F: So we think we're going to.\nSpeaker F: Okay, good.\nSpeaker F: Alright, going again.\nSpeaker F: So we're going to go around this before and do our digits.\nSpeaker F: Transcript, 1311-1330.\nSpeaker F: 323-4765.\nSpeaker F: 531-6241.\nSpeaker F: 677-890-94003.\nSpeaker F: 0158-17353.\nSpeaker F: 268-03624307.\nSpeaker F: 4.\nSpeaker F: 5069-4, 74857-9615-07802.\nSpeaker F: 090-90-604001.\nSpeaker F: 2.\nSpeaker B: I'm reading transcript.\nSpeaker B: 1391-1410-677-890-698.\nSpeaker B: 01319-16237-34.\nSpeaker B: 4.\nSpeaker B: 5.\nSpeaker B: 6.\nSpeaker B: 847-920-75.\nSpeaker B: 03696.\nSpeaker B: 0931003.\nSpeaker B: 1.\nSpeaker B: 2.\nSpeaker B: 3097.\nSpeaker B: 5267-983.\nSpeaker B: 6706.\nSpeaker C: Okay, I've got transcript 1271-1290.\nSpeaker C: 1710-281-207134.\nSpeaker C: 509-6080-7386-888818813.\nSpeaker C: 934-0394-042-021-2038232.\nSpeaker C: 42816-535-8308.\nSpeaker C: 659-1076.\nSpeaker C: 7662-879.\nSpeaker C: 0057804-141.\nSpeaker D: Transcript number 1291-1310.\nSpeaker D: 23902-38152-47467-5670-9401-00.\nSpeaker D: 07260.\nSpeaker D: 1604-645-23407-26440-74188-885-63989-00391.\nSpeaker A: Transcript number 1171-1190.\nSpeaker A: 83435-9439-0309691.\nSpeaker A: 203-556-086-304-0117882-89551-00214-36702-479-5905-7351.\nSpeaker A: 479-5905-7356-669-465-7.\nSpeaker H: Okay, this is Barry Chen and I'm reading transcript 1351-1370.\nSpeaker H: 4507-8928-71489-859-369-7159-047301-00106-243467-544-677-789-844-6389-00.\nSpeaker H: 544-677-899-899-001-1247-349-253-5008-48.\nSpeaker I: Transcript 1371-13905-6705-9263-305-0455-049162-30002-192-466-7208-5-0386-9088-7408-8408.\nSpeaker I: 3245-456-923.\nSpeaker F: Yeah, you don't actually need to say the name.\nSpeaker F: I'll probably bleep that out.\nSpeaker F: Okay.\nSpeaker F: These are not mine.\nSpeaker F: Oh.\nSpeaker F: Okay.\nSpeaker F: Not that there's anything defamatory about 857.\nSpeaker F: Okay.\nSpeaker F: So here's what I have.\nSpeaker F: I was just trying to add things I think that we should do today.\nSpeaker F: It's what I have for an agenda and so far.\nSpeaker F: We should talk a little bit about plants for the field trip next week.\nSpeaker F: A number of string field trip to OGI.\nSpeaker F: And mostly, first of all, about the logistics for it.\nSpeaker F: Then maybe later on, the meeting we should talk about, actually, might accomplish.\nSpeaker F: And then, kind of go around to see what people have been doing and talk about that progress record, essentially.\nSpeaker F: And then another topic I had was that Dave here said, give me something to do.\nSpeaker F: And I have failed so far on doing that and so we can discuss that a little bit, find some holes and some things that someone could use and help with these things.\nSpeaker F: I'm going to move into that current here.\nSpeaker F: Okay.\nSpeaker F: Always count on that.\nSpeaker F: That's a really good question.\nSpeaker F: So.\nSpeaker F: And then talk a little bit about disks and resource issues that started to work out.\nSpeaker F: And then anything else that he has that isn't in that list?\nSpeaker B: I was just wondering, does this mean the battery is dying and I should change it?\nSpeaker F: I think that means the battery is okay.\nSpeaker B: Oh, okay.\nSpeaker B: Yeah, that's good.\nSpeaker B: Because it's full.\nSpeaker F: All right.\nSpeaker F: Full of electrons.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker F: So I'm going to start this with this mundane thing.\nSpeaker F: It was kind of my bright idea to have us take a plane that leaves at 7.20 in the morning.\nSpeaker F: Oh, yeah.\nSpeaker F: This is a reason I did it was because otherwise for those of us who have to come back the same day is really not much of a visit.\nSpeaker F: So the issue is how we ever accomplished that.\nSpeaker F: What part of the time do you live in?\nSpeaker H: I live in the corner of campus.\nSpeaker H: Okay.\nSpeaker H: South East corner.\nSpeaker F: Okay.\nSpeaker F: So would it be easier?\nSpeaker F: Those of you who are not used to this area can be very tricky to get to the airport at 6.30.\nSpeaker F: So would it be easier for you if you came here and I drove you?\nSpeaker F: Yeah, bridge.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker F: So if everybody can get here at 6.\nSpeaker F: Yeah.\nSpeaker F: I'm afraid you can do that.\nSpeaker F: I guess.\nSpeaker F: Yeah.\nSpeaker F: Anyway.\nSpeaker C: So was that the enough time?\nSpeaker F: Yeah.\nSpeaker F: So I'll just pull up in front at 6.\nSpeaker F: And yeah, that'll be plenty of time.\nSpeaker F: It won't be bad traffic that time of day.\nSpeaker C: I guess once you get past the bridge.\nSpeaker C: That would be the opening.\nSpeaker C: Yeah.\nSpeaker C: Once you get past the turn off to the day bridge.\nSpeaker F: Well, the turn off bridge.\nSpeaker F: Yeah.\nSpeaker F: Won't even do that.\nSpeaker F: I mean, just go down.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: And the mountain with the King 988, 888.\nSpeaker F: So it's about 30 minutes to get there.\nSpeaker F: So it leaves us 50 minutes before the plane.\nSpeaker F: Yeah.\nSpeaker F: Great.\nSpeaker F: Okay.\nSpeaker F: So that'll, I mean, still not going to be really easy.\nSpeaker F: Well, particularly for the Burian, we're not staying overnight.\nSpeaker F: We're not bringing anything.\nSpeaker F: We're going to take a little bit.\nSpeaker F: We're going to have a paper.\nSpeaker F: We're going to do a little bit.\nSpeaker F: Don't bring a foot locker and we'll be okay.\nSpeaker F: So staying overnight, I figured he wouldn't need a great big.\nSpeaker F: Oh, yeah.\nSpeaker F: Anyway.\nSpeaker F: Okay.\nSpeaker F: Six a.m.\nSpeaker F: Front.\nSpeaker F: Six a.m. in front.\nSpeaker F: I'll be here.\nSpeaker F: I'll give you my phone number.\nSpeaker F: I'll give you a few minutes.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Wait a minute.\nSpeaker F: Okay.\nSpeaker F: That was the real, real important stuff.\nSpeaker F: I figured maybe wait on the potential goals for the meeting until we talk about what's been going on.\nSpeaker F: So what's been going on?\nSpeaker F: Let me start over here.\nSpeaker I: Well, the duration of the French test data.\nSpeaker I: Well, this is a digital French database, which is a microphone speech.\nSpeaker I: Don't suffer to make it worse.\nSpeaker I: I've had a noise to one part, which is actually the Aurora 2 noises.\nSpeaker I: So this is the training part.\nSpeaker I: And the remaining part are useful testing with one or two kind of noises.\nSpeaker I: So this is almost ready.\nSpeaker I: I'm preparing the HTK baseline for this task.\nNone: Okay.\nSpeaker F: So the HTK baseline, so this is using Melcadstrand.\nSpeaker F: Yeah.\nSpeaker F: Again, I guess the plan is to then give in this.\nSpeaker F: What's the plan again?\nSpeaker F: The plan we've been doing.\nSpeaker F: So just remind me of what you were going to do.\nSpeaker F: You just described what you've been doing.\nSpeaker F: Yeah.\nSpeaker F: So if you could remind me of what you're going to be doing.\nSpeaker I: Yeah.\nSpeaker I: Well, I'm like a cube.\nSpeaker I: I should definitely shoot.\nSpeaker I: Actually, we want to analyze three dimensions, the feature dimension, the training data dimension, and the test data dimension.\nSpeaker I: Well, what we want to do is first we have number for each task.\nSpeaker I: So we have the integer task, the Italian task, the French task, and the Finnish task.\nSpeaker I: So we have numbers with systems.\nSpeaker I: I mean, I mean, you run the work strain on the task data.\nSpeaker I: And then we have systems with neural networks trained on data from the same language, if possible, but using a more generic database, which is for the tickly balance.\nSpeaker F: So we had talked, I guess we had talked at one point about maybe the language ID corpus.\nSpeaker I: Yeah, but this corpus, there is the call OM and the callFranco.\nSpeaker I: The callFranco is for language identification.\nSpeaker I: Anyway, these corpus are telephone speech, so this could be a group of four.\nSpeaker I: Because the speech database are not telephone speech.\nSpeaker I: They are not sample to eight kilohertz, but they are not telephone bandwidth.\nSpeaker F: So it's funny, isn't it? I mean, because this whole thing is for telephone.\nSpeaker I: Yeah, but the idea is to compute the feature before sending them to the...\nSpeaker I: Well, you don't send speech use and features, compute, not the...\nSpeaker F: Yeah, I know, but the reason is...\nSpeaker F: So the point is that it's the features are computed locally, and so they aren't necessarily telephone bandwidth or telephone.\nSpeaker C: Did you happen to find out anything about the OGI multilingual database?\nSpeaker F: Yeah, that's what I meant.\nSpeaker F: I said, there's an OGI language ID, not the callFranco.\nSpeaker F: Yeah, there are also two other databases.\nSpeaker I: One, they call the multilingual database, and another one is 22 language.\nSpeaker I: But it's also a different speech. Oh, they are.\nSpeaker F: Well, but I'm not sure...\nSpeaker F: I mean, the bandwidth shouldn't be such an issue, right?\nSpeaker F: Because this is down sample filtered, right?\nSpeaker F: So it's just the fact that it's not telephone.\nSpeaker F: And there are so many other differences between these different databases.\nSpeaker F: I mean, some of the stuff's recorded in the car, some of the... I mean, there's many different acoustic differences.\nSpeaker F: So I'm not sure...\nSpeaker F: I mean, else we're going to include a bunch of car recordings in the training database.\nSpeaker F: I'm not sure if it's completely rules it out.\nSpeaker F: If our major goal is to have the metacontext, and you figure that there's going to be a mismatch in acoustic conditions, does it make it much worse to sort of add another mismatch if you will?\nSpeaker F: I guess the question is, how important is it to get multiple languages in there?\nSpeaker I: Yeah, but...\nSpeaker I: Yeah.\nSpeaker I: Well, actually, for the moment, if we do not want to use these different databases, we already have an English and French microphone speech.\nSpeaker F: So that's why I think we're using a sort of multilingual...\nSpeaker I: Yeah, for the multilingual part, we were thinking about using these three databases,\nSpeaker F: and the difference in the metacontext.\nSpeaker I: Actually, these three databases are generic databases. So for Italian, which is close to Spanish, French, the attitudes we have, digits training data, and also more general training data.\nSpeaker F: Well, we also have this broadcast news that we were talking about taking off the disk, which is this microchannel data for language.\nSpeaker I: Yeah, brups.\nSpeaker I: Yeah, there is also a timid.\nSpeaker I: Yeah, I would use timid.\nSpeaker F: Yeah, so it's probably a stuff around.\nSpeaker F: Okay, so anyway, the basic plan is to test this cube.\nNone: Yes, thank you.\nSpeaker F: The fillet filled in, yeah.\nSpeaker I: Okay, yeah, and brups, we were thinking that, perhaps, the cross-language issue is not so big of an issue, well-reviewed, brub-fusion, not focused too much on that cross-language stuff.\nSpeaker I: I mean, training on that on the language and testing for another language.\nSpeaker I: Perhaps the most important is to have neural networks trained on the target languages, with general database, general databases.\nSpeaker I: So that, well, the guy who has to develop an application with one language can use the net train on that language, or a generic net, but in other terms,\nSpeaker F: That's how you mean using the net. So if you're talking about for producing these discriminative features, you can't do that, because what they are asking for is a feature set.\nSpeaker F: Right, and so we're the ones who have been weird by doing this training.\nSpeaker F: But if we say, no, you have to have a different feature set for each language, I think this is very, very bad.\nSpeaker F: Oh, yeah.\nSpeaker F: Yeah, I mean, in principle, I mean, concessually, sort of like they want, well, they want a replacement for Melcafster.\nSpeaker F: So they say, okay, this is the year 2000.\nSpeaker F: We've got something much better than Melcafster, it's, you know, Gavaldi-Gook.\nSpeaker F: And so Gavaldi-Gook features, but these Gavaldi-Gook features are supposed to be good for any language.\nSpeaker F: Because you don't know who's going to call, and you know, I mean, so it's, it's, it's, how do you know what language?\nSpeaker F: So it picks up the phone.\nSpeaker F: So this is their English.\nSpeaker F: So it picks up the phone, right?\nSpeaker I: And you pick up the application, there is a target, the English one, the application.\nSpeaker F: So, yeah, you pick up the phone.\nSpeaker F: Yeah, you talk the phone and it sends features in.\nSpeaker F: Okay, so the phone doesn't know what your language is.\nSpeaker I: If it's the phone, but, but that's the image that they could be the, the server side.\nSpeaker F: It could be, but that's the image they have.\nSpeaker F: So that's, that's, I mean, one could argue all over the place about how things really will be in 10 years.\nSpeaker F: But the particular image that the cellular industry has right now is that it's distributed to be recognition where the probabilistic part and the semantics and so forth are all on the servers, and you compute features on the phone.\nSpeaker F: So that's, that's what we're involved.\nSpeaker F: You might, you might not agree that that's the way it will be in 10 years, but that's, that's what they're asking for.\nSpeaker F: So, so I think that the, it is an important issue whether it works cross-language.\nSpeaker F: Now it's the OGI folks perspective right now that probably that's not the biggest deal.\nSpeaker F: And that the biggest deal is the, you know, acoustic environment mismatch.\nSpeaker F: And they may very well be right, but I was hoping we could just do a task and determine if that was true.\nSpeaker F: That's true. We don't need to worry so much. Maybe, maybe we have a couple of languages in the training set, and that gives us enough breadth that, that, that the rest doesn't matter.\nSpeaker F: The other thing is this notion of training to, which I guess they're starting to look at up there, training to something more like articulate-toy features.\nSpeaker F: And if you have something that's just good for distinguishing different articulate-toy features, that should just be good across, you know, either into languages.\nSpeaker F: Yeah, so I don't, I don't, unfortunately, I don't, I see what you're coming, where you're coming from, I think, but I don't think we can, you know,\nSpeaker I: so we really have to do tests with real cross-language, I mean, for instance, training on English and testing in Italian, or we can train, or else, can we train a net on a range of languages, which can include the test, the target language.\nSpeaker F: Yeah, so there's, there's, this is complex. So, ultimately, is this saying, I think it doesn't fit within their image that you switch nets based on language.\nSpeaker F: Now, can you include the target language?\nSpeaker F: From a pure standpoint, it'd be nice not to, because then you can say, because surely someone is going to say at some point, okay, so you put in the German and the Finnish, now what do you do when somebody has Portuguese?\nSpeaker F: You know, and however, you aren't, it isn't actually a constraint in this evaluation.\nSpeaker F: So I would say, if it looks like there's a big difference to put it in, then we'd make note of it, and then we'd probably put in the other, because we have so many other problems in trying to get things to work well here, that it's not so bad as long as we know it, and we say, look, we did do this.\nSpeaker C: So ideally, what you'd want to do is you'd want to run it with and without the target language in a training set for a wide range of languages.\nSpeaker C: And that way you can say, well, you know, we're going to build it for what we think are the most common ones, but if that somebody is with a different language, you know, here's what, here's what's likely to happen.\nSpeaker F: And the truth is that it's not like there are, I mean, although there are thousands of languages, from the point of view of cellular companies, there aren't.\nSpeaker F: There's, you know, there's 50, there's something, you know, so, and they aren't, you know, the exception of Finnish, which I guess is pretty different from most things.\nSpeaker F: It's, most of them are like at least some of the others, I guess, that's why I guess this vanishes like the town.\nSpeaker F: I guess Finnish is a little bit like Hungarian, I suppose, right?\nSpeaker F: I don't know.\nSpeaker F: I don't know.\nSpeaker F: I know that, I mean, I'm not like this, but I guess Hungarian and Finnish, and one of the languages from the former Soviet Union is sort of the same family, but it's just this film.\nSpeaker F: The countries that are pretty far apart from one another, people are rode in on horses.\nSpeaker H: Okay.\nSpeaker H: Oh, my turn.\nSpeaker H: Okay.\nSpeaker H: Let's see, I spent the last week looking over Stefan Schulder, and understanding some of the data.\nSpeaker H: I reinstalled the HTK, the free version.\nSpeaker H: So, everybody's now using 3.0, which is the same version that OGI is using.\nSpeaker H: Oh, good.\nSpeaker H: Yeah, so, without any licensing big deals or anything like that.\nSpeaker H: And so, we've been talking about this cube thing, and it's beginning more and more looking like the, the Borg cube thing.\nSpeaker H: It's really gargantuan.\nSpeaker H: But, I...\nSpeaker F: I'm not going to be a similar one.\nSpeaker H: I'm not a resistant person.\nSpeaker H: Exactly.\nSpeaker H: Yeah, so, I've been looking at a Timmit stuff.\nSpeaker H: The stuff that we've been working on with Timmit, trying to get a, a labels file so we can train up a net on Timmit, and test the difference between this net train on Timmit and a net trained on digits alone, and seeing if it hurts or helps.\nSpeaker F: And again, just to clarify, when you're talking about training, have you been talking about training, have you been net for Tandem?\nSpeaker F: Yeah, yeah, I'm very...\nSpeaker F: And the inputs are POP and Delta.\nSpeaker H: Well, the inputs are one dimension of the cube, which we've talked about.\nSpeaker H: It being PLP, MFCCs, J-Rasta, LVA.\nSpeaker F: Yeah, but your initial things you're making one choice there.\nSpeaker H: Yeah, just PLP.\nSpeaker H: I haven't decided on an initial thing.\nSpeaker H: Probably something like PLP.\nSpeaker F: Yeah.\nSpeaker F: So, you take PLP and you...\nSpeaker F: You use HDK with it with the transform features using the neural net that's trained, and the training could either be from digits itself or from Timmit.\nSpeaker F: Right.\nSpeaker F: And then the testing would be these other things which might be foreign language.\nSpeaker F: Right.\nSpeaker F: I see.\nSpeaker F: I get in the picture about the cube.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker F: I mean, those listening to this will not have a picture either, so I guess I'm not in worse off.\nSpeaker F: Somebody just told me the cube.\nSpeaker F: It sounds...\nSpeaker F: I get it.\nSpeaker F: I think I get it.\nSpeaker C: Yeah.\nSpeaker C: So, when you said that you're getting the labels for Timmit, what do you mean?\nSpeaker H: Oh, I'm just transforming them from the standard Timmit transcriptions to do a nice long, huge P-file.\nSpeaker C: Were the digits hand labeled for phones?\nSpeaker C: Or were they those labels automatically?\nSpeaker H: Oh, yeah. Those were automatically derived by Dan using embedded training in the alignment.\nSpeaker F: Oh, but which Dan?\nSpeaker H: Alice.\nSpeaker H: Okay.\nNone: Yeah.\nSpeaker G: So.\nSpeaker C: I was just wondering because that test...\nSpeaker C: I think you're doing this test because you wanted to determine whether or not having general speech performs as well as having specific speech.\nSpeaker C: That's right.\nSpeaker F: Well, especially when you go over the different languages again, because you would...\nSpeaker F: The different languages have different words from the physics of it.\nSpeaker C: I was just wondering if the fact that Timmit, who is in a hand labeled stuff from Timmit, might be confused the results that you get.\nSpeaker F: I think it would, but on the other hand, it might be better.\nSpeaker C: But if it's better, it may be better because it was handed.\nSpeaker F: Yeah, so probably use it.\nSpeaker F: I mean, you know, I guess I'm Sonic Cavalier, but I mean, I think the point is you have a bunch of labels and hand-marked...\nSpeaker F: I guess actually Timmit was not entirely hand-marked.\nSpeaker F: It was automatically first and then corrected.\nSpeaker F: But it might be a better source.\nSpeaker F: So, you're right. It would be another interesting scientific question to ask.\nSpeaker F: Is it because it's a broad source or because it was, you know, carefully?\nSpeaker F: And that's something you could ask.\nSpeaker F: But given limited time, I think the main thing is if it's a better thing for you to run across languages on this training tandem system.\nSpeaker C: What about the differences in the phone sets?\nSpeaker H: Between languages?\nSpeaker C: No, between Timmit and the physics.\nSpeaker H: Oh, right.\nSpeaker H: Well, there's a mapping from the 61 phonemes in Timmit to 56, the XE56.\nSpeaker H: And then the digits phonemes, there's about 22 or 24 of them.\nSpeaker H: Out of that 56?\nSpeaker H: Out of that 56.\nSpeaker H: So, it's definitely broader.\nSpeaker I: But actually the issue of phonemes, phonemapics, will arise when we will do several languages.\nSpeaker I: Because some phonemes are not in every languages.\nSpeaker I: So, we plan to develop a subset of phonemes that includes all the phonemes of training languages.\nSpeaker I: Using the two words, kind of 100 of boots.\nSpeaker I: Yeah, super set.\nSpeaker I: Yeah, super set.\nSpeaker A: I look for some per form.\nSpeaker A: For English, American English.\nSpeaker A: And the language who have more phonemes are the English of the language.\nSpeaker A: But, for example, in Spain, in the Spanish half, several phonemes that does not appear in the English and we are too complete.\nSpeaker A: But for that, in this, we must do a lot of work because we need to generate the transcription for the database that we have.\nSpeaker E: Other than the languages, there are reasons not to use Timmit phonemes.\nSpeaker E: Because it's larger.\nSpeaker E: Is it supposed to be X-E?\nSpeaker H: Oh.\nSpeaker H: I mean, why map the 61 to 56?\nSpeaker H: I don't know.\nSpeaker F: I figured if that happened to start with you or was it, or was there, yeah, so it's what you did that.\nSpeaker F: But I think basically there were several phonemes that were just hardly over there.\nSpeaker C: Yeah, and I think some of them, they were making distinctions between silence at the end and silence at the beginning.\nSpeaker C: And really, most silence.\nSpeaker C: And it was things like that that got it mapped down.\nSpeaker F: Yeah, especially a system like ours, which is a discriminative system.\nSpeaker F: You know, you really ask him to map the learn.\nSpeaker C: There's not much different.\nSpeaker C: The ones that are gone, I think, or they also, they also, and Timmit had like a glottal stop, which basically a short period of silence.\nSpeaker C: Well, we have that now too.\nSpeaker C: I don't know.\nSpeaker F: It's actually pretty common that a lot of the recognition systems people use have things like, like say, 39, simple symbols, right?\nSpeaker F: And then they get the variety by by bringing in the context.\nSpeaker F: And that context.\nSpeaker F: So we actually have them usually large number.\nSpeaker F: We'll be 10 to use here.\nSpeaker F: So actually, maybe now you've got me sort of intrigued, but there's...\nSpeaker F: Can you describe what's on the cube?\nSpeaker H: I think that's a good idea to talk about the whole cube, and maybe we can cut out sections in the cube for people to work on.\nSpeaker H: Okay.\nSpeaker F: So even the one in the quarter doesn't, since we're not running a video camera, we'll get this.\nSpeaker F: If you use a board, it'll help us anyway.\nSpeaker F: Okay, point out one of the limitations of this.\nSpeaker F: You've got to worry less, right?\nSpeaker H: Yeah, worry less.\nSpeaker H: Can you walk around too? No.\nSpeaker H: Okay, well, basically the cube will have three dimensions.\nSpeaker H: First dimension is the features that we're going to use.\nSpeaker H: And the second dimension is the training corpus.\nSpeaker H: And that's the training on the discriminant neural net.\nSpeaker H: And the last dimension happens...\nSpeaker F: Yeah, so the training for HDK is always...\nSpeaker F: That's always set up for the individual test, right?\nSpeaker F: There's some training data and some test data.\nSpeaker F: Right, right.\nSpeaker H: This is for ANN only.\nSpeaker H: And the training for the HDK models is always fixed for whatever language you're testing on.\nSpeaker H: And then there's the testing corpus.\nSpeaker H: So then I think it's probably instructive to go and show you the features that we were talking about.\nSpeaker H: So let's see.\nSpeaker H: It's a healthy L.\nSpeaker H: Okay, what?\nSpeaker H: BLP.\nSpeaker H: FST.\nSpeaker I: MSG.\nSpeaker H: MSG.\nSpeaker I: J-Rasta.\nSpeaker H: J-Rasta.\nSpeaker H: And J-Rasta L.E.\nSpeaker I: J-Rasta L.E.\nSpeaker I: J-Rasta L.E.\nSpeaker H: Multiband.\nSpeaker H: Multiband.\nSpeaker I: So there would be multiband before...\nSpeaker I: Before the tour, I mean...\nSpeaker H: Yeah, just the multiband features, right?\nSpeaker H: Yeah.\nSpeaker I: So something like the CT with advanced.\nSpeaker I: And then multiband after networks, meaning that you would have...\nSpeaker I: You run networks, the script that you run, the script for each band.\nSpeaker I: And using the outputs of these networks or the linear outputs.\nSpeaker I: Something like that.\nSpeaker C: Yeah.\nSpeaker C: What about no?\nSpeaker C: Oh.\nSpeaker C: You don't include that because it's part of the base.\nSpeaker F: We do have a baseline system that's the smell capture, right?\nSpeaker I: But...\nSpeaker I: Not for the A and M.\nSpeaker I: Okay.\nSpeaker I: So yeah, we could add a CT.\nSpeaker F: Probably should.\nSpeaker F: At least conceptually, you know, it doesn't mean you actually have to do it.\nSpeaker F: Conceptually, it makes sense, so it's a baseline.\nSpeaker C: It'd be an interesting test just to have...\nSpeaker C: Just to do MFCC with the neural mat.\nSpeaker C: And everything else the same.\nSpeaker C: Compare that with just that MFCC without the net.\nSpeaker C: Yeah.\nSpeaker H: I think Dan did some of that in his previous Aurora experiments.\nSpeaker H: And with the net, it's wonderful.\nSpeaker H: And now the net's just baseline.\nSpeaker F: I think OGI folks have been doing it too.\nSpeaker F: Because I think they're for a bunch of their experiments.\nSpeaker F: They used a no-cam study.\nSpeaker F: Yeah, actually.\nSpeaker F: Of course, that's there.\nSpeaker F: It's here.\nSpeaker H: Okay.\nSpeaker H: Okay.\nSpeaker H: For the training corpus, we have the digits from the various languages.\nSpeaker H: English, Spanish, French, what else do we have?\nSpeaker I: The finish.\nSpeaker I: The finish.\nSpeaker A: Where did that come from?\nSpeaker A: Italy, no.\nSpeaker A: Italy, yes.\nSpeaker A: Italian.\nSpeaker C: Is that distributed with Aurora or Aurora?\nSpeaker I: So English, the finish and Italian are Aurora.\nSpeaker I: And Spanish and French is something that we can use in addition to Aurora.\nSpeaker I: What?\nSpeaker F: Yes, the German brother, Spanish and stuff on both the French.\nSpeaker H: Okay.\nSpeaker H: And...\nSpeaker H: Oh, yeah.\nSpeaker H: Is it French French or Belgian?\nSpeaker H: It's French French.\nSpeaker A: French French.\nSpeaker A: French French.\nSpeaker A: French French.\nSpeaker F: I think that is more important than French French.\nSpeaker F: Yeah, probably.\nSpeaker F: Yeah.\nSpeaker F: Yeah, everybody always insists the Belgium is absolutely pure French.\nSpeaker F: But he says those pre-easions talk about it.\nSpeaker F: He likes Belgian fries too.\nSpeaker H: And then we have a broader corpus like Timit.\nSpeaker H: Timit's so far.\nSpeaker H: Spanish.\nSpeaker H: Spanish stories.\nSpeaker A: I buy you tea.\nSpeaker H: What about TI digit?\nSpeaker H: All these Aurora data is derived from TI digits.\nSpeaker H: Basically, they corrupted it with different kinds of noises at different SNR levels.\nSpeaker F: And I think Stefan was saying there's some broader material in the French also.\nSpeaker F: Yeah, we could use it.\nNone: Okay.\nSpeaker A: French data.\nSpeaker A: Spanish stories.\nSpeaker H: Spanish.\nSpeaker H: Spanish something.\nSpeaker H: Yeah.\nSpeaker H: Okay.\nSpeaker D: The Aurora people actually corrupted themselves or just specifies a signal on signal noise ratio.\nSpeaker H: They corrupted it themselves, but they also included the noise files for us, right?\nSpeaker H: Yeah.\nSpeaker H: So we can go ahead and corrupt other things.\nSpeaker F: I'm just curious to come here.\nSpeaker F: I mean, I couldn't tell if you were joking or is it Mexican Spanish?\nSpeaker F: No, no, no.\nSpeaker A: Oh, no, no.\nSpeaker F: Spanish is Spanish.\nSpeaker F: Spanish is Spanish.\nSpeaker F: Spanish is Spanish.\nSpeaker H: Okay.\nNone: Spanish is Spanish.\nSpeaker F: Spanish is Spanish.\nSpeaker F: Spanish is Spanish.\nSpeaker F: Yeah, we're really covered.\nSpeaker F: Okay.\nSpeaker F: Yeah, no different.\nSpeaker I: Yes.\nSpeaker I: From Paris.\nSpeaker I: Oh, from Paris.\nSpeaker I: Yeah.\nSpeaker H: Timit's from lots of different places.\nSpeaker F: It's from Texas.\nSpeaker E: It's from Texas.\nSpeaker F: It's not really from the US either.\nSpeaker H: Okay.\nSpeaker H: And within the training corpus, we're thinking about training with noise.\nSpeaker H: So incorporating the same kinds of noises that Aurora is incorporating in their training corpus.\nSpeaker H: I don't think we were given the unseen noise conditions.\nSpeaker F: I think what they were saying was that for this next test, there's going to be some of the cases where they have the same type of noises you were given beforehand, and in some cases where you're not.\nSpeaker H: Okay.\nSpeaker F: So presumably that'll be part of the topic of analysis of the test results.\nSpeaker F: How will you do when it's matching noise?\nSpeaker F: How will you do when it's not?\nSpeaker F: Right.\nSpeaker H: I think that's right.\nSpeaker H: So I guess we can't train on the unseen noise conditions.\nSpeaker H: Well, it matters.\nSpeaker F: Not seen.\nSpeaker H: It matters, yeah.\nSpeaker G: Yeah.\nSpeaker F: I mean, it does seem to me that a lot of times when you train with something that's at least a little bit noisy, it can help you out in other kinds of noise, even if it's not matching, just because there's some more variance that you've built into things.\nSpeaker F: But exactly how well your work will, how near it is to what you have at the time.\nSpeaker F: Okay, so that's your training corpus and then your testing corpus?\nSpeaker H: The testing corpus are just the same one that's oral testing.\nSpeaker H: And that includes the English, Italian, Finnish.\nSpeaker H: We're going to get German, right?\nSpeaker F: Well, so yeah, the final test on the English.\nSpeaker H: The Spanish perhaps?\nSpeaker H: Oh, yeah, we can test on the Spanish.\nSpeaker I: But the oral test.\nSpeaker F: Oh, yeah.\nSpeaker F: Oh, there's Spanish testing in Europe.\nSpeaker I: Not yet, but...\nSpeaker A: Yeah, it's providing.\nSpeaker I: They're preparing it.\nSpeaker I: Well, according to Hineki, it would be good at this at the end of November.\nSpeaker F: Okay, so I think like seven things in each column, so that's 343 different systems that are going to be developed.\nSpeaker F: There's three of you.\nSpeaker B: 100.\nSpeaker B: What about noise conditions?\nSpeaker B: What?\nSpeaker F: Don't we need to put in the column for noise conditions?\nSpeaker H: You're just trying to be difficult.\nSpeaker H: Well, when I put these testings on there, I'm assuming there's three tests.\nSpeaker H: Type A, type B and type C.\nSpeaker H: And they're all going to be tested with one training of the HTK system.\nSpeaker H: Test all three different types of noise conditions.\nSpeaker H: Test A is like a match, noise.\nSpeaker H: Test B is a slightly mismatched.\nSpeaker H: And test C is a mismatched channel.\nSpeaker B: And do we do all our training on clean data?\nSpeaker H: No, no, no.\nSpeaker H: We're going to be training on the noise files that we do have.\nSpeaker F: So, yeah, so I guess the question is how long does it take to do a training?\nSpeaker F: I mean, it's not totally crazy.\nSpeaker F: I mean, a lot of these are built-in things.\nSpeaker F: We know we have programs that compute POP, we have MSG, we have JVET.\nSpeaker F: You know, a lot of these things, which is kind of how people take in a huge amount of developments, just trying it out.\nSpeaker F: So we actually can't be quite a few experiments.\nSpeaker F: But how long does it take?\nSpeaker F: We think one of these trainings.\nSpeaker C: That's a good question.\nSpeaker C: What about combinations of them?\nSpeaker F: Oh, yeah, that's right.\nSpeaker F: I mean, because so, for instance, I think the major advantage of MSG, yeah.\nSpeaker F: What's the point?\nSpeaker F: The major advantage of MSG, I see that we've seen in the past, is combined with POP.\nSpeaker H: Now this is turning into a fourth dimensional queue.\nSpeaker C: Oh, you just select multiple things on the one dimension.\nSpeaker H: Oh, yeah, okay.\nSpeaker F: Yeah, so I mean, you don't want to see seven cheeses too.\nSpeaker F: POP is 21.\nSpeaker F: Different combinations.\nSpeaker D: It's not a complete set of combinations, don't we?\nSpeaker D: What?\nSpeaker D: It's not a complete set of combinations, though.\nSpeaker G: Yeah, I hope not.\nSpeaker H: Yeah, that would be...\nSpeaker F: Yeah, so POP and MSG, I think we definitely want to try, because we've had a lot of good experience with putting this together.\nSpeaker C: When you do that, you're increasing the size of the inputs to the net.\nSpeaker C: Do you have to...\nSpeaker C: Well, so it doesn't increase the number of trainings.\nSpeaker C: I'm just wondering about number of parameters in the net.\nSpeaker C: Do you have to worry about keeping that the same length?\nSpeaker F: I don't think so.\nSpeaker E: There's a computation limit, doesn't it?\nSpeaker E: Yeah, I mean, just more computation.\nSpeaker E: Excuse me?\nSpeaker E: Isn't there like a limit on the computation, or latency, or something like that, for a large amount?\nSpeaker F: Oh, yeah, we haven't talked about that in a new then, and all that.\nSpeaker F: Yeah.\nSpeaker F: So it's not really a limit.\nSpeaker F: What it is is that there's...\nSpeaker F: There's a...\nSpeaker F: Just...\nSpeaker F: If you're using a megabyte, then I'll say that's very nice, but of course, you will remember growing a cheap cell phone.\nSpeaker F: And I think the computation isn't so much a problem.\nSpeaker F: I think it's more than that, right?\nSpeaker F: And the expensive cell phones, expensive handhelds, and so forth, are going to have lots of memory.\nSpeaker F: So it's just that, as people see, the cheap cell phones is being the biggest market.\nSpeaker F: So...\nSpeaker F: But yeah, I was just realizing that actually it doesn't explode out.\nSpeaker F: It's not only through the cell, but it's...\nSpeaker F: But it doesn't really explode out the number of trainings, because these are all trained individually.\nSpeaker F: And so if you have all of these nets trained in some place, then you can combine their outputs and do the care transformation and so forth.\nSpeaker F: And so what it uploads out is the number of test things.\nSpeaker F: And the number of times you do that last part, that last part I think is so...\nSpeaker F: It's got to be pretty quick.\nSpeaker F: And it's just running the data through...\nSpeaker C: Well, you got to do the care transformation.\nSpeaker C: What about a net that's trained on multiple languages?\nSpeaker C: Is that just separate nets for each language, then combined, or is that actually one net trained?\nSpeaker I: Probably one net.\nSpeaker I: One would think one.\nSpeaker F: But I don't think we tested that.\nSpeaker I: So in the broader training course, we can use the three or combinations of two languages.\nSpeaker C: In one net.\nSpeaker F: Yeah, so I guess the first thing is, if we know how long a training takes, if we can train up all these combinations, then we can start working on testing of them individually and in combination.\nSpeaker F: And putting them in combination, I think is not as much a computationally as they're training with the net in the first place.\nSpeaker I: It's not too much.\nSpeaker I: But there is a testicle, so it's nice training.\nSpeaker F: It's a lot of things to learn.\nSpeaker F: How long does it take for an APK training?\nSpeaker I: It's around six or so, I think.\nSpeaker A: For training at this thing?\nSpeaker A: For the Italian, maybe one day.\nSpeaker A: Running on what?\nSpeaker F: I don't know what is the value.\nSpeaker F: I don't know what the value is.\nSpeaker E: I don't know.\nSpeaker E: I don't know.\nSpeaker I: It's not so long because the data is about 30 yards of speech.\nSpeaker F: There's no way we can begin to do any significant amount here unless we use multiple machines.\nSpeaker F: What machines are fast, what machines are used a lot, are we still using P-Make?\nSpeaker F: Once you get the basic thing set up, you have all these combinations.\nSpeaker I: I would say two days.\nSpeaker A: I think you folks are probably the only ones using it.\nSpeaker F: It's faster to do it on the spurred board.\nSpeaker F: It's still a little faster on the spurred board.\nSpeaker H: Adam did some testing.\nSpeaker H: You run on a spurred and then you can do other things on your computer.\nSpeaker F: You could set up 10 different jobs on spurred boards and have 10 other jobs running on different computers.\nSpeaker F: It's going to take that sort of thing.\nSpeaker F: I kind of like this because we have very limited time.\nSpeaker F: We have quite a bit of computational resource available.\nSpeaker F: We can look across the institute and now a little things are being used.\nSpeaker F: We've gotten before about voice-down voice silence, detection features, and I think it's a great thing to go to.\nSpeaker F: I like about this.\nSpeaker F: This is what you're thinking of doing in short terms.\nSpeaker F: Adam sort out about what's the best way to really attack this as a mass problem in terms of using many machines.\nSpeaker F: We can then present to them what it is that we're doing.\nSpeaker F: We can pull things out of this list that we think they are doing sufficiently.\nSpeaker H: How they go to the net trade region?\nSpeaker H: For the net trade on digits, we have been using 400 border hidden units.\nSpeaker H: The digit's nets will be correspond to about 20 phonemes.\nSpeaker H: We're actually broader classes, actually finer classes.\nSpeaker A: Carmen, did you have something else to add?\nSpeaker A: I tried to do a different thing with the HTG program.\nSpeaker F: I don't know what is better if you look at us or J. Rasta.\nSpeaker F: J. Rasta is more complicated.\nSpeaker F: J. Rasta is more complicated.\nSpeaker F: There are more ways that it can go wrong.\nSpeaker A: I think to recognize the Italian digit with the NetWorps and also to try another NetWorps with the Spanish digit.\nSpeaker A: The data base was at difficult work last week with the level of time that I have the difference with the level of time.\nSpeaker F: I'm sorry.\nSpeaker A: The Spanish level was in different formats for the program to train the NetWorps.\nSpeaker A: You just have to be properly converting the labels.\nSpeaker A: I don't know what I asked to do.\nSpeaker A: I think that with LiveCat I can transfer to ACS format.\nSpeaker A: I want to put ACS format to ACS format and then use LiveCat to do that.\nSpeaker F: It seems like there are some peculiarities of the dimensions that are getting sorted out.\nSpeaker F: We have a lot more computation.\nSpeaker F: I was thinking two things.\nSpeaker H: I thought of this as not in stages but more on the time axis.\nSpeaker F: I was thinking of how much you can realistically do.\nSpeaker F: I was thinking of how much you can do.\nSpeaker F: I still think we could do a lot of it.\nSpeaker H: The second thing was about scratch space.\nSpeaker F: I want to clarify my point about that.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker C: I want to clarify my point about how much you can do.\nSpeaker I: I want to clarify my point about how much you can do.\nSpeaker I: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: I want to clarify my point about how much you can do.\nSpeaker F: The last topic I had here was Dave's fine offer to do something.\nSpeaker F: I was thinking perhaps if additionally to all this experiment with this not really research,\nSpeaker I: it's not really research, but it's running programs. The closer look at the speed noise detection or voice sound detection.\nSpeaker C: The thing that Sue Neill was talking about with the labels, the thing that was running through these models very quickly,\nSpeaker F: and maybe that's the only problem I have with it is the same reason why I thought it would be a good thing to do. Let's fall back to that.\nSpeaker F: That's good.\nSpeaker F: What an additional clever person could help with when we're really in crunch or time.\nSpeaker F: So over the years, if he's interested in voice and voice silencing, he could do a lot.\nSpeaker F: I think it's a good thing to do with the holidays and the middle of it to get a lot done.\nSpeaker F: The very fact that it is just work and it's running programs and so forth is exactly why it's possible that some piece of it could be handed to someone to do because it's not.\nSpeaker F: That's a question. We don't have to solve it right this second, but we can think of some piece that's well defined that he could help with.\nSpeaker C: What about training up a multilingual map?\nSpeaker A: It's good to have the label between them and Spanish or something like that.\nSpeaker F: So what we were just saying was that I was arguing for if possible coming up with something that really was development and wasn't research because we have a time crunch.\nSpeaker F: So if there's something that would save some time for someone else to do, then we should think of that first.\nSpeaker F: So I think that's the only way to do a core job is to do a core job.\nSpeaker F: So I think that's the only way to do a core job is to do a core job.\nSpeaker F: And the other tricky thing is that we are, even though we don't have a strict prohibition on memory size and computation and complexity, clearly there's a limitation to it.\nSpeaker F: So I think that's one of the very important things to do with the organization, at least some kind of harmonic, or something.\nSpeaker F: This is another whole thing, take a while to develop.\nSpeaker F: And then one of the things along with current speech recognition is that we really use the whole way of the harmonious information.\nSpeaker F: So I think the other suggestion just came up was what about having worked on the multi-lingual super-set and coming up with that and training and that on that.\nSpeaker F: Is that our multiple database? What would you think? What would this task consist of?\nSpeaker I: Yeah, it would consist in creating the super-set, modifying the labels for matching the super-set.\nSpeaker F: So you're creating the changing labels on timid or on multiple languages? Yeah, with the treated languages.\nSpeaker H: So you have to create a mapping from each language to the super-set.\nSpeaker H: So you have a machine readable IPA sort of thing. And they have a website that Stefan was showing us has all the English phonemes and their Sampa correspondent phonemes.\nSpeaker H: And then they have Spanish and German have all sorts of languages mapping to the Sampa phonemes.\nSpeaker A: You're comparing the Sampa's version to Albaist.\nSpeaker C: international. No, it's it's saying that you use a special die critical stuff which you can't you can't print out a husky. So the sample is just mapping. Got it. What does OGI\nSpeaker F: have done anything about this issue? Do they have do they have any kind of\nSpeaker I: superset they already have? I don't think so well. They they're going actually the other way defining funny blisters. So they just throw the speech from all\nSpeaker C: different languages together then clustered in the 60 or 50 or whatever. I think they've\nSpeaker I: not done it during multiple language yet but what they did is to training English nets with all the phonemes and then training English nets with kind of 17 I think it was 70 automatic root glasses. Yeah automatically dried but yeah I think so. And the result was that apparently when testing on cross language it was better but you didn't add didn't have all the results when you showed me that\nSpeaker F: but so that doesn't make an interesting question though. Is there some way that we should tie into that? Right I mean if if in fact that is a better thing to do should we have the training with our own categories and now we're saying well how do we have to cross language in one way is to come to the superset but they're trying to come up with clusters.\nSpeaker F: Do we think there's something wrong with that? I think it does something wrong with\nSpeaker I: okay well because well for a moment we're testing on digits perhaps you using broad phonemes classes it's okay for classifying your digits but as soon as you will have more words words can differ with only a single phonemes which could be the same class. So\nSpeaker F: right although you are not using this for the future generation. Yeah but you will ask the net to put\nSpeaker C: a one for the plus so you're saying there may not be enough information coming out of a net to\nSpeaker D: help discriminate the words. Like most confusions are with it from classes. I think\nSpeaker E: Larry was saying like obstetricians are only confused with obstetricians. So maybe we could look\nSpeaker H: at articulatory. Did they not do that? I don't think so. They were looking at both\nSpeaker F: phonemes but they were talking about it but that sort of a question of what they did because that's the other route to go. So suppose you don't really market it to really market your features you really want to look at the acoustics and see where we can get them. So the second class way of doing it is to look at the phones that are labeled and translate them into acoustic and articulatory features. It won't really be right you won't have these\nSpeaker C: old and laughing things. So the targets of the net are these articulatory features. Right. But that implies that you can have more than one honor at a time.\nSpeaker F: That's right. You either do that or you have an fulfillment.\nSpeaker F: I see. And I don't know if our software, if the version of the quick net that we're using allows for that.\nSpeaker H: Do you know? It allows for multiple targets being one.\nSpeaker H: We have gotten soft targets to work.\nSpeaker F: To work that way. Yeah. Okay. So that's another thing that could be that.\nSpeaker F: We could just translate instead of translating to a superset just translate to a articulatory feature. So the articulatory features are training that.\nSpeaker F: The fact even though it's a smaller number, it's still fine because you have the combinations.\nSpeaker F: So in fact it has every distinction in it.\nSpeaker F: But you should go across on your just.\nSpeaker C: We do an interesting thing. It's very not a fact too.\nSpeaker C: We could, if you had the phone labels you could replace them by their articulatory features and then feed in a vector with those things turned on based on what there's supposed to be for each phone.\nSpeaker C: Let's see if it's a big win. Do you know what I'm saying?\nSpeaker C: So I mean if your net is going to be outputting a vector of basically well it's going to have probabilities but let's say they were ones and zeros then you know for each.\nSpeaker C: I don't know if you know this for your testing data but if you know for your test data what the string of phones is and you have them aligned then you can just instead of going through the net just create the vector for each phone and feed that in to see if that data helps.\nSpeaker C: Let me think about this as I was talking with TNIC and he said that there was a guy at AT&T who spent 18 months working on a single feature and because they had done some cheating experiments.\nSpeaker F: This was the guy that we were just talking that we saw in campus who's the lawyer's fault.\nSpeaker F: He was on our hands.\nSpeaker F: Right okay.\nSpeaker C: So he was doing it.\nSpeaker C: And they had done a cheating experiment or something right?\nSpeaker C: He didn't mention that part.\nSpeaker C: But he said that I guess before they had him work on this they had done some experiment where if they could get that one feature right it dramatically improved the results I was thinking you know if you think about this that it would be interesting experiment just to see you know if you did get all of those right.\nSpeaker F: Should be because if you get all of them in there that defines all of the phones that's the squirtle I'm saying.\nSpeaker F: Right and you've got all the phones right so that doesn't help us.\nSpeaker F: Oh yeah it would be an interesting cheating experiment because we are using it in this funny way we're converting it into features.\nSpeaker C: And then you also don't know what error they've got on the HTK side you know it sort of gives you the best you could hope for.\nSpeaker E: The soft training of the nets still requires the vector to sum to one.\nSpeaker E: This one up to one.\nSpeaker E: So you can't really feed it like two articulatory features that are on at the same time with ones because it will kind of normalize them down to one half or something like that.\nSpeaker I: But the rest you have destroyed the binon.\nSpeaker I: Not many areas.\nSpeaker I: You know you can use it.\nSpeaker H: It's a six.\nSpeaker H: No it's actually sigmoid x.\nSpeaker I: So if you choose sigmoids.\nSpeaker H: I think apparently the linear outputs.\nSpeaker E: Linear outputs?\nSpeaker E: No what you want.\nSpeaker E: If you're going to do a KL transfer mod.\nSpeaker H: Right.\nSpeaker H: Right.\nSpeaker H: But during the training we're trained on sigmoid x and then at the end just chop off the final non-miniarity.\nSpeaker F: We're up there.\nSpeaker F: Oh no.\nSpeaker F: OK.\nNone: Really fun.\nNone: Yes.\n"}, {"meeting_id": "TS3011c", "summary": "This is the third meeting of the design group. At first, Marketing revealed that ease of use, innovation and a fancy look and feel could be important to customers and the team discussed what colour and material they should use. Then Industrial Designer introduced three casing possibilities: uncurved, single-curved and double-curved. Next, the team talked about whether to use kinetic energy supply as an energy source. In terms of user interface controls, Industrial Designer proposed to add a scroll-wheel that could be integrated with buttons. Finally, the team summed up their decisions and Project Manager asked User Interface Designer to create a prototype next.", "dialogue": "Speaker D: Do you need to change anything on it?\nSpeaker D: Because otherwise I will already open it.\nSpeaker B: I don't think so.\nSpeaker B: Unless things have suddenly changed again.\nSpeaker B: Much changes?\nSpeaker B: I don't know. Maybe...\nSpeaker D: No, I don't like the last time.\nSpeaker D: I only need the same information.\nNone: Okay.\nSpeaker D: Hello, Mr. Pian.\nSpeaker C: Hello, Mr. Pian.\nSpeaker D: I believe Ms. Dental is with us as well.\nSpeaker D: In the control room.\nSpeaker C: Well, that's where the thinking goes on.\nSpeaker D: It's the true again, always late.\nSpeaker D: Bonjour, no.\nSpeaker C: I think you should punish him.\nSpeaker C: I see some interesting...\nSpeaker C: You wish.\nSpeaker D: People, welcome back.\nSpeaker C: The third meeting.\nSpeaker D: I have some points.\nSpeaker D: I would like to point out.\nSpeaker D: First of all, if you make minutes yourself as well, like Sebastian does, could you put them on the shared folder?\nSpeaker D: If you do not make minutes, no problem.\nSpeaker D: But it's easy for me to see what you wrote down, so I can use that in the report.\nSpeaker D: The second thing...\nSpeaker D: I was thinking to myself, I have this little remote control, and I'm talking to it, but I still need to point to the television, because it works with infrared.\nSpeaker D: That's quite strange.\nSpeaker D: Yes.\nSpeaker D: Okay. So, we'll come to that later, I think.\nSpeaker D: The agenda for now, are there any pre-discussion questions?\nSpeaker C: No, no.\nSpeaker D: Okay.\nSpeaker D: We will have your individual presentations than the decision on the remote control concept and the closing, 40 minutes in total for this.\nSpeaker D: So, I think we can immediately start with the individual presentations.\nSpeaker D: Okay.\nSpeaker D: The progress you made.\nSpeaker D: I think it might be smart to look at Root's information first, because I understood there are some significant changes in the market situation.\nSpeaker B: All right.\nSpeaker C: So, just pressed OK button.\nSpeaker D: Okay.\nSpeaker B: My method.\nSpeaker B: How surprising.\nSpeaker B: Oh, findings.\nSpeaker B: These views are important, but innovation is more important.\nSpeaker B: And the fancy look and feel is even more important.\nSpeaker B: And some fashion watchers found out that the young target group likes fruit and vegetables in their clothes shoes and furniture, and that they want spongy material.\nSpeaker B: Probably too much sponge ball.\nSpeaker B: The older group still prefers dark colors, simple shapes from material, but since we're concentrating on the younger group.\nSpeaker D: Why this?\nSpeaker D: Why this?\nSpeaker D: Could you go to the previous slide?\nSpeaker D: Because I'm taking minutes and...\nSpeaker D: will be important in themes and clothes.\nSpeaker D: Okay.\nSpeaker D: The field of measurement is expected to be spongy.\nSpeaker C: So, do you think when fruit and vegetables are important for clothing and shoes, that they are in remote controls also?\nSpeaker B: Well, one example given was this.\nSpeaker B: So, I assume they just want something colorful, not just an apple as a remote control.\nSpeaker B: But like dark colors, you said.\nSpeaker B: The younger group likes more colorful.\nSpeaker B: Okay.\nSpeaker C: But then I suggest the corporate colors are gray and yellow.\nSpeaker C: Can you go back to the slide?\nSpeaker C: Which one?\nSpeaker C: Just one slide back.\nNone: No, no, no.\nSpeaker C: Yes.\nSpeaker C: Okay.\nSpeaker C: And the field of the material has to be spongy.\nSpeaker C: Yes.\nSpeaker C: Has it something to do with that natural feeling also, do you think?\nSpeaker B: Well, it might.\nSpeaker B: Personally, I would like a sponge as a remote control, but maybe some soft material or something.\nSpeaker B: Okay.\nSpeaker D: So, yeah, it might not be...\nSpeaker D: It shouldn't be too hard.\nSpeaker D: It may be rubber or...\nSpeaker D: Yeah, exactly.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker B: Yeah.\nSpeaker B: And like...\nSpeaker B: Oh, wait, the old one.\nSpeaker B: The old one.\nSpeaker B: It likes familiar materials, but that doesn't mean we should use wood.\nSpeaker B: So...\nSpeaker C: Okay.\nSpeaker B: Oh, this is an example of what they would like.\nSpeaker B: But since we're concentrating on the younger group, I think we should use soft materials and make it colorful or like cell phones, exchangeable covers.\nNone: Yeah.\nSpeaker B: So we could provide both for the young and the old, what they like.\nSpeaker D: Well, that's interesting.\nSpeaker C: It's quite interesting.\nSpeaker D: You could make a few very colorful ones.\nSpeaker D: Yeah.\nSpeaker D: And a very traditional...\nSpeaker C: I'm thinking about the banana ram, a telephone, a telephone from Siemens.\nSpeaker C: The yellow rubber...\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: The phone.\nSpeaker C: It's the rubber...\nSpeaker C: Yeah, it is.\nSpeaker C: It is.\nSpeaker C: Yeah.\nSpeaker C: And it's colorful.\nSpeaker C: It looks like some banana.\nSpeaker C: We have the fruit, we have the colors.\nSpeaker C: Do you know the phone?\nSpeaker C: I don't know the phone.\nSpeaker C: I don't know the phone.\nSpeaker C: It's the Siemens C25, I believe.\nSpeaker D: It's the one that Postbank gave away.\nSpeaker C: Oh, that one.\nSpeaker C: The very...\nSpeaker C: Oh, I know.\nSpeaker C: It's also a yellow.\nSpeaker C: Yes, I've seen it.\nSpeaker C: I've seen it.\nSpeaker D: You know, Root as well?\nSpeaker D: I've seen it.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Do you have...\nSpeaker D: That's about it.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: So the important findings are innovation is more important than ease of use for our target group.\nSpeaker D: And coloring is important and...\nSpeaker D: And soft materials.\nSpeaker C: Soft materials.\nSpeaker C: Okay.\nSpeaker C: So ease of use is important, but technology is twice as important.\nSpeaker C: And what was even more important?\nSpeaker B: The fancy looking fuel.\nSpeaker C: Okay.\nSpeaker C: So that's the most important thing for our customers.\nSpeaker C: Yes.\nNone: Okay.\nNone: Okay.\nNone: The rule.\nNone: Yeah.\nSpeaker D: Could you do your presentation?\nSpeaker D: Mm-hmm.\nSpeaker C: Oh, I don't really have much to add because most of the things we already said in the previous discussion.\nSpeaker C: Okay.\nSpeaker D: Well, it's good to sum up the things we already thought about.\nSpeaker C: The previous ideas were a voice recognition and the round button for the channel programming and volume.\nSpeaker C: Well, a few interfaces I found on the internet are these.\nSpeaker C: These are both with voice recognition, but they're very advanced and very high tech and just, well, a weird...\nSpeaker C: Shape.\nSpeaker C: Shape.\nSpeaker C: So I suggest I couldn't...\nSpeaker C: I had a small mock-up sign on the...\nSpeaker C:... but it didn't work.\nSpeaker C: But pen didn't load the information.\nSpeaker C: So I made a really simple shape in PowerPoint, but we could make around an oval...\nSpeaker C:... which is more control.\nSpeaker C: It's kind of organic, so that's very good.\nSpeaker D: Yeah.\nSpeaker D: And what I'm thinking about, maybe we should make very light, but a grip.\nSpeaker D: I mean, this is how you hold a remote control.\nSpeaker D: Yes.\nSpeaker D: It shouldn't be too...\nSpeaker D:... a lot to...\nSpeaker D: Slippery.\nSpeaker D: Slippery.\nSpeaker D: Because...\nSpeaker C: We didn't have something like the Siemens phone.\nSpeaker C: Yes.\nSpeaker C: Yeah, exactly.\nSpeaker C: It's easier in your hand.\nSpeaker C: There are also remote controls who have a little shape underneath...\nSpeaker B:... where you can put your fingers.\nSpeaker C: So you can get a really good grip on it, so you don't have to attain much pressure to it.\nSpeaker C: It's a lot easier.\nSpeaker D: It grips automatically.\nSpeaker D: Yes.\nSpeaker C: Okay.\nSpeaker C: I'll go into that deeper in my presentation.\nSpeaker C: Okay.\nSpeaker C: Good.\nSpeaker C: Right.\nSpeaker C: Oh, and to add on the route information.\nSpeaker C: In this interface, we can have a high tech with a force recognition and...\nSpeaker C:... well, the defensive colors and so on.\nSpeaker C: And still have the ease of use because we have an easy interface.\nSpeaker C: And all the other remote controls are high tech and buttons.\nSpeaker C: Yeah.\nSpeaker D: Way too much, I think, for our goal.\nSpeaker D: Yeah.\nSpeaker C: So if you have the force recognition, you can program like 30...\nNone: Okay.\nSpeaker C: Okay.\nSpeaker C: I'll go into that because there are some possibilities and some impossibilities.\nSpeaker C: I had a talk with our manufacturing division.\nSpeaker C: And we had a talk about all the different components who are in this design.\nSpeaker C: These are the things we've looked at.\nSpeaker C: And of course, I use the web to find my information.\nSpeaker C: About casing, we have three different casing possibilities.\nSpeaker C: We have the uncurved or flat case.\nSpeaker C: That's the most common remote control form we're used to.\nSpeaker C: It's just a box.\nSpeaker C: I'm sorry, I don't have any pictures of this thing.\nSpeaker C: We have a curved one.\nSpeaker C: It's curved in two dimensions.\nSpeaker C: You have to imagine it's a bit like a wave form.\nSpeaker C: So it's a little more advanced in its shape.\nSpeaker C: And we have an even more advanced shape which is curved in three dimensions.\nSpeaker C: I think you can compare it a little bit with the big gray image you had in your presentation.\nSpeaker C: You can compare the big remote control, something like that.\nSpeaker C: But it's quite advanced and it's quite daring to use, I think.\nSpeaker C: For these casings, we have different types of finishing.\nSpeaker C: We can use plastic, which is very slippery and maybe not so nice.\nSpeaker C: But you can give it any color, which is the same for rubber, but it's not slippery.\nSpeaker C: We can use wood and titanium.\nSpeaker C: We cannot use the titanium on the double curve cases.\nSpeaker C: These latex cases, just the plastic ones, won't allow the use of solar cells as an energy source, which brings me to the different energy sources.\nSpeaker C: We even seem to have a head dynamos for powering our remote control.\nSpeaker C: You really have to imagine winding up the rate.\nSpeaker D: It would be very new to the market.\nSpeaker C: It would be very new, but it's kind of a retro style.\nSpeaker C: This is quite interesting. There's also a kinetic energy supply.\nSpeaker C: When you're watching TV, you have to make some kind of energy, kinetic energy, by shaking the remote or throwing it against the wall.\nSpeaker C: It has to move. That's the sense of it.\nSpeaker C: And you can store the energy in the thing.\nSpeaker D: I think if I can, who can come to that?\nSpeaker D: The kinetic thing is very funny.\nSpeaker D: It's very funny indeed.\nSpeaker D: I mean, solar is, of course, it's nice, but it's, well, your calculator has a solar panel.\nSpeaker C: But if you're watching a movie, how many times you take the remote control and...\nSpeaker C: Well, maybe if you have a watch, you have the kinetic idea in a watch also.\nSpeaker C: Yeah, I know.\nSpeaker C: You walk, but you sit on a couch.\nSpeaker C: You know from your own watch.\nSpeaker C: Your watch uses a minimum, it consumes a minimal amount of energy.\nSpeaker C: The shaking of your body, which is almost every activity, makes your body shake, charges it.\nSpeaker C: But the problem here is that it supplies a very little amount of kinetic energy.\nSpeaker C: So I think you have a problem when you're watching a movie and you haven't moved the remote control in an amount of time.\nSpeaker C: And you want to switch the channel or something.\nSpeaker C: Well, it might not work. So that's something you have to keep in mind.\nNone: Okay.\nSpeaker C: But maybe there's a possibility to combine it with traditional batteries so we can save on the batteries.\nSpeaker C: And when there's a near kinetic energy, used kinetic energy and otherwise use the batteries.\nSpeaker D: Okay, because I guess the voice recognition feature consumes quite a lot of power.\nSpeaker C: Yes, it does.\nSpeaker C: Yeah.\nSpeaker C: I'll come to that later.\nSpeaker C: Okay.\nSpeaker C: And we, of course, have the traditional solar power, which is just a piece of material on the...\nSpeaker C:...remode control, which transfers light energy into electricity.\nSpeaker C: The user interface controls.\nSpeaker C: Of course, we have the push buttons.\nSpeaker C: And we also have scroll wheels.\nSpeaker C: And the scroll wheels can also be integrated with buttons.\nSpeaker C: So it's just like a mouse.\nSpeaker C: You can scroll them. You can also push it.\nSpeaker C: In the indicators, we have the LCD displays, which means you can watch in a display which channel you have chosen or something like that of the amount of volume, which is currently.\nSpeaker C: And we also have double scroll buttons, which are just two of these things.\nSpeaker C: So we don't really have the kind of button we had in mind in our last meeting.\nSpeaker C: The thing with the round.\nSpeaker C: We can just make four push buttons in the form of a round.\nSpeaker C: Yes.\nSpeaker C: It's disposable too.\nSpeaker C: Yes.\nSpeaker C: But I don't think the scroll wheels, I had some information about it too.\nSpeaker C: I don't think there's any possibility for us to use scroll wheels.\nSpeaker C: I can't see any scenario where you would use a scroll wheel with a button on it.\nSpeaker C: Maybe, well, maybe the...\nSpeaker C: Maybe it's a bit of a remote.\nSpeaker D: With maybe the channel selector.\nSpeaker D: What about integrating a scroll bar on this side?\nSpeaker D: That's the possibility.\nSpeaker C: Because this is how you keep it.\nSpeaker C: The floor is quite good.\nSpeaker C: Think about a scenario where you would use the scroll button.\nSpeaker C: Volume.\nSpeaker C: Well, what he means is there's a button integrated in the scroll wheel.\nSpeaker C: There's no scenario where you use the button in the scroll wheel.\nSpeaker C: You just use the wheel.\nSpeaker C: Well, what about mute?\nSpeaker C: About mute.\nSpeaker D: I guess this is my volume button.\nSpeaker D: Yes.\nSpeaker D: And I can either on this side or this side.\nSpeaker C: That's one possibility.\nSpeaker D: And click it to mute the device.\nSpeaker D: And it makes it different from the traditional devices on this market.\nSpeaker D: So I'm looking for a way to make it a little different than the traditional ones.\nSpeaker D: So maybe I guess that's something you need to think about.\nSpeaker D: What do you think about a scroll bar?\nSpeaker D: A scroll wheel.\nSpeaker B: Well, it's obviously new.\nSpeaker B: So it might attract the young customers.\nSpeaker C: But it's done before.\nSpeaker C: There are many auto devices like telephones and our radio, pocket radios.\nSpeaker C: We use this.\nSpeaker C: And well, it's been done years ago and I don't see it anywhere now.\nSpeaker D: Well, all the Sony telephones use it, for example, for volume.\nSpeaker D: Nokia has it, well, okay, it's not really a scroll wheel.\nSpeaker D: But on their side, the volume button is on the side because you grab it like this.\nSpeaker C: Yes, but it uses two separate buttons.\nSpeaker C: Yeah, I know.\nSpeaker C: It's not really a scroll wheel.\nSpeaker C: There's something for you too.\nSpeaker C: I believe if you have what we've already said, the grip places in the remote control, you have your hand on one place on the remote control.\nSpeaker C: So you have to place all the buttons in the range of your thumb.\nSpeaker C: Yes, you have to.\nSpeaker C: So in that case, the volume button on the side of the remote control would be perfect.\nSpeaker C: Okay.\nSpeaker D: Sebastian, then.\nSpeaker C: Okay.\nSpeaker C: We have to know, if you want to use these rubber double curved case, you must use these push buttons.\nSpeaker C: There's no way you can integrate LCD displays.\nSpeaker C: There's no way you can integrate scroll wheels because it's all curved.\nSpeaker C: There are no flat areas where you can incorporate these things.\nSpeaker C: So that's a limitation about the components, just the hardware.\nSpeaker C: We basically have three types of chips we can use.\nSpeaker C: And these chips incorporate all the largeica and hardware that is needed to send signal.\nSpeaker C: We have a simple regular and advanced chip.\nSpeaker C: And there is something like a sample sensor and sample speaker, which is a little cryptic to me.\nSpeaker C: But I think that's the voice recognition thing that we are thinking about.\nSpeaker C: Yeah, you can have some information about it.\nSpeaker C: In the voice recognition, you say a word.\nSpeaker C: You can program a word like volume up.\nSpeaker C: Okay.\nSpeaker A: So mute.\nSpeaker C: Let's say mute.\nSpeaker C: Okay.\nSpeaker C: You program it, you mute, and you give an action to it.\nSpeaker C: Just really the mute function.\nSpeaker C: And when you speak in the remote control, it repeats your saying.\nSpeaker C: So that's the sample sensor.\nSpeaker C: Okay.\nSpeaker C: So if you say mute, it says mute again.\nSpeaker C: And then it's, when I believe it.\nSpeaker C: It performs the action.\nSpeaker C: Yeah.\nSpeaker C: And then your piece is action which you believe it is.\nSpeaker C: You say mute, you repeat mute, and it makes the computer sound mute.\nSpeaker C: And then goes to the mute function.\nSpeaker C: Okay.\nSpeaker C: So that's basically the voice recognition item we were searching for.\nSpeaker C: Okay.\nSpeaker C: This sample sensor requires regular chip, I thought.\nSpeaker C: No, well, I'm not sure.\nSpeaker C: No, it's not in here.\nSpeaker C: If we want to use the LCD display, we really need the advanced version, which is a little bit more costly.\nSpeaker C: If we want to use the screw wheels, we need the regular version.\nSpeaker C: And if we don't want to use any of these more advanced functions, we can keep with the simple chip, which is a bit cheaper.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Did we already decide on the display?\nSpeaker C: No, but I think that's something for the whole here.\nSpeaker C: Yeah.\nSpeaker C: Well, I don't have a look for information about it.\nSpeaker C: But I don't think information, I don't think you need it on the display.\nSpeaker C: Especially when we have to look at the cost, I don't think you need it.\nSpeaker C: Any TV can view it on screen.\nSpeaker C: On screen, yeah.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Well, my conclusion, unfortunately, the market has decided a little other than I thought.\nSpeaker C: I thought the market would like a sort of titanium casing, but they seem to like natural stuff.\nSpeaker C: So maybe we should think about wood finish.\nSpeaker C: I've chosen the more battery with solar cell solution for the energy.\nSpeaker C: It's more reliable, it's cheaper.\nSpeaker C: Okay.\nSpeaker C: So I don't think we should use the diagonal thing.\nSpeaker C: Okay.\nSpeaker C: It's more advanced, but I think you should combine it with batteries.\nSpeaker C: Okay.\nSpeaker C: It's maybe a bit too flashy too.\nSpeaker C: Well, yeah.\nSpeaker C: In some way, it can give us an offense because you will save on your batteries.\nSpeaker D: Yeah, but that's the same with the solar cell.\nSpeaker D: Yes.\nSpeaker C: I think it's more robust.\nSpeaker C: But what about the market, what colorful designs?\nSpeaker C: So if we use titanium.\nSpeaker D: No, no.\nSpeaker D: That's what Sebastian said.\nSpeaker D: He said this is my personal preference, but yet I understood that the market is different.\nSpeaker D: Oh, sorry.\nSpeaker C: I think we should skip the titanium stuff and we should use wood or something like that.\nSpeaker B: And I would think that the rubber with colors.\nSpeaker B: Yeah, the older people liked wood.\nSpeaker C: Oh, okay, sorry.\nSpeaker C: It needs to be rubber.\nSpeaker C: Call it the younger people like soft material.\nSpeaker C: Okay.\nSpeaker C: Spongy materials.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Well, these scroll wheels, I think they can be handy.\nSpeaker C: And they can be implemented with a regular chip?\nSpeaker C: Yes, they can.\nSpeaker C: But they really need a regular chip.\nSpeaker C: You cannot use the same software.\nSpeaker C: Okay.\nSpeaker C: But also, we already need a regular chip for the sample.\nSpeaker C: Well, I'm not so sure.\nSpeaker C: Maybe that's a big thing.\nSpeaker D: No, but do you want the curved design or?\nSpeaker C: I think so.\nSpeaker C: If you stick with the simple, straightforward, not curved design.\nSpeaker C: It's too dull.\nSpeaker C: It's too dull.\nSpeaker C: I don't think our customers will like it.\nSpeaker C: And if you take the double curved, then you cannot only use the rubber buttons.\nSpeaker C: You cannot use the scroll wheels.\nSpeaker C: Okay.\nSpeaker C: I think this is the best of two wheels.\nSpeaker D: Okay.\nSpeaker D: What about the issue I addressed at the beginning of the meeting?\nSpeaker D: The voice function with the infrared issue.\nSpeaker D: Yes.\nSpeaker D: Because it's, of course.\nSpeaker C: I can imagine.\nSpeaker C: It's very dull to talk to a device, if you have to point to another device.\nSpeaker C: Yeah.\nSpeaker C: It's very logical.\nSpeaker C: But most infrared remote controls don't have to be pointed at all.\nSpeaker C: Well, there has to be some pointing.\nSpeaker C: It depends also on your hand before.\nSpeaker C: Okay.\nSpeaker C: It won't work.\nSpeaker C: But you can point it just.\nSpeaker C: Well, it depends on your walls, actually.\nSpeaker D: If you have smooth walls, you're probably right.\nSpeaker D: But if you have carpets on the wall, which are natural, loving friends, probably do have, then, yeah, it might be a bit more of an issue.\nNone: Yes.\nSpeaker C: Because the walls, they reflect the infrared light.\nSpeaker C: Yeah.\nSpeaker C: So, it's easier.\nSpeaker C: I suggest we use the sensor sample.\nSpeaker C: The sample sensor and sample speaker.\nSpeaker C: With a regular chip.\nSpeaker C: I think it gives us the advantage of the scroll wheel.\nSpeaker C: Yes.\nSpeaker C: I like the scroll wheels.\nSpeaker C: I skipped the LCD part.\nSpeaker C: I don't think it's any value added.\nSpeaker C: No.\nSpeaker C: Thanks.\nSpeaker D: Well, it looks, yeah.\nSpeaker D: Well, according to the market likes new flashy technology.\nSpeaker D: And I mean, else today is...\nSpeaker C: It's not very flashy.\nSpeaker D: Well, I know.\nSpeaker D: It's less...\nSpeaker D: Standard?\nSpeaker D: Standard than...\nSpeaker D: Well, we are not very...\nSpeaker D: We do not know much about the financial part.\nSpeaker D: That's the problem.\nSpeaker D: Because if we do have enough space in our finance, I would say do integrate it because it adds a little extra high-tech feeling to it.\nSpeaker C: But we all have the scroll wheels, the speaker recognition, the rubber, I think our customer is so good.\nSpeaker C: The fancy color.\nSpeaker C: Okay.\nSpeaker D: It's too much.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: I agree.\nSpeaker D: I think...\nSpeaker D: Do you have anything you would like to add or maybe thought or...\nSpeaker B: No, I don't think so.\nSpeaker D: No.\nSpeaker D: No, no.\nSpeaker C: Nothing more.\nSpeaker D: Nothing more.\nSpeaker C: The next phase will be...\nSpeaker C: What's the next phase to the floors?\nSpeaker D: Well, we need to describe decisions now.\nSpeaker D: Okay.\nSpeaker D: So, on energy, well, we decided...\nSpeaker D: Okay.\nSpeaker D: The case, rubber with one curve.\nSpeaker D: Use the interface.\nSpeaker C: But it can't go with...\nSpeaker C: With one curved case?\nSpeaker C: Yes, they can work.\nSpeaker C: Sorry.\nSpeaker C: They cannot work with double curved.\nSpeaker C: Oh, sorry.\nSpeaker C: Just a problem.\nSpeaker C: I'll check it for you.\nSpeaker D: Then the corporate identity should be in the product.\nSpeaker D: I guess that is something for Ruhl and Sebastian.\nSpeaker D: You talked about it before the colors.\nSpeaker D: Gray and yellow.\nSpeaker D: Yeah.\nSpeaker D: Keep it in mind.\nSpeaker D: And the buttons, well, we talked about it now.\nSpeaker D: The next phase, Sebastian, is the design of the look and feel.\nSpeaker D: Mm-hmm.\nSpeaker D: The user interface design.\nSpeaker D: And for you, the product evaluation, I'm sure your personal coach will give you.\nSpeaker D: More information on that.\nSpeaker D: And the ID and the UID need to work together on the prototype drawing on the smartboard.\nSpeaker C: Okay.\nSpeaker D: And I'm going to plan my holidays.\nSpeaker D: Okay.\nSpeaker C: So, you will build the house for the next...\nSpeaker D: Yeah.\nSpeaker D: It's...\nSpeaker D: When we come back in 30 minutes, you will have a prototype ready.\nSpeaker C: Okay.\nSpeaker C: So, can you give us a summary of all decisions we've made?\nSpeaker D: Yeah.\nSpeaker D: I can.\nSpeaker D: Maybe one of you could write it down.\nSpeaker D: I'll do.\nSpeaker D: Great.\nSpeaker D: You need to help me.\nSpeaker C: Okay.\nSpeaker C: Start with the casing.\nSpeaker D: The casing is curved.\nSpeaker C: Single curved.\nSpeaker C: Single curved.\nSpeaker C: Okay.\nSpeaker C: What about the energy source?\nSpeaker D: Traditional batteries and solar.\nSpeaker C: But can I be...\nSpeaker C: Can I work together?\nSpeaker C: When you have to choose between them.\nSpeaker C: No, they can be complementary.\nSpeaker C: I think they can.\nSpeaker C: Oh, okay.\nSpeaker C: Well, what...\nSpeaker C: Every device.\nSpeaker C: It should be.\nSpeaker C: There should be a real problem.\nSpeaker C: There can be a supplementary.\nSpeaker C: There's no problem.\nSpeaker D: Yeah.\nSpeaker D: So, okay.\nSpeaker C: So, just...\nSpeaker C: The energy source is...\nSpeaker C: The battery.\nSpeaker C: And the solar.\nSpeaker C: Yeah.\nSpeaker C: What about the finishing of the case?\nSpeaker C: We have decided we wanted to use the rubber.\nSpeaker D: Yeah.\nSpeaker D: Colorful rubber.\nSpeaker D: If it's possible with different covers.\nSpeaker D: But I'm not sure if our suppliers can help us with such a wish.\nSpeaker C: Okay.\nSpeaker C: And I think we should use the company color.\nSpeaker C: Something like black and red.\nSpeaker C: Or black and yellow.\nSpeaker D: Gray and yellow.\nSpeaker D: Gray and yellow.\nSpeaker C: Yeah.\nSpeaker C: Yellow case and gray buttons.\nSpeaker D: Yeah.\nSpeaker D: Although, I don't think that's very colorful.\nSpeaker D: Except for the yellow, of course.\nSpeaker D: But I could think of a more attractive set of colors.\nSpeaker C: I think it's...\nSpeaker C: It's not very dull.\nSpeaker C: It's quite modern, actually.\nSpeaker C: Don't you think?\nSpeaker D: Well, I was more thinking about the fruit colors that the root show just before.\nSpeaker C: But you have already...\nSpeaker C: You must have a red on and off button.\nSpeaker C: And...\nSpeaker D: Well, it doesn't have to be red.\nSpeaker D: Well...\nSpeaker D: I mean, I think these colors are really what our young people are looking for.\nSpeaker D: So maybe it shouldn't even be two colors.\nSpeaker D: It should be a full color cover with such an image.\nSpeaker D: Or...\nSpeaker D: I mean...\nSpeaker D: Okay.\nSpeaker D: Thinking in two colors is too black and white for our market group, I guess.\nSpeaker B: Oh, black and yellow.\nSpeaker B: Black and yellow.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker C: But I'm afraid it's not possible to print a picture on the device.\nSpeaker C: Because I agree it would be nice to have something like that on the device itself.\nSpeaker C: Well, there is.\nSpeaker C: Just a week ago.\nSpeaker C: Keyboard manufacturer.\nSpeaker C: Oh, I've read.\nSpeaker C: With print.\nSpeaker C: Yes, fine art and...\nSpeaker C: Yes, but our manufacturing department is not so advanced in its techniques.\nSpeaker C: They're actually very slow-heads techniques.\nSpeaker D: Okay, so we have to deal with what's possible here.\nSpeaker C: So, I'm afraid it's not possible.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: More decisions we made.\nSpeaker B: Um...\nSpeaker C: The scroll wheel?\nSpeaker D: The scroll wheel, yes.\nSpeaker D: The voice recognition.\nSpeaker D: We already decided.\nSpeaker C: Of course.\nSpeaker C: Okay.\nSpeaker C: So scroll wheel, but there will be some additional buttons, I guess.\nSpeaker C: And they should be spongy also.\nSpeaker C: Because they're rubber too.\nSpeaker D: I'm going to leave that up to you too.\nSpeaker C: What did you say?\nSpeaker C: Well, you can use, well, when you use the buttons, they'll be made of rubber too.\nSpeaker C: So it has the spongy, fuel also.\nSpeaker C: Yeah.\nSpeaker C: So I think that's okay.\nSpeaker C: Or you could use plastic buttons in the rubber.\nSpeaker D: I think rubber is nice.\nSpeaker D: Because I mean, what do you touch the button and what do they want, spongy devices?\nSpeaker C: But then you just have the fact that the trunks on the buttons will disappear eventually.\nSpeaker D: It's really?\nSpeaker C: What I said in the first discussion, the digit 6 on the button, it will disappear when it's from rubber.\nSpeaker D: Is that...\nSpeaker D: Does our virus say something?\nSpeaker C: It's not something that no information I read about it or so.\nSpeaker B: But did we include the digits?\nSpeaker B: No, but...\nSpeaker B: Because things like volume could be placed next to the button.\nSpeaker D: You could place a...\nSpeaker D: This would be the button.\nSpeaker D: The scroll wheel, I mean.\nSpeaker D: And you could place the signals.\nSpeaker D: So you don't touch it?\nSpeaker C: That's possible.\nSpeaker C: But then you have still the images on the rubber of the case.\nSpeaker C: Still then, if you feel your remote control, you just rub on the cover.\nSpeaker C: So you rub on the painting.\nSpeaker D: No, no, there's no painting.\nSpeaker D: Only yellow or the digits into the rubber.\nSpeaker C: But the plus or the minus?\nSpeaker C: Yes, the sign.\nSpeaker C: If this is on the cover.\nSpeaker C: It's on the cover.\nSpeaker D: I think... I know what you understand.\nSpeaker D: I know what you mean.\nSpeaker D: Just move the problem.\nSpeaker D: What about making this rubber and making this plastic?\nSpeaker C: I see what you mean.\nSpeaker C: Well, maybe that's possible because our manufacturing division also offers plastic finishing.\nSpeaker C: So maybe they can combine these two.\nSpeaker D: Yeah.\nSpeaker D: Well, actually we should have it the other way around, I guess.\nSpeaker D: A plastic cover with rubber finishing.\nSpeaker D: I mean, this is the finishing.\nSpeaker D: This is what's on the edge, what you feel.\nSpeaker D: The front on which the buttons are doesn't have to be rubber.\nSpeaker C: I'm not so sure.\nSpeaker C: Can you separate these?\nSpeaker C: Well, I'm not sure.\nSpeaker C: I have to ask with manufacturing.\nSpeaker C: I'm not sure that's what you want because our customers specifically ask for the spongy feel.\nSpeaker C: And that's what you get with rubber.\nSpeaker C: So if you want a spongy feel, you need to make these buttons all of rubber.\nSpeaker D: I know, but do you touch this or do you touch this?\nSpeaker D: I mean, I never touch between the buttons.\nSpeaker C: I do.\nSpeaker C: I do.\nSpeaker C: The sideways or?\nSpeaker C: Yeah, the sideways.\nSpeaker D: Exactly.\nSpeaker D: The sideways.\nSpeaker D: The side.\nSpeaker D: But do you touch between these buttons?\nSpeaker C: Yes, especially when there are a few buttons on it.\nSpeaker C: You have a lot of space to touch.\nSpeaker C: So you just have it in your hand completely or...\nSpeaker D: Okay.\nSpeaker D: Well, we do not have very much time left.\nSpeaker D: I guess you two have to figure that out.\nSpeaker D: I'm going to leave the decision to you because you have to make its prototype.\nSpeaker D: And yeah, you have the most knowledge about the suppliers, the possibilities.\nSpeaker D: So I'm going to leave the decision with you.\nSpeaker D: So, whilst, did you write enough decisions down?\nSpeaker C: Not quite.\nSpeaker C: What about the chips?\nSpeaker C: We use the regular chip.\nSpeaker C: Regular line.\nSpeaker C: Okay.\nNone: Yeah.\nNone: And...\nSpeaker C: Well, I think that's about it.\nSpeaker B: Yes.\nSpeaker B: Okay.\nSpeaker D: With those regular chips, we can still control advanced functions.\nSpeaker D: The chip is not really...\nSpeaker C: The only difference between the advanced and the regular version is that the advanced version of the chip supports LCD.\nSpeaker C: Okay.\nSpeaker C: Well...\nSpeaker C: And we've decided not to use LCD so...\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And I think we are quite finished.\nSpeaker D: I'll see you guys in 30 minutes.\nSpeaker C: The finishing touch.\nSpeaker D: Yeah.\nSpeaker D: If you wrote anything down, could you put it on the shared folder?\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: But Sebastian has everything.\nSpeaker D: I know, but...\nSpeaker D: Well...\nSpeaker C: I'll put it online.\nSpeaker D: Great.\nSpeaker D: Okay.\nNone: Okay.\nNone: I'm going to demonstrate.\nNone: I'm like...\nNone: Oh.\nNone: Okay.\nNone: Okay.\nNone: Yeah.\nNone: I can put it on a floor.\nNone: Okay.\nNone: Yeah.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Hi.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Vima.\nNone: I cut it away from online, and I'm doing it, and I'm going to add to it.\nSpeaker C: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "IS1003c", "summary": "This meeting was about conceptual design. The user interface designer first introduced three concepts in the user interface - Google controller, fancy controller, and intelligent controller. The first two were approved by the project manager, but the last one was seriously questioned. The group finally decided to mix the Google controller and fancy controller with some simple vocal commands. The discussion about product industrial design covered three aspects \u00e2\u0080\u0094material, alimentation, and chip. The industrial designer suggested using wood and having buttons with LCD, applying the classical battery and solar energy alimentation, and using low level chips, which was agreed by the group. The marketing expert gave three points from the market analysis. Users would like to have a fancy look and feel and the product should be technologically innovative. At the same time, being user-friendly was also important. Then the group discussed how to make these requirements into practice.", "dialogue": "None: You see that? You can see that.\nNone: Okay, I'll Berry do just pick on you for a little while guess what be tat scrwhat?\nSpeaker E: So let's start our second meeting on conceptual design.\nSpeaker E: So as the previous meeting I will be the secretary and we will have three presentations and we will have to decide on the remote control concept and finally we will close the session.\nSpeaker E: So I will first...\nSpeaker A: You do the minutes first or?\nSpeaker F: No.\nSpeaker E: I think I will let our user interface on designers speak first.\nSpeaker E: Mr. David Jordan.\nSpeaker E: So I will...\nSpeaker D: Yes.\nSpeaker E: Technical... that's this one?\nSpeaker E: Interface.\nSpeaker E: This one?\nSpeaker B: So first I will present the concept of user interface.\nSpeaker B: There are three concepts in the user interface.\nSpeaker B: The first one is the Google controller.\nSpeaker B: The second is the fancy controller.\nSpeaker B: The last one is the intelligent controller.\nSpeaker B: So there are three concepts in our controller.\nSpeaker B: Next I will explain my outline.\nSpeaker B: The first is the Google controller.\nSpeaker B: So I want the controller to be easy to use but with sophisticated functions.\nSpeaker B: So it's a combination of easy to use and sophisticated functions.\nSpeaker B: This is the first concept of our controller.\nSpeaker B: The second concept is the fancy controller.\nSpeaker B: So without giving the customer the impression that our controller is very attractive.\nSpeaker B: We can easily recognize our controller among a lot of products.\nSpeaker B: So the user interface would be very cute.\nSpeaker B: Very attractive.\nSpeaker B: Such as these are several examples in these slides.\nSpeaker B: I'm not sure.\nSpeaker B: They want it to be so extremely cute.\nSpeaker B: It's very big.\nSpeaker B: It's easy for me to remember it.\nSpeaker B: I need to recognize it.\nSpeaker E: Okay.\nSpeaker E: Why not?\nSpeaker E: We'll have big discussions with our experts after that.\nSpeaker B: So the last concept is the intelligent.\nSpeaker B: We want our controller to be smart.\nSpeaker B: So maybe we should use some technology such as speech recognition technology and gesture recognition technology.\nSpeaker B: So we need to have some cooperation with some research institute on speech recognition and on gesture recognition.\nSpeaker B: With these advanced features, I think we can attract a lot of users.\nSpeaker E: Okay.\nSpeaker B: Something else?\nSpeaker B: No.\nSpeaker B: This is the three concepts of our...\nSpeaker E: I just have one question because for the intelligent controller, you said that we can use the voice recognition or the gesture recognition.\nSpeaker E: But as the expert told us, most of the people want to use the remote control to zap between channels.\nSpeaker E: Do you think they will be able to use gestures because if they do all the time the same gesture, as you said previously in the last meeting, maybe they will get injuries because of that?\nSpeaker E: Or if you say channel 3, channel 3, 2, 3, 4, 6, 5, I think they will be born after a while.\nSpeaker E: You don't think so?\nSpeaker B: I think sometimes it's very convenient to use voice interface and gesture interface.\nSpeaker B: For example, if you cannot find your controller, you can't use it.\nSpeaker B: Just speak something such as one.\nSpeaker E: Yeah, but suppose you got a cold. You have a mute remote controller.\nSpeaker B: So you can use the gesture. That's a good problem.\nSpeaker A: How is it very easy to trust like speech recognition or gesture recognition?\nSpeaker B: For limited vocabulary speech recognition, it's very reliable and for limited vocabulary gesture recognition, it's also very easy.\nSpeaker E: Yeah, but suppose you have a family watching TV and if they want to use their private remote control in the same time, do you think it will work?\nSpeaker E: Everybody wants to change channels at the same time?\nSpeaker A: But this would never happen anyway.\nSpeaker B: Yeah, they cannot speak at the same time.\nSpeaker E: If you have one brother and one sister, they want to watch their favorite TV program, so they say, oh, channel 4, channel 3, channel 4.\nSpeaker C: Yeah, so all the time.\nSpeaker C: The same can happen even if you use it. It's kind of remote control.\nSpeaker E: Yeah, but you have the remote control. So maybe you can keep it with you.\nSpeaker E: You're not a blind gesture.\nSpeaker A: Okay, you mean it could be a problem for this guy.\nSpeaker B: Yeah, that's the advantage of intelligent controller. Even you have the controller. I can say channel 3.\nSpeaker B: So this is a disadvantage.\nSpeaker E: I think it's a disadvantage.\nSpeaker E: This is a advantage.\nSpeaker C: Maybe you can have a switching mode to pass from voice controller to...\nSpeaker E: But one other question, how much will it cost?\nSpeaker C: How much?\nSpeaker E: Yeah, because I suppose we need to do research. Do you have something working?\nSpeaker B: No, we just...\nSpeaker B: I'll use some efficient.\nSpeaker B: We just have some cooperation with some research institute. We don't have to do some basic research on this.\nSpeaker E: Oh, you think it won't cost in... not a lot for us?\nSpeaker B: I think it's because this technology is limited or lack-seeking recognition.\nSpeaker B: It's very...\nSpeaker A: Yeah, but it's changing how the remote control is going to be built.\nSpeaker A: Because then you need...\nSpeaker A: I mean, this doesn't have the power to do recognition, for example.\nSpeaker C: You have also the language protocol.\nSpeaker B: Because it has to be universal.\nSpeaker B: Yeah, I agree with some.\nSpeaker B: But there's not a problem.\nSpeaker E: Yeah, but there's one problem that Baba talked about.\nSpeaker E: It's the international remote control.\nSpeaker E: We need something that is international.\nSpeaker E: Suppose we want to set it in France.\nSpeaker E: The recognition system will be able to understand French.\nSpeaker E: If you want to go to England, it will be able to understand English.\nSpeaker E: Yeah, the key...\nSpeaker B: This could be downloaded by the webinar.\nSpeaker B: The key feature of our control is that it has some adaptation.\nSpeaker B: It means when you sell this controller in China, it can...\nSpeaker B: If you sell this controller, you can't use it.\nSpeaker B: It's a very smart...\nSpeaker C: Yeah, it's a smart control device.\nSpeaker E: Yeah, and with no increase in the production price of the remote control?\nSpeaker B: Because of this product, this technology has already been developed.\nSpeaker B: Yeah, but the problem is how to...\nSpeaker C: If this is a push button controller, you can send this remote control everywhere in the world.\nSpeaker C: The same one.\nSpeaker C: If you have the language, you have to develop for each count.\nSpeaker C: Yeah, that's why we have to do language adaptation.\nSpeaker C: In the French country, you have to do one.\nSpeaker C: Yeah.\nSpeaker B: Because for example, for even for different family, we have to do...\nSpeaker B: Really?\nSpeaker B: Yeah, we have to do adaptation tools.\nSpeaker B: Yeah, but it seems to be quite complex.\nSpeaker A: Yeah, we have to take care of the 12 euros...\nSpeaker E: And what about voice recognition?\nSpeaker E: Do we have microphones and where we'll be there?\nSpeaker E: Do you think if we are far from television, it will work?\nSpeaker B: I think that's not a problem, because you don't have to wear a microphone.\nSpeaker B: The microphone is embedded in the controller.\nSpeaker E: Yeah, but where is the controller?\nSpeaker B: There is a controller.\nSpeaker B: Yeah.\nSpeaker B: Is it your family?\nSpeaker B: Your home?\nSpeaker E: Yeah, but here it's an object.\nSpeaker E: But here you say you want to use a technology...\nSpeaker B: Yeah, you can embed it in a microphone here.\nSpeaker E: Yeah, but what is the use of voice or just recognition if you have remote control like this, if you have an object?\nSpeaker E: If you want to use voice or gesture, you need to be free without any objects.\nSpeaker E: You just want to interact with the device.\nSpeaker B: Yeah, just put the controller here.\nSpeaker B: If you say one, you see your command and you do your gesture.\nSpeaker E: Yeah, but you can lose it.\nSpeaker B: No, it's...\nSpeaker C: If you lose it.\nSpeaker C: So, but for example, if it is somewhere in the room, if it is in the table there, you can always say it's channel one and the remote control gives the audio to the TV.\nSpeaker E: Yeah, so you can build to kind of like box and put it on TV just to recognize gestures and voice.\nSpeaker C: Yeah, but so you need a camera and a microphone.\nSpeaker A: Oh.\nSpeaker A: So you would still have the buttons or do you think it should be only voice recognition and gesture recognition?\nSpeaker A: Or you still have the possibility to use buttons?\nSpeaker B: I think we should give the flexibility to the user.\nSpeaker B: Okay, so you...\nSpeaker B: Yeah, but you can see the gas switch from bi-modality to another.\nSpeaker A: Yeah, I don't know. It's a bit risky.\nSpeaker E: I think so. Maybe it will be quite...\nSpeaker E: Quite attractive.\nSpeaker C: I think that switching from one counter to another will be a problem.\nSpeaker B: Because if you do language adaptation, there should be no problem.\nSpeaker B: Yeah, but...\nSpeaker A: I don't know.\nSpeaker E: Okay.\nSpeaker B: We should have confidence in technology.\nSpeaker B: Yeah, we should.\nSpeaker E: So, what do you think? Which are the controllers you'd prefer?\nSpeaker F: What?\nSpeaker E: Which kind of controller would you prefer to use?\nSpeaker E: You as a remote control user?\nSpeaker A: I mean, I'm sure if the user pays the same price, he's happy to have recognition.\nSpeaker A: But if it like doubles, no one will...\nSpeaker C: I think he would prefer that he's very reliable.\nSpeaker E: So I think it would be better not to do near-the-lute-gen controller and to stay with the Google controller or fancy controller.\nSpeaker E: Maybe try to mix the Google controller and the fancy controller.\nSpeaker E: Yeah.\nSpeaker E: So, if you stick to the first two parts, so what's the difference between our controller with other products in the market?\nSpeaker E: There's no features of our controller.\nSpeaker B: So, is there any necessary to design new controller without...\nSpeaker B: I think it works through features.\nSpeaker B: No, I mean, do you replace your controller?\nSpeaker B: I think it works through the other two parts.\nSpeaker B: So, what's the difference between our controller with other products in the market?\nSpeaker B: No, I mean... do you replace your controller with a controller with similar function?\nSpeaker B: If you do not have some function...\nSpeaker C: I had for example some function like for browsing in the internet or something like that.\nSpeaker C: But I think a user needs...\nSpeaker B: That's not the function of the controller.\nSpeaker B: That's the function of TV.\nSpeaker B: You can replace your TV with a new TV with internet browser function.\nSpeaker A: No, but you need a new remote controller then.\nSpeaker A: Because if you want to browse internet...\nSpeaker A: You don't have any...\nSpeaker A: If you want to type something...\nSpeaker C: Okay, if you want to send a mail from...\nSpeaker B: It's not only the problem.\nSpeaker B: It's also the issue of the TV.\nSpeaker C: The problem I can see is the voice or gesture.\nSpeaker C: Because we can't have it in a family.\nSpeaker B: We cannot rely 100% on these features to use the controller.\nSpeaker B: But the features of our controller, such as you have the feature of voice recognition in your mobile.\nSpeaker B: But you seldom use it in your mobile.\nSpeaker B: But you choose a new mobile.\nSpeaker B: You choose a VAMB voice recognition.\nSpeaker B: The feature is not 100% reliable.\nSpeaker B: But it's a feature to distinguish our product from our products.\nSpeaker E: We want something that works all the time.\nSpeaker E: Every day, every hour.\nSpeaker E: For everyone.\nSpeaker C: The person of the family.\nSpeaker B: If you don't need to do this...\nSpeaker B: If you already have a product, it works 100% reliable.\nSpeaker B: Would you replace it with another one?\nSpeaker E: Yeah, why not?\nSpeaker A: For example, we would like to have a Google-like controller.\nSpeaker A: I don't see how adding speech or gesture recognition would make the remote control look more like Google.\nSpeaker A: Google is simple, works fine.\nSpeaker A: I guess if we can have a remote control that is really basic, simple and works fine, it's already in a lot.\nSpeaker B: There's no big difference between the traditional controller.\nSpeaker A: I mean, the user is not only interested in having speech or gesture recognition.\nSpeaker A: If he has something that works fine and is really fancy, looks nice.\nSpeaker B: Not enough motivation for them to replace their old controller with a new one.\nSpeaker B: It does no key feature in the new controller.\nSpeaker A: That's the problem.\nSpeaker A: I don't know it's more interesting to develop a remote controller with speech and gesture and whatever.\nSpeaker A: But you have to think the user is the one who is going to buy the product then.\nSpeaker E: Okay, so let's go to the industry old designer.\nSpeaker E: Maybe we'll be able to take a decision after that.\nSpeaker E: So, do you want to?\nSpeaker C: Yeah, pass it on to working.\nSpeaker C: So, I can...\nSpeaker C: Can you go to the next one?\nSpeaker C: It's not this one.\nSpeaker C: It's not working this way.\nSpeaker C: Sorry.\nSpeaker C: Come on, honey.\nSpeaker C: So, this is what describes using.\nSpeaker C: I don't see.\nSpeaker E: I think there's something wrong with the user.\nSpeaker B: Did you receive it?\nSpeaker E: Maybe you record it somewhere else.\nSpeaker D: I don't think so.\nSpeaker D: Pass it on to one.\nSpeaker B: Pass it on to one.\nSpeaker E: To face concept.\nSpeaker E: No.\nSpeaker C: Maybe I record it directly on the computer.\nSpeaker A: Yeah.\nSpeaker E: No.\nSpeaker F: Okay.\nSpeaker E: It seems that we have a problem with the...\nSpeaker A: I don't know if you remember what you had to say or...\nSpeaker C: I can say it without...\nSpeaker B: Yeah, so...\nSpeaker B: Maybe we can first come to...\nSpeaker E: No, I think it would be more interesting to start with...\nSpeaker B: I think it's more interesting.\nSpeaker E: Yeah.\nSpeaker E: We will have some more information.\nSpeaker E: I think it would be interesting after your presentation to have...\nSpeaker E: Baba's presentation.\nSpeaker C: In fact, I don't know.\nSpeaker C: Okay.\nSpeaker C: It was in fact to design use to show you the design of what is inside and what are the different components of the...\nSpeaker C: of the remote control.\nSpeaker C: Okay.\nSpeaker C: So, it would be interesting so I could show you some picture of what is inside.\nSpeaker C: So, in fact, something I want to discuss is which kind of material are you going to use?\nSpeaker C: So, it would be a wooden remote control or plastic remote control like this one.\nSpeaker C: So, in which...\nSpeaker C: which kind will be the different button?\nSpeaker C: So, it can be some classic push button like this one or you have also some button like LCD where, you know, the buttons are unlighted during the night or you can see them in the darkness.\nSpeaker C: Okay.\nSpeaker C: And the other thing I want to discuss also is which kind of electric elementation we want to have.\nSpeaker C: So, will it be, for example, solar energy elementation or will it be a battery like the classical battery?\nSpeaker C: So, and I think that for example for the elementation it would be good to have both of them.\nSpeaker C: So, for example, in some country where you are in the countryside and you are far from the cities.\nSpeaker C: For example, in some place in Senegal. So, if you have solar elementation, you just, when you want to have recharge your remote control power, you just put it on the sun and after one hour you can comment.\nSpeaker C: So, it can be interesting for people to have this kind of...\nSpeaker C: It can be something interesting to make people buy.\nSpeaker E: Yeah, I think it's an added value to the remote control and maybe it can attract all the electrical consumers.\nSpeaker E: And, but apparently the price of adding this solar battery, would it be something really that will increase the price of prediction or no?\nSpeaker C: In fact, having them both will, if you want to have battery, regular battery and the solar energy battery, it could be, it will add a little bit of the price.\nSpeaker C: So...\nSpeaker C: But it will be an added value also that will be compensated.\nSpeaker C: Okay.\nSpeaker E: And what about the materials?\nSpeaker C: And the materials, it depends. For example, if you have a wooden material, it can be more...\nSpeaker C: The plastic material is more common. It's very resistant.\nSpeaker C: But, you know, something wooden will be like, I don't know, special high, special high class.\nSpeaker E: Yeah, and if you, we want to put fashion and electronics, maybe we can try to do something we would.\nSpeaker C: Yeah, even if it is not completely wood, but just a part of the wood, it will be wooden, in wood and it can be interesting.\nSpeaker C: Okay, it's interesting.\nSpeaker C: So the last point is, also, we do want to have some very cheap integrated circuit chips, or we have low level or very expensive.\nSpeaker C: It depends.\nSpeaker C: But I think that low level will be, you know, it is an entry material.\nSpeaker E: Yeah, I want something easy to use, and so I think maybe something very low level would be enough.\nSpeaker E: Yeah.\nSpeaker E: And you think that we will be...\nSpeaker C: Yeah, I think it will fit on the price you want, 12 year or so.\nSpeaker E: So wood and what about the buttons?\nSpeaker C: I think the buttons, I prefer, you know, the LCD, you know, lighted buttons, because, you know, it's...\nSpeaker C: That's a special, yeah.\nSpeaker C: It's fashion and it's related to how beautiful it is.\nSpeaker C: So if you want to watch TV in the darkness, or if you want to find your remote control that is lost in the darkness, it's very easy.\nSpeaker A: What about the touch screen, for example?\nSpeaker A: It's expensive, I guess.\nSpeaker C: I think the touch screen will be as expensive as the LCD button, so...\nSpeaker C: And it is a kind of design.\nSpeaker C: It can be also interesting to have this kind of...\nSpeaker B: So you got the email?\nSpeaker E: I don't know.\nSpeaker E: I think we have only five minutes left.\nSpeaker D: Okay.\nSpeaker E: What have you been there for?\nSpeaker E: Functional requirements?\nSpeaker A: No train watching.\nSpeaker A: The other one.\nSpeaker E: This one?\nSpeaker A: Yeah.\nSpeaker A: I think so.\nSpeaker A: Just...\nSpeaker A: Yeah.\nSpeaker A: So just to meet the user requirements, I would just do a short analysis of the remote control market.\nSpeaker A: And to kind of have a better overview of what the fashion in general have checked more than only the remote control market.\nSpeaker A: So next.\nSpeaker A: So again, it's pretty much similar to what I've said in the previous meeting.\nSpeaker A: You really want a fancy look and feel. They are not so interested in a functional look and feel like the one you've shown David with all the buttons.\nSpeaker A: I mean, it sounds good technically, but it's not what they want.\nSpeaker A: So second point, they still want it to be technologically innovative.\nSpeaker A: So maybe it's sort of related to what you've said with the speech recognition and so on.\nSpeaker A: At the same time, it's important that it's easy to use.\nSpeaker A: So that were the three first points from the remote control analysis.\nSpeaker A: Now, if we look at fashion in general...\nSpeaker A: Okay.\nSpeaker A: What we really see this year is that everything from clothes to shoes and furniture is inspired by fruits and vegetables.\nSpeaker A: So I think we really have to take this into account for the design of the thing.\nSpeaker A: Because it's really what people want, even if it's in general fashion, we want it to be in the remote control.\nSpeaker A: And then if we take the ordering or the ranking of all the points, fancy look and feel on the score of seven would have six as important.\nSpeaker A: The remote control has to be technologically innovative.\nSpeaker A: It's three, then easy to use.\nSpeaker A: It's not so important, actually, with respect to other ones.\nSpeaker A: So we see fancy looking feel is the most important one.\nSpeaker A: And then if we combine this with the fashion from Milan and Paris...\nSpeaker C: It's fruits and vegetables.\nSpeaker A: We go to the fruits and vegetables.\nSpeaker A: And the other point I haven't mentioned is people want to have a spongy touch.\nSpeaker A: So this maybe doesn't really fit with the wooden design.\nSpeaker A: I don't know.\nSpeaker C: But the point is which kind of material do you need to be spongy with?\nSpeaker A: Yeah, this would be like plastic-like, but rubber.\nSpeaker A: Very smart.\nSpeaker A: Maybe not rubber-like device.\nSpeaker A: So...\nSpeaker A: Okay, that was the main point, I think, from the train in fashion.\nSpeaker D: Okay.\nSpeaker E: So we have to take decisions about the confidence concept about the energy.\nSpeaker E: So as you say, you want something technologically innovative.\nSpeaker E: Yeah.\nSpeaker E: Maybe using solar energy.\nSpeaker E: Yeah, solar, I think.\nSpeaker E: With battery would be something interesting.\nSpeaker E: Maybe we'll attract...\nSpeaker C: Yeah.\nSpeaker E: Pro-ecology consumers.\nSpeaker E: Chip and Prince, so you propose low-level chips would be enough to have something working well.\nSpeaker E: Case.\nSpeaker D: Leave it.\nSpeaker E: So you think...\nSpeaker E: Something is spongy?\nSpeaker E: No wood.\nSpeaker A: Maybe not wood, but I mean wood.\nSpeaker A: So maybe not the part you're touching.\nSpeaker C: Maybe you can have wood, for example, in the bottom.\nSpeaker C: Yeah, maybe the device.\nSpeaker C: The design you want.\nSpeaker C: It's natural.\nSpeaker E: It's natural.\nSpeaker E: It's natural.\nSpeaker C: Maybe we can stay with wood.\nSpeaker C: And it can be correlated to energy, solar energy.\nSpeaker C: So for the marketing aspect, you know, saying...\nSpeaker A: It's not exactly right for the spongy.\nSpeaker E: Yeah, it's not right.\nSpeaker E: But it's still fashion.\nSpeaker A: We could maybe have both like part of wood and some rubber for the buttons or...\nSpeaker A: Okay.\nSpeaker E: Yeah, something that you can enter.\nSpeaker E: And what about the user interface concept?\nSpeaker C: Google and...\nSpeaker E: Google and fancy?\nSpeaker C: And fancy.\nSpeaker C: How about the voice?\nSpeaker E: Because I think that with the voice and gesture recognition, there are still some disadvantages with this.\nSpeaker C: Maybe we can do some marketing studies asking people if they're interested and how...\nSpeaker C: It is an interesting concept to see to have voice control.\nSpeaker C: Yeah.\nSpeaker B: Smart control.\nSpeaker C: And the point is we can enter the...\nSpeaker C: If you have voice control, why not to put it directly on the TV?\nSpeaker C: You speak directly to the TV.\nSpeaker C: And you don't need a remote control actually.\nSpeaker E: Yeah, but you need a receiver to recognize the gestures and the voice.\nSpeaker C: It will be embedded on the TV and not on a remote control.\nSpeaker C: That's true.\nSpeaker E: Okay.\nSpeaker E: So maybe we'll just focus on the Google controller plus the fancy controller.\nSpeaker E: Maybe try to mix these two concepts together just in one and...\nSpeaker E: Do a remote control with solar energy and batteries and with low-level chips and wood.\nSpeaker E: Yep.\nSpeaker E: Good.\nSpeaker E: And LCD buttons.\nSpeaker E: Yeah, I think for these supplements, the solar energy would be something quite interesting and not maybe too difficult to add.\nSpeaker C: Yeah.\nSpeaker C: And what can we think as a supplement to what interface?\nSpeaker E: Yeah, for the interface something added value.\nSpeaker C: I think the supplement can be the voice.\nSpeaker C: It is not the most important way.\nSpeaker C: It can be a part of...\nSpeaker E: With a module, you're in a remote control with a module if you want.\nSpeaker E: You can just use commands, words and use them when you don't want to use their fingers.\nSpeaker C: Even if it is for just some keys like switching channels, one, two, three, four...\nSpeaker C: Turning the TV on or off.\nSpeaker C: Yeah.\nSpeaker C: Not very complex commands but easy commands.\nSpeaker E: So adding some vocal commands.\nSpeaker D: Yeah.\nNone: Simple ones.\nSpeaker E: Yeah, simple ones.\nSpeaker C: Okay.\nSpeaker E: So next meeting will start in 30 minutes.\nSpeaker E: So you will all have to work in your direction.\nSpeaker E: So you will have to work on the look and feel design to add the easy to use, powerful and fancy remote control with some added value such as the simple vocal commands recognition.\nSpeaker E: You will have to work more on the spongy way to add spongy touch to the buttons and try to find maybe a nice shape for the wooden remote control.\nSpeaker E: And I think we will have to evaluate the product.\nSpeaker E: Yeah.\nSpeaker A: Not forgetting about the fruits and vegetables, trends.\nSpeaker A: Yeah.\nSpeaker A: It's possible.\nSpeaker C: Okay.\nSpeaker E: And remember as I said last meeting, we really have to build to fashion remote control and the color of the society will be really...\nSpeaker E: It will be seen in the remote control.\nSpeaker E: So you will have Baba and David Jordan.\nSpeaker E: You will have to work together on the prototype.\nSpeaker E: Yeah.\nSpeaker E: And you will have next time to show us modeling a clay remote control.\nSpeaker E: So you will have to model something.\nSpeaker E: Yeah.\nSpeaker E: And I think that some specific instructions will be sent to you by your personal coach.\nSpeaker E: Okay.\nSpeaker E: So no more questions.\nSpeaker D: We can close the session.\nSpeaker D: Sounds good.\nNone: Okay.\nNone: Okay.\nNone: Cool.\nNone: So.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "TS3007c", "summary": "This meeting was about conceptual design. Firstly, groupmates had three presentations on components, trend watching, and interface contents. Their target age group was below forty, so they suggested personal preferences on the remote, like the material, the battery, the speech recognition technology, etc. Then, they discussed the trendy features and decided to include parental control as well as the gaming features in the remote control. Finally, Project Manager shared some master class knowledge with group mates and suggested applying a marketing campaign, large provider marketing approach, and real reaction product idea.", "dialogue": "Speaker D: Hello again.\nSpeaker B: Hi.\nSpeaker B: Hey, Project Manager.\nSpeaker E: Project Manager, I'm talking to tell you I have little problems with my laptop.\nSpeaker E: Okay.\nSpeaker E: So I had a little less time to complete.\nSpeaker D: Okay.\nSpeaker E: A little problem.\nSpeaker E: What was it?\nSpeaker E: The laptop, the entire window.\nSpeaker E: It's hand.\nSpeaker E: Project Manager.\nSpeaker B: Yes.\nSpeaker F: Your Project Manager.\nSpeaker B: Welcome to the conceptual design meeting for real reaction.\nSpeaker D: Hello again.\nSpeaker D: It's the same as the last time.\nSpeaker D: Also this time there will be three presentations.\nSpeaker D: And we must reach the decision on the remote control concepts.\nSpeaker D: And at the end, when I finish it off, I have some info from a computer.\nSpeaker D: I'm from a master class.\nSpeaker C: Master.\nSpeaker B: Master.\nSpeaker D: The concepts where we must reach the decision on are from two sorts.\nSpeaker D: Component concepts and user interface concepts.\nSpeaker D: The first one is really about the total package with the, well, we have decided to do it with the touch screen.\nSpeaker D: But there must be a case around it, so it won't be as breakable.\nSpeaker D: How about the energy?\nSpeaker D: The interface concept with the type and the supplements were to put one button.\nSpeaker D: And I would say yes, can you begin?\nSpeaker D: Yeah, sure.\nSpeaker D: Okay.\nSpeaker D: At the end, I will take notes.\nSpeaker D: Okay.\nSpeaker D: At the end, the minutes will be at the shared corner.\nSpeaker A: And let me see.\nSpeaker A: I'll be talking about the components.\nSpeaker A: What I did was I reviewed previously used items by two different age groups below 45 and above.\nSpeaker A: And I just watched what the differences are.\nSpeaker A: I checked what to do we want and how can we aim at the target group.\nSpeaker A: Well, what I found was that a senior citizen, sorry, it's okay.\nSpeaker A: They like the traditional materials like wood and such more, wood and chrome.\nSpeaker A: And they like straightforward shapes.\nSpeaker A: They like luxurious styles, whereas the young and dynamic, they like more soft materials, think of the tilly-tilly-tobish, for instance, soft and fluffy and colorful.\nSpeaker A: Well, shapes are curved and round.\nSpeaker A: You get to have the picture.\nSpeaker A: And they like sports and gaming and it gives them vitality.\nSpeaker D: So, well, one little question about the material, a soft material for remote control.\nSpeaker A: No, I'll get to that.\nSpeaker A: You'll see.\nSpeaker A: Yeah.\nSpeaker A: First off, let's start with the energy.\nSpeaker A: I had a choice between a few different sources.\nSpeaker A: But the two basic sources that I found were the best possible were the battery versus kinetic.\nSpeaker A: When you move something, it gets energy.\nSpeaker A: I figured, well, that's kind of high-tech.\nSpeaker A: When you have any melt control, when you pick it up, it has power.\nSpeaker A: That would be actually very nice.\nSpeaker A: I figured.\nSpeaker A: We could also use a battery that's a bit...\nSpeaker B: When the power gets low, you have to shake it.\nSpeaker A: Yeah, you just have to shake it around a bit.\nSpeaker A: And then it has some more energy.\nSpeaker A: Oh, you could just go for a battery or you could go for both.\nSpeaker E: Have you considered the option of using a solar panel?\nSpeaker A: Yeah, I actually did.\nSpeaker A: But the thing is about solar panels, you have to have light.\nSpeaker A: Some light preferably.\nSpeaker A: You could use normal light, but you wouldn't get the same amount of energy that you would from a battery or something.\nSpeaker A: So, ultimate use of solar panels.\nSpeaker A: You could use solar panels, but you'd have to implement them into the remote control, leaving you a bit less space for the interface.\nSpeaker A: So, what's actually, I figured that seeing that you'll always be within the distance of TV from the TV comes a whole bunch of light.\nSpeaker A: So, it would actually power itself from the TV, but you would just take up all the space and you wouldn't have the full amount of power actually used.\nSpeaker A: But you prefer kinetic.\nSpeaker A: I prefer kinetic because it's...\nSpeaker A: Well, the costs aren't that much higher.\nSpeaker A: And it's just a bit more high-tech than a normal battery.\nSpeaker D: Yeah, if you don't move a remote, it could not.\nSpeaker A: No, but...\nSpeaker A: So, you pick it up and you press buttons and...\nSpeaker D: That's enough to keep the energy level...\nSpeaker A: Yeah, well, actually it is.\nSpeaker A: And if it isn't, you just shake it a bit and...\nSpeaker A: And it would power up again.\nSpeaker A: But like I said, you could use a battery that you'd just keep it on the recharging whenever it moves.\nSpeaker A: And for us, you'd just juice it on the battery.\nSpeaker B: And when the battery doesn't work, I usually shake it too.\nSpeaker B: I slam it on it.\nSpeaker A: Exactly.\nSpeaker A: Exactly.\nSpeaker A: Thank you, Tim.\nSpeaker A: Furthermore, you check the cases.\nSpeaker A: We have different options concerning the cases.\nSpeaker A: You simply add a basic standard model that was kind of square and I figured that's a bit boring.\nSpeaker A: So, you could go for the single curve or the double curve.\nSpeaker A: A single curve, it's just a nice curve.\nSpeaker A: But you could go in a double curve and that's like several different dimensions.\nSpeaker A: That gives you a whole new...\nSpeaker D: Dynamic loop.\nSpeaker A: Yeah, that gives you a younger and more high-tech look, I figured.\nSpeaker A: We'll discuss later.\nSpeaker B: Are you going to draw it?\nSpeaker A: What?\nSpeaker A: Do you want me to draw?\nSpeaker A: Yeah.\nSpeaker A: In 3D.\nSpeaker A: Yeah.\nSpeaker A: I can't imagine.\nSpeaker A: Yeah, I could show you.\nSpeaker A: I could show you.\nSpeaker A: Well, let's say you...\nSpeaker B: Yeah.\nSpeaker A: Let's say that's your standard...\nSpeaker B: Design?\nSpeaker B: Yeah.\nSpeaker A: That's a bit your standard design.\nSpeaker A: But you could actually go like something like this.\nSpeaker A: And then in 3D effect you could go...\nSpeaker A: So you just...\nSpeaker A: Yeah, this is a bit difficult to him.\nSpeaker A: I didn't take a major in art.\nSpeaker A: But you could have a whole new back.\nSpeaker A: The depth you could just play around a bit with.\nSpeaker A: You don't have to use standard...\nSpeaker D: A little artistic.\nSpeaker A: Yeah, it leaves more space for creativity.\nSpeaker A: It might be an idea, but just...\nSpeaker A: Well, plastic versus rubber, we could choose what's better.\nSpeaker A: Plastic or rubber.\nSpeaker A: I prefer rubber because it feels...\nSpeaker A: It's soft.\nSpeaker A: Yeah, it's soft.\nSpeaker A: I like soft.\nSpeaker A: You're from people, aren't they?\nSpeaker A: Yeah.\nSpeaker A: Plastic also has a cheap feeling to it.\nSpeaker A: Plastic... I usually associate plastic with something as cheap.\nSpeaker A: Maybe just me.\nSpeaker A: We could talk about a layer.\nSpeaker A: Furthermore, buttons, traditional touch screen.\nSpeaker A: Well, we discussed it in a previous meeting.\nSpeaker A: So I figured just leave it at LCD.\nSpeaker A: And chipset.\nSpeaker A: Well, if we are going to use traditional buttons, we could go with a simple chipset.\nSpeaker A: But if we decide to go on an LCD screen, we would use an advanced chipset, and that would bring the necessary cost with it.\nSpeaker A: So that's something we have to keep in mind.\nSpeaker A: If it isn't manageable budget-wise, we'd have to go off to simple buttons.\nSpeaker B: Well, I think we're going to sell 10 millions of them.\nSpeaker B: So I bet a big company in Korea or Taiwan, like Samsung, think of a big discount on chips.\nSpeaker B: Yeah, probably.\nSpeaker B: Usually chips aren't more expensive than $1.\nSpeaker A: Yeah, probably.\nSpeaker A: But yeah, that shouldn't be a real issue.\nSpeaker A: That shouldn't be that big issue.\nSpeaker A: I just had a... I put a big summary here, so we could discuss it a bit.\nSpeaker A: What are your ideas when concerning battery versus kinetic?\nSpeaker E: I think you should use kinetic as a backup.\nSpeaker A: Yeah, combination.\nSpeaker A: Yeah, combination.\nSpeaker A: You use the best thing and charge it up with kinetic.\nSpeaker A: When you pick it up, it charges up.\nSpeaker B: Like an...\nSpeaker B: Acu...\nSpeaker B: Yeah, yeah, yeah.\nSpeaker D: Just like the watch from Saika.\nSpeaker A: Yeah, exactly.\nSpeaker A: What do you think?\nSpeaker D: I would prefer both to combine them.\nSpeaker A: Okay.\nSpeaker A: Well, that would bring some more cost, but...\nSpeaker A: I mean...\nSpeaker A: Okay, right?\nSpeaker A: That's the project manager's problem too.\nSpeaker E: Fifty-cent.\nSpeaker E: Of course.\nSpeaker A: Yeah.\nSpeaker A: Fifty-cent.\nSpeaker A: And then we have single curve versus double curve.\nSpeaker E: Maybe I have something in my presentation to...\nSpeaker A: Okay.\nSpeaker A: No, we'll just wait.\nSpeaker A: Plastic versus rubber.\nSpeaker A: Rubber.\nSpeaker E: Rubber.\nSpeaker E: Isn't it possible to make combination what kind of rubber is?\nSpeaker E: It's an abandoned ball, so you've got to sit there around.\nSpeaker A: I figured it would be rather hard.\nSpeaker A: Rubber casing.\nSpeaker A: This is a casing, yeah.\nSpeaker A: Rubber casing because...\nSpeaker A: Or if you're using a touch screen, it's just a casing around it.\nSpeaker A: Yeah.\nSpeaker A: You could go for plastic, but I figured...\nSpeaker A: Yeah.\nSpeaker A: I would choose rubber.\nSpeaker E: Maybe I can ask it right now, the question that I have.\nSpeaker E: Is it possible...\nSpeaker E: Is it necessary to make a touch screen square?\nSpeaker E: It isn't, I think.\nSpeaker B: I don't know.\nSpeaker B: No.\nSpeaker B: I think touch screens are generally square.\nSpeaker B: But it's the case you put around it that makes it shape.\nSpeaker E: Especially in the trunnies.\nSpeaker E: No, maybe we can...\nSpeaker B: Yeah, okay, but if you have a square LCD screen, you put a case around it that has bulbs or that covers part of the LCD screen.\nSpeaker A: Yeah, that would solve...\nSpeaker A: That would solve...\nSpeaker A: Customizable.\nSpeaker A: Okay, I guess.\nSpeaker A: So, what are the opinions?\nSpeaker A: Rubber or plastic?\nSpeaker A: I prefer rubber.\nSpeaker A: Me too.\nSpeaker A: Me too.\nSpeaker A: Yeah, you too.\nSpeaker A: You sure?\nSpeaker E: You seem to hesitate a bit.\nSpeaker E: As long as it's firm and you don't...\nSpeaker E: Yeah.\nSpeaker E: It's not bendable or something.\nSpeaker E: I think that goes too far.\nSpeaker A: Yeah, it shouldn't flop over when you hold it.\nSpeaker A: Yeah.\nSpeaker A: No, no, no, no, no.\nSpeaker A: That's...\nSpeaker A: The chipset will hold it from place and the LCD screen will...\nSpeaker E: I think it bows back where you'd drop the...\nSpeaker A: Yeah, it might.\nSpeaker A: Traditional versus LCD, well, I figured we all...\nSpeaker A: Yeah.\nSpeaker A: And then also the simple versus advanced.\nSpeaker A: I figure if we go for LCD, we should have the advanced.\nSpeaker A: So that shouldn't be a problem.\nSpeaker A: Okay, well, that's my...\nSpeaker D: Thank you.\nSpeaker B: You're welcome.\nSpeaker B: Can I do my thing?\nSpeaker B: Yeah.\nSpeaker B: Do your thing.\nSpeaker B: Bring it on.\nSpeaker B: Expert map.\nSpeaker B: Okay.\nSpeaker B: Last week.\nSpeaker B: I went to Paris and Milan, which I'm trans-watching, from marketing.\nSpeaker B: Our research and development department and I went to Paris and Milan.\nSpeaker B: In Paris and Milan, we asked different people differing in age and in income, the amount of money to spend.\nSpeaker B: What they like in design and material nowadays.\nSpeaker B: Findings.\nSpeaker B: Our main audience, so that's people below 40, prefer the following.\nSpeaker B: At first, the colors of fruit, very basic colors, like Jana's explained.\nSpeaker B: Fresh colors, green, red, strawberry, red, yellow, banana, yellow.\nSpeaker B: Considering material, they like spongy material, like in a sponge bar, like soft material.\nSpeaker B: Jana mentioned it also.\nSpeaker B: I think he did some findings himself.\nSpeaker B: You were in Paris.\nSpeaker B: Like this.\nSpeaker B: Like big fleshy colors.\nSpeaker B: Fresh.\nSpeaker B: It's nice, I think.\nSpeaker B: Okay.\nSpeaker B: Our secondary audience, people above 40 years in age, they like the dark traditional colors.\nSpeaker B: Yeah.\nSpeaker B: Materials like wood.\nSpeaker C: Yeah.\nSpeaker C: Not a material which will build a remote controller.\nSpeaker B: Oh, you could.\nSpeaker B: No.\nSpeaker B: You could.\nSpeaker B: Yeah, but never seen.\nSpeaker B: No, okay, but just a wooden, okay.\nSpeaker B: Look.\nSpeaker B: Yeah.\nSpeaker B: Like you have those fake panels on the floor.\nSpeaker B: That isn't wood anyway.\nSpeaker B: Okay.\nSpeaker B: But that's our secondary audience.\nSpeaker B: So we decided to take the people below 40.\nSpeaker B: That doesn't apply.\nSpeaker B: Okay, the third.\nSpeaker B: There are some overall thoughts about new remote controls.\nSpeaker B: They have to look fancy.\nSpeaker B: They should be technological innovative.\nSpeaker B: That means they have to be fancy things in and easy to use, but that's common.\nSpeaker B: My personal preferences.\nSpeaker B: We have to aim at our main audience and therefore use fresh colors.\nSpeaker D: Would you prefer that you can choose the color of your remote control?\nSpeaker B: Oh, I'll come to that in a second point.\nSpeaker B: Yeah, think about removable covers as you know, a mobile telephone market, customize your own remote control.\nSpeaker B: Like the Nokia, the removable covers, just put a red on it and go to the shop and buy a green one.\nSpeaker B: A third material being used could well be stuff like rubber.\nSpeaker B: Two advantages.\nSpeaker B: It fits within the current market trends and it adds safety to your remote when you drop it.\nSpeaker B: So to come back to your question, I think any people in Milan and Paris also think that the rubber should be pretty hard.\nSpeaker B: Like seeing on regular mobile phones.\nSpeaker B: Actually, they have a lot in common.\nSpeaker A: Actually, it's all a phone that you could use in your remote.\nSpeaker D: Maybe you could use your remote as a phone.\nSpeaker B: That'd be a next step.\nSpeaker B: Exactly.\nSpeaker B: My second sheet of personal preferences.\nSpeaker B: We have to reconsider the speech function recognition.\nSpeaker B: It's very innovative.\nSpeaker C: So that you say SPSS and it goes through?\nSpeaker B: Yeah, like something.\nSpeaker B: You can see that a market is just screaming for new technologies like speech recognition and stuff.\nSpeaker B: But we have to keep the cost in mind.\nSpeaker B: Well, 50 years.\nSpeaker B: It can be very expensive.\nSpeaker B: Okay, second, building games like Snake or Tetris.\nSpeaker D: Also, the link with mobile phones.\nSpeaker D: Yeah, but you don't use the games when you're watching television.\nSpeaker B: No, okay.\nSpeaker B: But you don't use games when you're making a telephone call.\nSpeaker B: It's just the same.\nSpeaker E: When your parents are watching some boring program, you can take your remote and do something else.\nSpeaker B: Yeah.\nSpeaker B: Yeah, okay.\nSpeaker B: When you're at college.\nSpeaker A: You can take your remote to school.\nSpeaker B: You also take your iPad and go play games.\nSpeaker B: Oh, I do that.\nSpeaker B: Okay, and at third, I stick with it.\nSpeaker B: The login functionality with the slogan, take parental control to a new dimension.\nSpeaker F: Very nice.\nSpeaker D: Oh, so great.\nSpeaker E: Okay.\nSpeaker E: You interface content.\nSpeaker E: Well, it's just a simple PowerPoint markup.\nSpeaker E: I want to make it more dynamic for you, so I'll draw there.\nSpeaker E: But the main concept is take one big touch screen.\nSpeaker E: Always display the primary buttons, clear and visible.\nSpeaker E: Maybe even like this with uses of a lot of space.\nSpeaker E: And make the menu structure changeable.\nSpeaker E: So if you press system, yeah, well, multiple system options can maybe five or ten or one can fit in.\nSpeaker E: Or maybe even a step further when you want to have some other options that are not programmable with one horizontal button.\nSpeaker E: Other buttons can maybe display it here if that's necessary.\nSpeaker E: Well, how do we want to look at?\nSpeaker E: Yeah, that's a remote look.\nSpeaker E: Well, you've got to hold it in one hand.\nSpeaker E: So the middle, like all remotes have to be a little bit small so you can put it in inside your hands.\nSpeaker E: And some remotes you can use with a thumb, but I think that's too difficult for this one.\nSpeaker E: So you can make it a little broader here and here.\nSpeaker E: And maybe use your thumb on the main buttons and use your other hand on the menu structure.\nSpeaker E: So here are going to be the program buttons, one, two, three, four, etc.\nSpeaker E: And the volume control and program control maybe.\nSpeaker E: And down here, this could be one big LCD menu structure.\nSpeaker E: So you can use it in this way.\nSpeaker E: One thing you've got to keep in mind, keep the remote clear of too much functions at the same time.\nSpeaker E: Note that if you are changing the menu structure here, and well, I still believe you should keep displaying the buttons, but if you're programming the color of the TV, do not display 20 other options that are possible.\nSpeaker E: Just keep it as simple as possible.\nSpeaker E: And do not use too many levels.\nSpeaker E: Do not have to select a screen first and then brightness and then color and on plus and on push plus 20 times.\nSpeaker E: But just in one button, if possible.\nSpeaker E: And well, in all noted TV levels, channel one, two, and when you get to ten and want to go back, well, you have a problem, just most modern TV, you press one, zero and you go to ten and not to one and zero after that.\nSpeaker E: So a button less and things like teletext put them in the menu, things like, yeah.\nSpeaker B: What do you think about a back and forth club buttons?\nSpeaker B: Yeah, yeah, like in the index bar.\nSpeaker E: I was trying to tell that it's volume plus and minus and this is program plus a minus.\nSpeaker B: Okay, but that is for going from four to five.\nSpeaker B: Yes, but if you go from two to eight to one.\nSpeaker E: Oh, you're watching two sports on seventeen and your wife is watching.\nSpeaker E: You can switch on two easy.\nSpeaker E: Well, yes, I think that's a good idea.\nSpeaker E: But well, my question is would you be put it in the menu?\nSpeaker E: Yeah.\nSpeaker E: And if you're using that button a lot of times, well, of course, that menu will still be displayed on screen.\nSpeaker E: You just don't have to play games in between.\nSpeaker E: But if you're really switching between two channels, you won't have time to use the other options.\nSpeaker E: Yeah, well, already told that.\nSpeaker E: That's why I conclude here.\nSpeaker D: Okay.\nSpeaker D: I think we can discuss a little about the three kind of revolutionary things Tim came with.\nSpeaker D: This parental control games and the voice recognition.\nSpeaker D: There's not too much decision on that one.\nSpeaker D: So I think the parental control is a good function to put in the remote.\nSpeaker D: I don't know how you think about it.\nSpeaker E: Well, I agree, just put it in the menu structures somewhere.\nSpeaker E: Yeah, but properties and parents.\nSpeaker B: What I see, how I see it is you put it on the remote.\nSpeaker B: And then you have like a Windows log on screen with parents, children, something like that.\nSpeaker B: When you want to use a parents option.\nSpeaker E: It has to be fast. You don't want to go down and watch the news at eight o'clock and turn it on.\nSpeaker E: And wait 20 seconds for the log on screen and then remember your code and all kinds of settings.\nSpeaker D: Yeah, but I think most people find it much more important that their children don't watch sexual violence on the television.\nSpeaker D: And wait 10 or 15 seconds longer so they can finally watch it.\nSpeaker E: And then they use like Windows XP simple log on to just push one or two or three.\nSpeaker E: And if you push parents then you have to log in, go to three digits and if you log in, like two, one, three.\nSpeaker B: And it's in.\nSpeaker E: And if you push children, you don't have to automatically go.\nSpeaker E: But you can all have children's shadows over there.\nSpeaker A: I don't know if it's worth the time and effort we're going to spend on it because it's a simple function but it's going to take a lot of resources and a lot of time into programming because you'll have to start working with profiles and such.\nSpeaker A: And I'm not sure if it's actually worth investing that much time and effort into it.\nSpeaker D: Well, I think that there's big markets for it because you're in many times in magazines and stuff and you're in the news that they believe that children are influenced by television and we're aiming to below 40 years.\nSpeaker D: But there are a lot of people below 40 who have children in young age who want them to not watch violence or violence.\nSpeaker E: Well, maybe some ID on that. Just make the remote as it is.\nSpeaker E: But make an option to insert profiles because if my granddad would buy this remote, he wouldn't want to bother with all the obligations and things to do.\nSpeaker E: Just make it an option in the menu, install profiles or something.\nSpeaker B: Yeah.\nSpeaker D: That's a better idea.\nSpeaker B: It just has to be when it's delivered out of the store.\nSpeaker B: It has to be just simple and plain.\nSpeaker B: If you want to install a person, if I got kids, I could choose between two remote controls, one with parent control and one without.\nSpeaker B: And I would buy the one with.\nSpeaker E: I have still some questions I have about how to incorporate it.\nSpeaker E: As you were thinking about some channels that you cannot see.\nSpeaker E: Well, when I think you have all the standard channels and only maybe after 10 o'clock in the evening, the films and movies will come.\nSpeaker E: Maybe some timing will be needed instead of channels because if you're watching it, I don't know.\nSpeaker E: At day cartoons will come up, maybe Friday night.\nSpeaker E: I don't know.\nSpeaker D: Maybe it can work with showview.\nSpeaker D: You can control your video recorder with showview when you take in the number.\nSpeaker D: It will start and end recording.\nSpeaker D: But maybe there's some option that the kind of showview numbers are violent and they are blocked out.\nSpeaker E: This one just font is just that someone has to send all the showview numbers on the program's day to three modes.\nSpeaker E: And it will have to decide.\nSpeaker B: Yeah, but that isn't possible.\nSpeaker A: Well, if we incorporate the parental control, that's how we do it.\nSpeaker A: Well, whatever charges go up to the TV and pressures up, for instance.\nSpeaker A: Well, nobody stopping the child from checking the channel.\nSpeaker E: Well, yeah, you could say if parental control is on, it will go from 14 to 16 with that page up.\nSpeaker A: Oh, I'm not sure because for that to happen, you'll have to receive a signal from the remote control.\nSpeaker A: So it would have to be constant signaling.\nSpeaker A: But what maybe better is to incorporate a separate device that you can program with the remote control.\nSpeaker A: And that provides parental control, for instance.\nSpeaker A: And that's just an optional device.\nSpeaker A: So besides the remote control, you'll have a separate...\nSpeaker D: I wouldn't put it in an optional device.\nSpeaker D: And that it becomes too much, I think.\nSpeaker D: If we do it, we must do it in a kind of way, like a parent profile, and other ways.\nSpeaker B: And you know, when you install another device, children can still go up to the TV, pop open the thing.\nSpeaker A: Yeah, that's true.\nSpeaker A: But in the same instance, when you have remote control.\nSpeaker A: Yeah, of course.\nSpeaker A: But only difference is that people buying the remote control will now get added feature of parental control.\nSpeaker A: And those people wouldn't necessarily want it.\nSpeaker A: Yeah, it is.\nSpeaker A: Yeah, okay.\nSpeaker B: But it's just an added feature.\nSpeaker D: Okay, and what do you guys think of the games in the voice recognition?\nSpeaker D: I personally think that it becomes too much.\nSpeaker D: Yeah, well, yeah.\nSpeaker E: It's more like it gets you to the functionality.\nSpeaker E: Email to from management that it's very popular to use voice recognition.\nSpeaker E: But I don't think when you're watching TV and hearing loud noises from the TV, someone screaming, Ah, one! And you're channel smetches.\nSpeaker B: Yeah, like, voice recognition is too much, I think.\nSpeaker B: Okay, in games?\nSpeaker B: In games.\nSpeaker B: In those...\nSpeaker B: Yeah, I can see games happening.\nSpeaker E: Yeah, I put it on the chip anyway.\nSpeaker E: That doesn't look as good as it is.\nSpeaker E: It's a recent primary feature of the remote.\nSpeaker B: And that doesn't cost a lot of extra resources.\nSpeaker D: So that will... that must be in it, do you think?\nSpeaker A: Yeah, that would be nice.\nSpeaker D: Okay.\nSpeaker D: So the games are in it, the voice recognition are not an apparent control or...\nSpeaker B: Oh, it's in it, but...\nSpeaker B: But how do we do it?\nSpeaker B: I think so.\nSpeaker E: Well, I think also it's a good idea, but it's very difficult to incorporate.\nSpeaker E: So we should make it too complex.\nSpeaker E: Some manual function, you choose parental control and maybe profiles will emerge when you put it on.\nSpeaker E: But how it's going to work?\nSpeaker D: Yeah, it'll be perfect.\nSpeaker D: But will there...\nSpeaker D: Like the first idea from... you can buy it without...\nSpeaker D: With parental control?\nSpeaker D: Or are we going to put it in and just...\nSpeaker E: I think best would be to put it in and to make it in the menu option.\nSpeaker E: Yeah, to put it in always.\nSpeaker E: Yeah.\nSpeaker E: And you can just... when you buy it, you can select personal preference and prepare the control home and prepare for it or something.\nSpeaker A: Well, I don't know. I figured if you had two different remotes, you could choose one with...\nSpeaker A: Well, a receiver in it.\nSpeaker A: So you could actually easily build in parental control, but it would be more expensive.\nSpeaker A: So in that way, you could make cheap model and expensive model.\nSpeaker A: You could make the simple model and do large model for instance.\nSpeaker B: Oh, that's it.\nSpeaker B: Yeah, but...\nSpeaker B: DVs aren't capable of sending.\nSpeaker A: Yes, they are.\nSpeaker A: Yeah.\nSpeaker B: I thought they were just able to receive it.\nSpeaker D: Yeah, some, but most of them...\nSpeaker A: Yeah, that is true.\nSpeaker E: Maybe you just have to restrict it to what program.\nSpeaker E: So the parent says you cannot watch channel 7, 9 and 10.\nSpeaker E: Yeah.\nSpeaker E: And you cannot watch all channels after 10 o'clock and just some little clock in the...\nSpeaker E: Yeah, you could easily...\nSpeaker A: Just some rules.\nSpeaker A: You could easily do that with the remote control, but you still have the problem about the television itself.\nSpeaker E: Oh, no, it's very simple.\nSpeaker E: I have seen some of the remotes who have a clock in it.\nSpeaker E: Yeah, yeah, yeah.\nSpeaker E: So the remote doesn't transmit when it's off the clock.\nSpeaker E: Yeah, okay.\nSpeaker B: But on the TV, at least my TV, is a compartment which you can press.\nSpeaker B: And there are buttons behind it which you can use.\nSpeaker B: If you don't have a...\nSpeaker E: Food is very blunt, that's not our problem.\nSpeaker E: Just have the parents buy some glue and it's not part of the remote.\nSpeaker A: Yeah, of course.\nSpeaker A: Yeah, that is true.\nSpeaker A: Yeah, that is true.\nSpeaker A: Yeah, you could go...\nSpeaker A: Yeah, that would actually make things a lot more easy.\nSpeaker A: You could just blame it on television.\nSpeaker A: Make it that problem.\nSpeaker F: Yeah.\nSpeaker A: I think we could do that.\nSpeaker A: Yeah, sure.\nSpeaker A: I'm not sure what marketing thinks about it, but...\nSpeaker A: I have to...\nSpeaker B:...consolve my legal advice.\nSpeaker D: Okay, so I think we have decided on the things that...\nSpeaker D: From Janus, the energy, the combination between battery and kinetic.\nSpeaker D: The case will be double-curved and rubber in a flashy, fruity color.\nSpeaker D: And with cover is removable.\nSpeaker D: The buttons will be touchscreen, but there may not be too many buttons.\nSpeaker D: And in the menu structure, there must not be five minutes to go.\nSpeaker D: Too many levels and it must be easy to use.\nSpeaker D: The parental control will be incorporated, but it must be not too complex.\nSpeaker D: And the games are in it.\nSpeaker D: So, I think we have...\nSpeaker D: Consensus.\nSpeaker D: Okay.\nSpeaker B: A little more...\nSpeaker B: Oh, I have one thing left.\nSpeaker B: Yeah.\nSpeaker B: Maybe for Jerome.\nSpeaker B: Yeah.\nSpeaker B: The fuse.\nSpeaker B: Maybe it's handy to build in an expert view and a simple view.\nSpeaker D: Like a mode or something.\nSpeaker D: Yeah.\nSpeaker B: Like in an expert view, you have a lot of...\nSpeaker B: Yeah, if you have anything...\nSpeaker E:...thinking about it is just a menu structure when you don't use it.\nSpeaker E: It's simple.\nSpeaker D: Yeah.\nSpeaker E: When you put system properties, entire list pops up with...\nSpeaker E:...all kind of stuff you can put.\nSpeaker B: Yeah, okay.\nSpeaker D: It's already incorporated a little in a concept.\nSpeaker A: You could actually build in a function that you can program it yourself for the more advanced users.\nSpeaker A: Yeah.\nSpeaker A: Which buttons do you want to in it?\nSpeaker A: Because you can build on the back-forward button and somebody would just want to watch two channels.\nSpeaker A: Yeah.\nSpeaker B: Just leave the other numbers away.\nSpeaker A: Yeah.\nSpeaker A: You could have people want to...\nSpeaker A: We take it to the other meeting, okay?\nSpeaker D: Okay.\nSpeaker D: A little chat to do and then we finish.\nSpeaker D: I went to a master class and two things can come in handy for us.\nSpeaker D: First thing is we talked about knowledge management, knowledge engineering and the fact that the idea of knowledge sharing and learning knowledge from other companies is all like that.\nSpeaker D: It's very hot at the moment. So it's mostly for the management expert to look what other marketing...\nSpeaker D: Marketing.\nSpeaker B: What did I said, management?\nSpeaker D: Management.\nSpeaker D: Just talking about myself.\nSpeaker D: Yeah, that's helpful.\nSpeaker B: Okay.\nSpeaker D: Go on.\nSpeaker D: What other companies had to do also with the marketing campaign when they put on a newer remote, just some...\nSpeaker D:...amort expert.\nSpeaker D: And another one.\nSpeaker D: There are... yeah, benchmarking.\nSpeaker D: That's the word I saw.\nSpeaker D: I couldn't...\nSpeaker D: Okay.\nSpeaker D: Another thing is there was a survey and it came out that people like to buy things from a single large provider instead of those who are partnering with us.\nSpeaker D: So we must bring it as if a real reaction is a big company, a trustful company and...\nSpeaker B: I know a marketing name for a product.\nSpeaker B: Okay.\nSpeaker B: R to the third power.\nSpeaker B: R3.\nSpeaker B: Real reaction remote.\nSpeaker D: I had another idea to put the idea of a real reaction and a single brand and that kind of thing.\nSpeaker D: When you put on the remote, you see kind of just like when you tell a little animation.\nSpeaker D: Real reaction remote.\nSpeaker D: And with the motto, we put fashion in electronics.\nSpeaker D: And then you go...\nSpeaker B: Okay, but it has to be like a split second because you have to put in a code also.\nSpeaker E: You can show somebody a logo for.5 seconds.\nSpeaker E: They'll still remember it.\nSpeaker D: Yeah.\nSpeaker D: But lingo on every time you see it.\nSpeaker D: The idea from this thing is that we must present it as we are a single large provider who will stand on our own.\nSpeaker A: If you have the LCD screen, you can have a small logo at the bottom and you could just stay there.\nSpeaker A: That spins around.\nSpeaker B: Yeah, that spins around.\nSpeaker D: Also, but we are...\nSpeaker B: It's just like a globe in the internet.\nSpeaker B: Yeah, yeah.\nSpeaker B: Yeah, yeah.\nSpeaker B: Yeah, something like that.\nSpeaker D: Okay, but the small icon.\nSpeaker D: Think about that kind of thing.\nSpeaker D: That's what they said in the master club.\nSpeaker D: And she never.\nSpeaker D: Next meeting starts in 30 minutes.\nSpeaker D: So maybe you can go to Paris and Milana.\nSpeaker B: What's the game?\nSpeaker B: Who gave you the master class?\nSpeaker B: The master class.\nSpeaker B: Ronald Bathenberg.\nSpeaker D: French men, actually.\nSpeaker B: Okay, thanks.\nSpeaker D: Very special.\nSpeaker D: Next session.\nSpeaker D: ID, the industrial development center.\nSpeaker D: We'll work together on a prototype.\nSpeaker D: And we'll drawing it on the smart board.\nSpeaker D: So that's a kind of new one, I think.\nSpeaker D: And the marketing expert will be keeping busy with the product evaluation.\nSpeaker D: But the most specific instructions will be sent to you by email.\nSpeaker A: So we're going to work together right now?\nSpeaker D: Yeah, well, this is what I heard.\nSpeaker D: In the master class.\nSpeaker D: So you just wait for the specific instructions.\nSpeaker D: But I think it was a very good session.\nSpeaker D: Gentlemen.\nSpeaker D: Definitely.\nSpeaker E: No new.\nSpeaker E: Me too.\nSpeaker D: Well, thank you all.\nSpeaker B: Oh, thank you.\nSpeaker B: Thank you.\nSpeaker B: Thank you very much.\nSpeaker D: Thank you too.\nSpeaker D: Give me a good graduation.\nSpeaker D: Okay, guys.\nSpeaker D: You were a success.\nSpeaker D: Cheers.\nSpeaker D: See you in a whole hour.\nSpeaker D: But keep an eye on your laptop.\nSpeaker A: I'm not sure if we only, because I saw something about individual actions.\nSpeaker D: Yeah, but there's still my instructions that you will work together.\nSpeaker E: Yeah.\nSpeaker E: Here?\nSpeaker E: That's not my problem.\nSpeaker D: Bye bye.\nSpeaker D: No, I'm still managing.\nSpeaker D: The product manager always works alone.\n"}, {"meeting_id": "ES2015c", "summary": "This was the third meeting of conceptual design, including three presentations and a conclusive discussion to make some specific decisions. First, Industrial Designer introduced and led to a discussion about different options for the components of the circuit board, the energy sources, and the materials of casing and button. Next, User Interface gave the second presentation about interface concept and two types of user interfaces: the graphical user interface and the command interface, and strongly recommended the latter. The last presentation by Marketing reported the trend-watching for remote control and the relevant discussion centralized the trendy theme of vegetable and fruit. Given all of the information, the group eventually decided on some special designs, such as the kinetic energy, the regular chip, the plastic rubber coat, the interchangeable plate, the command line interface, and the basic buttons.", "dialogue": "Speaker B: Is everyone ready to start?\nSpeaker B: Yeah.\nSpeaker B: Yes.\nSpeaker B: Okay, great.\nSpeaker B: Well, welcome to the third meeting of conceptual design.\nSpeaker B: Let's get the PowerPoint presentation up and running.\nNone: Okay.\nSpeaker B: On the agenda for this particular meeting, we'll have your three presentations on what you've done since our last meeting after we came up with some general ideas of our design.\nSpeaker B: And then we have to make some key decisions on our remote control concept and how we're going to make you what material we're going to use and that sort of thing.\nSpeaker B: The meeting will be 40 minutes long.\nSpeaker B: And we'll once again have Poppy as our first presenter.\nSpeaker B: Okay.\nSpeaker B: All right.\nSpeaker B: And now we'll switch up PowerPoint.\nSpeaker B: Thank you.\nNone: Okay.\nSpeaker D: Should be just loading.\nNone: Okay.\nNone: Oh, I can't see it.\nNone: Oh, okay.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker A: Oh, I can't see it on my screen.\nSpeaker A: That's just going here.\nNone: Okay.\nSpeaker A: We've been doing some research into the different components that we could use.\nSpeaker A: What's available to us to actually make the remote control.\nSpeaker A: First of all, we have to look at how the remote control is actually made and what is it happens inside the casing, which is more your field.\nSpeaker A: The main internal feature is a circuit board, which contains all the electronics and also the contacts with the power source, which is not necessarily a battery is without to see.\nSpeaker A: There are several components of the circuit board that we need to consider where we'll be getting them from, what they'll be made of, including the integrated circuit, which is also known as the chip, which is where all the main information is contained.\nSpeaker A: Diode, transistors, resonators, resistors and capacitors all need to be considered as well and all their positioning in the circuit.\nSpeaker B: Are they all included, like mandatory or are these different options?\nSpeaker A: These, they're all different options.\nSpeaker A: They're all separate apart from the chip, which we will probably decide whether we buy a simple or regular advance.\nSpeaker A: We'll go into that later.\nSpeaker A: All the other things, the individual components that we'd have to get in separately and work out the most effective circuit, including all the wires and that and the LED of course, that's the light emitting diode.\nSpeaker A: So we've got flexibility with colours and things with that as well.\nSpeaker A: Okay, there are lots of different possibilities for the energy source.\nSpeaker A: We could use a basic battery, but that brings with it, like, the need to be recharged and the bulk, the size of it as well.\nSpeaker A: And they're not so great to dispose of environmentally.\nSpeaker A: There's the hand-powered dynamo, which is the sort of thing that was used for torches 50 years ago, a bit out of date.\nSpeaker A: Kinetic energy is something that's been recently developed.\nSpeaker A: What is a hand-powered dynamo?\nSpeaker A: Where you manually charge up the power, like you do everything.\nSpeaker A: Just every once in a while?\nSpeaker A: Sorry?\nSpeaker A: Just every once in a while.\nSpeaker B: It'd be kind of strange to always be kind of strange.\nSpeaker B: Yeah, be like going a step back in time.\nSpeaker A: I don't think it would really be with cutting edge technology.\nSpeaker A: No.\nSpeaker A: Whereas kinetic energy is a new idea that's been used by some watches and other devices where you just shake the device and it gives it the power.\nSpeaker A: The kinetic energy is transformed into power to make the circuit work.\nSpeaker A: Or the solar power, which we're considering inside a building, which is where it's going to be used, might not be quite so useful.\nSpeaker A: But good to look into, renewable energy, always there.\nSpeaker A: Lots of considerations for the case, like what sort of shape it would be curved or flat.\nSpeaker A: That's got a lot to do with the ergonomics.\nSpeaker A: I think how it's comfortable and sits in the hand.\nSpeaker A: We don't want something that's huge and you can't pick up or too small or too slidey.\nSpeaker A: I know I've had a remote control before, which you couldn't tell, which was the front and the back.\nSpeaker A: It had so many buttons on and the shape was so symmetrical that I'd be pressing like a volume button instead of the on button because you can't really see what you're around it.\nSpeaker A: We also can choose what materials.\nSpeaker A: We could use metal, we could use rubber, which might be more ideal for the anti-RSI.\nSpeaker A: It's like the same sort of rubber that's used in stress balls and things like that.\nSpeaker A: So it's very soft, not so stressful, on your hands.\nSpeaker A: Wood.\nSpeaker A: Again, stepping back inside, we get there.\nSpeaker A: I don't think that's quite up to date with what we're looking for here.\nSpeaker A: Titanium is not going to be possible, it's beyond our budget really, but would have been maybe for future projects.\nSpeaker A: I have a choice for buttons as well.\nSpeaker A: We've developed some good advances in technology with our research team.\nSpeaker A: We've found some new multiple options, scroll buttons.\nSpeaker A: I think that was brought up before.\nSpeaker A: They're basically quite a flexible design, modern.\nSpeaker A: You don't have to use individual buttons.\nSpeaker A: You can just slide up and down.\nSpeaker A: I'm sure we're all quite familiar with those on mobile or computer laptop.\nSpeaker B: One thing with the scroll buttons though, we would have to have an LCD.\nSpeaker B: Yeah, that's true.\nSpeaker A: Display and the glow-in-the-dark thing.\nSpeaker A: Yeah, if we have.\nSpeaker A: Might be difficult.\nSpeaker A: Yeah.\nSpeaker A: We're going on to that later with the LCD.\nSpeaker A: That means we need a really advanced chip.\nSpeaker A: It's unlikely that that's going to be in our budget.\nSpeaker A: Also, we've got the integrated push button, which is what we're most familiar with.\nSpeaker A: It's most straightforward.\nSpeaker A: You can incorporate that with the scroll button as well.\nSpeaker A: Good.\nSpeaker A: Good.\nSpeaker A: Stations to make there.\nSpeaker A: This is what I was just saying.\nSpeaker A: Four linked into the different depends on what type of buttons we have and the inputs.\nSpeaker A: Simple.\nSpeaker A: The settings button, regular, you could link with the scroll button and the advanced would have to go with a LCD display.\nSpeaker A: My personal preference is I think we could go for the kinetic energy source.\nSpeaker A: I think that's quite an advanced kind of technology.\nSpeaker A: It's not been seen before so it could be quite a novelty factor, attractive as well.\nSpeaker A: So energy savings, you're producing the energy.\nSpeaker A: You don't need an external battery supply or solar panels.\nSpeaker A: So you just give it a shake.\nSpeaker A: Yeah.\nSpeaker A: Sounds brilliant to me.\nSpeaker A: Rubbing casing, I think we're probably the best if we're going for the anti-RSI and more choice with aesthetics.\nSpeaker A: It could be pretty much any colour we want.\nSpeaker A: It gives you more flexibility there.\nSpeaker A: Probably the regular chip is supposed to be simple and we could possibly have the scroll and the push but no LCD because we probably can't afford that one.\nSpeaker B: One concern with the rubber casing is that it would be rubber encapsulating all of these chips and diodes and delicate technology as the exterior.\nSpeaker B: This is the one thing that's protecting its innards.\nSpeaker A: I think that would be more internal casing and the rubber would just be the what's in contact with the human hand.\nSpeaker B: Another thing is it might be more difficult if it's a rubber exterior.\nSpeaker B: We're talking about putting on interchangeable plates.\nSpeaker B: I don't see how a rubber plate going on top of it would stay there.\nSpeaker B: If it was a clip-on plastic plate, it would work that way.\nSpeaker A: Maybe if it was just more of a rubber coating which was on to a case, the whole thing would be removable.\nSpeaker D: Plastic with rubber.\nSpeaker D: I can't think of what.\nSpeaker A: There is a certain phone that has a rubber casing.\nSpeaker A: Or you can get outer casing for iPods and something that's just protective as well.\nSpeaker A: It stops it, it will reduce the impact if it was dropped or something.\nSpeaker A: It wouldn't damage its help.\nSpeaker A: Maybe a mixture of both.\nSpeaker C: The actual remote would be hard plastic.\nSpeaker B: That sounds good.\nSpeaker B: Is it possible to put designs onto this type of rubber?\nSpeaker A: As far as I know, it should be.\nSpeaker B: I like the kinetic energy source idea.\nSpeaker B: I don't know when people will be moving around a lot.\nSpeaker B: But I think that it's worth it.\nSpeaker B: Kinetic energy source it can make.\nSpeaker B: We can have any kind of style that would be as heavy or bulky.\nSpeaker C: We can have a battery pack.\nSpeaker C: It's a battery pack.\nSpeaker C: There is a one battery.\nSpeaker C: It was running off of one battery.\nSpeaker A: It's an alternative just in case.\nSpeaker D: Maybe we could.\nSpeaker D: That's true.\nSpeaker A: Maybe that could be the backup.\nSpeaker A: Yeah.\nSpeaker C: Instead of a battery.\nSpeaker C: It's so light.\nSpeaker C: You can't watch TV in the dark then.\nSpeaker A: But you don't need the solar all the time.\nSpeaker A: It can be stored.\nSpeaker A: It can be like you can have the solar energy and then it can store that energy.\nSpeaker A: It just needs to be in light for a certain amount of time per day, like a few hours a day.\nSpeaker B: I think that might be a little impractical.\nSpeaker C: I think sometimes it's just shoved under a cushion.\nSpeaker C: That's true.\nSpeaker B: People don't want to have to worry about that.\nSpeaker A: I suppose it would be really annoying if you get to think, oh no, I've got to charge my remote.\nSpeaker A: It's a kinetic thing I think that's the best about that.\nSpeaker A: It's instant energy.\nSpeaker A: You can shake it a few times or whatever.\nSpeaker A: Or just like a big it up when you're going to use it.\nSpeaker A: Instead of you don't have to make sure it's in the right place to charge.\nSpeaker B: Cool.\nSpeaker B: Is that the end of your presentation?\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker A: Thank you.\nSpeaker A: Thank you.\nNone: Oh.\nSpeaker D: It's not on my screen.\nSpeaker A: I didn't know.\nSpeaker A: I don't know why.\nSpeaker A: I don't know why.\nSpeaker A: I think I just use the mouse.\nSpeaker D: Oh, okay.\nSpeaker B: Yeah.\nSpeaker D: Awesome.\nSpeaker D: I'm just going to go through the interface concept with use noise.\nSpeaker D: First of all, explain what a user interface is.\nSpeaker D: It's the aspects of the computer system which can be seen or otherwise seen for example, a pair or a shelf, maybe a by the human user.\nSpeaker D: And it's also the commands that the user uses to control its operation and to input data.\nSpeaker D: There are two types of user interfaces.\nSpeaker D: There's the graphical user interface which emphasizes the use of pictures for our fit and the point device for example, a mouse for input control.\nSpeaker D: Is that sort of like the schools I knew were talking about?\nSpeaker B: Oh, okay.\nNone: Yeah.\nSpeaker D: I'm not really sure about the pictures that maybe that's on an LCD screen.\nSpeaker D: Or maybe it's the buttons and pictures.\nSpeaker D: Yeah.\nSpeaker A: So I suppose second time.\nSpeaker B: I'm sorry.\nSpeaker B: Because command interface requires you to type textual commands and input at a keyboard.\nSpeaker B: So the numbers are sort of like a keyboard.\nSpeaker B: Yeah.\nSpeaker B: You're pressing the numbers for what you want.\nSpeaker D: For the graphical user you must use some kind of presentation for the graphics.\nSpeaker A: I suppose on some buttons you would have the power would be some kind of symbol.\nSpeaker A: And if you wanted to go on to teletext or I know we're not having that.\nSpeaker A: But I mean the similar thing is they have this little picture of a screen with lines across it which I suppose it's that sort of thing like the symbol on the button.\nSpeaker A: But if we're having a simplified display anyway, that we would have to focus so much on that and be more than the numbers in the volume.\nSpeaker A: It would be more a command interface.\nSpeaker A: So it's a need to think of symbols for the volume to display.\nSpeaker C: Yeah.\nSpeaker C: So it's just graphical for the pointing aspect.\nSpeaker C: The infrared is like that's considered.\nSpeaker D: Hmm.\nSpeaker D: No, I think it did you with the actual symbols that are on the buttons that are in the control.\nSpeaker C: Okay.\nSpeaker C: So when it says pointing, it invites that doesn't include.\nSpeaker D: Well, it could be a school or something and something could come up in the school.\nSpeaker B: Yeah.\nSpeaker D: I think they're talking about LCD type things.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But um.\nSpeaker D: I think we're going to go with the command interface anyway.\nSpeaker D: Yeah.\nSpeaker D: But the, we could incorporate some of the graphical user points as in just to make it nicer to look at maybe.\nSpeaker D: What do you mean?\nSpeaker D: Like, I can't think of an example, but sort of like little pictures rather than.\nSpeaker D: Oh yeah.\nSpeaker D: Like how the buttons signed instead of same volume.\nSpeaker D: Like a little speaker.\nSpeaker D: Yeah.\nSpeaker B: As a button though.\nSpeaker B: So it's a keyboard in the shape of it.\nSpeaker B: Right.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker D: I like that.\nSpeaker D: Um.\nSpeaker D: The.\nSpeaker D: We've started with the command interface.\nSpeaker D: This is the most useful.\nSpeaker D: The little control.\nSpeaker D: The less complicated.\nSpeaker D: And the controls is the more than user.\nSpeaker D: The more.\nSpeaker D: Um.\nSpeaker D: The little control is the cheaper than you think.\nSpeaker D: That would have more money in the budget to, um.\nSpeaker D: Target the design.\nSpeaker D: Yeah.\nSpeaker D: To face.\nSpeaker D: Make it more trendy.\nSpeaker D: And then it's no longer.\nSpeaker D: Yeah.\nSpeaker B: I'm sure like kinetic energy would probably dip into the budget.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Quite a new technology.\nSpeaker D: Um.\nSpeaker D: We.\nSpeaker D: We also have to keep in mind when we're designing our.\nSpeaker D: Um.\nSpeaker D: More users and like.\nSpeaker D: A lot of interfaces.\nSpeaker D: The clutter buttons that.\nSpeaker D: Um.\nSpeaker D: That the functions, colors and forms.\nSpeaker D: Aren't always helpful.\nSpeaker D: Yeah.\nSpeaker A: That's true.\nSpeaker A: Yeah.\nSpeaker A: The buttons is all the different colors for different choices.\nSpeaker A: Yeah.\nSpeaker A: Things can be a bit.\nSpeaker A: Yeah.\nSpeaker A: And that.\nSpeaker D: All.\nSpeaker D: All interfaces are different.\nSpeaker D: Um, that doesn't improve the product.\nSpeaker D: So we need to come up with something that's easy to understand.\nSpeaker D: And maybe learn from these takes of all our interfaces that can be too complicated for people to use.\nNone: Mm.\nSpeaker D: Does anyone have any questions?\nSpeaker B: Do you think that we should keep.\nSpeaker B: All the buttons to one same color just to keep it give it a simplistic look.\nSpeaker C: I think if we go with the design plate thing, we'll have to.\nSpeaker C: Just because of color clashing and.\nSpeaker B: Yeah.\nSpeaker C: If you want to.\nSpeaker C: So I mean, depending on what comes out of the design, we might have to stick to just black buttons.\nSpeaker B: But what about the lighting up effect?\nSpeaker C: I mean different colors for the lighting or.\nSpeaker B: Um, well, um, I thought we had decided that we would.\nSpeaker B: If you touch one of the buttons, they'd all light up.\nSpeaker B: And so if they were black, it would be possible for them to light up.\nSpeaker C: Oh, I see what you're saying.\nSpeaker B: Well, if they were white, they would glow probably if they were made of rubber.\nSpeaker C: Oh, so you're picturing the light is coming from the back.\nSpeaker C: I kind of pictured it as kind of coming from the sides and lighting it up frontwards.\nSpeaker C: Oh, oh, we're like coming from the back.\nSpeaker C: Okay.\nSpeaker A: I'd seem like an internal light.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: That comes through.\nSpeaker A: So there would have to have to be some parts, maybe transparent around.\nSpeaker A: Right.\nSpeaker A: But it's something.\nSpeaker B: A little rubber is a more translucent product too.\nSpeaker B: So it should be able to.\nSpeaker B: Yeah.\nSpeaker B: Is that taken care of?\nSpeaker D: And the phone that I was thinking of as well, when you change the, um, covers of it, the, the, the little buttons that actually, you know, that.\nSpeaker D: Yeah.\nSpeaker D: Are behind the rubber.\nSpeaker D: So you can change the buttons when you're changing the faces.\nSpeaker C: Oh, yeah.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: I mean, I don't have to decide on one color.\nSpeaker D: Each face could have a.\nSpeaker B: Well, if they're raised up buttons.\nSpeaker B: Yeah.\nSpeaker B: So that you can feel them, we were talking about it being more.\nSpeaker B: A lot more tangible.\nSpeaker B: It might be more difficult to do.\nSpeaker B: Yeah.\nSpeaker B: If they're sticking up, if it's flat, then like, I don't know, like a cell phone or a mobile phone.\nSpeaker B: It's like all very flat and you just have to sort of press down on these tiny little buttons.\nSpeaker B: But.\nSpeaker B: Yeah.\nSpeaker A: I think it would be possible. I don't think it would make that much difference.\nSpeaker A: I mean, the dimensions of it, because if it's just like constructed in the same way as like the front cover of a mobile phone, you can take off the hard cover and then there's the, like, so the buttons and then you get to the circuit.\nSpeaker A: I don't think it would matter that the buttons are bigger through the, the top casing.\nSpeaker A: I'm sure you could work out to fit in the casing without causing too much.\nSpeaker A: It's fine.\nSpeaker B: All right.\nSpeaker B: If it's doable, you can.\nSpeaker B: Yeah.\nSpeaker B: That was good to me.\nNone: Yeah.\nSpeaker D: That's everything.\nSpeaker D: Okay.\nSpeaker B: All right.\nSpeaker B: Thank you.\nNone: Okay.\nNone: Is that my turn?\nSpeaker B: That means you're out.\nSpeaker C: All right.\nSpeaker C: I have a shorter report for you today.\nSpeaker C: It took a while to get this.\nSpeaker A: You're not plugged in yet.\nSpeaker C: Oh, that's a very good point.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: I'll go through some issues.\nSpeaker C: real remote. We can work off of them and see how the fashion has changed. I'll list the three most important aspects that I've come across. They're each more important than the one that comes after it. The first one is that there should be a fancy look and feel instead of the current functional look and feel. This should be our priority, as we've been saying.\nSpeaker C: The second most important aspect is that the remote control should be technologically innovative. I think we've done a lot of talking about that, just with lighting and the buttons in the face plates. It looks like we'll be able to keep on track with that. The third most important aspect is that their remote control should be easy to use. Pretty basic there.\nSpeaker C: The recent fashion update, according to fashion washers in Paris and Milan, is that fruit and vegetables will be the most important theme for clothes, shoes, and furniture.\nSpeaker C: Also, in contrast to last year, the field and materials is expected to be spongy. Again, we've already discussed that with rubber versus hard plastic buttons. My personal preference is here. Of course, we've already talked about the personal face plates in this meeting, and I'd like to stick to that. The fruits and vegetable themes, I don't know if that's going to work for us. It sounds something that you do use on kitchenware. I don't know if we want to do it on our moats. It could be one of the options, maybe for the television that people have in their kitchen. The temporary light-up idea sounds like we're going to stick to that, and then tying in a trendy look with user-friendliness.\nSpeaker B: The fruits and vegetables is the only area that I find rather jarring. Everything else we can really do, according to our plans already, given the market. But fruits and vegetables seems a very strange idea for the world. It's everywhere.\nSpeaker C: I've seen a lot of persons with olives on them. I think we stick to TV-based TV shows. They still need to fit into\nSpeaker A: people's decor, though. I think we could possibly take a more abstract design, look at the basic shapes of different fruits and vegetables, and then just really strip it down to really basic shapes. We don't have to make something in the shape of a strawberry, but it could have the curves of a strawberry, or a strawberry seed, or a really make it quite abstract if that fits in more with what we're doing instead of fruits and vegetables. If you look at it straight forward, it's a bit... It doesn't fit in with the trendy... Well, obviously, it does, if that's the correct theme, but maybe we could go more directly. I don't know. But initially, I don't know. I think if we just sort of tone it down a bit.\nSpeaker B: Not photos of fruits.\nSpeaker C: One thing I was thinking, I don't know if you'll remember from our kickoff meeting, we talked about our favorite animals. So maybe they could be animal-based, you know, because a lot of people have a house cat or a dog. That might be getting too specific, and we should see what the success of the first faceplates are. But it doesn't like to keep in mind.\nSpeaker C: Did you have any questions?\nSpeaker C: Pretty straightforward.\nSpeaker B: All right, given that information, we need to start making some more specific decisions.\nSpeaker B: So I'll need to hook up the PowerPoint again.\nSpeaker C: Have you guys been saving your PowerPoint presentations to the...\nSpeaker A: I didn't for the first one, but I have now. It's still around, right?\nSpeaker A: Yeah.\nSpeaker B: Okay.\nSpeaker B: I'm just trying to make this pop up now.\nSpeaker B: All right, here we go. Okay, so we have to decide now exactly what we are going to do. So energy we owe.\nSpeaker B: Oh no, I can't write it in when it's in this setting.\nSpeaker B: So I don't know how to take it out of...\nSpeaker B: The PowerPoint key, I think.\nSpeaker B: Yeah, okay. So back to the decision. Energy, we've decided on kinetic range. Okay, so that's good.\nSpeaker A: Are we going to have a backup?\nSpeaker A: Hmm. Or do we just...\nSpeaker B: But would a backup really be necessary?\nSpeaker B: I mean, will people just use the battery if there's no?\nSpeaker A: I think maybe we could just go for the kinetic energy and be bold and innovative and hope this works and well, hope that it works.\nSpeaker A: Yeah, there's no reason why it wouldn't work, right?\nSpeaker A: I think we should just take advantage of using this with full potential. It can even be one of our selling points.\nSpeaker B: It could be fully kinetic energy. Yeah, environmentally conscious.\nSpeaker C: They're so just how important they just have to shake it a bit and that will be very light. Yeah, yeah.\nSpeaker B: All right, the next decision is chip on print.\nSpeaker A: I don't exactly know what that means. It's whether we went for the simple, the regular O of the advanced chip and that linked in with what buttons we were going to have. Right, and we were going for more simplistic style, right?\nSpeaker A: Yeah, but that was without the LCD, so that means we are not doing the advanced. So it depends on whether we wanted the scroll buttons or just the push buttons. I think we decided on the push buttons, right?\nSpeaker B: So that's the simple... Would we need a more advanced one for the lighting, the interior lighting system? Yeah, possibly. So it's probably going to\nSpeaker A: be the regular chip that we're going to need. So it's a medium. So regular chip. That's called medium or regular? Regular, sorry.\nSpeaker A: Okay. Regular chip. Oh, it was regular not simple?\nSpeaker A: Yeah, I think. Yeah, if we... Because of the lighting that we've decided to\nSpeaker B: put in as well. Okay, and cases. So is this dependent on shape or what\nSpeaker A: it's made of or what? I think this is just going to be a very outside case which we both decided on rubber.\nSpeaker B: Yeah, well I guess plastic and rubber. Coated in rubber.\nSpeaker B: Plastic with rubber coating and inter-changeable.\nSpeaker A: Inter-changeable, yeah, still going for that.\nSpeaker B: Yeah, inter-changeable, please.\nSpeaker B: User interface concept. This is your time to shine.\nSpeaker D: We decided on the component. I'm sorry, I've lost my um...\nSpeaker D: Part one thing.\nSpeaker D: So what happened? I think it was called command interface.\nSpeaker B: Oh yeah, yeah. Command interface. Command interface.\nSpeaker D: Command line interface here. Does it command line? Yeah.\nSpeaker B: Okay. Line interface.\nSpeaker B: All right, and supplements. What's that all about?\nSpeaker D: Um, I think that is... Well, we haven't really made any decisions about what we're going to do, like the cluster of button functions, collage and forms, and that and consistent use. Like what are our ideas to combat these problems?\nSpeaker D: Different interfaces are very different and can be confusing because the causes are different and the cause um, the different cluster chip buttons that they have.\nSpeaker D: We haven't really decided what to do like that.\nSpeaker B: Um, what are our choices here? Well, it's just um,\nSpeaker D: where, where shall we locate the buttons? What kind of functions will it shall we have? We like will have the numbers of the\nSpeaker B: channels and we'll have power, a channel changer, and volume, and power. So, just basic. Yeah, like I don't know if we should go into like adjusting light levels, things like that, because\nSpeaker A: different televisions will have. Yeah, tone, contrast, and things\nSpeaker C: that's a bit... That was on um, one of my presentations about how often it was used. Do you remember that? Yeah, it was minimal.\nSpeaker C: Well, it wasn't that I was hardly ever used. Should I bring it up?\nSpeaker A: Yeah, yeah. I was told, visions will come with a remote. Yeah, and surely that would be like quite specific to the individual television.\nSpeaker B: Okay, well we know we want numbers. The screen settings was used\nSpeaker C: 0.5 times an hour. So twice every once every two hours. Um, and it was considered a 1.5 relevance on the scale of 10.\nSpeaker C: That's brightness color tone and all that.\nSpeaker A: I think most of that comes like on the individual television set itself, isn't that? I'm sure it has its own buttons, so it don't necessarily need to have it on remote. Yeah, and different televisions, like we, I don't\nSpeaker B: know if we can make a remote that will be universal to all the different kinds of\nSpeaker A: Yeah, so we're just going for power, channels, volume.\nSpeaker C: The other one was audio settings, mono stereo pitch. I mean, I sometimes use that. Some TVs will have the option of like living room style, movie style.\nSpeaker C: Yeah. And that they said that's used 0.8 times an hour, which is actually somewhat high, like almost once an hour.\nSpeaker C: Yeah, um, relevance of two.\nSpeaker B: Oh, we have five minutes to finish our media.\nSpeaker B: All right, yeah. Okay, um, let's do this fast.\nSpeaker B: Um, well that didn't come up on mind. Should we have audio?\nSpeaker D: It only comes up on mind usually.\nSpeaker B: Yeah, I don't, it's a problem with the international appeal.\nSpeaker B: I think if we have audio because we don't know how other televisions work, but we know that everyone has this and it's the same.\nSpeaker C: Yeah, I guess it affects the marketing because it, it is a good sales point to say, aren't you annoyed with the remote controls that have all these buttons?\nSpeaker A: Yeah, yeah, that's it. And you can just go for it, make it a selling point that is just the basic. Yeah. All right.\nSpeaker B: All right. Good. And okay, in closing, because we'll be at five minutes, we'll be meeting again in 30 minutes.\nSpeaker B: Um, you'll be working, probably we'll be working on the look and feel design.\nSpeaker B: Wait a minute, is that right? Yep. And, um, the user interface design, so this is where the trendy stuff comes in and you'll be evaluating the product.\nSpeaker B: Um, Poppy and Tara will have to work together, zoom modeling clay, and um, your personal coach will give you the best of the information of what needs to happen. Okay. All right. So,\nSpeaker A: anyone else have something to say? Um, I just have one question about the whole fruit and vegetable aspect. Yeah. Are we seeing as that was the most popular thing that came up out of your market research? I think we should keep to that rather than moving to animals or something, because yeah, yeah. Even if that may not seem obvious to us, if that's what the surveys brought out, I think we should probably go along with that. Yeah. Um, so I suppose that I'll come out in our design development, but happy to go\nSpeaker C: ahead with the fruit. No, do you guys need one idea of how many, uh, are you gonna come over casing ideas? Yeah.\nSpeaker D: We're not five. What are you thinking? What do you think of this? Um, having the numbers kind of like, not a bunch of grapes, but you know, like purple and kind of in a triangle. Mm hmm. Like grape. Yeah, you can have some fun with the buttons. It's\nSpeaker C: true. Yeah. We can have a nice idea. Yeah. I guess ideally you kind of think of age markets as well. So, yeah. Something that will, you know, appeal to the 15 to 25 year olds, something that your granny would want on her remote control. Yeah. Yeah. And international taste as well.\nSpeaker B: A lot of things to consider. Lots of, all right. Okay. Yeah, we'll stick to the fruit veggie theme. All right. And, um, we'll reconvene in 30 minutes. Okay.\nSpeaker B: Thank you. I'll see you later.\nSpeaker B: Oh, what I just did.\n"}, {"meeting_id": "TS3011a", "summary": "This is the kick-off meeting of the new television remote control project. The team got acquainted with each other and Project Manager described the scope of the project. Industrial Designer claimed that he could support any practical ideas technically and proposed that they could make an interoperable remote control. User Interface agreed and pointed out that people changed the state of the TV by pressing the button. Then he proposed that people should be able to feel the buttons without having to look at them and the buttons should be visible in dark. In terms of the way to connect the remote control to the TV, Industrial Designer suggested that infrared could be substituted by Bluetooth.", "dialogue": "None: Good morning.\nSpeaker D: Good morning.\nSpeaker B: Marketing experts.\nNone: All right.\nSpeaker B: Are you ready?\nSpeaker A: You should put the laptop right into the square for the cameras.\nSpeaker A: Good morning.\nSpeaker C: Good morning.\nSpeaker C: Mr. P.M.\nSpeaker A: How are you today?\nSpeaker C: I'm fine.\nSpeaker C: How was your business trip to Boston?\nSpeaker A: Well, actually, I didn't go.\nSpeaker A: I didn't feel like it.\nSpeaker A: Do you want to open it as read-only?\nSpeaker A: I guess I should close it here.\nSpeaker D: You have the same message of Windows connection?\nSpeaker D: Yeah, we will have a tip that I would like to play.\nSpeaker A: Oh, stand by and close the window.\nSpeaker A: Okay. The waiting is for our marketing expert.\nSpeaker A: Root.\nSpeaker A: Project kick-off.\nSpeaker C: Is there a schedule for this meeting?\nSpeaker A: Yes, there is, actually.\nSpeaker A: I will... this is the agenda for today, for this meeting.\nSpeaker A: Good morning, Root.\nSpeaker A: It's important.\nSpeaker A: Shall I close the door?\nSpeaker A: Yeah, great.\nSpeaker A: It's important that the laptops are exactly on the square for the cameras.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: We're here to develop a new product.\nSpeaker A: I'm sure you've had a mail from our account manager about it.\nSpeaker A: And this is the first meeting to generate some ideas about it.\nSpeaker A: You are here in a specific role.\nSpeaker A: Root is here as the marketing expert.\nSpeaker A: Root is here as the user interface designer, and Sebastian is here in the role of industrial designer.\nSpeaker A: Is that correct?\nSpeaker A: That's correct.\nSpeaker A: Okay.\nSpeaker A: We're going to do a little tool training for the tools we are going to use during the meetings we are going to have here.\nSpeaker A: Then I will tell you a little bit about my idea of the project plan, and we will have a discussion.\nSpeaker A: This meeting should take no more than 25 minutes, so we should keep that in mind.\nSpeaker C: Is there any room for a little presentation?\nSpeaker C: Maybe during the discussion session?\nSpeaker C: There is.\nSpeaker A: Yeah, there is.\nSpeaker C: Okay.\nSpeaker A: No problem.\nSpeaker A: Okay, this new product we are going to develop.\nSpeaker A: It's remote control, a television remote control.\nSpeaker A: First of all, it should be original, it should be trendy and user friendly.\nSpeaker A: Those are kind of easy goals.\nSpeaker A: And I'm sure we can find more goals for the product we are going to develop.\nSpeaker A: We will discuss later on more ideas about how the remote should look and how it should function.\nSpeaker A: All those kind of things.\nSpeaker A: The market, we should have a look at the market.\nSpeaker A: We are going to use a project method during this development, which consists of three different design stages.\nSpeaker A: The functional design, the conceptual design and the detailed design.\nSpeaker A: All of these stages mean that we do some individual work, prepare, and then meet to discuss the programming.\nSpeaker A: The progressions.\nSpeaker A: The first stage, the functional design.\nSpeaker A: We are going to search for the user requirements and we will make a specification of that.\nSpeaker A: The second is the technical functional design.\nSpeaker A: What effect should the remote have?\nSpeaker A: Well, in this case, control the television, I think.\nSpeaker A: The last one is the working design. How exactly does it work in the technical sense?\nSpeaker A: The other design stages, we will discuss that later.\nSpeaker A: We will keep it with the functional design.\nSpeaker A: Before we think about remote control, we will work with some of the tools we have here.\nSpeaker A: As you see now, I can give a presentation.\nSpeaker A: It is also possible to use this one as well. You can display two presentations as you want to.\nSpeaker A: To present it, to show us a file, you will need to place it in your project documents folder, which is on your desktop.\nSpeaker A: Then we have this electronic whiteboard system.\nSpeaker A: I will show that now.\nSpeaker A: You can draw on the board using this pen.\nSpeaker A: There are little sensors.\nSpeaker A: Do not grab it here, but a little bit more to the end.\nSpeaker A: It is on the eraser now. We click the pen button.\nSpeaker A: Not too fast writing.\nSpeaker A: You can insert a new slide or whiteboard file by either using the insert function or by clicking the next button or the blank button.\nSpeaker A: All our smartboard notes should be kept in the same file.\nSpeaker A: Do not make a new file. Just use this one during the day.\nSpeaker A: You can use the eraser to make something go away.\nSpeaker D: You can use the same whiteboard file.\nSpeaker D: We can work together on it while we are going to only use it in the meetings.\nSpeaker A: In the meetings, only in the meetings. It is really like the regular whiteboard.\nSpeaker A: You can choose the format.\nSpeaker A: Let me see.\nSpeaker A: I guess it is maybe because I am not pen selected.\nSpeaker A: Current color. You can choose another color.\nSpeaker A: You can choose the line width.\nSpeaker A: Now I have a different line width and a color.\nSpeaker A: It is quite easy if you do have any questions.\nSpeaker A: I wrote down the documents.\nSpeaker A: It should be in the project documents folder if you want to discuss it with us.\nSpeaker A: As a little training, I will ask Root first to draw your own animal on a new slide with a different color and a different line width than the one now selected.\nSpeaker B: Okay.\nSpeaker B: It is a funny rabbit.\nSpeaker A: A rabbit.\nSpeaker A: Could you do the same?\nSpeaker D: Sebastian is thinking about the animal.\nSpeaker D: It is a cat.\nSpeaker A: It should be a cat.\nSpeaker B: It should be a cat.\nSpeaker A: I am guessing a horse.\nSpeaker D: With a very small legs.\nSpeaker C: I guess you get the idea.\nSpeaker A: Beautiful.\nSpeaker A: Beautiful.\nSpeaker A: You can use this at any time during the presentation if you want to.\nSpeaker A: Back to our project.\nSpeaker A: The remote control we are going to develop.\nSpeaker A: We will have a selling price of about 25 euros.\nSpeaker A: This product is 50 million euros.\nSpeaker A: The product will be sold if there is market interest in more than one country.\nSpeaker A: The production costs should not be more than 12 euros 50.\nSpeaker A: We should keep that in mind during the development because those are important numbers.\nSpeaker A: Then the discussion may be the time for Sebastian to show his presentation.\nSpeaker C: I have some technical issues which I would like to present to you before we start the discussion.\nSpeaker C: There might be some.\nSpeaker C: First, my role of the industrial designer.\nSpeaker C: I would like to think about the implementation of things and the technical possibilities and impossibilities.\nSpeaker C: If someone comes up with ideas, I will try to translate them in technical functions.\nSpeaker C: There might be some impossibilities.\nSpeaker C: I will propose some implementations for that.\nSpeaker C: I will remind people of some new technical possibilities which are available and which might be interesting to implement in our product.\nSpeaker C: I have some initial ideas about some things which are maybe nice to take with you in the upcoming discussion.\nSpeaker C: One thing about interoperability.\nSpeaker C: I think modern remote control shoots control a diverse subset of equipment.\nSpeaker C: For instance, DVD player, cell phones, video and audio equipment.\nSpeaker C: One remote control for all your equipment.\nSpeaker D: For cell phone?\nSpeaker C: There should be some interoperability between them.\nSpeaker C: I think it can come in handy.\nSpeaker C: We should discuss that.\nSpeaker C: We should think about the way how these things should communicate with each other.\nSpeaker C: We are not living in the 1980s anymore.\nSpeaker C: You should think about Bluetooth.\nSpeaker D: The infrared is a little old-fashioned.\nSpeaker D: But all the fees are equipped with infrared.\nSpeaker D: You cannot know.\nSpeaker D: That is the most of them.\nSpeaker C: That is something we should discuss about everybody should think about it.\nSpeaker C: That is just my role.\nSpeaker C: I will give you some technical input.\nSpeaker C: I think now the time is to have a little discussion about what the product should be and how it should look.\nSpeaker C: Take these things into account when you start the discussion.\nSpeaker A: The main point you are telling us is focus on the interoperability.\nSpeaker C: One remote control shoots control one or more pieces of equipment.\nSpeaker A: The way of communicating with these equipment.\nSpeaker A: That was your presentation.\nSpeaker A: I will go back to my own presentation.\nSpeaker A: I do think it is time now to discuss some things.\nSpeaker A: Sebastian told us a few things about the technical implications.\nSpeaker A: There are other things like how to make it trendy, which is, I think, most routes role.\nSpeaker A: The way how it should be controlled by the user, which is a rule during this part.\nSpeaker A: Let us start with you.\nSpeaker A: How do you think the remote should function for the user?\nSpeaker D: A few things in mind.\nSpeaker D: The remote should work on different TVs or different devices.\nSpeaker D: It is a universal remote control for customers to buy a store.\nSpeaker D: It is not for Philips or whatever to buy our remote for their own product.\nSpeaker A: We are going to business to consumer.\nSpeaker C: We are not developing this product for a specific vendor.\nSpeaker C: We are developing this product and we want to set it to a very broad public.\nSpeaker C: You should fit to every device.\nSpeaker A: Do you agree?\nSpeaker D: The technical function, like what you said, is just by pressing the button.\nSpeaker D: You should change the state of the TV.\nSpeaker D: That is just the basic technical function.\nSpeaker D: That is my part for this session.\nSpeaker A: Is that ease of use?\nSpeaker D: The technical function is pure.\nSpeaker D: What should the remote control do?\nSpeaker D: What is his task?\nSpeaker D: It just should change the TV's state.\nSpeaker C: You see the buttons as a means of doing this.\nSpeaker D: Maybe you want a touchscreen.\nSpeaker C: I have seen these remote controls with this little stick, which you can move forward.\nSpeaker C: It is very easy for a user to switch between channels or between tracks on a CD or on a DFT player.\nSpeaker C: Maybe that is an idea.\nSpeaker D: For other user interface, I had more industrial point to get a DFT.\nSpeaker D: It is quite disturbing when you always have to point directly to the DFT.\nSpeaker D: You must point it everywhere.\nSpeaker D: Maybe not even point it.\nSpeaker D: In that case, infrared should be a restriction to that.\nSpeaker C: It is a restriction for the operating range too.\nSpeaker C: When you are not able to point at the device, the range is fairly important.\nSpeaker D: For TV, you are always in the neighborhood of a TV.\nSpeaker D: I don't think the range should be a problem to that.\nSpeaker D: If you want to get it working with a radio and you are in outside your garden with just one speaker, then maybe the range should be a...\nSpeaker A: Okay, gentlemen.\nSpeaker A: Just to remind you, we have five minutes left for this meeting.\nSpeaker D: Two more things.\nSpeaker D: You are able to feel the buttons without...\nSpeaker D: You have to know what you do without looking at the buttons.\nSpeaker D: It should be a user interface for feeling.\nSpeaker D: It should be good to understand.\nSpeaker D: You must use the buttons for volume or whatever.\nSpeaker D: Of course, you don't always know where all buttons are.\nSpeaker D: It should be visible in dark too.\nSpeaker D: Those are two really user interface.\nSpeaker A: I will write down glow in the dark.\nSpeaker A: Yeah, that's perfect.\nSpeaker A: Okay.\nSpeaker A: I just want to hear Ruth's input for this meeting.\nSpeaker A: Do you have anything already idea about how the market will respond to such a product or what we should take in account when developing such a product?\nSpeaker B: I think most things have already been said like control multiple devices.\nSpeaker B: And infrared might be an issue because...\nSpeaker B: Well, he said about pointing, but lots of devices already use infrared.\nSpeaker B: We'll probably have to implement that.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Sebastian, did you have any other ideas?\nSpeaker C: Well, yes, I had about three minutes ago.\nSpeaker C: I seem to forget them.\nSpeaker C: I forgot them.\nSpeaker C: Oh, yes, I remember.\nSpeaker C: You said something about visibility in dark.\nSpeaker C: Would it be nice for a user to have display on this remote control on which you can see functions which makes it easier to operate this?\nSpeaker D: Yeah, well, maybe I...\nSpeaker D: It can be quite simple.\nSpeaker D: You can just have white buttons with black mark on it.\nSpeaker D: The digits in black, then it's already visible in dark.\nSpeaker D: So it don't have to be lightning or fancy.\nSpeaker D: Well, it can be for design.\nSpeaker A: Yeah, okay, because we want to develop a trendy product.\nSpeaker C: So there's a cost limitation too.\nSpeaker C: Yeah.\nSpeaker A: Twelve year or fifty.\nSpeaker C: So, yes.\nSpeaker C: A big problem, I think.\nSpeaker C: I think the financial part of this project implicates that it's not going to be high end product.\nSpeaker C: The cost price is very low.\nSpeaker D: Just remind something.\nSpeaker D: The digits of the painting on the buttons should not fade.\nSpeaker D: Okay.\nSpeaker D: Always have the soft buttons always clear the paint on it.\nSpeaker D: The mark.\nSpeaker A: Okay, I have one point which comes in mind now.\nSpeaker A: I think the device should either be rechargeable very easily, or should not consume too much power.\nSpeaker A: Because it's very annoying if you need to change the batteries every other week.\nSpeaker C: Yeah.\nSpeaker A: So maybe we could, for example, only light the buttons that are applicable at that moment.\nSpeaker A: Or, yeah, I don't know.\nSpeaker A: That's more, Sebastian's.\nSpeaker D: But then Bluetooth might be a problem.\nSpeaker D: Because I think Bluetooth uses a lot of...\nSpeaker C: Yeah.\nSpeaker C: I'm not sure.\nSpeaker C: I know cell phones have integrated Bluetooth also.\nSpeaker C: Well, it seems to work quite okay.\nSpeaker D: But you can use Bluetooth all the time, 24 hours a day.\nSpeaker A: Okay.\nSpeaker A: Gentlemen.\nSpeaker A: It's over.\nSpeaker A: I'm afraid we do not have any more time.\nSpeaker A: So we will go back to our own work.\nSpeaker A: Next meeting starts in 30 minutes.\nSpeaker A: And well, you know your individual actions or your personal coach will probably email you about it.\nSpeaker A: So, this was it.\nSpeaker A: Okay.\nSpeaker A: See you in 30 minutes.\nNone: Okay.\n"}, {"meeting_id": "Bro028", "summary": "The meeting occurred very close to the finalization of the Aurora model. The team discussed how various components of the model could be moved around to improve results. They had to decrease system latency as well to meet the Aurora requirements. The team thought that LPC could solve the problems they were having. The team also began discussing the Wall Street Journal data and task, which focused on speech recognition. The team was studying the task and learning from Mississippi State's existing model on how to perform better on it. Finally, the team discussed how they could histogram equalization as an alternative to some of the techniques they were using in their current models.", "dialogue": "Speaker D: So next week we'll have both Berger and Michael, Michael Clansmith, Berger Comire will join us.\nSpeaker D: And you're probably going to go up in a couple of three weeks or so, when are you thinking of going up to...\nSpeaker F: Yeah, like, not next week, but maybe the week after.\nSpeaker D: Good, so at least we'll have one meeting with you with your store and that's good.\nSpeaker F: Yeah, well, maybe you can start with this.\nSpeaker F: All today, huh?\nSpeaker F: Yeah, so there was this conference call this morning and the only topic on the agenda was just to discuss and to get a decision about this latency problem.\nSpeaker F: No, this is a conference call between different rural people or just the conference call between the overall group.\nSpeaker F: Yeah, there were like two hours of discussions and then suddenly people were tired, I guess, and they decided on a number of 220.\nSpeaker F: Including everything.\nSpeaker F: It means that it's like 80 milliseconds less than before.\nSpeaker E: And what are we sitting at currently?\nSpeaker F: So currently we have a system that has 230, so that's fine.\nSpeaker F: 230?\nSpeaker F: Yeah, so that's the system that's described on the second point of...\nSpeaker E: So we have to reduce it by 10 milliseconds.\nSpeaker F: Yeah, but that's not a problem, I guess.\nSpeaker E: Okay, it's primarily determined by the VAD at this point, right?\nSpeaker D: Yeah, so we can make the VAD a little shorter.\nSpeaker D: We probably should do that pretty soon so that we don't get used to it being a certain way.\nSpeaker D: Was Harri on the phone?\nSpeaker F: Yeah, sure.\nSpeaker F: Well, it was mainly a discussion between Harri and David who was like...\nSpeaker F: Yeah.\nSpeaker F: So the second thing is the system that we have currently, we have like a system that gives 62% improvement, but if you want to stick to this latency...\nSpeaker F: Well, it has a latency of 230, but if you want also to stick to the number of features that's limited to 60, then we go a little bit down, but it's still 61%.\nSpeaker F: And if we drop the tandem network, then we have 57%.\nSpeaker E: But the 230 includes the tandem network?\nSpeaker F: Yeah.\nSpeaker E: Okay.\nSpeaker D: And is the tandem network small enough that it will fit on the terminal size?\nSpeaker F: No, I don't think so.\nSpeaker E: No.\nSpeaker F: Okay.\nSpeaker F: In terms of computation, if we use their way of computing the MIPS, I think it fits, but it's a linear problem of memory.\nSpeaker F: And I don't know how much this can be discussed or not, because it could be in ROM, so it's maybe not that expensive.\nSpeaker F: How much memory do you have?\nSpeaker F: I don't remember exactly, but yeah, to check that.\nSpeaker D: Yeah, I'd like to see that. Maybe I could think a little bit about it, because maybe we could make it a little smaller, or...\nSpeaker D: I mean, it'd be neat if we could fit it all.\nSpeaker D: I'd like to see how far off we are.\nSpeaker D: But I guess it's still within their rules to have it on the server side.\nSpeaker D: Yeah.\nSpeaker E: Okay.\nSpeaker E: This is still...\nSpeaker D: Oh, well, you're saying here.\nSpeaker F: Yeah, there were small tricks to make this tandem network work.\nSpeaker F: And one of the tricks was to use some kind of hierarchical structure, where the silence probability is not computed by the final tandem network, but by the VAD network.\nSpeaker F: So apparently, it looks better when we use the silence probability from the VAD network, and we rescale the other probabilities by 1 minus the silence probability.\nSpeaker F: So it's some kind of hierarchical thing that Sonilo also tried on Spine, and apparently it helps a little bit also.\nSpeaker F: Yeah, the reason why we did that with the silence probability was that...\nSpeaker D: I'm really sorry. Can you repeat what you're saying about the silence probability only in my mind?\nSpeaker F: So there is a tandem network that estimates the phone probabilities and the silence probabilities also.\nSpeaker F: And things get better when instead of using the silence probability computed by the tandem network, we use the silence probability given by the VAD network.\nSpeaker F: The VAD network is smaller, but maybe...\nSpeaker F: So we have a network for the VAD which has 100 in the unit, and the tandem network has 500.\nSpeaker F: So it's smaller, but the silence probability from this network seems better.\nSpeaker E: Okay.\nSpeaker E: Well, it looks strange, but...\nSpeaker F: Maybe it's something to do with the fact that we don't have infinite training data.\nSpeaker F: We don't.\nSpeaker F: So things are not optimal.\nSpeaker C: You were going to say why what made you...\nSpeaker F: Yeah, there was a problem that we observed...\nSpeaker F: That there were many insertions in the system.\nSpeaker F: Actually, plugging the tandem network was increasing, I think, the number of insertions.\nSpeaker F: And...\nSpeaker F: So it looked strange, and then just using the other silence probability apps.\nSpeaker F: So the next thing we will do is train the tandem on more data.\nSpeaker D: So you know in a way what it's a little bit like combining knowledge sources, right?\nSpeaker D: Because the fact that you have these two nets that are different sizes, means they behave a little differently, they find different things.\nSpeaker D: And if you have the distribution that you have from speech sounds, you have sort of one source of knowledge, and rather than just taking one minus that to get the other, which is essentially what's happening.\nSpeaker D: You have this other source of knowledge that you're putting in there.\nSpeaker D: So you make use of both of them, which you're ending up with.\nSpeaker D: Maybe it's better.\nSpeaker D: Anyway, probably justify anything.\nSpeaker F: And the features are different also.\nSpeaker F: Do you need a new, the same features there?\nSpeaker D: Oh, that might be a key, actually.\nSpeaker D: Thinking about speech versus non-speech.\nSpeaker D: That's a good point.\nSpeaker F: Well, there are other things that we should do, but...\nSpeaker F: to require time.\nSpeaker F: We have ideas like, so these things are like having a better VAD.\nSpeaker F: We have some ideas about that, to probably implies working a little bit on features that are more suited to a voice activity detection.\nSpeaker F: Working on the second stream, of course we have ideas on this or so, but we need to try different things.\nSpeaker F: Better noise estimation.\nSpeaker D: I mean, back on the second stream, and that's something you've talked about for a while.\nSpeaker D: I think that's certainly a high hope.\nSpeaker D: Yeah.\nSpeaker D: So we have this default idea about just using some sort of purely spectral thing.\nSpeaker F: For a second stream.\nSpeaker F: We need to first try with this, and it clearly hurts.\nSpeaker D: But how was the stream combined?\nSpeaker F: It was just combined by the acoustic model.\nSpeaker F: So there was no neural network.\nSpeaker D: Right, so I mean, if you just had a second stream, it was just Spectro and had another neural net.\nSpeaker D: Combined there.\nSpeaker F: Might be good.\nSpeaker F: Yeah.\nSpeaker F: Yeah, and the other thing about noise estimation.\nSpeaker F: Maybe try to train the training data for the tandem network right now.\nSpeaker F: It's using the noises from the Aurora task.\nSpeaker F: I think that people might try to argue about that because then in some cases we have the same noise for training the network.\nSpeaker F: The noise is that are used for testing.\nSpeaker F: We have to try to get rid of these.\nSpeaker D: Maybe just put in some other noise.\nSpeaker D: Something is different.\nSpeaker D: I mean, it's probably helpful to have a little noise there.\nSpeaker D: But maybe something else they could say it was, if it doesn't hurt too much, that's a good idea.\nSpeaker F: Yeah, the last thing is that I think we are getting close to human performance.\nSpeaker F: Well, that's something I would like to investigate further.\nSpeaker F: But I did like, I did listen to the most noisy attorneys of the speech that Caritalian tried to transcribe them.\nSpeaker D: So this is a particular human.\nSpeaker F: Yeah, so that's the experiment.\nSpeaker F: It's just one subject.\nSpeaker F: But still, what happens is that the digit error rate on this is around one person while our system is currently at seven person.\nSpeaker F: But what happens also is that if I listen to the recentized version of the speech, and I recentize this using a white noise that's filtered by LPC filter, well, you can argue that this is not speech, so the era is not trained to recognize this, but actually it sounds like whispering.\nSpeaker D: Well, there's two problems there.\nSpeaker D: The first is that by doing LPC 12 with synthesized speech, like you're saying, you're adding other degradation.\nSpeaker D: So it's not just the noise, but you're adding in fact some degradation because it's only an approximation.\nSpeaker D: The second thing is, which is maybe more interesting, is that if you do it with whispered speech, you get this number.\nSpeaker D: What if you had done analysis, recentesis, and taken the pitch as well?\nSpeaker D: So now you put the pitch in.\nSpeaker D: What would the percentage be then?\nSpeaker D: That's a question.\nSpeaker D: So you see if it's, let's say it's back down to 1% again.\nSpeaker D: That would say at least for people having the pitch is really, really important, which would be interesting in itself.\nSpeaker D: If it's stayed up near 5%, then I'd say, boy, LPC 12 is pretty crummy.\nSpeaker D: So I'm not sure how we can conclude from this anything about that our system is close to.\nSpeaker F: The point is that what I listen to when I recent this idea at BC 12 Spectrum is in a way what the system is hearing.\nSpeaker F: Because all the excitation is not taken into account.\nSpeaker F: That's what we do with our system.\nSpeaker F: Well, it's not doing LPC.\nSpeaker D: What if you did LPC 20?\nSpeaker D: 20.\nSpeaker D: The thing is LPC is not a really great representation of speech.\nSpeaker D: All I'm saying is that you have in addition to the removal of pitch, you also are doing a particular parameterization, which...\nSpeaker D: So let's see, how would you do that?\nSpeaker F: That's what we do with our systems.\nSpeaker D: We don't, because we do Mel filter bank.\nSpeaker F: Is it that different?\nSpeaker D: I don't know what Mel based synthesis would sound like, but certainly the specter are quite different.\nSpeaker B: Could you test the human performance on the state of original?\nSpeaker F: This is the one person's number.\nSpeaker D: Yeah, it's 1%. He's trying to remove the pitch information and make it closer to what we're seeing as the feature vectors.\nSpeaker B: So your performance was 1%.\nSpeaker B: And then when you recent the size of LPC 12, it went to 5.\nSpeaker D: I mean, it's a little bit still in the apples and oranges because we are choosing these features in order to be the best for recognition.\nSpeaker D: And if you listen to them, they still might not be there, even if you made something closer to what we're going to do.\nSpeaker D: It might not sound very good.\nSpeaker D: And the degradation from that might actually make it even harder to understand than the LPC 12.\nSpeaker D: So all I'm saying is that the LPC 12 puts in synthesis, puts in some degradation.\nSpeaker D: That's not what we're used to hearing.\nSpeaker D: And it's not just the question of how much information is there as if you will always take maximum advantage of any information that's presented to you.\nSpeaker D: In fact, you hear something is better than others.\nSpeaker D: But I agree that it says that the kind of information that we're feeding it is probably a little bit minimal.\nSpeaker D: There's definitely some things that we've thrown away.\nSpeaker D: And that's why I was saying it might be interesting if you, an interesting test of this would be if you actually put the pitch back in.\nSpeaker D: So you just extract it from the actual speech and put it back in and see does that, does that make the difference?\nSpeaker D: If that takes it down to 1% again, then you say, okay, it's in fact having not just the spectral envelope but also the pitch that has information that people can use anyway.\nSpeaker B: From this it's pretty safe to say that the system is with either 2% to 7% away from the performance of a human.\nSpeaker D: Well, or it's, yeah, so it's 1.4 times 2% to 7 times the error for Stefan.\nSpeaker D: But I don't know, I don't want to take you away from other things, but that's the first thing that I would be curious about is, you know, when you...\nSpeaker F: But the signal itself is like a mix of a periodic sound and a voiced sound and the noise which is mostly noise, I mean not periodic.\nSpeaker F: So what do you mean exactly by putting back the pitch in the LPC?\nSpeaker D: Yeah, you did LPC recentuses.\nSpeaker D: LPC recentuses.\nSpeaker D: So, and you did it with a noise source rather than with periodic source.\nSpeaker D: So if you actually did real recentuses like you do in an LPC synthesizer where it's on voice, do you use noise, where it's voice, do you use periodic pulses?\nSpeaker F: Yeah, but it's neither purely voiced or purely unvoiced.\nSpeaker F: Especially because there is noise.\nSpeaker D: Well, it might be hard to do it, but the thing is that if you detect that there's a very strong periodic components, then you can use a voiced thing.\nSpeaker D: Yeah, I mean, it's probably not worth your time. It's a side thing and it's allowed to do.\nSpeaker D: But I'm saying, at least as a thought experiment, that's what I would want to test. I would want to drive it with a two-source system rather than a one-source system.\nSpeaker D: And then that would tell you whether in fact it's because we've talked about like the harmonic tunneling or other things that people have done based on pitch.\nSpeaker D: Maybe that's really a key element. Maybe without that it's not possible to do a whole lot better than we're doing.\nSpeaker F: That's what I was thinking by doing this experiment.\nSpeaker D: But I mean, other than that, I don't think it's, I mean, other than the pitch-steed information, it's hard to imagine that there's a whole lot more in the signal that we're throwing away.\nSpeaker F: Yeah, but yeah.\nSpeaker D: I mean, we're using per-number filters in the filter bank.\nSpeaker E: Yeah.\nSpeaker D: Yeah. That's, I mean, 1% is sort of what I would figure if somebody was paying really close attention, you might get.\nSpeaker D: I would actually think that if you look to people in various times of the day in different amounts of attention, you might actually get up to 3% or 4%.\nSpeaker D: So we're not incredibly far off. On the other hand, with any of these numbers except maybe the 1%, it's not actually usable.\nSpeaker D: A commercial system with a full telephone number or something. Yeah, but these numbers never see.\nSpeaker F: Yeah. Right. Well, yeah, these numbers I mean.\nSpeaker D: Good. While we're still on Aurora stuff, maybe we can talk a little about the status with the, well, three journal things for it.\nSpeaker B: So I've downloaded a couple of things from the SIP state. One is their software, their OVC SR system, downloaded latest version of that, got it compiled and everything.\nSpeaker B: Downloaded the scripts. They wrote some scripts that sort of make it easy to run the system on a one-street journal data.\nSpeaker B: So I haven't run the scripts yet. I'm waiting. There was one problem with part of it. I wrote it and I know to Joe asking him about it.\nSpeaker B: So I'm waiting to hear from him. But I did print something out just to give you an idea about where the system is.\nSpeaker B: They, on their website, they did this little table of where their system performs relative to other systems that have done this task.\nSpeaker B: And the SIP state system using a diagram grammar is about 8.2%.\nSpeaker B: They're comparable systems from, we're getting from like 6.9628%.\nSpeaker B: So they're, this is unclean. This is on Corsa.\nSpeaker B: Yeah, they've started the table where they're showing their results on various different noise conditions, but they don't have a lot of it filled in.\nSpeaker B: And I didn't notice until after I printed out that they don't say here what these different testing conditions are going to have to click on it on the website to see them.\nSpeaker D: So I don't know what those, what kind of numbers are they getting on these on them.\nSpeaker B: Well, I was a little confused because on this table, I'm, they're showing word error rate, but on this one, I don't know if these are word error rates because they're really big.\nSpeaker B: So under condition one here is 10%, and under three goes to 64.6%.\nSpeaker B: That's probably a error. I mean, so I guess maybe there are error rates, but they're really high.\nSpeaker D: I don't find that, but I mean, what's, what's some of the lower error rates on, on, on some of the higher error rates on some of these highly mismatched, difficult conditions?\nSpeaker F: Yeah, it's around 15 to 20%.\nSpeaker F: Right. And the baseline error rate.\nSpeaker F: Yeah. 20% error rate.\nSpeaker D: Yeah. So 20% error rate on digits.\nSpeaker D: So if you're doing, so if you're doing, you know, 60,000, yeah, if you're saying 60,000 word recognition, getting 60% error on some of these noise conditions, not at all surprising.\nSpeaker F: The baseline is 60% on some digits on the more mismatched conditions.\nSpeaker B: Okay.\nSpeaker F: Yeah.\nSpeaker B: So yeah, that's probably what it is then.\nSpeaker B: Yeah, so that's a lot of different conditions.\nSpeaker D: It's a bad sign when you're looking at the numbers you can't tell whether it's accuracy or error rate.\nSpeaker B: Yeah, it's going to be hard.\nSpeaker B: I'm still waiting for them to release the multi CPU version of their scripts, because right now their script only handles processing on a single CPU, which will take a really long time.\nSpeaker B: So this is for the training.\nSpeaker B: I believe yes, for the training also.\nSpeaker B: Okay.\nSpeaker B: Yeah, so it gets started because I'll go ahead and fill with just the single CPU one.\nSpeaker B: They released like a smaller data set that you can use.\nSpeaker B: It only takes like 16 hours to train.\nSpeaker B: Oh, good.\nSpeaker B: Just to make sure with that.\nSpeaker B: Yeah.\nSpeaker D: That's cool.\nSpeaker D: I guess the actual evaluation will be in six weeks or something.\nSpeaker D: So, so all right.\nSpeaker F: We don't know yet.\nSpeaker E: I think we don't know.\nSpeaker B: It wasn't on the conference call this one.\nSpeaker B: Did they say anything on the conference call about how the Wall Street Journal part of the test was going to be run?\nSpeaker B: I thought I remember hearing that some sites were saying that they didn't have the compute to be able to run the Wall Street Journal stuff at their place.\nSpeaker B: So there was some talk about having Mississippi State run assistance for them.\nSpeaker B: Did that come up at all?\nSpeaker F: I know.\nSpeaker F: Well, this first this was not the point at all of this meeting today.\nSpeaker F: And frankly, I don't know because I didn't read also the most recent maize about the large vocabulary task.\nSpeaker F: But did you still get the maize?\nSpeaker F: You're not in the main English.\nSpeaker B: Only male I get is from Mississippi State.\nSpeaker B: So we should about this.\nSpeaker D: I have to say there's something funny sounding about saying one of these big companies doesn't have enough compute power to do that.\nSpeaker D: So they're having to have it done by Mississippi State.\nSpeaker D: Yeah.\nSpeaker D: It just sounds funny.\nSpeaker D: Yeah.\nSpeaker B: Anyway.\nSpeaker B: Yeah, I'm wondering about that because there's this whole issue about, you know, simple tuning parameters like word insertion penalty.\nSpeaker B: And whether or not those are going to be tuned or not.\nSpeaker B: It makes a big difference.\nSpeaker B: You change your front end.\nSpeaker B: You know, the scale is completely empty completely different.\nSpeaker B: It seems reasonable that that at least should be quick to match the front end.\nSpeaker F: You didn't get any answer from Joe.\nSpeaker B: I did, but Joe said, you know, what you're saying makes sense.\nSpeaker B: I don't know.\nSpeaker B: He doesn't know what the answer is.\nSpeaker B: I mean, we have this back in front a little bit about, you know, our site is going to, are you going to run this data for different sites?\nSpeaker B: And well, if Mississippi State runs it, then maybe they'll do a little optimization on that parameter.\nSpeaker B: But then he wasn't asked to run it for anybody.\nSpeaker B: So it's just not clear yet what's going to happen.\nSpeaker B: He's been putting this stuff out on their website and the people to grab that.\nSpeaker B: I haven't heard too much about it.\nSpeaker D: So it could be, I mean Chuck and I had actually talked about this a couple of times in some lunches, I think, that one thing that we might want to do, there's this question about, you know, what do you want to scale?\nSpeaker D: Suppose you can't adjust these word insertion penalties and so forth.\nSpeaker D: So you have to do everything at the level of the features. What could you do?\nSpeaker D: And one thing I suggested earlier time was maybe some sort of scaling, some sort of root or something of the features.\nSpeaker D: But the problem with that is that isn't quite the same. It occurred to me later because what you really want to do is scale the range of the likelihoods rather than.\nSpeaker D: But what might get at something similar just occurred to me is kind of an intermediate thing is because we do this strange thing that we do with the tandem system, at least in that system, what you could do is take the values that come out of the net, which are something like log probabilities and scale those.\nSpeaker D: And then, then at least those things would have the right values or the right range. And then that goes into the rest of it and it's used as observation.\nSpeaker D: So it's not the way to do it.\nSpeaker F: But these values are not directly used as probabilities.\nSpeaker D: So what we're doing is pretty strange and complicated. We don't really know what the effect is. So my thought was maybe, I mean, they're not used as probabilities. But the log probabilities were taking advantage of the fact that something like log probabilities has more of a Gaussian shape than probabilities.\nSpeaker D: And so we can model them better. So in a way, we're taking advantage of the fact that there are probabilities because they're this quantity that looks kind of Gaussian when you take its log self. Maybe, maybe you would have a reasonable effect to do that. I don't know.\nSpeaker D: But I mean, I guess we still haven't had a ruling back on this. And we may end up being a situation where we just really can't change the written session penalty. But the other thing we could do is also be could, I mean, this may not help us in the evaluation, but it might help us in our understanding, at least.\nSpeaker D: And it might just run it with different insertion penalties and show that, well, okay, not changing it, playing the roles the way we wanted. We did this, but in fact, if we did that, it made a big difference.\nSpeaker B: I wonder if it might be possible to simulate the backend with some other system. So we get our front end features. And then as part of the process of figuring out the scaling of these features, and I hope we're going to take it to a reader to power or something.\nSpeaker B: We have some backend that we attach on our features that sort of simulates what would be happening.\nSpeaker D: And just adjusted until the best number.\nSpeaker B: And just adjusted until that our version of the backend and decides that.\nSpeaker D: Well, we can probably use the real thing, can't we? And then just just using out a reduced test set or something.\nSpeaker D: And then we just use that to determine some scaling factor. Yeah, so I mean, I think that that's a reasonable thing to do. And the only question is what's the actual knob that we use and the knob that we use should.\nSpeaker D: Unfortunately, I don't know the analytic solution to this because what we really want to do is change the scale of the likelihoods, not the scale of the observations, but.\nSpeaker C: Yeah. Yeah. I'm curious to hear what we kind of recognize or.\nSpeaker C: What do you mean we say what kind of.\nSpeaker C: Is it like I got to. Yeah.\nSpeaker B: It's the same system that they use when they participate in the hot five evals. It's sort of came out of.\nSpeaker B: Looking a lot like hdk. I mean, they started off with when they were building their system, they were always comparing to hdk and make sure they were getting similar results.\nSpeaker B: So it's a Gaussian mixture system.\nSpeaker D: They have the same sort of mixed down sort of procedure where they start off with a small number of.\nSpeaker B: Yeah, and then. Yeah.\nSpeaker B: I don't know.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Do you know what kind of tying they use or they sort of some sort of under Gaussian's they share across everything or.\nSpeaker B: I have a whole system description that describes exactly what their.\nSpeaker B: It's some kind of a mixture of the out.\nSpeaker D: So the other or I think maybe I don't know if any of this is going to come in in time to be relevant, but we had talked about.\nSpeaker D: Playing around over in Germany and and possibly coming up with something that would fit in later. I saw that other mail where he said that he wasn't going to work for him to do CBS.\nSpeaker F: Yeah. So no, he has a version of the software.\nSpeaker D: So he just has it all sitting there. Yeah.\nSpeaker D: So if he'll he might work on improving the noise estimate or on some histogram thing. Yeah.\nSpeaker D: Yeah, I just saw the USB. We didn't talk about it at our meeting, but I just saw the just read the paper.\nSpeaker D: Someone I forget the name and they about histogram equalization.\nSpeaker D: Did you see that one?\nSpeaker F: It was a poster.\nSpeaker D: Yeah, I mean, I just read the paper. I didn't see the poster.\nSpeaker F: It was something similar to online organization.\nSpeaker D: Yeah, but it's a little more it's a little finer, right?\nSpeaker D: So they had like 10 quantiles.\nSpeaker D: And they just the distribution. So you have the distributions from the training set.\nSpeaker D: And then this is just a histogram of the amplitudes, I guess, right?\nSpeaker D: And then people do this image processing some you have this kind of histogram of levels of brightness or whatever.\nSpeaker D: And then when you get a new thing that you want to adjust to be better in some way, you adjust it so that the histogram of the new data looks like the old data.\nSpeaker D: You just kind of piecewise linear or some kind of piecewise approximation. They did a one version of piecewise linear and another that had a power law thing between them between the points.\nSpeaker D: And they said they said they sort of see it in a way is for the speech case of being kind of a generalization of spectral subtraction in a way because you know, spectral subtraction, you're trying to get rid of this excess energy.\nSpeaker D: It's not supposed to be there. And this is sort of adjusting it for a lot of different levels. And then they have some kind of floor or something.\nSpeaker D: So if it gets too low, you don't do it. And they claim very nice results.\nSpeaker B: So is this a histogram across different frequency pins?\nSpeaker D: Or I think this, you know, I don't remember that. Do you remember?\nSpeaker F: I think they have different histograms.\nSpeaker F: So it's like one per frequency band.\nSpeaker F: So what I did to gram per frequency band. Yeah, I guess.\nSpeaker F: But I should read the paper. I just went to the question.\nSpeaker D: Yeah. And I don't remember whether it was filter bank things or whether it was FFT bands or.\nSpeaker B: That histogram represents the different energy levels that have been seen and have that frequency.\nSpeaker D: And how often they have seen them. Yeah.\nSpeaker D: Yeah. And they do it. They said that they could do it for the test.\nSpeaker D: So you don't have to change the training. You just do a measurement over the training.\nSpeaker D: And then for testing, you can do it for one per utterance.\nSpeaker D: Even relatively short utterances. They claim it works pretty well.\nSpeaker B: So they is the idea that you run test utterance through some histogram generation thing.\nSpeaker B: And then you compare the histograms and that tells you what to do to the utterance to make it more like.\nSpeaker D: In principle, I didn't read carefully how they actually implemented.\nSpeaker D: Yeah.\nSpeaker D: Or whether it was a second pass or what, but they know sort of the idea.\nSpeaker D: So that seemed different. We were sort of curious about what are some things that are conceptually quite different from what we've done.\nSpeaker D: Because one thing that Stephanie and Sennielson defined was they could actually make a unified piece of software that handled a range of different things that people were talking about.\nSpeaker D: And it was really just sort of setting a different constants.\nSpeaker D: And it would turn one thing into another, turn winner filtering into spectral subtraction or whatever.\nSpeaker D: But there's other things that we're not doing. So we're not making any use of pitch.\nSpeaker D: Which again might be important.\nSpeaker D: Because the stuff between the harmonics is probably schmutz and the transcribers will have fun with that.\nSpeaker D: And the stuff at the harmonics isn't so much.\nSpeaker D: And there's this overall idea of really sort of matching the distribution somehow.\nSpeaker D: Not just subtracting off your estimate of the noise.\nSpeaker D: So I guess Gunther is going to play around with some of these things now over this next.\nSpeaker F: I don't know. I don't have feedback from him.\nSpeaker D: I guess he's got it anyway. So potentially if he came up with something that was useful, like a better noise estimation module, something he could ship it to you guys up there.\nSpeaker D: And we could put it in.\nSpeaker D: Yeah. So that's good.\nSpeaker D: So we just, I think starting, starting a couple weeks now, especially if you're not going to be around for a while, we'll be shifting more over to some other territory.\nSpeaker D: But not so much in this meeting about Aurora.\nSpeaker D: But maybe just quickly today about maybe to say a little bit about what you've been talking about with Michael.\nSpeaker D: And then very can say something about what we're talking about.\nSpeaker A: So Michael Klanchman, who's a PhD student from Germany, showed up this week.\nSpeaker A: He'll be here for about six months. And he's done some work using an auditory model of human hearing and using that to generate speech recognition features.\nSpeaker A: And he did work back in Germany with a toy recognition system using isolated digit recognition as the task. It was actually just a single layer neural network that classified words, classified digits. In fact, and he tried that.\nSpeaker A: I think on some Aurora data and got results that he thought seemed respectable.\nSpeaker A: And he's coming here to use it on a real speech recognition system.\nSpeaker A: So I'll be working with him on that.\nSpeaker A: Maybe I should say a little more about these features.\nSpeaker A: Although I don't understand in that. Well, I think it's a two-stage idea.\nSpeaker A: And the first stage of these features correspond to what's called the peripheral auditory system.\nSpeaker A: And I guess that is like a filter bank with a compressive nonlinearity.\nSpeaker A: And I'm not sure what we have in there that isn't already modeled in something like PLP.\nSpeaker A: I should learn more about that. And then the second stage is the most different thing, I think, from what we usually do. It's, it computes features which are based on sort of like based on different different wavelet basis functions used to analyze the input.\nSpeaker A: But so the, he uses analysis functions called GABAR functions, which have a certain extent in time and in frequency.\nSpeaker A: And the idea is these are used to sample the signal represented as a time frequency representation.\nSpeaker A: So you're sampling some piece of this time frequency plane.\nSpeaker A: And that is, is interesting because for, for one thing, you could use it in a multi-scale way.\nSpeaker A: You could have these instead of having everything like we use a 25 millisecond or so analysis window typically.\nSpeaker A: And that's our time scale for features. But you could using this basis function idea.\nSpeaker A: You could have some basis functions which have a lot longer time scale and some which have a lot shorter.\nSpeaker A: And so it would be like a set of multi-scale features. So he's interested in that.\nSpeaker A: This is because it's, there are these different parameters for the shape of these basis functions.\nSpeaker A: There are a lot of different possible basis functions. And so he actually does an optimization procedure to choose an optimal set of basis functions out of all the possible ones.\nSpeaker B: So he do to choose this.\nSpeaker A: The method he uses is kind of funny is he starts with, he has a set of M of them.\nSpeaker A: And then he uses that to classify. And he tries using just M minus one of them.\nSpeaker A: So there are M possible subsets of this length M vector.\nSpeaker A: He tries classifying using each of the M possible sub-vectors.\nSpeaker A: Whichever sub-vector works the best I guess he says the feature that didn't use was the most useless feature.\nSpeaker A: So we'll throw it out and we're going to randomly select another feature from the set of possible basis functions.\nSpeaker B: So it's actually a little bit like a genetic algorithm.\nSpeaker D: It's much simpler. But it's, there's a number of things I like about, I'm going to say.\nSpeaker D: So first of all, you're absolutely right. I mean, in truth, both pieces of this have their analogies and stuff we already do.\nSpeaker D: But it's a different take at how to approach it and potentially one that's maybe a bit more systematic than what we've done.\nSpeaker D: And a bit more inspiration from auditory things. So it's, so I think it's a neat thing to try.\nSpeaker D: The primary features are in fact, yeah, essentially it's, you know, POP or Melkepster or something like that.\nSpeaker D: You've got some compression. We always have some compression. We always have some, you know, the, the kind of filter bank with kind of quasi log scaling.\nSpeaker D: And if you put in, if you also include the Rasta in it, Rasta, the filtering being done in the log domain has an A.G.C. like characteristic, which, you know, people typically put in these kind of auditory front-end.\nSpeaker D: So it's very, very similar, but it's not exactly the same.\nSpeaker D: I would agree that the second one is somewhat more different, but it's mainly different in that the things that we've been done like that have been had a different kind of motivation and have ended up with different kinds of constraints. So for instance, if you look at the LDA, Rasta stuff, you know, basically what they do is they, they look at the different Eigenvectors out of the LDA and they form filters out of it, right?\nSpeaker D: And those filters have different kinds of temporal extents and temporal characteristics. And so in fact, they're multi-scale.\nSpeaker D: But they're not sort of systematically multi-scale. Like let's start here and go to there and go to there and go to there and so forth. It's more like you run it on this, you do discriminate analysis and you find out what's helpful.\nSpeaker D: It's multi-scale because you use several of these in parallel, so I'm going to use several of them, yeah. I mean, I don't have to, but, but, but, but, but, but, but, but, but, but, but, but, but, but, he can't.\nSpeaker D: But it's also, even he and X had people do this kind of LDN analysis, they've done it on frequency direction and they've done it on the time direction.\nSpeaker D: I think he may have had some people sometimes doing it on both simultaneously, some 2D, and that would be the closest to these good board function kind of things.\nSpeaker D: But I don't think they've done that much of that.\nSpeaker D: And the other thing that's interesting, the feature selection thing, it's a simple method, but I kind of like it.\nSpeaker D: There's an old old method for feature selection.\nSpeaker D: I mean, I remember people referring to his old one I was flying with it 20 years ago, so I know it's pretty old, called stepwise linear discriminant analysis, in which I think it's used in social sciences a lot.\nSpeaker D: So you pick the best feature, and then you find the next feature that's the best in combination with it, and then so on and so on.\nSpeaker D: And what Michael's describing seems to me much, much better, because the problem with the stepwise discriminant analysis is that you don't know if you've picked the right set of features, just because something's a good feature doesn't mean you should be adding it.\nSpeaker D: Here, at least, you're starting off with all of them, and you're throwing out useless features.\nSpeaker D: I think that seems like a lot better idea.\nSpeaker D: You're always looking at things in combination with other features.\nSpeaker D: So the only thing is, of course, there's this artificial question of exactly how you assess it, and if your order had been different in throwing them out, I mean, it still isn't necessarily really optimal, but it seems like a pretty good heuristic.\nSpeaker D: So I think it's kind of neat stuff, and the thing that I wanted to add to it also was to have us use this in a multi-stream way. So that when you come up with these different things, these different functions, you don't necessarily just put them all into one huge factor, but perhaps you have some of them in one stream, some of another stream, and so forth.\nSpeaker D: And we've also talked a little bit about shia pshamas stuff, in which the way you look at it is that there's these different mappings and some of them emphasize upward moving energy and frequency and some are emphasizing downward and fast things and slow things and so forth. So this is a bunch of stuff to look at, but we're sort of going to start off with what he came here with and branch out from there.\nSpeaker D: And his advisor is here too, at the same time, so it'll be another interesting source of wisdom.\nSpeaker C: As we're talking about this, I was thinking whether there's a relationship between my close approach to some sort of optimal brain damage or optimal brain surgeon on the neural nets.\nSpeaker C: We have our roster features and presumably the neural nets are learning some sort of nonlinear mapping from the features to this probably posterior space. And each of the hidden units is learning some sort of some sort of pattern, it could be like these auditory patterns that Michael is looking at, and then when you're looking at the best features, you can take out the brain surgery by taking out\nSpeaker D: hidden units that don't really help at all. Or the four features, right? I mean, actually, you make me think of a very important point here is that if we again try to look at how is this different from what we're already doing, there's a nasty argument that we made that it's not different at all because if you ignore the selection part because we are going into a very powerful nonlinearity that in fact is combining over time and frequency and is coming up with its own better than good bore functions, its neural net functions, whatever it finds to be best.\nSpeaker D: So you could argue that in fact it, but I don't actually believe that argument because I know that you can computing features is useful even though in principle you haven't added anything, in fact, you subtracted something from the original waveform. If you've processed it in some way, you've typically lost something, some information. And so you've lost information and yet it does better with features than it does with the waveform. So I know that sometimes it's useful to constrain things. So that's why it really seems like the constraint in all this stuff, it's the constraints that actually matters because it wasn't the constraints that mattered then we would have completely solved this problem long ago because long ago we already knew how to put waveforms into\nSpeaker F: powerful statistical mechanisms. Yeah, well, if we had infinite processing power and data,\nSpeaker D: then it would work, yeah, I agree. Yeah, there's the problems. Yeah, that would work, but I mean, it's with finite those things. I mean, we have done experiments where we literally have put waveforms in and kept number parameters the same and so forth and used a lot of training data and it's not infinite, but in a lot compared to the number parameters and it just doesn't do it nearly as well. So the point is that you want to suppress, it's not just having the maximum information, you want to suppress the aspects of the input signal that are not helpful for for the discrimination they're trying to make. So I'm just briefly.\nSpeaker C: Well, that's sort of segues into what I'm doing. So the big picture is come up with a set of intermediate categories, the intermediate category class drives and the recognition and groups between recognition that way. So right now in the phase we're looking at and deciding on an initial set of intermediate categories and I'm looking for data-driven methods that can help me find a set of intermediate categories that must be each that helped me to discriminate better down the line. And one of the ideas that was to take a neural net, train an ordinary neural net to learn the posterior probabilities of phones. And so at the end of the day you have this neural net and it has hidden hidden units and each of these hidden units is is learning some sort of pattern. And so what are these patterns? I don't know.\nSpeaker C: And I'm going to try to look at those patterns to see from those patterns presumably those are important patterns for discriminating between phone glasses. Maybe some intermediate categories can come from just looking at the pattern of that neural net. Before we go on next part,\nSpeaker D: let me just point out that there's a pretty nice relationship between what you're talking about doing and what you're talking about doing there, right? So it seems to me that if you take away the the difference of this primary features and say you just as we talked about me be doing use Rasta PLP or something for the primary features then this feature discovery thing is just what he's talking about doing too except that he's talking about doing them in order to discover intermediate categories that correspond to these what these sub features are showing you. And the other difference is that he's doing this in a multi band setting which means that he's constraining himself to look across time in some relatively limited spectral extent, right? And whereas in this case you're saying let's do it on constraint. So they're really pretty related maybe they'll be at some point where we'll see the connections a little better and connect them.\nSpeaker C: Yes, so that's the first part one of the ideas to get at some pattern of intermediate categories. The other one was to come up with a model, a graphical model that treats the intermediate categories as hidden variables, like variables that we don't know anything about but that through statistical training and the EM algorithm at the end of the day we have we have learned something about these latent variables which happen of course farmed to intermediate categories. So those are the two directions. I'm looking at you now.\nSpeaker D: Okay, we do our digits and get our treats. Yeah, it's kind of like the little rats with the little thing dropping down to and we do the digits and then we get our treats.\nSpeaker D: Okay, transcript L-371. 205453359. 445104230. 3317556710. 8, 831172630. 5606049727.\nSpeaker C: 8183298057. 9857208893. 6534585475. Transcript L-370. 9868763723. 492168110. 6171438.\nSpeaker C: 4883. 4184. 731. 7191. 3165. 4378. 9669. 231. 7. 0. 564. 460. 411. 611. 741. 890. 750. 449.\nSpeaker B: Transcript L-369. 026. 0. 9985. 4494. 395. 756. 403. 496. 266. 7000. 942. 120. 767. 177. 468. 289. 2519. 37. 285. 2.\nSpeaker F: 1. 73577. 2. 864. 788. 313. 576. Transcript L-368. 596. 9. 3. 890. 615. 609. 303. 0. 703. 880. 2. 0. 0. 0. 6.\nSpeaker F: 413. 683. 743. 737. 966. 933. 666. 7. 845. 308. 576. 332. 021. 828. 960. 475. 2. 3.\nSpeaker A: Transcript L-367. 987. 231. 9211. 620. 722. 238. 9. 475. 738. 970. 393. 587. 540. 5. 765. 8. 542. 0.\nSpeaker A: 1. 846. 515. 9. 680. 9. 3. 1. 2. 8. 5. 313. 639. 7. 480. 1. 6. 0. 4. 9. 460. 0.\n"}, {"meeting_id": "ES2002d", "summary": "This was the final meeting of detailed design. To start with, Project Manager introduced the planned scheme of the meeting, followed by User Interface and Industrial Designer indicating the possible changes into an LCD screen and a jog-dial. After that, the group continued to talk about the slogan, the button size, and the button color. However, considering the budget, they consistently abandoned some of the designs. These included kinetic power, rubber material, LCD panel, advanced chip, and special colored buttons. Marketing moved the discussion to evaluate the current design through collectively rating from one to seven. The final part was the project evaluation, including the system, leadership, teamwork, and tools given.", "dialogue": "Speaker B: Okay, we're all set.\nSpeaker B: Right.\nSpeaker B: Well, this is the final detailed design meeting.\nSpeaker B: We're going to discuss the look and feel design, the easier interface design, and we're going to evaluate the product.\nSpeaker B: And the end result of this meeting has to be a decision on the details of this remote control, like absolute final decision.\nSpeaker B: And then I'm going to have to specify the final design in the final report.\nSpeaker B: So just from last time to recap, we said we were going to have a snowman shaped remote control with no LCD display, no need for talk back.\nSpeaker B: It was hopefully going to be kinetic power and battery with rubber buttons, maybe back lighting the buttons with some internal LEDs to shine through the casing, hopefully a job dial, incorporating the slogan somewhere as well.\nSpeaker B: Anything I missed?\nSpeaker B: Okay.\nSpeaker B: Sue, if you want to present your prototype, go ahead.\nSpeaker G: That's it.\nSpeaker G: Ninja Homa made in Japan.\nSpeaker C: There are a few tunes we've made.\nSpeaker C: We're looking at the expansion.\nSpeaker C: And it's going to be quite a lot expensive to have open up now what's the buttons and stuff inside.\nSpeaker C: So instead of this is going to be an LCD screen.\nSpeaker C: Just a very, very basic one, very small, with access to the menu through the screw wheel and confirm.\nSpeaker C: And button.\nSpeaker C: Okay.\nSpeaker C: Part of my, there's pretty much the same thing we discussed last time.\nSpeaker G: And there isn't a, it doesn't open up to the advanced functions.\nSpeaker G: So the advanced functions are still hidden from you, but they're hidden in this inset.\nSpeaker G: They're not in use.\nSpeaker G: Where are they?\nSpeaker G: They're in the LCD panel and the job dial.\nSpeaker G: All right.\nSpeaker B: So what kind of thing is going to be?\nSpeaker G: The LCD panel just displays functionally what you're doing.\nSpeaker G: If you're using an advanced function, right, like brightness, contrast, whatever, it would just say, you know, it's like, it only has four columns.\nSpeaker G: It's a very simple LCD.\nSpeaker G: Like, or as many, the minimum amount we need that they usually automatically know, like, this is brightness, or this is contrast.\nSpeaker G: It might even be one, a bit more complex LCD panel, which pictures like maybe the sun, or, you know, the symbols of the various.\nSpeaker G: And what is this here?\nSpeaker G: That's a number pad.\nSpeaker G: Okay.\nSpeaker B: So the number pad is where we're going to have the slogan.\nSpeaker G: Um, just like along the side.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker G: Cool.\nSpeaker G: You have this space here and then you have this thing on the side as well, or at the bottom.\nSpeaker G: So, the slogans are usually quite small, right?\nSpeaker G: They're not like huge.\nSpeaker G: So, there's, say, buttons about this size, right?\nSpeaker G: So you would still have plenty of space for a slogan.\nSpeaker B: So if this isn't to scale, what kind of dimensions are you thinking about here?\nSpeaker C: Well, I want the buttons to be bigger enough to push easily with my fingers.\nSpeaker C: So we're going to, maybe that'll be about the same size as in Pam and, yeah.\nSpeaker G: So that would be about centimeter for a button.\nSpeaker G: So one, two, three, four centimeters plus maybe half, five, six, seven, eight.\nSpeaker B: Yeah.\nSpeaker B: That sounds good.\nSpeaker B: That would be good.\nSpeaker B: So 10 centimeters in height.\nSpeaker B: Okay.\nSpeaker B: Um, that would be good.\nSpeaker B: In fact, a pen is about 10 centimeters usually.\nSpeaker B: So that would be, that sounds like a really good sign.\nSpeaker B: Yeah.\nSpeaker D: You see it there.\nSpeaker D: That's great.\nSpeaker B: And it's very bright as well.\nSpeaker B: So, um, okay.\nSpeaker D: Is it possible?\nSpeaker D: Uh, just going to bring up the idea of colors.\nSpeaker D: Is these, are these the colors that have production?\nSpeaker D: Um, is this just what we had available?\nSpeaker C: We're going to have again the sort of the foggy, um, yellow, masking that the app when you beat the button, right?\nSpeaker B: Okay.\nSpeaker B: I can do this list of all the things that it does.\nSpeaker C: So I can write them in the report.\nSpeaker C: Um, this button.\nSpeaker C: Um, because it's red, it's sort of very small.\nSpeaker C: I'm going to use it as, uh, it's going to be the power button if you hold it for maybe two seconds.\nSpeaker C: I'll send it in the mic if you want to make a change, or be used to be more popular and can use this as a jog though.\nSpeaker B: Okay. So that's like an okay button?\nSpeaker B: Yeah.\nSpeaker G: Mm-hmm.\nSpeaker G: Oh, we discussed how high it is, but how wide is it?\nSpeaker G: Okay.\nSpeaker G: How high is it?\nSpeaker G: No, as in the height.\nSpeaker G: Yeah.\nSpeaker G: But what about the width?\nSpeaker B: Oh, like depth of the...\nSpeaker B: We need five.\nSpeaker G: Yeah, we need about three and a half.\nSpeaker G: Okay.\nSpeaker B: Oh, is this to get an idea of scale from your...\nSpeaker B: Sure.\nSpeaker B: Sure.\nSpeaker B: Sure.\nSpeaker B: Okay.\nSpeaker C: No, what else can you do?\nSpeaker C: You can skip straight to the channel using these buttons.\nSpeaker B: Okay.\nSpeaker C: We're going to have the volume control here.\nSpeaker C: But because we've got the LCD and the jog dial, we just start with the new start as the volume.\nSpeaker B: Okay, jog dial for volume.\nSpeaker B: What else do you do as the jog dial?\nSpeaker C: You can use it for more advanced functions like contrast, color, and...\nSpeaker B: Contrast.\nSpeaker B: Contrast.\nSpeaker B: Brightness.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: And then if not...\nSpeaker C: Um, it's whatever else we wanted to include as the advanced functions.\nSpeaker C: Um, but then I still got to inspect by the...\nSpeaker B: Well, as the designers, what are they?\nSpeaker C: Uh, what can a TV do?\nSpeaker G: Okay, things like, um, brightness contrast.\nSpeaker G: Uh-huh.\nSpeaker G: Um, maybe tuning the channels.\nSpeaker G: Okay, channels.\nSpeaker B: Um, what else?\nSpeaker B: Um, the various inputs.\nSpeaker G: So, you're having VCR, you're having, you know, which input you have.\nSpeaker B: Okay.\nSpeaker B: Exactly.\nSpeaker G: Probably color sharpness.\nSpeaker G: Yeah.\nSpeaker G: Color sharpness.\nSpeaker G: Um, a lot of these things will have to be, um, free and open for users to define them.\nSpeaker B: Okay, what about, uh, sign settings?\nSpeaker B: Or can you change any of those at all?\nSpeaker G: Yeah.\nSpeaker G: Audio.\nSpeaker G: We have like your basic...\nSpeaker G: Your base, your mid-range, your high-range.\nSpeaker G: Um, they have the bounce.\nSpeaker G: Yep.\nSpeaker G: Left-right-balance.\nSpeaker G: Um, maybe even pre-programmed.\nSpeaker G: Okay.\nSpeaker G: Um, the user could determine, like, a series of sound modes, and then what could happen would be, um, when you click on that, then we'll go to that setting.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Is there anything else that it can do?\nSpeaker B: Cause that's fine.\nSpeaker B: You just need to know so I can write it down.\nSpeaker B: Okay.\nSpeaker B: Um, right?\nSpeaker B: I guess that's it.\nSpeaker B: So we can now, um, we can now have a little look at, uh, the Excel sheet and price this thing and see if we need to, um, if we need to rethink anything at all.\nSpeaker B: So, um, for this first part here, par-wise, have we got...\nSpeaker B: Battery.\nSpeaker B: Battery.\nSpeaker B: Do we have kinetic as well?\nSpeaker B: No.\nSpeaker B: No?\nSpeaker B: Okay.\nSpeaker B: Um, just battery.\nSpeaker G: We need...\nSpeaker B: That's because of cost restraints, is it?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Um, what about the electronics here?\nSpeaker G: We need a...\nSpeaker G: The advanced chip.\nSpeaker G: Let me just confirm that.\nSpeaker G: Yes, I think so.\nNone: Okay.\nSpeaker B: Um, the case, what does it mean by single and double, do you know?\nSpeaker C: I think single just be sort of one, or a double, where it doubles this sort of thing.\nSpeaker B: Double car.\nSpeaker B: Yep.\nNone: Okay.\nSpeaker B: Um...\nSpeaker B: Plastic.\nSpeaker B: Is there any rubber at all in the button?\nSpeaker B: I think we're gonna have to skip a rubber.\nSpeaker B: Okay.\nSpeaker B: Um, and we want it to special colors, don't we?\nSpeaker B: So, I'll have to pop that.\nSpeaker B: Oh, no, wait.\nSpeaker B: How many colors have we got there?\nSpeaker G: For the case itself, one color.\nSpeaker G: It's one special color.\nSpeaker B: Just one color, okay.\nSpeaker G: Because the case unit itself, the rest of it.\nSpeaker B: Okay.\nSpeaker B: So, interface-wise, is it this third option we have?\nSpeaker G: It's one of the other ones.\nSpeaker G: Okay.\nSpeaker B: Um, then buttons.\nSpeaker B: I mean, we have one.\nSpeaker B: Two colors.\nSpeaker B: We've got push buttons as well.\nSpeaker G: Like, oh wait, so push button and integrated screw.\nSpeaker B: Okay.\nSpeaker C: We've got one button for this thing, because it's just one big sheet of rubber.\nSpeaker C: Uh-huh.\nSpeaker C: Um, not sure if that counts.\nSpeaker G: Okay.\nSpeaker G: So, let's just be safe and put, like, say, four buttons for that one.\nSpeaker B: Okay.\nSpeaker G: Okay.\nSpeaker G: Um, and maybe a special color for the months.\nSpeaker G: So, maybe four again.\nSpeaker G: Four.\nSpeaker B: So, why are we arriving at the number four?\nSpeaker B: Where does the number four come from?\nSpeaker G: Because that's one button, but it's the complexity of 12 buttons.\nSpeaker B: Okay.\nSpeaker B: All right.\nSpeaker G: So, we're just estimating that, yeah, it'll be less.\nSpeaker B: Okay.\nSpeaker B: Are we wanting them?\nSpeaker B: No.\nSpeaker B: Is everything going to be faster?\nSpeaker B: Okay.\nSpeaker B: So, we're all quite far over.\nSpeaker B: Something's going to have to go.\nSpeaker B: Um, we're at 16.8.\nSpeaker B: And...\nSpeaker D: How are we going to achieve this high end product?\nSpeaker B: Well, something has to go to the tune of 2.3 euro.\nSpeaker B: So, let me see.\nSpeaker B: What are we?\nSpeaker B: I mean...\nSpeaker B: Oh, yeah, sorry, 4.3.\nSpeaker C: Oh, we could take it once when we can get a single curve.\nSpeaker C: Just fill in those bits.\nNone: Yeah.\nSpeaker F: And then we're...\nSpeaker F: How much did that say?\nSpeaker F: How much did that say?\nSpeaker F: That one say if you want one.\nSpeaker G: The other thing could be that, um, you could take away the LCD panel and advance chip together.\nSpeaker G: Um, because when you do something on the TV, the TV responds and reacts as well.\nSpeaker G: So, the user could be looking at the TV and pushing this thing so they may not need to...\nSpeaker G: So, when we scroll, we need just some way to get the TV to respond.\nSpeaker G: Okay, so...\nSpeaker G: Which I think is a technically doable thing, so...\nSpeaker B: So, what's our review suggestion?\nSpeaker B: Um, take away the LCD display.\nSpeaker B: Yep.\nSpeaker G: And the advance chip goes away as well.\nSpeaker B: To be replaced with a...\nSpeaker B: Regular chip.\nSpeaker B: Regular chip.\nSpeaker B: Okay.\nSpeaker G: So, what that means is that...\nSpeaker G: Oh.\nSpeaker B: So, we've got 0.3 to get rid of.\nSpeaker B: Um, and we have...\nSpeaker B: Where are the four...\nSpeaker B: The four push buttons are where exactly now?\nSpeaker G: The 12 buttons that you see there.\nSpeaker B: 12 buttons?\nSpeaker C: That's one piece of rubber, but it's going to have 12 buttons.\nSpeaker C: Yeah.\nSpeaker G: Functionally, you have to intercept.\nSpeaker G: So, 4 is a good estimate for...\nSpeaker B: Do you think?\nSpeaker G: Yeah.\nSpeaker G: So, you can't actually cut...\nSpeaker G: It's like three times the number of buttons.\nSpeaker B: Okay.\nSpeaker B: 4, 8, 12.\nSpeaker B: Is that one big button or is it 12 buttons?\nSpeaker B: How can it be something in between?\nSpeaker G: It needs to be more than one big button because if you open up your phone and then there's actually one button underneath, it's just that the panel itself is a single panel.\nSpeaker B: Okay.\nSpeaker B: Well, we have 0.3 to get rid of somewhere.\nSpeaker G: Just report that it has to be over budget.\nSpeaker G: Or the colors.\nSpeaker G: No, there's no...\nSpeaker G: Take away colors for the buttons.\nSpeaker G: Yeah, we could just go with normal color buttons.\nSpeaker B: Well, do you want color differentiation here?\nSpeaker G: No, that's not the button with all of this.\nSpeaker G: Oh, sorry.\nSpeaker G: The buttons only refer to the pads.\nSpeaker B: Right.\nSpeaker B: So...\nSpeaker B: Take that off.\nSpeaker B: Aww.\nSpeaker G: It's back to the original.\nSpeaker G: So then these just become normal color buttons.\nSpeaker G: So that might be some way of cutting the cost.\nSpeaker B: Okay.\nSpeaker B: Off that's the sheet.\nSpeaker B: And...\nSpeaker B: Right.\nSpeaker B: So take away that completely.\nSpeaker B: Aww.\nSpeaker B: And now we're under budget.\nSpeaker B: So we do have 0.5 euro to play with if we wanted.\nSpeaker D: How about with embossing the logo?\nSpeaker D: Isn't that going to cost us some money?\nSpeaker B: It doesn't say so.\nSpeaker C: I can't afford your current as a special form for the buttons.\nSpeaker B: Yeah.\nSpeaker B: Yeah, that's a good idea.\nSpeaker B: There's one.\nSpeaker B: Does that mean that one button has a special form or...?\nSpeaker C: I think one button.\nSpeaker C: There's just one button.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Well, there we go.\nSpeaker B: So I'm just going to have to re-draw this nice.\nSpeaker B: We're not going to have the LCD anymore.\nSpeaker B: We're just going to have an on the TV.\nSpeaker B: It'll show you what you're doing.\nSpeaker B: But I think it's fair enough.\nSpeaker B: And so this is going to be one big thing here.\nSpeaker B: Erm...\nSpeaker D: Was the goal in your prototype design that it be as low profile as possible?\nSpeaker D: What do you mean by profile?\nSpeaker D: Sort of flat as possible.\nSpeaker C: No.\nSpeaker C: Yeah.\nSpeaker C: It's quite deep.\nSpeaker C: Yeah.\nSpeaker C: It's quite deep.\nSpeaker G: Yeah.\nSpeaker G: It's quite light and black.\nSpeaker G: Sure.\nSpeaker G: We didn't have enough play to make it.\nSpeaker G: Yeah.\nSpeaker G: Alright.\nSpeaker G: Okay.\nSpeaker G: Just...\nSpeaker G: So there's one more dimension to the thing which we need to add.\nSpeaker B: And you might want to add in a report length width and height.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: So just to...\nSpeaker B: Well, to be thorough then, width wise, we're looking at about what three centimeters or something?\nSpeaker B: Yeah.\nSpeaker B: So height wise...\nSpeaker B: High toward envisaging.\nSpeaker B: I like that.\nSpeaker B: I like two centimeters.\nSpeaker B: Okay.\nSpeaker D: Two's not very high at all though.\nSpeaker D: This is about two.\nSpeaker G: Slightly more than two.\nSpeaker G: Yeah.\nSpeaker G: I'm about that thick.\nSpeaker D: Okay.\nSpeaker B: Maybe closer to three.\nSpeaker B: Okay.\nSpeaker B: We'll say two point five.\nSpeaker B: Okay.\nSpeaker B: So we have it within cost anyway.\nSpeaker B: So yeah.\nSpeaker B: Project evaluation is this point.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: So can we close that?\nSpeaker B: This is what it's the final spec that's going to be.\nSpeaker B: So I'm just going to have to...\nSpeaker B: Yeah.\nSpeaker D: That's fine.\nSpeaker D: It's probably just...\nSpeaker D: It's worth getting into.\nSpeaker D: But just in that we want this to be stylish, should we think a little bit more out of the box in terms of a button grid?\nSpeaker D: Because I've seen there's lots of devices out there that instead of taking your standard nine square grid and they have it sort of stylized or in different concept.\nSpeaker G: I think that's something that's very hard to catch.\nSpeaker G: Yeah.\nSpeaker G: So you restrict the number of people when they try something.\nSpeaker G: Sure.\nSpeaker G: Okay.\nSpeaker G: The look and the color is something which is cool.\nSpeaker G: Yeah.\nSpeaker G: So there's also that factor of if it's too unfamiliar.\nSpeaker G: Okay.\nSpeaker G: Sure.\nSpeaker D: Because when you put it on the shelf...\nSpeaker G: Or button shape.\nSpeaker G: Square buttons.\nSpeaker G: Yeah.\nSpeaker G: Bunch shape might be a good idea to change.\nSpeaker G: Okay.\nSpeaker G: Rather than positioning.\nSpeaker G: Because I think positioning is...\nSpeaker G: Sure.\nSpeaker G: We're kind of ingrained into the telephone kind of.\nSpeaker B: Okay.\nSpeaker B: Right.\nSpeaker B: So at this point we...\nSpeaker B: Is it?\nSpeaker B: Discuss how satisfied we all are with these four points with the room for creativity in the project and leadership and teamwork and the stuff we have.\nSpeaker B: Or on this I guess.\nSpeaker D: Do you want me to...\nSpeaker D: Do you want me to do my design evaluation last?\nSpeaker D: I'm not really sure what that was.\nSpeaker D: Yeah.\nSpeaker B: Evaluation.\nSpeaker B: Yeah.\nSpeaker B: Go for that first.\nSpeaker B: I wasn't entirely sure what he was supposed to be doing that but you go for it.\nSpeaker D: All right.\nSpeaker D: So the way this works.\nSpeaker D: I'm going to need to plug in the PowerPoint.\nSpeaker D: Okay.\nSpeaker D: I'll try and do it as quick as possible.\nSpeaker D: Okay.\nSpeaker D: So I'll just go over your head if that's okay.\nSpeaker D: Yeah.\nSpeaker G: I don't think you need the power.\nSpeaker G: So what's that?\nSpeaker G: It's okay.\nSpeaker G: It's okay.\nSpeaker G: I don't need the power point.\nSpeaker G: The power cord itself.\nSpeaker D: Oh, of course.\nSpeaker G: Yeah.\nSpeaker G: Freedom.\nSpeaker G: A bit more.\nSpeaker D: So what this is is a sub-frust to use kind of like a...\nSpeaker D: I'm not sure what the idea is that I've set up, I've reviewed all of the points of discussion from the beginning and use that as a criteria of evaluation for the...\nSpeaker D: For the current design or the plan and so we can review that.\nSpeaker D: I think it's going to end up being sort of elementary because we're not going to properly use it to change anything but...\nSpeaker D: Okay, great.\nSpeaker D: And I'm going to write up our results on the board so this will be a way for us to go through and decide if we're...\nSpeaker D: To review where we're standing.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker D: So sort of bringing together two things for design goals and also the market research that we had.\nSpeaker D: When we write this, one is high in succeeding or fitting to our original aim and seven is low.\nSpeaker D: Okay.\nSpeaker D: So these are... and then we've been asked to... to collectively write this so we can use try and just keep work on a consensus system.\nSpeaker D: So we just come to an agreement.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: So the first one, the stylish look and feel.\nSpeaker G: I rate that pretty high.\nSpeaker B: Well, yeah.\nSpeaker B: I mean compared to most remote controls you see, that's pretty good.\nSpeaker B: I didn't know like a six or something.\nSpeaker B: Yeah.\nSpeaker D: What does anybody else say?\nSpeaker D: My only reservation was that we basically went with yellow because it's a company's color and I don't know if yellow is going to really be a hit.\nSpeaker B: But...\nSpeaker B: Okay.\nSpeaker B: I should apply that.\nSpeaker D: What do you guys think?\nSpeaker B: I would say five or six.\nSpeaker B: Okay.\nSpeaker G: Yep.\nSpeaker G: I'm fine with that.\nSpeaker G: Okay.\nSpeaker B: Okay.\nSpeaker D: Okay.\nSpeaker D: Oh, it's actually one to seven.\nSpeaker B: Oh, yes.\nSpeaker B: Sorry.\nSpeaker G: Then I would say two or three.\nSpeaker G: Okay.\nSpeaker G: Wait.\nSpeaker G: What's the scale?\nSpeaker G: One to seven.\nSpeaker G: One is high.\nSpeaker D: Okay.\nSpeaker C: Yeah.\nSpeaker C: Two is free.\nSpeaker C: Okay.\nSpeaker D: Okay.\nSpeaker D: So that's a bit more than five then.\nSpeaker D: Okay.\nSpeaker D: Control high tech innovation.\nSpeaker D: Well, how do we jog down?\nSpeaker D: So we had to remove a few of our features we wanted, but jog down.\nSpeaker D: I'd say it's more three or four.\nSpeaker G: Maybe, yeah, but going towards a little bit higher than medium.\nSpeaker G: Okay.\nSpeaker D: Okay.\nSpeaker D: Style reflects a fruit inspired colored line.\nSpeaker D: Let me say color.\nSpeaker G: The blue colors and don't actually represent the color except for the red button.\nSpeaker G: Okay.\nSpeaker G: For one of the yellow.\nSpeaker B: I mean, it could be a yellow color.\nSpeaker G: Yeah.\nSpeaker G: The yellow is more representative of the color, but the button itself, the blue can be anything else.\nSpeaker D: Okay.\nSpeaker D: So two.\nSpeaker D: Okay.\nSpeaker D: And design is simple to use, simple and features.\nSpeaker D: Yeah.\nSpeaker B: And it's really basic.\nSpeaker B: I mean, I'd give that nearly a one.\nSpeaker D: You guys think?\nSpeaker D: Yeah.\nSpeaker D: One.\nSpeaker D: Yeah.\nSpeaker D: One.\nNone: That's fine.\nSpeaker D: Soft and spongy.\nSpeaker D: We achieved that.\nSpeaker D: We've used mostly plastic in the end.\nSpeaker D: It's going to be quite a bit of a compromise for price.\nSpeaker D: Bye.\nSpeaker D: Bye.\nSpeaker B: That's really low.\nSpeaker B: Well, I have to use a plastic.\nSpeaker B: Yeah.\nSpeaker B: It's closed.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: Could we use an entirely rubber frame?\nSpeaker D: Is that an option?\nSpeaker D: I think it would be close.\nSpeaker C: It would be more than plastic.\nSpeaker C: And then it seems like logo.\nSpeaker D: We've got it in there.\nSpeaker D: Yeah.\nSpeaker B: Gonna have that on the side.\nSpeaker D: And it's within budget.\nSpeaker D: Yep.\nSpeaker D: Okay.\nSpeaker D: So we can say then.\nSpeaker D: Out of a possible.\nSpeaker B: Or.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: 49 with zero being highest.\nSpeaker D: And.\nSpeaker D: 15.5.\nSpeaker D: Pretty good.\nSpeaker D: Translate to something like.\nSpeaker D: About approximately 72% efficacy.\nSpeaker D: Of our original goal.\nSpeaker D: Right.\nSpeaker D: I think because if you turn that into.\nSpeaker D: A hundred.\nSpeaker D: It would be about.\nSpeaker B: Twice that.\nSpeaker B: About 31.\nSpeaker D: And then invert that.\nSpeaker B: So yeah, well, you have a 69.\nSpeaker B: Oh, right.\nSpeaker D: That's pretty good.\nSpeaker D: Okay.\nSpeaker D: That was just a little formality for us to go through.\nSpeaker D: Okay.\nSpeaker D: Oh, a hundred pound pen.\nSpeaker D: Sorry.\nSpeaker B: Nobody saw that on a stick.\nSpeaker G: Cameras did.\nSpeaker B: Is that your finish?\nSpeaker D: Yeah, that's that's me.\nSpeaker D: I did have one other one other frame.\nSpeaker D: I mean, I'm not knowing how we would deal with this information.\nSpeaker D: I thought, okay.\nSpeaker D: In theory, this kind of process would be, but refining our design.\nSpeaker D: Re-visiting our original goals.\nSpeaker D: It's not something I need to push through, but I thought.\nSpeaker D: We're thinking more about the dimensions.\nSpeaker D: Sort of like more of a three dimensional shapes as well as opposed to just that.\nSpeaker D: Flat.\nSpeaker D: Could our design of all the series of colors that it's more of like a line where we have like sort of the.\nSpeaker D: I don't know, like the harvest line or the vibrant.\nSpeaker D: Whatever, just some theme and we have different tones, line, green.\nSpeaker D: Okay.\nSpeaker D: It's just discussion.\nSpeaker D: I mean, obviously we can just abandon this.\nSpeaker D: It's fine.\nSpeaker D: Just thinking about what we originally said it to do.\nSpeaker D: Yep.\nSpeaker D: Okay.\nSpeaker D: Great.\nSpeaker B: Are you submitting the.\nSpeaker B: The evaluation criteria for my, I don't know what your instructions are being.\nSpeaker D: I think they record it and I haven't been asked to submit it yet.\nSpeaker B: Okay.\nSpeaker B: Just wondering if I need to shoot it in the minutes because if you're submitting it anyway.\nSpeaker B: I will.\nSpeaker B: Okay.\nSpeaker B: Great.\nSpeaker B: Right.\nSpeaker B: Well, next up then, because we've done finance is the project evaluation.\nSpeaker G: I'm listening.\nSpeaker G: I'm just trying to incorporate the logo into the.\nSpeaker G: So I'm playing with the play.\nSpeaker G: Just in case you're wondering.\nSpeaker G: Just about right.\nSpeaker B: Okay.\nSpeaker B: Do you want to just individually say you think about about these four points and not those four points.\nSpeaker B: My four points.\nSpeaker B: I don't know.\nSpeaker B: I like this printer cables.\nSpeaker D: I just have the two little buttons like clips.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: So here we are.\nSpeaker B: As I know, we'll do this alphabetically and to the start.\nSpeaker D: So what is it you're asking me now?\nSpeaker B: I don't know.\nSpeaker B: In your opinion on those four, these four points really and how we.\nSpeaker D: I sort of are working on setting this up.\nSpeaker D: Well, okay.\nSpeaker D: I'll just go through your system.\nSpeaker D: The room.\nSpeaker D: Is.\nSpeaker D: Is.\nSpeaker D: The main thing is, I think.\nSpeaker D: Our use of the space is more just to report on things as opposed to be creative and.\nSpeaker D: Constructive and would probably help to.\nSpeaker D: Have a sort of a cumulative effect of we have ideas and we come back and then the ideas are still in discussion.\nSpeaker D: In other words, this room is sort of a center point of creativity.\nSpeaker B: But do you feel that you were able to have quite a lot of creative input into this?\nSpeaker D: Yeah, yeah.\nSpeaker D: But that's just the thing is the question in terms of the first point there, the room.\nSpeaker D: It feels that the creativity goes on.\nSpeaker D: We leave and then we come here and then we kind of put out our ideas and then.\nSpeaker B: I don't think it means the room is in this room.\nSpeaker B: It means like, you know, all right, all right.\nSpeaker D: All right.\nSpeaker D: I just looked up and said, okay, wait for a digital pen to the room.\nSpeaker B: Oh, yeah.\nSpeaker B: Of course.\nSpeaker B: Yeah.\nSpeaker B: I think it means, I think it means did you feel you were able to give creative input.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: If answering the question in those terms, I'd say that actually there's sort of a tease of creativity because we're asked to work through this, but actually the guidelines are fairly contrived in terms of.\nSpeaker D: Okay.\nSpeaker D: Fashion trends, a fruit and vegetable color scheme, but then we're told that he's company company colors.\nSpeaker D: So what do we do?\nSpeaker D: We're told, okay.\nSpeaker D: I think in terms of style and looking feel and technology, but build something for 12 and a half pounds.\nSpeaker D: Okay.\nSpeaker D: So actually the creativity was more of like a sort of a formality than.\nSpeaker G: You feel like caged within whatever you strength.\nSpeaker G: It's like a balloon in a cage. You can go so big. If not hit the site.\nSpeaker B: The constraints do come in.\nSpeaker B: Actually, let's take each point and everybody discuss that.\nSpeaker B: Yeah.\nSpeaker B: So still on the topic of room for creativity.\nSpeaker B: Next up is great.\nSpeaker C: I agree with his point.\nSpeaker C: It's quite like fun to go to and then you have to have the end and go right.\nSpeaker C: Go everything out because we don't have enough money.\nSpeaker C: Yeah.\nSpeaker B: Yeah.\nNone: Yeah.\nSpeaker G: I think another point is that the meetings are more brainstorming sessions than meetings.\nSpeaker G: So time is also a very strong factor and structure.\nSpeaker G: Yeah.\nSpeaker G: Because for brainstorming meeting, you want to structure that allows you to allows ideas to get tossed to be evaluated and to be reviewed and to get feedback and come back.\nSpeaker G: And I guess that point about the room not being very friendly to that, I think that's a very big thing. And I think the fact that we're wearing these things restricts.\nSpeaker G: I feel it because I wear my glasses right.\nSpeaker G: But that irritates me.\nSpeaker G: It does actually affect how, whether you feel comfortable to communicate.\nSpeaker G: I feel like I'm hiding behind the equipment rather than the equipment is helping me.\nSpeaker G: So you think a more relaxed atmosphere would be more kind of, not so much an atmosphere is very relaxed, but the actual environment creates boundaries to that.\nSpeaker G: And the time given also restricts.\nSpeaker B: Very good.\nSpeaker B: What about leadership?\nSpeaker B: I don't know if I did a good job or something.\nSpeaker D: I mean, my sense on that is sort of what kind of guidance and direction encouragement.\nSpeaker B: So I'm like your personal coach person and stuff like that.\nSpeaker D: Yeah, from in you as well.\nSpeaker D: I think I'm sort of acting as a team leader.\nSpeaker D: Okay.\nSpeaker D: Yeah, I think it's huge.\nSpeaker D: I think it's good.\nSpeaker D: I mean, my personal views on leadership is that effective leadership sort of gives people a certain room for freedom and delegation, but then to come back to something that they take great ownership and innovative thought with.\nSpeaker D: In reality, I think here the different elements of leadership, such as the original briefing and then the personal coach and then having you with the meeting agenda is actually quite a quite a, quite a confining framework to work with.\nSpeaker D: And so it is leadership, most of the point of sort of disempowering the team member.\nSpeaker D: Okay. But it's not bad leadership is just sort of fairly strong.\nSpeaker D: It turns it turns the individual into more of like a sort of predetermined mechanism as opposed to.\nSpeaker B: So they may be able to control him or.\nSpeaker F: Yeah.\nSpeaker F: I think controlling is not the right word.\nSpeaker F: I mean, the interaction is very structured.\nSpeaker G: I think structure is probably what you're saying that each individual is structured to one particular task and one rather than controlling.\nSpeaker G: I don't mean as a sense of control because decisions have been made in terms of a consensus, right?\nSpeaker G: We go around and we think about it, but that, you know, process actually says you have to do it in a certain way.\nSpeaker G: Okay.\nSpeaker G: It doesn't tell you, you know, sometimes you might want to be a bit more creative in terms of the process.\nSpeaker B: Yeah.\nSpeaker B: Not that.\nSpeaker B: Okay.\nSpeaker B: What about teamwork?\nSpeaker C: Did you want to correct? That was a bit hard because we could only discuss things in the meeting.\nSpeaker C: If we could just go up somebody outside meeting and have a quick talk to them, I would have been that a lot easier.\nSpeaker B: Okay.\nSpeaker G: You tried to use the common share folder to communicate, but it just comes back to us so slow in the email.\nSpeaker G: Yeah.\nSpeaker G: It doesn't have a messenger.\nSpeaker B: Did you guys get the email I sent you?\nSpeaker B: Yeah.\nSpeaker B: I was wondering if that got very good.\nSpeaker B: Okay.\nSpeaker B: So to summarize the teamwork issue, saying that if we could communicate outside the meeting, you know, just like quick questions, quick thoughts, whatever, it probably would be.\nSpeaker G: I think the tools that they were given, the tools that they were given to us are fancy, but they don't support collaboration.\nSpeaker G: I think that's the word.\nSpeaker G: They don't support the team working together.\nSpeaker D: Yeah.\nSpeaker G: They're still very individual tools.\nSpeaker D: Yeah.\nSpeaker D: I mean, sort of taking upon that idea, the way I see this is that it's the structure in which we've approached this whole task is quite contrary to the principle of teamwork because the tasks were sort of divided and then the work went on in isolation.\nSpeaker D: I don't know what you guys did while you were together.\nSpeaker D: Maybe that was different.\nSpeaker D: They don't find it.\nSpeaker D: But actually, if you imagine not entirely completely same task given to us, but us said, okay, first thing we have to do is come up with, let's say, a design concept and we sit here together and do it.\nSpeaker D: Yeah.\nSpeaker D: Well, that's what teamwork is, to say, okay, go off and don't talk to each other.\nSpeaker D: It actually sort of predisposes you to quite the contrary of teamwork.\nSpeaker D: Okay.\nSpeaker D: Not that we haven't done, I think, the best we could have done.\nSpeaker D: Okay.\nSpeaker D: I'm satisfied with it.\nSpeaker B: Right.\nSpeaker B: Anything else?\nSpeaker B: Teamwork tool?\nSpeaker B: No, okay.\nSpeaker B: Okay.\nSpeaker B: What about the, you know, how we use the whiteboard, the digital pens, the projector, stuff like that.\nSpeaker B: And did anybody think anything was really useful?\nSpeaker B: Anything was pretty un-\nSpeaker D: I think the whiteboard, for me, is the kind of thing I would use all the time, but it's not quite as useful to us as it could have been. Maybe just in the way that we use it in the sense that once we have an idea out there or while work was going on in between meetings, that could have been up on a board, you know, opposed to in like, in text.\nSpeaker D: And then we could then keep our ideas sort of building on that.\nSpeaker D: I know that people who design cars and, you know, in aviation, they quite often just have a simple, like, fiberglass prototype and it's completely, you know, abstract from the final product, but they use it as a kind of a product.\nSpeaker D: Yeah.\nSpeaker D: And so we're just, we're actually just, each of us discussing something in each of our own minds.\nSpeaker D: It wasn't until we had this year, you know, like at one point, I peaked across and looked at Craig's paper and I'm like, now I know what he's thinking because I saw his book.\nSpeaker D: But the, the whiteboard could have actually been this kind of continuing.\nSpeaker D: So do you think producing a prototype earlier in the process would have been a good idea.\nSpeaker G: I think the focus of it a lot was the PowerPoint as opposed to the, to the whiteboard.\nSpeaker G: And I think that that is also does.\nSpeaker G: Yeah.\nSpeaker G: The PowerPoint or maybe the whiteboard and the PowerPoint in the same place.\nSpeaker D: Yeah.\nSpeaker D: Because the PowerPoint was provided to us while we had time to prepare.\nSpeaker D: Whereas I can imagine if I've been encouraged to use paintbrush, for example, or whatever, I would have actually used it.\nSpeaker D: Just because that's sort of how we were set up to you go.\nSpeaker B: Okay.\nSpeaker B: Too many PowerPoint.\nSpeaker G: Yeah.\nSpeaker G: The plugin and the plugging spent, we spent a lot of time doing that and a lot of information on the PowerPoint.\nSpeaker G: I don't think, you know, we need it to actually, it could have, we could have gone through them very verbally, especially my slides.\nSpeaker G: I felt that, you know, as opposed to having to present them.\nSpeaker B: What about the digital pens?\nSpeaker B: Did you find them easy enough to use?\nSpeaker G: Yeah.\nSpeaker G: Yeah.\nSpeaker G: Conkey.\nSpeaker G: Agreed.\nSpeaker B: Okay.\nSpeaker G: Having to take it before you go off was a bit hindering as well.\nSpeaker G: Because you're halfway through a thought and then you run out of paper and then you have to jump.\nSpeaker B: I know I think at the very start of today, I like wrote a whole load of stuff.\nSpeaker B: Didn't click note on one.\nSpeaker B: Then went back and wrote one tiny thing on the another page.\nSpeaker B: But then did click note.\nSpeaker B: So I'm quite worried that I've just written over the top of it or something, but they'll have my paper anyway.\nSpeaker B: And I haven't done that.\nSpeaker G: I think the pen is very intuitive.\nSpeaker G: Everybody knows how to use it.\nSpeaker G: You don't have to worry.\nSpeaker G: Yeah.\nSpeaker G: I think the pen is good.\nSpeaker G: It's one of the best things.\nSpeaker D: On the topic of the technology, it just occurred to me that we actually didn't need to move our computers because each computer has all of the files.\nSpeaker D: Yeah.\nSpeaker A: It just occurred to me that.\nSpeaker D: Yeah.\nSpeaker D: We only needed one computer.\nSpeaker D: We only actually needed one computer.\nSpeaker D: Yeah.\nSpeaker G: There had been a fifth that could have just been sitting there ready to go.\nSpeaker G: And the computer may not be conducive to a meeting because you tend to look at your computer and want to have the urge to check something.\nSpeaker G: Yeah.\nSpeaker B: It's useful.\nSpeaker B: But do you think the computers just provide distraction in the meeting?\nSpeaker G: I think too many computers are just distracting.\nSpeaker G: Okay.\nSpeaker B: And I like to have things written down in front of me actually.\nSpeaker B: Like a lot of the stuff that was emailed to me.\nSpeaker B: I ended up being like riding down there or something.\nSpeaker B: So I could look at it really quickly and not have the distraction of all of that.\nSpeaker B: And I'll buy any videos.\nSpeaker B: What else?\nSpeaker B: I'm not really sure what they're looking for in the same new ideas.\nSpeaker B: I don't know.\nSpeaker B: Can you think of that?\nSpeaker B: Anything else that would have been helpful today at all?\nSpeaker D: The main one for me is that the process and the natural context would not have been interrupted by this necessity to discommunicate ourselves.\nSpeaker D: That's kind of a new idea for me is like just sort of that idea while you make it.\nSpeaker D: It's kind of hard to keep working forward on a team-based project when you're told you must now work away from the team.\nSpeaker B: Yeah.\nSpeaker B: I think it's quite good that we had time limits on the meetings because they really could have run on.\nSpeaker B: And my experience of meetings is that they really do.\nSpeaker B: I think it's been a lot of time to talk about the only thing is that when we had our meeting about the conceptual design, I thought maybe another 15 minutes would have been useful there.\nSpeaker B: I think maybe if we all have been working in the one room and they just said, you know, like, every hour or something, everybody makes sure you just have a short meeting and just to have something written down, just like, you know, a milestone if you like.\nSpeaker B: I'm not having meetings, but there you go.\nSpeaker B: So, in closing, I haven't got my five minutes to go.\nSpeaker B: Oh, there it is.\nSpeaker B: Five minutes to go.\nSpeaker B: Wonderful.\nSpeaker B: Okay.\nSpeaker B: Are the courses in the budget?\nSpeaker B: Yes, they are.\nSpeaker B: And is the project evaluated?\nSpeaker B: Yes, it is.\nSpeaker B: So, I celebrate.\nSpeaker G: So, we have Ninja Homo.\nSpeaker G: So, no, yeah.\nSpeaker B: Well, apparently, now I write the final report.\nSpeaker D: What are you guys doing now?\nSpeaker D: I don't know.\nSpeaker D: Oh, wow.\nSpeaker D: That is lovely.\nSpeaker B: Yeah.\nSpeaker B: I said Ninja Homo.\nSpeaker C: What did you call it?\nSpeaker G: Ninja Homo.\nSpeaker G: See, it looks like Homer Simpson.\nSpeaker G: So, is that the Stilogu or is it doing a thing?\nSpeaker B: Yeah, it's just the logo.\nSpeaker B: Just the logo and then, like, Ninja Homo.\nSpeaker B: Right, okay.\nSpeaker G: The red is supposed to represent the whatever else you want to print on the side of it.\nSpeaker G: It's crashing technology or something.\nSpeaker G: You can throw Homer when you're frustrated.\nSpeaker B: It's cool.\nSpeaker B: I'm kind of slightly gutted that we couldn't get Plastigand rubber.\nSpeaker B: I think that would have been nice.\nSpeaker B: Yeah.\nSpeaker B: Oh, well, maybe for now on real reactions, you should give us more money.\nSpeaker G: Oh, I did learn something.\nSpeaker G: Plato is useful.\nSpeaker G: No, it is.\nSpeaker G: It is useful in conceptualizing, in being more interested.\nSpeaker G: It's like you say it's something you can put your hands on and feel and touch.\nSpeaker G: Yeah.\nSpeaker G: Yeah, the sense where we were playing with the Plato and the ideas came with the Plato rather than with everything else.\nSpeaker G: You might want to write that.\nSpeaker G: No.\nSpeaker G: Okay.\nSpeaker G: I'm just filling with the Plato.\nSpeaker C: Yeah.\nSpeaker C: It's not true.\nSpeaker C: I forgot how you just did a smile.\nSpeaker C: Yeah, it's not funny, doesn't it?\nSpeaker B: The Plato's are actually edible, aren't they?\nSpeaker B: The whole Plato is in the Gulf, Montauk.\nSpeaker B: It's just sweet.\nSpeaker G: It's the stuff that your mom could make with preservedism.\nSpeaker B: And so what are your summarizing words about Plato?\nSpeaker G: It's helpful to the creative process.\nSpeaker G: It engages all your senses, not just your sight, but your sense of feel, your sense of touch, and it helps you to understand the dimension as well.\nSpeaker G: I think that's very helpful because it's such to pop up where it's on a piece of paper on a computer on a board.\nSpeaker G: Yeah.\nSpeaker G: Even with a treaty graphic thing, it still requires a lot of time.\nSpeaker G: Yeah.\nSpeaker G: It's tangible.\nSpeaker G: It becomes tangible.\nSpeaker B: It's tangible.\nSpeaker B: Okay.\nSpeaker B: Er...\nSpeaker B: I don't know if there's anything else.\nSpeaker B: Nope.\nSpeaker B: I need to discuss.\nSpeaker B: I'll pass the board up really.\nSpeaker D: Do we retreat to our...\nSpeaker G: I think we could probably do it here as long as we do it.\nSpeaker B: I'm sure that the thing will pop up any minute now.\nSpeaker G: Can we turn off the microphones?\nSpeaker B: Yeah.\nSpeaker B: The meeting's over.\nSpeaker G: And my microphone has no signal.\nNone: Which would explain why you don't get into the sound.\nNone: Yeah.\nSpeaker C: Might be because of the power storm.\nSpeaker D: I think the time has to run out first to get any more instructions.\nSpeaker F: Yeah.\nSpeaker D: Okay.\nSpeaker D: That's really cool.\nSpeaker D: I can see the...\nSpeaker D: I can see my own voice.\nSpeaker D: You really want...\nNone: I can see my own voice.\nNone: You can see my own voice.\nNone: You can see my own voice.\nNone: You can see my own voice.\nNone: You can see my own voice.\nNone: You can see my own voice.\n"}, {"meeting_id": "Bro014", "summary": "The meeting began with an update on the HTK backend. The team was testing different numbers for Gaussian mixtures. The team then discussed how performance on the Aurora tasks was measured. The final score for Aurora was a weighted average, but it seemed that there was no practical implementation against which tasks were being measured. The VAD of the team's existing system for Aurora was better. The team was also working on feature classification, mainly to do voice-unvoice detection. The meeting concluded with a few more updates on the Aurora project.", "dialogue": "Speaker C: Channel 3.\nSpeaker I: Yes.\nSpeaker I: Okay.\nNone: Channel 3.\nSpeaker I: Channel 3.\nSpeaker G: Okay.\nSpeaker G: Did you sell speech recognition last week?\nSpeaker G: Alright.\nSpeaker G: That's the image processing.\nSpeaker G: Yes, again.\nSpeaker H: We did it again, Morgan.\nSpeaker G: Alright.\nSpeaker A: Okay.\nSpeaker G: It's April 5th.\nSpeaker G: Actually, he next to be getting back in town shortly, if he hasn't already.\nSpeaker G: Is he going to come here?\nSpeaker G: Well, we'll drag him here.\nSpeaker H: So when you sit in town, you mean, Oregon?\nSpeaker G: Hey, man, you know this end of the world.\nSpeaker G: That's really what I meant.\nSpeaker G: He's been in Europe.\nSpeaker G: So.\nSpeaker H: I have something just fairly brief to report on.\nSpeaker H: I did some experiments, just a few more experiments before I had to go away for that week.\nSpeaker H: Was it last week or whatever?\nSpeaker H: So what I was starting playing with was, again, this is the HTK back end.\nSpeaker H: I was curious because the way that they train up the models, they go through about four sort of rounds of training.\nSpeaker H: And in the first round, they do, I think it's three iterations.\nSpeaker H: And for the last three rounds, they do seven iterations of re-estimation in each of those three.\nSpeaker H: And so, you know, that's part of what takes so long to train the back end for this.\nSpeaker H: And so, you know, it's the first one that's been in the first three rounds of training.\nSpeaker H: I guess you could say iterations.\nSpeaker H: The first one is three, then seven, seven, and seven.\nSpeaker H: And what these numbers refer to is the number of times that the HMM re-estimation is run.\nSpeaker G: It's this program called HE Rest.\nSpeaker H: So what happens is, at each one of these points, you increase the number of Gaussian's in the model.\nSpeaker G: Oh, right. This was the mix up.\nSpeaker H: Yes, the mix up.\nSpeaker H: And so, in the final one here, you end up with, for all of the digit words, you end up with three mixtures per state.\nSpeaker H: In the final thing. So I had done some experiments where I was, I want to play with the number of mixtures.\nSpeaker H: But I wanted to first test to see if we actually need to do this many iterations early on.\nSpeaker H: And so, I ran a couple of experiments where I reduced that to be three, two, two, five, I think.\nSpeaker H: And I got almost the exact same results.\nSpeaker H: But it runs much, much faster.\nSpeaker H: So I think it only took something like three or four hours to do the full training.\nSpeaker H: As opposed to...\nSpeaker H: As opposed to what, 16 hours or something like that?\nSpeaker H: I mean, you have to do an overnight, basically, the way it is set up now.\nSpeaker H: So, even if we don't do anything else, doing something like this could allow us to turn experiments around a lot faster.\nSpeaker G: And then when you have your final thing...\nSpeaker H: And when you have your final thing, we go back to this.\nSpeaker H: So, and it's a real simple change to make.\nSpeaker H: I mean, it's like one little text file you edit and change those numbers.\nSpeaker H: And you don't do anything else. And then you just run.\nSpeaker H: So it's a very simple change to make and it doesn't seem to hurt all that much.\nSpeaker C: So...\nSpeaker C: You run with three, two, two?\nSpeaker H: I have to look to see what the exact numbers were.\nSpeaker H: I thought it was like three, two, two, five. But I'll double check.\nSpeaker H: It was over a week ago that I did it, so I can't remember exactly.\nSpeaker H: But it's so much faster.\nSpeaker H: It makes a big difference.\nSpeaker H: So we could do a lot more experiments and throw a lot more stuff in there.\nSpeaker H: That's great.\nSpeaker H: Oh, the other thing that I did was...\nSpeaker H: I compiled the HTK stuff for the Linux boxes.\nSpeaker H: So we have this big thing that we got from IBM, which is a five processor machine, really fast.\nSpeaker H: But it's running Linux. So you can now run your experiments on that machine.\nSpeaker H: And you can run five at a time and it runs as fast as five different machines.\nSpeaker H: So I've forgotten now what the name of that machine is, but I can send an email around about it.\nSpeaker H: And so we've got it now.\nSpeaker H: HTK is compiled for both the Linux and for the sparks.\nSpeaker H: You have to make sure that in your.cshrc it detects whether you're running on a Linux or a spark and points to the right executables.\nSpeaker H: And you may not have had that in your.cshrc before if you were always just running the spark.\nSpeaker H: So I can tell you exactly what you need to do to get all of that to work.\nSpeaker H: But it'll really increases what we can run on.\nSpeaker H: So together with the fact that we've got these faster Linux boxes and it takes less time to do these, we should be able to crank through a lot more experiments.\nSpeaker H: So after I did that, then what I wanted to do was try increasing the number of mixers just to see how that affects performance.\nSpeaker G: Yeah, in fact, you could do something like keep exactly the same procedure and then add a fifth thing onto it.\nNone: Exactly.\nSpeaker A: So at the middle where the arrows are showing, that's you're adding one more mixture per state?\nSpeaker H: Let's see, it goes from this, let's try to go backwards. At this point it's two mixtures per state.\nSpeaker H: So this just adds one except that actually for the silence model, it's six mixtures per state.\nSpeaker H: So it goes to two. And I think what happens here is...\nSpeaker G: Might be between a shared...\nSpeaker H: Yeah, I think that's what it is or something.\nSpeaker H: Yeah, it's...\nSpeaker H: I can't remember now what happens at that first one. I have to look it up and see.\nSpeaker H: Because they start off with an initial model, which is just this global model, and then they split it to the individuals.\nSpeaker H: And so it may be that that's what's happening here. I have to look it up and see. I don't exactly remember.\nSpeaker H: So that's it.\nSpeaker G: Right. So what else?\nSpeaker C: Yeah, there was a conference call this Tuesday.\nSpeaker C: I don't know yet what happened Tuesday, but the points that they were supposed to discuss is still things like the weights.\nSpeaker G: Oh, this is a conference call for Aurora participants, sort of thing. I see.\nSpeaker G: Do you know who was... since we weren't in on it, do you know who was in from OGI?\nSpeaker G: Was he involved or was it Sunil?\nSpeaker G: Yeah, right.\nSpeaker D: Oh, you don't know.\nSpeaker D: Okay.\nSpeaker C: All right.\nSpeaker C: Yeah.\nSpeaker C: So the points where the weights...\nSpeaker C: Oh, to weight, the different error rates that are obtained from different languages and conditions.\nSpeaker C: It's not clear that they will keep the same kind of weighting. Right now it's a weighting on improvements.\nSpeaker C: Some people are giving that it would be better to have weights on...\nSpeaker C: Well, to combine error rates before computing improvement.\nSpeaker C: And the fact is that right now for the English, they have weights... they combine error rates.\nSpeaker C: And the other language is they combine improvements. It's not the very consistent.\nSpeaker C: Yeah.\nSpeaker C: And so, well, this is a point.\nSpeaker C: And right now, actually, there is a thing also that happens with the current weight.\nSpeaker C: It's that very non-significant improvement on the well-matched case, resulting in huge differences in the final number.\nSpeaker C: So, perhaps they will change the weights to...\nSpeaker H: How should that be done?\nSpeaker H: I mean, it seems like there's a simple way...\nSpeaker H: This seems like an obvious mistake or something.\nSpeaker G: Well, I mean, the fact that it's inconsistent is an obvious mistake.\nSpeaker G: But the other thing... I don't know, I haven't thought of through, but one would think that each...\nSpeaker G: It's like, if you say, what's the best way to do an average, an arithmetic average or geometric average?\nSpeaker G: It depends what you want to show.\nSpeaker G: Each one is going to have a different characteristic.\nSpeaker H: So...\nSpeaker H: It seems like they should do like the percentage improvement or something, rather than the absolute improvement.\nSpeaker H: Well, they are doing that.\nSpeaker G: No, it is relative.\nSpeaker G: But the question is, do you average the relative improvements or do you average the error rates and take the relative improvement of that?\nSpeaker G: And the thing is, it's not just a period average because of these ratings.\nSpeaker G: It's a weighted average.\nSpeaker C: Yeah, and so, when you average the relative improvement, it tends to give a lot of importance to the when-match case, because the baseline is already very good.\nSpeaker H: Why don't they not look at improvements, but just look at your scores.\nSpeaker H: Figure out how to combine the scores with a weight or whatever, and then give you a score.\nSpeaker H: Here's your score.\nSpeaker H: And then they can do the same thing for the baseline system. Here's its score.\nSpeaker H: And then you can look at...\nSpeaker G: Well, that's what you're seeing as one of the things they could do.\nSpeaker G: Just when you get all done, I think that they...\nSpeaker G: I was in there, but I think they started off this process with an ocean that you should be significantly better than the previous standard.\nSpeaker G: And so they said, how much is significantly better and so they said, well, you should have half the errors or something that you had before.\nSpeaker G: So it's...\nSpeaker G: But it does seem like...\nSpeaker G: It does seem like it's more logical to combine them first and then do the...\nSpeaker C: Combining error rates?\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Well, but there is still this problem of ways.\nSpeaker C: When you combine error rates, it tends to give more importance to the difficult cases.\nSpeaker C: And some people think that they have different opinions about this.\nSpeaker C: The people think that it's more important to look at 12-10% relative improvement on well-matched cases than 12-50% on the mismatched.\nSpeaker C: And other people think that it's more important to improve a lot on the mismatch.\nSpeaker H: It sounds like they don't really have a good idea about what the final application is going to be.\nSpeaker G: Well, you know, the thing is that if you look at the numbers on the more difficult cases, if you really believe that was going to be the predominant use, none of this would be good enough.\nSpeaker G: Nothing anybody...\nSpeaker G: Whereas you sort of, with some reasonable error recovery, could imagine in the better cases these systems working.\nSpeaker G: So, I think the hope would be that it would work well for the good cases and it would have reasonable soft degradation as you got to worse and worse conditions.\nSpeaker H: Yeah.\nSpeaker H: I guess what I'm...\nSpeaker H: I mean, I was thinking about it in terms of if I were building the final product and I was going to test to see which front end I wanted to use.\nSpeaker H: Try to wait things depending on the exact environment that I was going to be using the system in if I...\nSpeaker G: But no, no, no. I mean, this isn't the operating theater.\nSpeaker G: I mean, they don't really know, I think.\nSpeaker H: So, if they don't know, doesn't that suggest a way for them to go?\nSpeaker H: You assume everything's equal. I mean, you...\nSpeaker G: Well, I think one thing to do is to just not rely on a single number to maybe have three numbers.\nSpeaker G: And say, here's how much you improve the relatively clean case.\nSpeaker G: And here's a real match case and here's how much you...\nSpeaker H: So, not try to combine them.\nSpeaker G: Yeah, actually it's true. I've forgotten this.\nSpeaker G: But, well matched, it's not actually clean. What it is is just that...\nSpeaker G: The training and testing are similar.\nSpeaker G: So, I guess what you would do in practice is you try to get as many examples of similar sort of stuff as you could.\nSpeaker G: And then...\nSpeaker G: So, the argument for that being the more important thing is that you're going to try and do that.\nSpeaker G: But you want to see how badly it deviates from that when the...\nSpeaker H: So...\nSpeaker H: So, you should wait those other conditions very, you know, really small.\nSpeaker H: But, no, that's more of an information kind of thing.\nSpeaker G: That's an argument for it. Let me give you the opposite argument.\nSpeaker G: The opposite argument is you're never really going to have a good sample of all these different things.\nSpeaker G: Meaning, are you going to have examples with Windows open, half open, full open, going 70, 60, 50, 40 miles an hour on what kind of roads, with what passing you, with...\nSpeaker G: I think that you could make the opposite argument that the well-matched case is a fantasy.\nSpeaker G: So, I think the thing is that if you look at the well-matched case versus the medium and the mismatched case, we're seeing really, really big differences in performance, right?\nSpeaker G: And you wouldn't like that to be the case.\nSpeaker G: You wouldn't like this. As soon as you step outside, you know, a lot of the cases...\nSpeaker H: A lot of the cases... A little teach him to roll their window up.\nSpeaker G: I mean, these cases, if you go from the... I mean, remember the numbers right off, but if you go from the well-matched case to the medium, it's not an enormous difference in the training testing situation.\nSpeaker G: And it's a really big performance drop.\nSpeaker G: So, yeah, I mean, the reference one, for instance, this is back old on Italian, was like 6% of the error for the well-matched and 18% for the medium matched and 60% for the...\nSpeaker G: for how they mismatched.\nSpeaker G: And, you know, these other systems, we helped it out quite a bit, but still there's something like a factor or two or something between well-matched and medium matched.\nSpeaker G: And so, I think that if what you're... if the goal of this is to come up with robust features, it does mean...\nSpeaker G: so you could argue, in fact, that the well-matched is something you shouldn't be looking at at all.\nSpeaker G: That the goal is to come up with features that will still give you reasonable performance.\nSpeaker G: You know, it's again, gentle degradation, even though the testing condition is not the same as the training.\nSpeaker G: So, you know, I could argue strongly that something like the medium mismatch, which is, you know, not pathological, but...\nSpeaker G: What was the medium mismatch condition again?\nSpeaker C: Yeah, medium mismatch is everything with the far microphone, but trained on low-noisy condition, low speed and stop-car, and tested on high-speed conditions.\nSpeaker G: Right, so it's still the same microphone in both cases, but there's a mismatch between the car conditions.\nSpeaker G: And that's... you could argue that's a pretty realistic situation, and I'd almost argue for waiting that highest, but the way they have it now, I guess it's... they compute the relative improvement first and then average that with a waiting.\nSpeaker G: And so then that makes the highly matched, the really big thing.\nSpeaker G: So, since they have these three categories, it seems like the reasonable thing to do is to go across the languages and to come up with an improvement for each of those.\nSpeaker G: And say, okay, in the highly matched case, this is what happens, in the... the... so the medium, if this happens, in the highly mismatched, that happens.\nSpeaker G: And you should see a gentle degradation through that.\nSpeaker G: But I think that... I gather that in these meetings, it's really tricky to make anything... can he policy change?\nSpeaker G: Because everybody has their own opinion.\nSpeaker E: Yeah.\nSpeaker C: So, yeah. But there is probably a big change that we made is that the baseline, they want to have a new baseline, perhaps, which is the MFCC, but with voice activity detector.\nSpeaker C: And apparently, some people are pushing to still keep this 50% number, so they want to have at least 50% improvement on the baseline, but it should be a much better baseline.\nSpeaker C: And if we look at the result that's summing the sound, just putting the VAD in the baseline improves like more than 20%, which would mean that 50% on this new baseline is like more than 60% improvement.\nSpeaker G: So, nobody would be there, probably, right?\nSpeaker C: Right, nobody would be there.\nSpeaker G: Good. What to do.\nSpeaker G: So, who's VAD? Is this...\nSpeaker C: They didn't decide yet. I guess this was one point of the conference.\nSpeaker C: But...\nSpeaker C: Yeah.\nSpeaker G: I think that would be good. I mean, it's not that the design of the VAD isn't important, but it does seem to be a lot of work to do a good job on that, as well as being a lot of work to do a good job on the feature design.\nSpeaker G: So, if we can cut down on that, maybe we can make some progress.\nSpeaker C: But I guess, perhaps...\nSpeaker C: Yeah. So, one told that perhaps it's not fair to do that because to make a good VAD, you don't have enough to do the business features.\nSpeaker C: So, you really need to put more in the different things.\nSpeaker G: Yeah.\nSpeaker H: Sure. But I'm confused. What do you mean?\nSpeaker G: Yeah. But, let's say, MFCC, for instance, doesn't have anything related to the pitch. So, just for example. So, suppose you've got what you really want to do is put a good pitch detector on there, and if it gets a non-imbeguous...\nSpeaker G: Oh, I see.\nSpeaker G:...if it gets a non-imbeguous result, then you definitely in a region with speech.\nSpeaker H: So, there's this assumption that the voice activity detector can only use the MFCC?\nSpeaker C: That's not clear, but...\nSpeaker G: Well, for the baseline.\nSpeaker G: Yeah. So, if you use other features, then it's just a question of what is your baseline?\nSpeaker G: What is it that you're supposed to do better than? And so, having the baseline be the MFCCs means that people could choose to pour their effort into trying to do really good VAD.\nSpeaker G: But they seem like two separate issues, right?\nSpeaker G: I mean, there's sort of separate. Unfortunately, there's coupling between them, which is part of what I think Stefan is getting to is that you can choose your features in such a way as to improve the VAD.\nSpeaker G: And you also can choose your features in such a way as to improve recognition.\nSpeaker G: But you should do both, right?\nSpeaker G: You should do both. And I think that this still makes...\nSpeaker G: I still think this makes sense as a baseline.\nSpeaker G: It's just saying, as a baseline, we know that we had the MFCCs before, lots of people have done voice activity detectors.\nSpeaker G: You might as well pick some voice activity detector and make that the baseline, just like you picked some version of HDK and made that the baseline.\nSpeaker G: And then let's try to make everything better. And if one of the ways you make it better is by having your features be better features for the VAD than that's so be it.\nSpeaker G: But at least you have a starting point that's...\nSpeaker G: Because some of the people didn't have a VAD at all, I guess, right? And then they look pretty bad.\nSpeaker G: And in fact, what they were doing wasn't so bad at all.\nSpeaker H: Yeah, it seems like you should try to make your baseline as good as possible.\nSpeaker H: And if it turns out that you can't improve on that, well, I mean, nobody wins and you just use MFCC, right?\nSpeaker G: Yeah, I mean, it seems like...\nSpeaker G: It should include sort of the current state of the art that you want to try and improve.\nSpeaker G: And MFCCs, or PLP or something, it seems like a reasonable baseline for the features.\nSpeaker G: And anybody doing this task is going to have some sort of voice activity detection at some level.\nSpeaker G: Some way they might use the whole recognizer to do it, rather than a separate thing.\nSpeaker G: But they'll have it on some level.\nSpeaker H: Seems like whatever they choose, they shouldn't purposefully brain damage a part of the system to make a worse baseline.\nSpeaker G: Well, I think people just... it wasn't that they purposefully brain damage, I think people hadn't really thought through about the VAD issue.\nSpeaker G: And then when the proposals actually came in, half of them had VADs and half of them didn't.\nSpeaker G: And half the did did well and half the didn't did poorly so.\nSpeaker C: Yeah, I'm sorry, we see what happened with this.\nSpeaker C: Yeah, so what happened since last week is quite from BOTGID's experiments on VAD on the baseline.\nSpeaker C: And these experiments also are using some kind of noise compensation, so spectra-suppraction.\nSpeaker C: And putting on nine normalization just after this.\nSpeaker C: So having spectra-suppraction, LDA filtering and on nine normalization.\nSpeaker C: So which is similar to the proposal one, but with spectra-suppraction in addition.\nSpeaker C: And it seems that on nine normalization doesn't help further when you have spectra-suppraction.\nSpeaker H: Is this related to the issue that you brought up a couple of meetings ago with the musical tones?\nSpeaker C: I have no idea because the issue I brought up was with a very simple spectra-suppraction approach.\nSpeaker C: The one that they use at OGI is one from the proposed or a proposal, which might be much better.\nSpeaker C: So yeah, I asked Sunil for more information about that, but I don't know yet.\nSpeaker C: And what's happened here is that we have this kind of new reference system which use a nice clean-down sampling of sampling, which uses a new filter that's much shorter, and which also gets the frequency bit of 64 hours, which was not done for the proposal.\nSpeaker C: When we say we have that, Sunil, have it now too or? No. No. Okay. Because we're still testing. So we have the result for just the features.\nSpeaker C: We are currently testing with putting the neural network into Kerti.\nSpeaker C: It seems to improve on the well-matched case, but it's a little bit worse on the mismatched, highly mismatched.\nSpeaker C: I mean, when we put the neural network, and with the current weight thing, I think it would be better because the one-matched case is better.\nSpeaker F: But how much worse since the weighting might change? How much worse is it on the other conditions when you say it's a little worse?\nSpeaker C: It's like... 10% relative.\nSpeaker G: Okay. But it has the latency is much shorter.\nSpeaker C: When I say it's worse, it's not... I compare proposal 2 to proposal 1.\nSpeaker C: Putting neural network compared to not having any neural network. This new system is better because as this 64 hours cutoff, clean, don't sampling, and what else? Yeah, good VAD.\nSpeaker F: But the latency... but you've got latency shorter now. Yeah. So it's better than the system that we had before.\nSpeaker C: Yeah, mainly because of the 64 hours and the good VAD.\nSpeaker C: And then I took this system and we put the old filters also. So we have this good system with good VAD, with the short filter and with the long filter.\nSpeaker C: And with the short filter it's not worse. So why is it?\nSpeaker G: Okay, so that's all fine. But what you're saying is that when you do these... let me try to understand. When you do these same improvements to proposal 1, that on the things are somewhat better in proposal 2 for the well-matched case and somewhat worse for the other two cases.\nSpeaker G: So now that these other things are in there, is it the case maybe that the additions of proposal 2 over proposal 1 are less important?\nSpeaker C: Yeah, probably.\nSpeaker C: Okay. So yeah, but it's a good thing anyway to have shorter delay. Then we try to do something like proposal 2, but using also have the G features.\nSpeaker C: So there is this G L T part which is just standard features and then junior 2 neural networks.\nSpeaker C: And it doesn't seem to help. However, we just have one result which is the Italian mismatch.\nSpeaker F: Okay. There was a start of some effort and something related to voicing or something.\nSpeaker C: Yeah. So basically we try to find good features that will be useful voicing detection.\nSpeaker C: But it's still on the... basically we're still paying with my laptop.\nSpeaker H: What sorts of features are you looking at?\nSpeaker C: So we would be looking at the variance of the spectrum of the excitation, something like this, which is really I have a voiced sound.\nSpeaker H: What does that mean? The variance of the spectrum of excitation?\nSpeaker C: Yeah. So basically the spectrum of excitation.\nSpeaker G: Okay. What you're calling the excitation is that recall is you're subtracting the male filter spectrum from the FFT spectrum.\nSpeaker C: So we have the male filter bank, we have the FFT.\nSpeaker G: So it's not really an excitation, but it's something that hopefully tells you something about the excitation.\nSpeaker C: Yeah, that's right.\nSpeaker B: Yeah.\nSpeaker B: There's some histograms.\nSpeaker B: Yeah, that's...\nSpeaker C: Yeah, so for a voiced portion we have something that has a mean around 0.3 and for a voiced portion the mean is 0.59.\nSpeaker C: But the variance seems quite...\nSpeaker H: How do you know... how did you get your voiced and unvoiced truth data?\nSpeaker C: We used a timet and we used the cannon guns between the phones and...\nSpeaker B: But if we look at one send, apparently it's good.\nSpeaker C: Yeah, but yeah.\nSpeaker C: So it's noisy timet, that's right.\nSpeaker C: It seems quite robust to know, so when we take mid-row, this parameter across time for the same values and that's very close.\nSpeaker C: Yes, so there is this, there will be also the...\nSpeaker C: Something like the maximum of the other relation functions.\nSpeaker H: Is this a trained system or is it a system where you just pick some thresholds?\nSpeaker C: How does it work?\nSpeaker C: Right now we're just trying to find some features.\nSpeaker C: Hopefully I think what we want to have is to put these features in kind of...\nSpeaker C: Well, to obtain a statistical model of these features, just to use an neural network.\nSpeaker C: Hopefully these features would help.\nSpeaker H: Because it seems like what you said about the mean of the voiced and the unvoiced, that seemed pretty encouraging.\nSpeaker G: Well, yes, except the variance was big.\nSpeaker H: Well, I don't know that I would trust that so much because you're doing these canonical mappings from timet lablings, right?\nSpeaker H: So really that's sort of a cartoon picture about what's voiced and unvoiced.\nSpeaker H: So that could be giving you a lot of variance.\nSpeaker H: It may be that you're finding something good and that the variance is sort of artificial because of how you're getting your truth.\nSpeaker G: Yeah, but another way of looking at it might be that, I mean, we are coming up with feature sets after all.\nSpeaker G: So another way of looking at it is that the melkepster, melke spectrum, melkepster, any of these variants, give you this smooth spectrum.\nSpeaker G: It's a spectral envelope.\nSpeaker G: By going back to the FFT, you're getting something that is more like the raw data.\nSpeaker G: So the question is, what characterization, and you're playing around with this, another way of looking at it, is what characterization of the difference between the raw data and the smooth version is something that you're missing that could help.\nSpeaker G: So I mean, looking at different statistical measures of that difference, coming up with some things and just trying them out and seeing if you add them onto the feature vector, is that make things better or worse in noise.\nSpeaker G: Where you really just, the way I'm looking at it is not so much you're trying to find the best world's best voiced unvoiced classifier, but it's more that, you know, try some different statistical characteristics of that difference back to the raw data.\nSpeaker G: Right.\nSpeaker G: And maybe there's something there that the system can use.\nSpeaker C: Yeah, but the more of use is that, the more of use is that, well, using the FFT, you just give you just information about if it's voiced or not voiced mainly.\nSpeaker C: Yeah.\nSpeaker C: This is why we started to look.\nSpeaker G: Well, that's the way what I'm arguing is that, yeah, I mean, what I'm arguing is that that that's gives you your intuition.\nSpeaker G: But in reality, it's, you know, there's all this, this overlap and so forth.\nSpeaker G: And, but what I'm saying is that maybe okay, because what you're really getting is not actually voiced or sound voiced, both for the fact the reason of the overlap and then, you know, structural reasons like the one the Chuck said, that in fact, well, the data itself is, that you're working with is not perfect.\nSpeaker G: So, I'm saying is maybe that's not a killer because you're just getting some characterization, one that's driven by your intuition about voiced and voiced certainly, but just some characterization of something back in the, in the, in the almost raw data rather than the smooth version.\nSpeaker G: Your intuition is driving you towards particular kinds of statistical characterizations of what's missing from the spectra envelope.\nSpeaker G: Obviously, something about the excitation.\nSpeaker G: And what is it about the excitation?\nSpeaker G: And, you know, and you're not getting the excitation anyway, you know, so, so I would almost take especially if these trainings and so forth or faster, but almost just take a scatter shot.\nSpeaker G: A scatter shot at a few different ways of look of characterizing that difference and you have one of them, but, and see, you know, which of them helps?\nSpeaker H: So, is the idea that you're going to take whatever features you develop and just add them on to the feature vector? Or what's the use of the voice, unvoiced detector?\nSpeaker C: I guess we don't know exactly yet, but, yeah. It's not part of a VAD system that you're doing?\nSpeaker C: Oh, okay.\nSpeaker C: Yeah, it could be, it could be a neural network that does voice and voice detection, so the big neural networks that does, on the specification.\nSpeaker G: But each one of the mixture components, I mean, you have variants only, so it's kind of like you're just multiplying together these, probably some individual features within each mixture.\nSpeaker G: So it's so...\nSpeaker H: I think it's an 8 thing. It seems like a good idea.\nSpeaker G: Yeah.\nSpeaker G: Yeah, I mean, I know that people doing some robustness things are always back, we're just doing, just being gross, just throwing in the FFT and actually it wasn't so bad.\nSpeaker G: So, and you know that it's got to hurt you a little bit to not have a smooth spectral envelope, so there must be something else that you get in return for that.\nSpeaker H: So, how does, maybe I'm going into much detail, but how exactly do you make the difference between the FFT and the smooth spectral envelope?\nSpeaker H: Well, yeah, how is that?\nSpeaker C: We just...\nSpeaker B: We have the 23 coefficient of after the FFT and we understand this coefficient between the other frequency range and the interpolation between the point is given for the triangular filter, the value of the triangular filter.\nSpeaker B: And these weight we obtained is mod.\nSpeaker G: So, you essentially take the values that you get in the triangular filter and extend them, just sort of like a rectangle.\nSpeaker G: That's at that value.\nSpeaker C: So, yeah, we have one point for one energy for a filter bank, which is the energy that's centroid on the surface.\nSpeaker H: So, you end up with a vector that's the same length as the FFT vector, and then you just compute differences and sum the differences.\nSpeaker C: And I think the variance is computed only from like 200 hertz to 1500, because...\nSpeaker C: Right.\nSpeaker C:...1500, because yeah.\nSpeaker C:...20000.\nSpeaker C: Above seems that some voices on can add also like noisy part on the African.\nSpeaker G: Yeah, but no, it's being sensed to look at.\nSpeaker G: So, this is...\nSpeaker H:...basically.\nSpeaker H: Basically, this is comparing an original version of a signal to a smoothed version of the same signal.\nSpeaker G: Right. So, this is...\nSpeaker G: I mean, you could argue about whether it should be linear interpolation or zero-thord or...\nSpeaker G:...anyway, something like this is what you're feeding your recognizer, typically.\nSpeaker G: Like which...\nSpeaker G: No, so the male capstream is the...\nSpeaker G:...capstream of this spectrum, or the long spectrum, whatever.\nSpeaker G: You're subtracting in...\nSpeaker G:...power domain or log domain.\nSpeaker G: Okay, so it's sort of like division.\nSpeaker G: You do that, you have this vector.\nSpeaker G: So, ratio?\nSpeaker G: Yeah.\nSpeaker G: But anyway...\nSpeaker H:...and that's...\nSpeaker H: So, what's the intuition behind this kind of a thing?\nSpeaker H: I don't really know the signal processing well enough to understand what...\nSpeaker H:...what is that doing?\nSpeaker C: Yeah, like the sub-exam.\nSpeaker C: What we would like to have is some spectrum of the excitation signal, which is for a very strong, ideally, a full strain.\nSpeaker C: And for a voiced, it's something that's more flat.\nSpeaker C: Right.\nSpeaker C: And the way to do this is that, well, we have the FFT because it's computing the system.\nSpeaker C: And we have the male filter bands.\nSpeaker C: And so, if we like remove the male filter band from the FFT, we have something that's close to the excitation signal.\nSpeaker C: Okay.\nSpeaker C: Something that's like a train of both strain, a voiced sound, and that's...\nSpeaker C: I see.\nSpeaker H: So, do you have a picture that...\nSpeaker H:...is this for a voiced segment, this picture?\nSpeaker H: Yeah.\nSpeaker H: What does it look like for unvoiced?\nSpeaker B: No, I'm voicing over.\nSpeaker B: Oh, thanks.\nSpeaker B: So, you know...\nSpeaker C: This is the...\nSpeaker B: This is another voiced accent.\nSpeaker B: You know, is this part between the frequency that we are considered for the situation, for the difference, and this is the difference.\nSpeaker H: This is the difference, okay?\nSpeaker C: Yeah, because we begin in 15.\nSpeaker H: So...\nSpeaker H: Does the periodicity of this signal say something about the pitch?\nSpeaker H: The pitch?\nSpeaker H: Yeah.\nSpeaker G: That's like fundamental frequency.\nSpeaker G: So, I mean...\nSpeaker G: Day-on-see.\nSpeaker G: I mean, to first order, what you're doing, you can ignore all the details and all the ways which is these are complete lies, that what you're doing in future extraction for speech recognition is you have, in your head, a simplified production model for speech, in which you have a periodic, a great periodic source of strivings and filters.\nSpeaker G: First order for speech recognition, you say, I don't care about the source, right?\nSpeaker G: So, you just want to find out what the filters are.\nSpeaker G: The filters, roughly act like an overall resonances and so forth, that's prositing the excitation.\nSpeaker G: So, if you look at the spectral envelope, just the very smooth properties of it, you get something closer to that.\nSpeaker G: And the notion is, if you have the full spectrum of all the little minigritty details, that that has the effect of both, and it would be a multiplication in frequency domain, so that would be like an addition in log, a spectrum domain.\nSpeaker G: And so, this is saying, well, if you really do have that, so a vocal tract envelope, and you subtract that off, what you get is the excitation.\nSpeaker G: I call that lies because you don't really have that.\nSpeaker G: You just have some kind of signal processing trickery to get something that's kind of smooth.\nSpeaker G: It's not really what's happening in the vocal tract.\nSpeaker G: So, you're really getting the vocal excitation.\nSpeaker G: That's why I was referring to it in a more conservative way when I was saying, well, it's the excitation.\nSpeaker G: It's not really the excitation.\nSpeaker G: It's whatever it is that's different between...\nSpeaker G: So, standing back from that, you sort of say there is this very detailed representation.\nSpeaker G: You go to a smooth representation.\nSpeaker G: You go to a smooth representation because it typically generalizes better.\nSpeaker G: But, whenever you smooth, you lose something.\nSpeaker G: So, the question is, have you lost something you can use?\nSpeaker G: Probably you wouldn't want to go to the extreme of just saying, okay, our features that will be the FFT.\nSpeaker G: Because we really think we do gain something in robustness from going to something smoother.\nSpeaker G: But maybe there's something that we missed.\nSpeaker G: So, what is it?\nSpeaker G: And then you go back to the intuition that, well, you don't really get the excitation, but you get something related to it.\nSpeaker G: And as you can see from those pictures, you do get something that shows some periodicity in frequency.\nSpeaker G: And sometimes.\nSpeaker G: That's really nice.\nSpeaker H: So, you don't have one for unvoiced picture?\nSpeaker B: No, I don't think so.\nSpeaker G: But presumably you'll see something that won't have this kind of regularity and frequency.\nSpeaker H: I would like to see those pictures.\nSpeaker H: Yeah.\nSpeaker G: Yeah.\nSpeaker H: And so, you said this is pretty doing this kind of thing.\nSpeaker H: It's pretty robust to noise.\nSpeaker C: It seems, yeah.\nSpeaker C: The mean is different.\nSpeaker B: Because the histogram is different.\nSpeaker C: And I know that the kind of robustness to noise, so if you take this frame from the noisy utterance, and the same frame from the king utterance,\nSpeaker H: do you end up with a similar difference over here? Yeah.\nSpeaker H: Okay.\nSpeaker B: Cool.\nSpeaker B: Because here the same frame for the clean speed.\nSpeaker B: Oh, that's clean.\nSpeaker B: Okay.\nSpeaker B: There are differences because here the FFT is only with 256 points.\nSpeaker B: And this is with 512.\nSpeaker B: Okay.\nSpeaker C: This is kind of interesting also because if we use the standard frame length of 25 milliseconds, it happens is that for low pitched voice because of the frame length, you don't really have.\nSpeaker C: You don't clearly see this period of structure because of the first logo for each each of you.\nSpeaker H: So this one include is a longer.\nSpeaker C: It's like 50 milliseconds.\nSpeaker C: 50 minutes.\nSpeaker C: Yeah.\nSpeaker C: But it's the same frame.\nSpeaker H: What's that time frequency trade off thing, right?\nSpeaker H: I see.\nSpeaker H: Sorry.\nSpeaker H: Oh, so is this the difference here?\nSpeaker B: No, this is the same frame.\nSpeaker H: Oh, that's the original.\nSpeaker C: Yeah, so with the short frame basically, the period is not enough to use these kind of neat things.\nSpeaker B: But yeah.\nSpeaker C: Yeah, so probably we'll have to use like long, long frames.\nSpeaker H: Oh, that's interesting.\nSpeaker G: Maybe.\nSpeaker G: Well, I mean, it looks better.\nSpeaker G: But the thing is if you're actually asking if you actually need to do very long and FFT and maybe pushing things.\nSpeaker H: Would you want to do this kind of difference thing after you do spectral subtraction?\nSpeaker G: Maybe.\nSpeaker G: The spectral subtraction is being done at what level is it being done at the level of FFT bins or at the level of a male spectrum or something.\nSpeaker C: I guess it depends.\nSpeaker G: I mean, how are they doing it?\nSpeaker C: I guess Erickson is still doing that.\nSpeaker C: So yeah, I'm not really...\nSpeaker G: So in that case, it might not make much difference at all.\nSpeaker H: It seems like you'd want to do it on the FFT bins.\nSpeaker H: Maybe.\nSpeaker H: I mean, if you were going to edit it for this purpose, that is.\nSpeaker F: Yeah.\nSpeaker G: Okay.\nSpeaker G: What else?\nSpeaker C: So we'll perhaps try to convince the JIP people to use a new FFT.\nSpeaker G: Okay.\nSpeaker G: Has anything happened yet on this business of having some sort of standard...\nSpeaker C: Sorry, sir.\nSpeaker C: Not yet, but I will go down.\nSpeaker C: No, they are.\nSpeaker C: I think they're more time because they're...\nSpeaker H: When is the next Aurora deadline?\nSpeaker F: Early June?\nSpeaker F: Late June?\nSpeaker F: Not early June?\nSpeaker G: Okay.\nSpeaker G: And he's been doing all the talking, but...\nSpeaker G: Yeah.\nSpeaker G: This is by the way a bad thing.\nSpeaker G: We're trying to get female voices in this record as well.\nSpeaker G: Make sure Carmen talks as well.\nSpeaker G: But is he pretty much been talking about what you're doing also?\nSpeaker B: I am doing this.\nSpeaker F: Yes.\nSpeaker B: I don't know.\nSpeaker B: I think that's for the recognition.\nSpeaker E: The meeting record that is better than that can be speaking to you.\nSpeaker G: Well, we'll get to Spanish voices sometime.\nSpeaker G: We do.\nSpeaker G: We want to recognize you too.\nSpeaker B: And the result for the TVG.\nSpeaker B: The video record that we did for each people.\nSpeaker G: No, we like...\nSpeaker G: We're in the...\nSpeaker F: We're in the Lord Hermansky work in the frame of mind.\nSpeaker F: We like higher rates.\nSpeaker F: That way there's lots of work to do.\nSpeaker F: That's...\nSpeaker F: Anything to...\nSpeaker I: Not much is new.\nSpeaker I: I talked about what I'm trying to do last time.\nSpeaker I: I said I was going to use Avandano's method of using a transformation to map from long analysis frames, which I used for removing reverberation to short analysis frames for feature calculation.\nSpeaker I: He has a trick for doing that involving viewing the DFT as a matrix.\nSpeaker I: But I decided not to do that after all because I realized to use that I'd need to have these short analysis frames get plugged directly into the feature computation somehow.\nSpeaker I: And right now I think our feature computation is set up to take audio as input in general.\nSpeaker I: So I decided that I'll do the reverberation removal on the long analysis windows and then just re-sensitize audio and then send that.\nSpeaker G: This is in order to use the SRI system.\nSpeaker I: Or even if I'm using R system I was thinking it might be easier to re-sensitize the audio because then I could just use FeeCalc as it is and I wouldn't have to change the code.\nSpeaker G: Yeah, I mean it's certainly in a short turn.\nSpeaker F: This sounds easier.\nSpeaker F: Yeah, I mean longer term if it turns out to be useful when I want to.\nSpeaker G: Right, that's true.\nSpeaker G: No, you may be putting other kinds of errors in from the recent analysis.\nSpeaker I: Okay, I don't know anything about recent analysis.\nSpeaker F: But I'm not sure if you're likely to think that is.\nSpeaker F: It's a reasonable way to go for an initial thing.\nSpeaker F: We can look at exactly what you end up doing and figure out if there's something that could be heard by the end part of the process.\nSpeaker I: Okay.\nSpeaker I: That's, that's it, that's it.\nSpeaker G: Anything to?\nSpeaker A: Well, I've been continuing reading.\nSpeaker A: I went off on a little tangent this past week looking at Modulations Spectrum stuff and learning a bit about what it is and the importance of it in speech recognition.\nSpeaker A: I found some neat papers, historical papers from Kanadera, Hermanski and Array.\nSpeaker A: And they did a lot of experiments where they take speech and they modify the, they measure the relative importance of having different portions of the Modulation Spectrum intact.\nSpeaker A: And they find that the spectrum between one and 16 hertz in the Modulation is important for speech recognition.\nSpeaker G: Sure, I mean this sort of goes back to earlier stuff by Drillman.\nSpeaker G: And the MSG features were sort of built up this notion.\nSpeaker G: But I guess I thought you had brought this up in the context of targets somehow.\nSpeaker G: Right.\nSpeaker G: But it's not, I mean they're sort of not in the same kind of category as a phonetic target or a syllabic target or a more like a feature or something.\nSpeaker A: I was thinking more like using them as the inputs to the detectors.\nSpeaker G: Oh, I see.\nSpeaker G: Well, that's sort of what MSG does.\nSpeaker G: Right. So it's, but, but, yeah.\nSpeaker G: We'll talk more about it later.\nSpeaker G: We can talk more about it later.\nSpeaker F: Would you dig it?\nSpeaker F: Let's do dig it.\nSpeaker F: You start.\nSpeaker I: Reading transcript L-5617686691.\nSpeaker I: 7921.\nSpeaker I: 20350125.\nSpeaker I: 40564334.\nSpeaker I: 9290-3114-8629.\nSpeaker I: 4136256690.\nSpeaker I: 4367-615298.\nSpeaker I: 76633377823.\nSpeaker I: 842-614627.\nSpeaker G: Transcript L-55.\nSpeaker G: Or transcript L-55.\nSpeaker G: 687-715-075.\nSpeaker G: 896-03865.\nSpeaker G: 566-2002-96.\nSpeaker G: 848-9164.\nSpeaker G: 1686-24013.\nSpeaker G: 3126-619960.\nSpeaker G: 837-08080.\nSpeaker G: 6236-4006-9743.\nSpeaker H: Transcript L-49.\nSpeaker H: 884259-7450.\nSpeaker H: 787-0106158.\nSpeaker H: 742-503970.\nSpeaker H: 858-034714.\nSpeaker H: 06044-2001.\nSpeaker H: 2293-3128.\nSpeaker H: 8558-558-691.\nSpeaker H: 358-294017.\nSpeaker A: Transcript L-50.\nSpeaker A: 9067-3933.\nSpeaker A: 08308-3481.\nSpeaker A: 21365-3159.\nSpeaker A: 4084305211.\nSpeaker A: 924-584-5504.\nSpeaker A: 2226168155.\nSpeaker A: 707-087-8402-803-160507.\nSpeaker C: Transcript L-53.\nSpeaker C: 5954-883914.\nSpeaker C: 860310-9753.\nSpeaker C: 55529-3365.\nSpeaker C: 337-074710.\nSpeaker C: 6418-3166.\nSpeaker C: 576-89596.\nSpeaker C: 228-3305595.\nSpeaker C: 355-059-615-025.\nSpeaker B: Transcript L-54.\nSpeaker B: 1431771032.\nSpeaker B: 9882-488812.\nSpeaker B: 1310576812.\nSpeaker B: 628875912.\nSpeaker B: 5272-8617498-0000709.\nSpeaker B: 862-458892.\nSpeaker B: 221-196783.\n"}, {"meeting_id": "IS1002d", "summary": "The group mainly talked about the detailed design of the product at this meeting. First, the industrial designer introduced the function design of the product. It had not only some basic functions, but also the locator function and provided users with a revolutionary way of zapping. Later, the group went on to check the controller's paging ability and talked about more details on buttons. Next, the marketing expert mentioned three things making the product marketable and one possible drawback of the product. Besides, the marketing recommended making it upgradable but the project manager pointed out the risk of doing that. When evaluating the cost of the product, the group discussed some details of the components and made some adjustments. They finally got an estimate of fifteen point eight Euros, which was within the budget. In the end, they did some self-assessment and celebrated the completion of the project.", "dialogue": "Speaker C: So, you forgot how this works again.\nSpeaker B: Boss.\nSpeaker B: Baby Boss.\nSpeaker D: Maybe, maybe, maybe, maybe.\nSpeaker A: Okay, so we're here to talk about the detailed design of the product.\nSpeaker A: Okay, yep.\nSpeaker A: And here's the agenda for this meeting.\nSpeaker A: I'm just going to open, see a few boring words to start with again, and start taking minutes afterwards.\nSpeaker A: You guys are going to give us a presentation of our wonder product that I can see some demonstrations of over there.\nSpeaker A: Looks cool.\nSpeaker A: And then we're going to evaluate it.\nSpeaker A: And we're going to talk about finance, and I've got a lovely Excel spreadsheet that knocked up in the last five minutes for this.\nSpeaker A: And...\nSpeaker A: You knocked it up.\nSpeaker A: Yep, and we're going to evaluate the product and close.\nSpeaker A: We've got 40 minutes to do this in.\nSpeaker A: We should be fine, let's try and keep this one in schedule.\nSpeaker A: Alright.\nSpeaker A: So, your thing is in...\nSpeaker A: Where is it?\nSpeaker A: Three.\nSpeaker C: Three.\nSpeaker C: Three.\nSpeaker B: The end product thingy.\nSpeaker B: Yeah.\nSpeaker A: Who wants it?\nSpeaker B: Peter can have it.\nNone: I'll talk.\nSpeaker C: Yeah, so what we ended up with.\nSpeaker C: Production co-estimated by our manufacturing department and the research department, which is ours.\nSpeaker C: It is a 15.8 era, as someone forgot.\nSpeaker C: Unit price.\nSpeaker C: And unit price.\nSpeaker C: Unit production price.\nSpeaker C: Cost.\nSpeaker C: Thank you.\nSpeaker C: We implemented the basic functions, which is just TV functions plus the locator, which was one of the marketing things.\nSpeaker C: Cradle, scroll wheel for the channels.\nSpeaker C: And now we implemented the way of putting the new and revolutionary zapping your favorite channels functionality in the scroll.\nSpeaker A: Zapping your favorite channels, eh?\nSpeaker B: Yeah.\nSpeaker B: Scrolling through your favorites list.\nSpeaker A: Okay, okay.\nSpeaker A: Zapping, you know.\nSpeaker A: Zapping.\nSpeaker A: Ah, okay, okay.\nSpeaker C: That's the favorite.\nSpeaker C: Maybe this is Portuguese thing.\nSpeaker C: And yeah, that was the result.\nSpeaker A: Okay.\nSpeaker A: I like the logo on there as well.\nSpeaker B: It's very prominent.\nSpeaker A: It's very prominent.\nSpeaker A: So this is the...\nSpeaker B: So here, give you the...\nSpeaker B: So this is the cradle unit, and this is the actual remote itself.\nSpeaker B: So the scroll bar is... or the scroll wheel is this green little scrolly guy here.\nSpeaker B: And then the volume controls are here and here.\nSpeaker B: Okay.\nSpeaker B: You've got the keypad, which is the numbers from zero to nine, and then ten.\nSpeaker B: This is the power button.\nSpeaker B: We have our... we have the enter button and what is the other button here?\nSpeaker C: This is the teletext.\nSpeaker C: This is the start to...\nSpeaker B: The program button.\nSpeaker B: Yeah.\nSpeaker B: The program.\nSpeaker B: So this button will be used both for the favorites and for programming your... the type of television you want to use.\nSpeaker B: So the plastic is the white area of this... of the model here.\nSpeaker B: And the red area is like a rubber covering.\nSpeaker B: So you can see that when it lays like this or like this, and the buttons are all going to be rubber.\nSpeaker B: So it's pretty hard to actually damage it.\nSpeaker A: Is that...\nSpeaker A: Could that be easy for the scroll wheel to be rotated if it ones on it?\nSpeaker B: Yeah.\nSpeaker B: That might be a possible... a possible problem.\nSpeaker B: Not helping.\nSpeaker A: No, I guess it depends on the stiffness or the level of it.\nSpeaker B: Yeah.\nSpeaker B: I don't know if it's sliding, but I think it's pretty ergonomic.\nSpeaker B: You can feel it.\nSpeaker A: It was good.\nSpeaker A: I think this is going to be slightly lighter in the...\nSpeaker A: Yeah, of course.\nSpeaker B:...the kind of thing.\nSpeaker B: Well, this is clay.\nSpeaker A: Yeah.\nSpeaker A: It's going to go... you have to reach it.\nSpeaker B: Yeah, the power button is a bit of a reach, but I think we might scale down the final model a bit.\nSpeaker B: Ah, yeah.\nSpeaker B: That would make sense.\nSpeaker B: This is a bit larger than it would be.\nSpeaker C: But...\nSpeaker A: It's cool.\nSpeaker C: I'm impressed.\nSpeaker C: It's all a low on the head.\nSpeaker A: In holder.\nSpeaker A: So, which was the market and perspective?\nSpeaker D: Oh, I like it.\nSpeaker D: I mean, you guys gave me more than I was asking for, so I'm happy.\nSpeaker D: Because we've got some really marketable features in this.\nNone: Yeah.\nSpeaker D: I think it's a good job.\nSpeaker B: And Pedro can demonstrate that...\nSpeaker B: Yeah.\nSpeaker C:...the page inability.\nSpeaker C: You have the base station with the little button for the... where's my remote?\nNone: Beep beep beep.\nSpeaker D: Oh, blah.\nSpeaker D: The more key to function.\nSpeaker D: That's great.\nSpeaker D: That's great.\nSpeaker D: Very future.\nSpeaker B: Beep beep beep.\nSpeaker A: So, let me get it.\nSpeaker A: If I press this button...\nSpeaker A: Beep beep beep.\nSpeaker A: I see.\nSpeaker A: That's weird.\nSpeaker A: It is.\nSpeaker A: Hang on.\nSpeaker A: Beep beep beep.\nSpeaker B: Shut up.\nSpeaker B: So, you can take this home.\nSpeaker D: Take this home with you tonight and you can push that and you'll be across town.\nSpeaker A: Beep beep beep.\nSpeaker A: Okay.\nSpeaker A: I find that.\nSpeaker A: No, no, no.\nSpeaker A: But the two blue...\nSpeaker A:...are those for the... to charge off of.\nSpeaker A: Exactly.\nSpeaker B: That's exactly what those are for.\nSpeaker B: And there's one other feature that we were debating but we decided to go against it is you could...\nSpeaker B: Beep beep beep.\nSpeaker B: We were thinking that it might be interesting to have a trigger button here because you have this finger.\nSpeaker B: Yeah.\nSpeaker B: It kind of feels like there should be something there but we couldn't figure out what button is important enough to put there.\nSpeaker B: And we don't want to accidentally be hitting the power button like that.\nSpeaker A: Maybe if you had a trigger plus the scroll and that would get past the problem of landing and scrolling because then it would need to be hitting both sides.\nSpeaker B: So maybe in the final design phase we might tweak that a little bit.\nSpeaker A: Okay.\nSpeaker A: But it's definitely got options for different types of models and things as well based on\nSpeaker D: that, haven't it? Yep, I like.\nSpeaker A: Good job.\nSpeaker A: So is that the final color scheme as well?\nSpeaker B: Oh no, this is just what we had to work with at the time.\nSpeaker B: Okay.\nSpeaker B: So we'll leave the color scheme up to the marketing people.\nSpeaker A: Okay.\nNone: Careful.\nNone: Yeah, it came off.\nSpeaker A: The scroll wheels is a problem with them not being.\nSpeaker A: I don't think he's out in the face guy who wants to touch it.\nSpeaker D: My suggestion is we're going to go to the silvers and blacks like most of the televisions.\nSpeaker D: You know, some blend of silvers and blacks.\nSpeaker A: Beep, beep, beep.\nSpeaker A: Okay.\nSpeaker A: Enough of that.\nSpeaker A: Wow, that's cool guys.\nSpeaker A: Okay, so we've done with the this presentation.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nNone: Okay.\nSpeaker D: So you're not going to find my folder up there.\nSpeaker D: I got to do mine up at the board.\nSpeaker A: Okay.\nSpeaker D: Yeah.\nSpeaker D: So.\nSpeaker A: Can I not get your get stuff from your shirt folder, no?\nSpeaker D: No, it's not in there because I had the computer problem and I couldn't create it.\nSpeaker D: I couldn't create it in the PowerPoint.\nSpeaker D: And I think I've got this really strange cable.\nSpeaker D: So what I had basically going from the PowerPoint format is that, yeah, yeah, I like this a lot.\nSpeaker D: Is this one of the tests is to see how we can adapt to changing situations in the meeting room.\nSpeaker D: So what we had is we had the method.\nSpeaker D: That's not how you spell method, is it?\nSpeaker B: No way.\nSpeaker D: No way.\nSpeaker D: Come here you.\nSpeaker D: So this doesn't go so fast this way.\nSpeaker D: And when I speak about method, I speak about the marketing of the product.\nSpeaker D: And to me with this product, we got basically three things to market.\nSpeaker D: We've got the features, we have the characteristics, and we have the, I don't know what we call the other part, would we call, you know, the corporate.\nSpeaker D: Help me.\nSpeaker D: The corporation stands behind the product.\nSpeaker D: So the features, I think we got the scroll, we've got the locator, we've got the durability, we've got the dependability, we've got, you know, the features that make this a unique product.\nSpeaker D: The characteristics I talk about, we have reliability, we have comfort, we have ergonomics, we have environmentally sensitive, and the corporation we're talking about, we're a new company.\nSpeaker D: We're wanting to make a name for ourselves.\nSpeaker D: We're wanting you to find our product.\nSpeaker D: So we're going to give you a good product at a fair price.\nSpeaker D: One thing I would want to see is can we get a lifetime guarantee on this product, a normal use guarantee, which means that this product for the life of the life use, if it should have a technical problem that we could replace it at no cost, that was something I would be interested in.\nSpeaker D: So yeah, without going into great details, we have a product that has the features and the characteristics and the background, I believe to make it marketable, I believe at a cost of 35 to 50 euros, we're going to be competitive and we're going to have a market niche.\nSpeaker A: Would you argue that we're better going for the higher cost than bringing it down to 25 as we probably could, but lowering a profit margin?\nSpeaker D: That would be, I think, a decision best paid by corporate.\nSpeaker D: For my evaluation, based on what our competition is, I think that we can go after this and go after more of the exclusivity sense than the mass market sense.\nSpeaker D: But I'm sure open to market this in either direction.\nSpeaker D: But you guys came up with a great product and at that cost, I think there's nobody else that's putting this combination of ingredients together.\nSpeaker D: The only limitations I see to this is that we're focused on television only.\nSpeaker D: That's the only drawback I see to this.\nSpeaker D: But with all of these other features, I think people, one thing I'm hoping for is people are not going to even notice.\nSpeaker D: There's going to be somebody going home and say, oh, this thing doesn't work for my DVD and my, but I like these other things that they keep it, they don't take it back.\nSpeaker A: They're actually quite open to be able to expand the product for a later version with those features.\nSpeaker A: Quite simply anyway, there's no fundamentally different thing.\nSpeaker D: One of the thoughts that I had is can this unit be produced in a way that makes it upgradable?\nSpeaker D: Like a SIM card in a telephone.\nSpeaker D: Is there a card?\nSpeaker D: Can we make a card?\nSpeaker C: And so after we need some more buttons if it were to work on some other stuff.\nSpeaker D: But you follow one of those?\nSpeaker C: Yeah, we could get another version of it that actually works.\nSpeaker D: Yeah, because if we can make this unit upgradable, then we're, but then we talk about changing the warranty concept and everything.\nSpeaker D: But that's, that was just an idea I had.\nSpeaker D: To me, the only additions.\nSpeaker A: The risk of making it unusable as well or making it less because at the moment it's actually very straightforward to look at all the buttons you know what they do.\nSpeaker A: It's very simple and it just works.\nSpeaker A: Yep.\nSpeaker A: Does the rest of that?\nSpeaker D: Anyway, that's...\nSpeaker A: Okay, so we can talk about finance now.\nSpeaker A: So I have a little spreadsheet for us where I was wondering, when you talked about the 15.8 euros, is wondering how you came up with that figure?\nSpeaker B: Well, that was just our technical team ended up the production cost of the individual units.\nSpeaker A: Okay, so I have a spreadsheet here for this.\nSpeaker A: Oh, very cool.\nSpeaker A: I've made a few assumptions here in that I'm assuming that our power adapter we can make for a cost of four euros equivalent to solar cells, which I think is probably fair considering that we have in-house manufacturing of power adapters already.\nSpeaker A: And I'm assuming that the locator beacon, the...\nSpeaker B: Yeah, BBB.\nSpeaker A: It can be made for...\nSpeaker A: It's only a different that time.\nSpeaker A: It can be made for a similar price to an L.C. display.\nSpeaker A: Do you think that's fair coming from the manufacturing?\nSpeaker B: Yeah, I do think that we don't need the advanced chip on print.\nSpeaker B: We only need the regular chip on print.\nSpeaker B: Okay.\nSpeaker B: So we have a miscalculation there.\nNone: Okay.\nSpeaker A: So we're down to 16.4.\nSpeaker B: And we have a single curved...\nSpeaker A: Is that a single curved rather than a double curved?\nSpeaker B: I think that...\nSpeaker A: We're not entirely sure what single curved...\nSpeaker D: We've got a curve and a droop, I don't know whether that...\nSpeaker B: It's single curved.\nSpeaker A: Okay, I'm not convinced.\nSpeaker A: We save one euro that way, yeah?\nSpeaker A: So we can bring it down to...\nSpeaker B: So it's a little bit more than single curved.\nSpeaker B: So yeah, it's 15.8.\nSpeaker B: That's where we came up with it.\nSpeaker B: Well, hang on.\nSpeaker A: Don't speak so.\nSpeaker A: It's in here.\nSpeaker A: And do we have any...\nSpeaker A: We have special form, don't we?\nSpeaker A: Yeah, we do.\nSpeaker A: So that's...\nSpeaker A: Yeah.\nSpeaker A: Ah!\nSpeaker A: But you know...\nSpeaker A: We haven't talked about any special colour though.\nSpeaker A: I forgot for grazing silvers.\nSpeaker B: That's very special.\nSpeaker B: It's pretty.\nSpeaker A: Okay, so we're...\nSpeaker A: Piss-whattern scroll wheel.\nSpeaker A: We're basically...\nSpeaker A: We have...\nSpeaker A: Is this intended to be a button as well?\nSpeaker A: Or just a...\nSpeaker A: Scroll scroll.\nSpeaker A: That's a scroll.\nSpeaker A: Just a scroll.\nSpeaker A: It's not one of the scrolls where, for example, with this one, you could push it down to be\nSpeaker B: a button. No, we just use it as a scroll.\nSpeaker A: Okay.\nSpeaker A: Then we have 15.8 euros.\nSpeaker B: It was a pretty accurate estimate.\nSpeaker C: It was a bit.\nSpeaker C: Oh, look at it.\nSpeaker A: Yeah, awesome.\nSpeaker A: Okay, so...\nSpeaker A: We're on to the...\nSpeaker A: Kind of frighteningly accurate.\nSpeaker A: We're on to the back part of the presentation where we have a look at the criteria that Paul, the criteria that Paul has given us.\nSpeaker A: And we can use that to tell.\nSpeaker A: How's it gone?\nSpeaker A: I don't get any thoughts.\nSpeaker A: How have we done today?\nSpeaker B: I think we did pretty well.\nSpeaker A: I think we did pretty well too.\nSpeaker A: It was pretty spectacular.\nSpeaker D: No, I think we come up with an attractive, marketable product and concept.\nSpeaker A: Any other changes or thoughts?\nNone: Okay.\nSpeaker A: So, we depart about...\nSpeaker A: You know, for creativity.\nSpeaker B: That was...\nSpeaker B: I think there was plenty of room.\nSpeaker B: I think we ended up being innovative.\nSpeaker D: Yeah.\nSpeaker D: Well, we...\nSpeaker D: We kind of broke...\nSpeaker D: We kind of at least adjusted every criteria that gave us.\nSpeaker D: Because we still have the teletext capability in this thing, right?\nSpeaker D: We raised the price of it.\nSpeaker D: We've added new technology to it.\nSpeaker D: So, you know, if this thing flies, then we've adjusted or broken every idea that gave us.\nSpeaker A: Every idea necessarily.\nSpeaker A: It's still a remote control.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: No, yeah.\nSpeaker A: But we did break with the specs a little bit, I guess.\nSpeaker A: But the...\nSpeaker D: But I like...\nSpeaker D: I mean, what I say, we have...\nSpeaker D: I believe we have room creativity because we did it.\nNone: Okay.\nSpeaker A: We just ship what we report back to the bosses.\nSpeaker A: Yeah.\nSpeaker A: I think they were reasonably flexible with this over the whole changing the specs thing.\nSpeaker C: It was more of a teamwork thing than really leadership as a glutter.\nSpeaker D: Yeah, I agree.\nSpeaker D: Just cohesive.\nSpeaker A: It's energy.\nSpeaker D: Yes.\nSpeaker D: It's energetic.\nSpeaker A: It's a lot of synergy.\nNone: Teamwork.\nSpeaker A: Yeah, yes.\nSpeaker A: What did we think of the meeting room, I guess, is an important thing here?\nSpeaker B: These cables suck.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: This falls off and...\nSpeaker A: Yep.\nSpeaker A: The whiteboard worked really well.\nSpeaker C: They must have very long shirts.\nSpeaker C: That's way far down lapel, like... lapel, lapel, lapel.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker C: That's almost a crotch.\nSpeaker C: Microphone.\nSpeaker A: That's quite close.\nSpeaker D: You know what they're going to have on the recording and they're from that microphone is your lunch digesting, you know?\nSpeaker A: Oh dear.\nSpeaker A: Oh dear.\nSpeaker A: Oh dear.\nSpeaker A: No more pizza for me.\nSpeaker A: So yeah, the pen came in all right and it's a little uncomfortable to use, but I guess\nSpeaker C: it works. You were using it to upset all.\nSpeaker C: It's still that.\nNone: That's our boss.\nSpeaker A: I'm not convinced of that at all.\nSpeaker A: I've been wondering about that all day, but CDs look like they're...\nSpeaker A: That looks like it would be that way around.\nSpeaker A: Natures right.\nSpeaker A: It feels more comfortable quite you could upside down.\nSpeaker A: I don't care.\nSpeaker A: Okay, so that data might be slightly involved.\nSpeaker A: What new ideas have we found?\nSpeaker D: New ideas for the product or for the environment or what are we...\nSpeaker A: Well, the list of both then for the product.\nSpeaker B: Well, we had the favorites list in the scroll bar and we have the cradle and the remote call feature.\nSpeaker C: And we came up with quite a bunch of new ideas.\nSpeaker A: And for the meeting room is anyone getting any more?\nSpeaker C: Comfortable headsets would be nice.\nSpeaker D: Well, I mean, clearly remote control microphones would be a nice solution to all these cables, but I'm sure that there's just some justification for these things that I don't know about.\nSpeaker D: And of course I didn't have so much fun with my computer this afternoon.\nSpeaker D: It's not been cooperating so well.\nSpeaker D: But I don't think that's available.\nSpeaker D: Could have been worse.\nSpeaker A: Okay, so are the costs within the budget?\nSpeaker A: Nope.\nSpeaker A: Oh, hang on.\nSpeaker A: It really did something that the costs were under 1250 euros.\nSpeaker A: Oh, crime rates have changed.\nSpeaker A: We're still under 20 euros to build.\nSpeaker A: So we're good.\nSpeaker A: The costs within the budget?\nSpeaker A: Not the original budget, but they are now.\nSpeaker A: Is the project evaluated?\nSpeaker A: I think so.\nSpeaker A: Yeah, then.\nSpeaker A: Hooray!\nSpeaker A: As it says.\nSpeaker C: Okay.\nSpeaker C: Freco provided out the cafeteria.\nSpeaker A: I don't know how that got there.\nSpeaker C: Anyway.\nSpeaker C: Who wrote that one?\nSpeaker D: So we need to close this meeting here for our role.\nSpeaker D: Congratulations.\nSpeaker C: Good job, guys.\nSpeaker D: I've got a lot of paperwork to catch up on, so I'll just close this one and come back.\nSpeaker A: Yeah, I've got a lot of paperwork to catch up with.\nNone: Mm!\nNone: Perfect.\n"}, {"meeting_id": "TS3006d", "summary": "The meeting began with the harsh budget problem that the group had to cut down something of the current conceptual remote control, otherwise they would be unable to handle the cost. They talked about reducing the buttons, integrating functions, or using cheaper materials. After the discussion, the Marketing offered a questionnaire for the group to check if they were on the right track. Through the questionnaire, they found that it would be better if they paid more attention to the external design so the Industrial Design embellished the current design on the color and logo. As the last thing left to the meeting, the group reviewed their cooperation and got a general plan of their coming group presentation.", "dialogue": "Speaker F: We still have to do our questionnaire.\nSpeaker D: Okay, first of all, I'll start with the costs.\nSpeaker D: Because that's going to influence our design.\nSpeaker D: Alright.\nSpeaker D: If you don't know if you already had a look or not.\nSpeaker C: Did you do your questionnaire already?\nSpeaker C: I already did it.\nSpeaker C: It's not much suggestion.\nSpeaker D: Because we have a problem.\nSpeaker D: If you look closely, you can see it.\nSpeaker D: And I already took the liberty to make some suggestions.\nSpeaker D: At the moment, we have 15 buttons.\nSpeaker D: One LCD screen.\nSpeaker D: One advanced chip on print.\nSpeaker D: We use a sensor, this is for the speech.\nSpeaker D: We use kinetic energy.\nSpeaker D: And we wanted the buttons in a special color.\nSpeaker D: Okay.\nSpeaker D: First thing we should do is draw a special color of the buttons.\nSpeaker C: No, that's for the trendy feeling.\nSpeaker D: Everything is.\nSpeaker D: But we switch to a hint dynamic.\nSpeaker D: Because that's still the same idea as kinetic energy.\nSpeaker D: Because you have to use it and do that.\nSpeaker B: Yeah, but I don't think people like it.\nSpeaker B: Just do a normal battery.\nSpeaker F: Just do a normal battery then.\nSpeaker B: It has to be 12 and a half.\nSpeaker B: Oh my goodness.\nSpeaker D: You're going to need these.\nSpeaker E: Oh no.\nSpeaker D: Okay.\nSpeaker D: So we're at 25.\nSpeaker D: Do we keep the shape doubly curved or do we go for single curved?\nSpeaker F: I guess we'll have to go for single curved.\nSpeaker F: I mean we have to drop on everything.\nSpeaker B: But we can keep it single curved with top view still curved.\nSpeaker B: From the side it's flat.\nSpeaker B: You just have to hold it like this then.\nSpeaker D: How about, sorry.\nSpeaker D: Another option I saw was to drop the buttons one through nine.\nSpeaker D: So you can directly access a channel.\nSpeaker D: But instead I use only the up and down errors.\nSpeaker C: And we don't have any basic options anymore.\nSpeaker C: Oh yeah.\nSpeaker F: And then they don't need special colors.\nSpeaker D: They don't need special colors.\nSpeaker D: Fine.\nSpeaker D: We like it.\nSpeaker D: Yeah.\nSpeaker D: You were saying.\nSpeaker F: That was exactly my point.\nSpeaker F: Like let's drop all the buttons and just make one.\nSpeaker F: I mean we're going to use the LCD screen anyway.\nSpeaker F: So we'll just have to use it for everything.\nSpeaker F: And then you can make an overview of channels in the screen and select the channel.\nSpeaker D: Click.\nSpeaker D: Okay.\nSpeaker D: So maybe we can drop a few more buttons.\nSpeaker D: Let's see.\nSpeaker C: We only need the manual arrow button thing.\nSpeaker C: Everything you can do with the menu.\nSpeaker C: Yeah.\nSpeaker F: We need one integrated button for everything.\nSpeaker F: Yeah.\nSpeaker B: A joystick.\nSpeaker F: Yeah.\nSpeaker F: Kind of.\nSpeaker F: Because.\nSpeaker D: If you go to our.\nSpeaker F: If you are in the sound system there.\nSpeaker F: And you want to adjust the treble for instance.\nSpeaker F: This is just an example.\nSpeaker F: You want to see a bar on which you can adjust it from zero to ten for example.\nSpeaker F: But you want a sound preview of how it's going to sound right.\nSpeaker F: So you want to click on it.\nSpeaker F: It's activated and when you move it here the difference of the treble coming out or going into the sound.\nSpeaker F: So you'll need a kind of a joystick.\nSpeaker B: Yeah.\nSpeaker B: Or a button.\nSpeaker B: Or a scroll wheel push button.\nSpeaker B: Yeah.\nSpeaker B: So that's kind of on your mouse and then you can click it.\nSpeaker B: Adjust it.\nSpeaker B: Click again and then you're out of it.\nSpeaker B: But you still.\nSpeaker B: But you then still need to have.\nSpeaker B: Well you can use the scroll wheel as well for maybe for the channels.\nSpeaker B: But you still have to have some button in the menu to go back.\nSpeaker C: You can do one integrated scroll wheel push button.\nSpeaker C: Yeah.\nSpeaker C: And then just drop all the other buttons.\nSpeaker C: Well not all.\nSpeaker B: Not sound I guess.\nSpeaker C: The cost of one integrated button is far more than a few extra push buttons.\nSpeaker D: Yeah.\nSpeaker D: It's one integrated button is five times the cost of a normal button.\nSpeaker C: So we have to make it more.\nSpeaker D: You could also drop three more of these without losing much functionality.\nSpeaker D: Yeah.\nSpeaker B: What is the sample sensor?\nSpeaker B: Sample speed.\nSpeaker B: Oh that's for the speech recognition.\nSpeaker B: Yeah.\nSpeaker B: Good.\nSpeaker D: Speech recognition.\nSpeaker F: No but.\nSpeaker D: Yeah that's possible.\nSpeaker F: Yeah it's expensive but we never considered the possibilities of speech recognition.\nSpeaker F: Because it can take the function of a lot of buttons.\nSpeaker D: But it's not very easy to use.\nSpeaker D: No.\nSpeaker B: It can be disturbed by noise.\nSpeaker F: Yeah.\nSpeaker B: Unless you say that speech recognition works.\nSpeaker B: Let me see what's more popular.\nSpeaker B: I guess the screen was more popular than the speed recognition.\nSpeaker B: Yeah.\nSpeaker B: I have to look on that.\nSpeaker B: Let me see.\nSpeaker B: Oh no.\nSpeaker B: There are more people who like speech recognition than LCD screen.\nSpeaker D: Okay.\nSpeaker D: But if the LCD screen.\nSpeaker B: But it's both.\nSpeaker C: A whole lot of concept.\nSpeaker C: Yeah.\nSpeaker B: No.\nSpeaker B: Yeah we keep the screen.\nSpeaker B: I mean it's about the same.\nSpeaker B: 81 to 91 percent.\nSpeaker F: Yeah we haven't really integrated the speech into the system so we can drop that.\nSpeaker B: Yeah.\nSpeaker D: Let's drop the speech.\nSpeaker D: For less years.\nSpeaker D: 16 years.\nSpeaker D: We still have three and a half year to lose.\nSpeaker D: We need to lose some buttons.\nSpeaker D: Yeah if you lose the back.\nSpeaker D: The okay button.\nSpeaker D: Let's say we only have the four arrows.\nSpeaker D: Yeah.\nSpeaker B: And then use.\nSpeaker B: Or in the power button.\nSpeaker B: Okay and the menu button does also does the okay function then.\nSpeaker B: And then when you in the menu.\nSpeaker B: So you activate menu.\nSpeaker C: If we do a two integrated scroll we will push buttons.\nSpeaker C: We can drop all the push buttons.\nSpeaker C: With one integrated button we can do the whole menu thing.\nSpeaker C: Yeah.\nSpeaker D: That would save 0.2 euros compared to.\nSpeaker D: No.\nSpeaker C: No it's three euros.\nSpeaker D: Yeah.\nSpeaker D: This together is more expensive.\nSpeaker D: Yeah.\nSpeaker D: It's almost the same as this.\nSpeaker B: Okay.\nSpeaker B: For example if you have four buttons.\nSpeaker B: The channel up and down for you left right.\nSpeaker B: That's basic.\nSpeaker B: That's what you need anyway.\nSpeaker B: And then for the menu.\nSpeaker B: You can have a button that activates menu.\nSpeaker B: Or do we just integrate and scroll wheel with a push button.\nSpeaker B: And then if the moment you use the scroll wheel the menu gets activated and then you can scroll to an option.\nSpeaker B: Click on it.\nSpeaker B: It goes into a feature.\nSpeaker B: Click on it against flex feature.\nSpeaker B: Scroll adjust it.\nSpeaker B: Click again.\nSpeaker B: It's okay.\nSpeaker B: Then you only need one button to move back.\nSpeaker B: Or under each option.\nSpeaker B: You set a screen thing with says back and you select that one.\nSpeaker B: Click again.\nSpeaker B: And then you scroll back and in that menu scroll.\nSpeaker B: Click one step back.\nSpeaker B: So then you need five buttons.\nSpeaker B: And one integrated scroll wheel push button.\nSpeaker B: But you can drop three buttons.\nSpeaker B: But I see that's one year or more expensive.\nSpeaker D: Yeah.\nSpeaker B: So I could have had the option.\nSpeaker B: Because which button do we have now?\nSpeaker B: Those five which I mentioned and then menu.\nSpeaker B: The four things.\nSpeaker B: Four arrows.\nSpeaker B: Yeah.\nSpeaker B: Power.\nSpeaker D: If you go to eight I don't know how to.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker B: So four arrows.\nSpeaker B: Yeah.\nSpeaker B: Power.\nSpeaker B: Power.\nSpeaker D: Yeah.\nSpeaker D: That's five.\nSpeaker D: We have a back and a okay button.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: That's seven.\nSpeaker B: I want to activate menu.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: That's eight.\nSpeaker B: And then you can read the menu.\nSpeaker B: We keep the display.\nSpeaker D: Okay.\nSpeaker D: Even if we drop three buttons from here we still have to make some adjustments around here.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: We need the chip for the for the LCD display.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker C: We need the advanced.\nSpeaker C: Okay.\nNone: It's plastic.\nNone: Instead of.\nSpeaker B: instead of...\nSpeaker D: Then we ought to make it wood, because that also is good in the market with the 45 plus people.\nSpeaker B: Yeah, but that's not our market.\nSpeaker D: No, that may be not, but maybe...\nSpeaker D:...and it's better than plastic anyway.\nSpeaker B: I don't know, hard plastic.\nSpeaker C: Plastic with a special color.\nSpeaker B: Yeah, a plastic with a special color.\nSpeaker B: Yeah, a special color.\nSpeaker B: Okay.\nSpeaker B: No, because we have to use the special color anyway, you forgot that.\nSpeaker B: Yeah.\nSpeaker C: Yeah.\nSpeaker D: So let's go for the plastic.\nSpeaker D: Yeah, okay. Plastic.\nSpeaker B: And since it's not kinetic, it doesn't have to flip around that much.\nSpeaker D: It's free.\nSpeaker B: We still have a problem of two euros.\nSpeaker D: Yeah, okay.\nSpeaker D: If you drop to...\nSpeaker C: The buttons, those are really neat.\nSpeaker C: Yeah, we can drop them.\nSpeaker C: It's an out-found strip on the front.\nSpeaker B: Yeah, you still need...\nSpeaker B: Do we really need that out-found strip for an LCD display?\nSpeaker F: Yeah.\nSpeaker F: So the other option would be to go for the sample speaker, which can use a regular chip, which is six euros in total.\nSpeaker F: That doesn't matter.\nSpeaker B: Oh.\nSpeaker B: I'd rather keep the display.\nSpeaker B: Yeah.\nSpeaker B: Because we've already designed for it.\nSpeaker B: The only option is in hand-dynamo.\nSpeaker D: Yeah, and something else.\nSpeaker D: All right, the other one.\nSpeaker D: The camera, right.\nSpeaker F: Can't we fit all the buttons in an integrated scroll push button?\nSpeaker F: Because that will save us one and a half euro already.\nSpeaker B: And then integrate it?\nSpeaker B: Yeah, but that would make it not so easy to use.\nSpeaker B: I mean, it's not that important, easy to use, but...\nSpeaker F: Then we have to scroll through a lot of menus to be able to get where you want.\nSpeaker D: Yeah, two additional buttons, I believe.\nSpeaker F: Yeah, at least one for power.\nSpeaker D: I am power.\nSpeaker D: Three buttons and this.\nSpeaker F: It's just as expensive as what we have now.\nSpeaker C: It's an integrated button.\nSpeaker C: How many functions can it have?\nSpeaker D: Three.\nSpeaker D: Up, down.\nSpeaker F: Yeah, endlessly.\nSpeaker F: I mean, it can be a power button.\nSpeaker F: As soon as it's powered on, you can...\nSpeaker F: You can press it for like three seconds.\nSpeaker F: Then you can choose a flip channel.\nSpeaker F: You can choose sound options.\nSpeaker B: Then you should do everything in the menu.\nSpeaker B: Yeah.\nSpeaker B: On the screen.\nSpeaker F: Maybe we should, because we don't have money and we want to screen.\nSpeaker D: You can choose this, drop this, then we have a half euro left.\nSpeaker D: Yeah.\nSpeaker D: So we can maybe still use power button.\nSpeaker B: Yeah, but we...\nSpeaker B: All right.\nSpeaker B: I guess we have...\nSpeaker B: It saves us four euros and it costs us two and a half.\nSpeaker B: So, let's see.\nSpeaker B: We need to drop the one and a half, but we still 13 left.\nSpeaker D: Yeah.\nSpeaker D: No, it doesn't miscalculate it.\nSpeaker B: 13, so still half.\nSpeaker A: There goes the special...\nSpeaker B: That would make it less appealing, so that's no option.\nSpeaker D: Okay, what else?\nSpeaker D: Uncurved?\nSpeaker B: We're sure about it at Fans' Chip.\nSpeaker B: We need for the display.\nSpeaker F: Yeah, it says right here.\nSpeaker D: Okay.\nSpeaker D: They made it very easy for us.\nSpeaker B: Yeah.\nSpeaker B: We made it hard for ourselves with the display, but it's a cool feature.\nSpeaker D: I don't think I can persuade the management to say this is better for the market, so we sell more.\nSpeaker C: Okay.\nSpeaker C: What we could do is drop the special color for the buttons.\nSpeaker D: Oh, yeah, since we only have one button.\nSpeaker B: Yeah, but what is meant by special color?\nSpeaker B: I guess something else.\nSpeaker D: I think black or white.\nSpeaker D: Yeah, I think it's gray regular.\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker B: Gray and rubber.\nSpeaker F: But we definitely want the thing to be a special color.\nSpeaker B: Yeah.\nSpeaker B: So, I rather have an hand-dynamo.\nSpeaker B: Yeah.\nSpeaker B: And then drop the color.\nSpeaker B: Yeah.\nSpeaker B: You can still play with it then, I guess.\nSpeaker B: I don't know.\nSpeaker B: Yeah, but a hand-dynamo costs a lot of energy to keep the stream working, so I guess that isn't an option.\nSpeaker C: Display.\nSpeaker C: Yeah.\nSpeaker F: You only have to power it up when you want to use it.\nSpeaker C: Yeah.\nSpeaker C: If you have to power it up for 10 minutes.\nSpeaker D: No, I don't think the current status of chips are pretty...\nSpeaker D: Yeah.\nSpeaker D: Any shikun is saving the...\nSpeaker B: Yeah, true.\nSpeaker B: Yeah.\nSpeaker B: Let's go for the hand-dynamo then.\nSpeaker D: Yeah, hand-dynamo.\nSpeaker D: Do we want an extra button?\nSpeaker B: Or do we uncurf and flat instead of...\nSpeaker C: No, no, it has to be curved.\nSpeaker B: Yeah, it has to be curved and you have that color in the stream.\nSpeaker B: Yeah, special form.\nSpeaker C: That's the most important thing.\nSpeaker D: Yeah.\nSpeaker D: Is that also supposed to be too special?\nSpeaker B: Yeah.\nSpeaker B: Well, what a special form.\nSpeaker B: Oh, a special form.\nSpeaker D: Maybe that's nicer.\nSpeaker D: For scroll.\nSpeaker D: Yeah.\nSpeaker B: Oh, no, we don't have any buttons.\nSpeaker B: Yeah.\nSpeaker C: So, do you have to make it a special color then?\nSpeaker C: It's going to be integrated button, I think, also.\nSpeaker D: Maybe make it a special color then.\nSpeaker D: Yeah, just scroll.\nSpeaker D: Yeah, push down.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Make it a special color and then it looks fancy.\nSpeaker D: So...\nSpeaker D: Whoa, we're in a bit yet.\nSpeaker D: Yeah.\nSpeaker C: Let's save it.\nSpeaker C: Two special colors.\nSpeaker C: We only have one button.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Let's do it like this.\nSpeaker B: I mean, because it does not lose our identity of the product as we...\nSpeaker D: Okay.\nSpeaker D: So...\nNone: Okay.\nNone: This was old.\nSpeaker D: Wow, we come back to the drawing board then.\nSpeaker F: Yeah, back to work.\nSpeaker F: Yeah, you're pretty much...\nSpeaker F: It's silly because we should have had this meaning before we start drawing.\nSpeaker B: Yeah, but that's a fun part of it.\nSpeaker D: I wanted that a bit.\nSpeaker D: I wasn't allowed.\nSpeaker D: Okay.\nSpeaker F: All right.\nSpeaker D: I just forgot to save this.\nSpeaker D: Just a minute.\nNone: Off.\nNone: Yeah.\nSpeaker B: What's the next phase?\nSpeaker D: Yeah, this is the last phase, of course.\nSpeaker B: The agenda by your humble PM.\nSpeaker B: Oh, first phase.\nSpeaker D: All right.\nNone: Yeah.\nSpeaker D: Okay.\nSpeaker D: Well, we first should have had a prototype presentation as well as you saw that hadn't made those sense because we had to...\nSpeaker D: Drop everything.\nSpeaker F: Drop.\nSpeaker F: Yeah.\nSpeaker F: We went straight into finance.\nSpeaker F: Yeah.\nSpeaker D: Yeah, it was more important.\nSpeaker B: Yeah.\nSpeaker D: Push it up to the agenda.\nSpeaker D: Evaluation criteria and you have to use something about that?\nSpeaker B: Yeah.\nSpeaker B: I sure did.\nSpeaker B: And it combines with product evaluation.\nSpeaker B: So you put it in the...\nSpeaker B: We all have to keep in mind what has changed now.\nSpeaker B: So what we have left on the...\nSpeaker B: Yeah.\nSpeaker B: Because the evaluation is based on the initial user requirements.\nSpeaker B: All right.\nSpeaker B: Let's make it big.\nSpeaker B: Well, we evaluate the design with a seven-point skill as following.\nSpeaker B: Well, the remote controls recording criteria blah, blah, blah.\nSpeaker B: Two are false and then on the scale of seven points.\nSpeaker B: The leakage skill.\nSpeaker B: So you all know it.\nSpeaker B: Well, the criteria are based on the user requirements, the trends from the marketing research and the marketing strategy of the company itself.\nSpeaker B: Well, there are any words, documents which are open now.\nSpeaker B: I don't know, it's open yet now.\nSpeaker B: We all have to agree on a certain level.\nSpeaker B: What's this?\nSpeaker B: I don't know.\nSpeaker B: Well, the first point, the remote control matches the operating behavior of the young user.\nSpeaker B: So that means the operating behavior was using quite a lot of functions.\nSpeaker B: How do you think about that?\nSpeaker F: I think it does.\nSpeaker F: Because the operating behavior of the young user was also defined in color and shape and the use of LCD, right?\nSpeaker B: Yeah, the user.\nSpeaker B: It's mainly constraining on a lot of functions in this question.\nSpeaker F: Well, we have extended menus.\nSpeaker B: Yeah, you can make a lot of extended menus.\nSpeaker B: That's true.\nSpeaker B: I mean, the possibilities are almost unlimited to build a menu in the screen.\nSpeaker B: Yeah.\nSpeaker B: So on the scale from one to seven, what do you think?\nSpeaker C: Two or three?\nSpeaker B: Two or three?\nSpeaker B: Two or three?\nSpeaker B: Something like that?\nSpeaker B: Well, we have to choose one.\nSpeaker B: So what do you say?\nSpeaker B: I agree on two.\nSpeaker B: You could build this in now.\nSpeaker B: Yeah.\nSpeaker B: Yeah, okay.\nSpeaker B: I say two, personally.\nSpeaker D: Yeah.\nSpeaker D: I don't know.\nSpeaker D: You would decide as a would say it's free, but now in order to decide.\nSpeaker B: Yeah, but we have to evaluate, I guess, what we have now.\nSpeaker D: Okay, then I say three.\nSpeaker B: Yeah, you say three.\nSpeaker B: Yeah.\nSpeaker B: You say also three.\nSpeaker B: Three.\nSpeaker B: I say still two, but it has to be three then.\nSpeaker D: You're marketing it.\nSpeaker B: Yeah, I know.\nSpeaker B: So it's made bold, but it's not very clear.\nSpeaker B: Maybe you like.\nSpeaker B: Maybe other color, yeah.\nSpeaker B: Better.\nSpeaker B: Red.\nSpeaker D: Yeah.\nSpeaker B: All right.\nSpeaker B: It doesn't have to be bold anymore.\nSpeaker B: All right.\nSpeaker B: Oh, yeah.\nSpeaker D: Are we true?\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah, we're true.\nSpeaker E: Well, the remote control.\nSpeaker B: Definitely one.\nSpeaker B: It has to be.\nSpeaker B: The remote control hides irrelevant or less use functions.\nSpeaker B: For example, audio settings and screen settings.\nSpeaker B: Yeah.\nSpeaker B: So it's a very true point.\nSpeaker B: I mean, it hides all those functions.\nSpeaker B: They're hidden in the screen if you don't want to use them.\nSpeaker B: You just scroll over them and you place them.\nSpeaker B: I don't know where.\nSpeaker B: So that's very true, I guess, for our case.\nSpeaker B: Yeah, the second point.\nSpeaker B: Not so much.\nSpeaker B: It shows the relevant, the most used functions.\nSpeaker B: Power button.\nSpeaker B: Do we have a still power button?\nSpeaker B: Well, yeah, the buttons integrate.\nSpeaker B: Yeah.\nSpeaker B: So it's all huge.\nSpeaker B: Yeah, we dropped it.\nSpeaker B: You just push it in for our...\nSpeaker B: It's integrated.\nSpeaker B: Yeah, just push it in for three seconds or something.\nSpeaker B: Exactly.\nSpeaker D: It just goes for all the other things.\nSpeaker B: Yeah.\nSpeaker B: It shows relevant and most used functions.\nSpeaker B: Yeah.\nSpeaker B: On one side, we would say yes.\nSpeaker B: On the other side, we would say no.\nSpeaker B: I don't know.\nSpeaker D: It shows the most used functions.\nSpeaker D: And they are relevant.\nSpeaker C: But each channel directly with just one button.\nSpeaker C: With the scroll button.\nSpeaker B: With the menu.\nSpeaker B: Yeah, and then say channel.\nSpeaker F: Well, we should build it so that if you don't kind of push into the menu or something, when it's on, yeah?\nSpeaker D: It's turned on.\nSpeaker D: You need double click on the...\nSpeaker F: It automatically has the program and the volume function.\nSpeaker F: But as soon as you click it, you go into the menu or something.\nSpeaker B: Yeah.\nSpeaker B: Or you double click it.\nSpeaker B: But how do you change from volume to channel?\nSpeaker F: No, because it has four arrows, right?\nSpeaker D: No, no.\nSpeaker D: Because you scroll wheel.\nSpeaker D: Yeah.\nSpeaker F: I was thinking still about our integrated joystick.\nSpeaker B: We have no balance left.\nSpeaker B: So the joystick was not an option.\nSpeaker B: Yeah, that's a bummer.\nSpeaker B: So you have to double click for volume and three double click for the menu or something.\nSpeaker B: Yeah.\nSpeaker B: Hold it 10 seconds.\nSpeaker B: Yeah.\nSpeaker B: We'll make it a Morse code.\nSpeaker B: But ease of use was not very important.\nSpeaker B: Yeah, it should be true.\nSpeaker B: Yeah, that's not a question.\nSpeaker B: And the remote control shows the relevant and most used functions.\nSpeaker B: Well, I think it's pretty much in the middle.\nSpeaker D: Yeah.\nSpeaker B: You have to search for them in the screen in the menu.\nSpeaker B: Yeah.\nSpeaker B: So maybe it's more like a five.\nSpeaker F: Yeah, I would go for five or six.\nSpeaker F: Yeah.\nSpeaker B: Five or six.\nSpeaker B: Five.\nNone: Okay.\nSpeaker F: Five.\nSpeaker F: Five.\nSpeaker F: That's not this hard remote.\nSpeaker D: This is the weird remote control I've ever seen.\nSpeaker B: Yeah.\nSpeaker E: Yep.\nSpeaker E: Just one button.\nSpeaker E: All right.\nSpeaker B: The remote control makes use of technological innovation like LCD screen and speech recognition.\nSpeaker B: Well, yeah, we dropped the speech recognition, but it has at least one innovation.\nSpeaker D: Yeah, I say to them.\nSpeaker F: We still have the fruit and vegetable print.\nSpeaker F: Yeah, but that's not this question.\nSpeaker B: Never mind.\nSpeaker B: The other question.\nSpeaker B: I'm a bit lost.\nSpeaker C: This one.\nSpeaker C: I think that too.\nSpeaker B: Too.\nSpeaker B: Yeah.\nSpeaker B: I think LCD is more usable than speech.\nSpeaker B: Yeah.\nSpeaker B: Not the bold one.\nSpeaker F: It's way more practical.\nSpeaker B: Right.\nNone: Okay.\nSpeaker B: It has a fancy look.\nSpeaker B: I feel by using fruit vegetable prints primary colors, the sponge like material.\nSpeaker B: It should have been two questions.\nSpeaker B: Can't realize now because sponge like material is dropped, but the look I feel.\nSpeaker F: Yeah.\nSpeaker F: So we still have the primary colors.\nSpeaker F: Yeah.\nSpeaker F: But only on the outside.\nSpeaker B: The button is also color.\nSpeaker B: Yeah.\nSpeaker D: Okay.\nSpeaker D: The one button is yes.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Still we dropped also on the double curve.\nSpeaker D: Yeah.\nSpeaker D: You can check with the cell sheet.\nSpeaker D: Yeah.\nSpeaker B: We have single curve now.\nSpeaker B: And color.\nSpeaker B: And no material.\nSpeaker B: Yeah.\nSpeaker B: So maybe in the middle or something.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I mean, it's.\nSpeaker B: Yeah.\nSpeaker B: Works really.\nSpeaker B: Works really.\nSpeaker F: Yeah.\nSpeaker F: Actually, we didn't do so well on this one.\nSpeaker F: Because it's basically an old one with little curve on the side and in a different color.\nSpeaker F: Still, it's still hard.\nSpeaker F: Yeah.\nSpeaker F: I mean, the sponge like and the 3D shape that would give it something young and fresh.\nSpeaker B: Yeah.\nSpeaker B: But then we would have to drop the screen.\nSpeaker B: Oh, red.\nSpeaker B: Yeah.\nSpeaker D: You like both.\nSpeaker D: Yeah.\nSpeaker B: I like both.\nSpeaker B: The remote control displays the corporate logo.\nSpeaker D: Oh, yeah.\nSpeaker E: Of course.\nSpeaker E: You just have to draw.\nSpeaker E: Just a minute.\nSpeaker E: Yeah, but it's a white part.\nSpeaker F: Because we couldn't hardly draw on the lower half of the screen.\nSpeaker F: We couldn't make it big enough to actually draw in the.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker F: So, I mean, I wanted to put digits on the buttons as well.\nSpeaker B: But there's enough space for the corporate logo.\nSpeaker B: I mean, if there's only one.\nSpeaker B: Yeah.\nSpeaker B: If we have only one button.\nSpeaker B: Yeah.\nSpeaker B: So, I will say that is very true.\nSpeaker F: And even the shape we have now simulates kind of an R and the reversed R, doesn't it?\nSpeaker B: The remote control is easy to use.\nSpeaker B: Well, I would say.\nSpeaker B: No.\nSpeaker B: Skill, I would say 6 or something.\nSpeaker B: I don't think it's easy to use.\nSpeaker F: The only thing easy about it is that you don't have to move your fingers.\nSpeaker F: Yeah.\nSpeaker C: Because there's only one button.\nSpeaker C: And you only have to control one button.\nSpeaker B: Yeah, exactly.\nSpeaker B: And it has a nice screen.\nSpeaker B: But, yeah.\nSpeaker C: It gives facial feedback.\nSpeaker B: Yeah.\nSpeaker B: I would say 5 or 6.\nSpeaker B: I would say 6.\nSpeaker B: What do you say?\nSpeaker B: It's really not easy to use.\nSpeaker B: Because you're putting everything.\nSpeaker B: So, 6 more.\nSpeaker B: Yeah.\nSpeaker B: Yeah, I would go for the 6 too.\nSpeaker B: Okay.\nSpeaker F: Most folks can't.\nSpeaker B: Well, another question.\nSpeaker B: The remote control is durable.\nSpeaker B: I don't know if that's the correct word.\nSpeaker B: Definitely in casing.\nSpeaker B: And you have a hard place.\nSpeaker B: Because you have a hard place.\nSpeaker B: Yes, casing.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Yeah, because the battery is toasting is less forever.\nSpeaker F: True.\nSpeaker F: True.\nSpeaker F: And the casing is less.\nSpeaker F: The casing is hard plastic also.\nSpeaker B: That's the last forever.\nSpeaker B: Yeah.\nSpeaker B: If you don't drop it too much, it's a shoot last pretty long.\nSpeaker F: I would go for one.\nSpeaker B: Yeah.\nSpeaker B: But I think rubber comparators better.\nSpeaker B: So, I think a 2 is more appropriate.\nSpeaker B: Okay, yeah.\nSpeaker C: Wow.\nSpeaker B: Yeah.\nSpeaker B: Logo.\nSpeaker B: Okay, I would go for two.\nSpeaker B: The last one.\nSpeaker B: The remote control is a good example for companies model.\nSpeaker B: It puts a fashion in electronics.\nSpeaker C: No.\nSpeaker C: Now, let me put the electronics in the fan.\nSpeaker E: Let's turn around.\nSpeaker E: Yeah.\nSpeaker E: I would go for four.\nSpeaker E: Yeah.\nSpeaker E: Because we kind of try to, but we kind of filled with the budget that we have.\nSpeaker B: Yeah.\nSpeaker D: True.\nSpeaker B: Yeah, okay.\nSpeaker B: A four.\nSpeaker B: It's...\nSpeaker B: It goes.\nSpeaker B: It's not the best we could do, I guess.\nSpeaker B: The last do with the budget because it's not the bad idea we had.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker B: All right.\nNone: Yeah.\nSpeaker E: Right.\nSpeaker E: So.\nNone: So.\nSpeaker B: That's it.\nSpeaker B: If I understood it right, we have to count these numbers.\nSpeaker D: Yeah.\nSpeaker D: What?\nSpeaker D: Yeah.\nSpeaker B: Mm-hmm.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nNone: Yeah.\nSpeaker C: Oh.\nSpeaker B: All right.\nSpeaker B: Yeah.\nSpeaker B: We have to count them.\nSpeaker D: Count them.\nSpeaker D: Add them?\nSpeaker D: Yeah.\nSpeaker B: Just add them and then...\nSpeaker B: Define them.\nSpeaker B: Start calculating.\nSpeaker B: Yeah.\nSpeaker B: I all made it.\nSpeaker B: I all made it possible of positive questions.\nSpeaker B: So we can count it.\nSpeaker B: I mean, if you have reversed questions, you have reversed the skill.\nSpeaker B: Okay.\nSpeaker D: Yeah.\nSpeaker D: So four.\nSpeaker F: Did you make this question here?\nSpeaker B: Yeah.\nNone: Thanks.\nSpeaker B: Nice work.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker C: Three plus.\nSpeaker F: I wouldn't be able to do that that fast.\nSpeaker C: Plus one.\nSpeaker D: Plus five.\nSpeaker D: Plus five.\nSpeaker D: Whoa.\nSpeaker B: Oh.\nSpeaker B: Easy.\nSpeaker B: Question number four.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker B: Four.\nSpeaker C: Wait a second.\nSpeaker C: Oh.\nSpeaker C: It's going wrong.\nSpeaker C: How hard is it?\nSpeaker B: Okay.\nSpeaker B: Pretty difficult.\nSpeaker A: Yeah.\nSpeaker A: Just start over.\nSpeaker A: Oh, it has no...\nSpeaker B: There's no numpets.\nSpeaker B: You can...\nSpeaker C: No.\nSpeaker C: That's why it's...\nSpeaker C: It's a bit...\nSpeaker F: Just type in the digits.\nSpeaker F: They're all one digit numbers, right?\nSpeaker F: And then you can count them together.\nSpeaker B: Yeah.\nSpeaker B: You can just count them by half.\nSpeaker B: Okay.\nSpeaker A: That's number four.\nSpeaker B: Okay.\nSpeaker B: Three.\nSpeaker B: Four.\nNone: Nine.\nNone: Nine.\nSpeaker B: Yeah.\nSpeaker B: Eleven.\nSpeaker B: Fifteen.\nSpeaker B: Fifteen.\nSpeaker D: Fifteen.\nNone: Fifteen.\nSpeaker F: Fifteen.\nSpeaker F: Fifteen.\nSpeaker F: Yeah.\nSpeaker F: Seventeen.\nSpeaker F: Oh, sixteen.\nSpeaker B: Sixteen plus six.\nSpeaker B: Twenty-two.\nSpeaker B: Twenty-two.\nSpeaker B: How hard is this?\nSpeaker B: Twenty-two, yeah.\nSpeaker B: Twenty-four.\nSpeaker F: Twenty-four.\nSpeaker F: Twenty-eight.\nSpeaker F: Twenty-eight.\nSpeaker F: Twenty-eight.\nSpeaker D: Twenty-eight.\nSpeaker D: Oh.\nSpeaker C: That was the last one.\nSpeaker C: Twenty-eight.\nSpeaker C: Okay.\nSpeaker C: Twenty-eight.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Divided by nine.\nSpeaker B: That's a three, yeah.\nSpeaker F: The lower the score, the better, right?\nSpeaker B: Yeah.\nSpeaker B: Twenty-eight.\nSpeaker B: Divided by nine.\nSpeaker B: Divided by nine.\nSpeaker B: Makes.\nSpeaker B: Three points.\nSpeaker B: One, one, one, one, one.\nSpeaker B: So we're better than the average.\nSpeaker B: Yeah.\nSpeaker F: Are you sure this number actually tells us some things?\nSpeaker F: No.\nSpeaker F: I don't think so.\nSpeaker F: Because some questions are related to positive issues and some two negatives.\nSpeaker F: So if you give a true to a positive, it actually means that the lower, the better.\nSpeaker F: But if you give true to a negative question.\nSpeaker F: Yeah.\nSpeaker B: But there are no negative questions, I guess.\nSpeaker B: Yeah.\nSpeaker B: Good example.\nSpeaker B: Do you want to use?\nSpeaker B: That's good.\nSpeaker B: Easy to use.\nSpeaker B: Easy to use.\nSpeaker B: This way score is good.\nSpeaker B: Fancy look and feel.\nSpeaker B: That's good.\nSpeaker B: Yeah.\nSpeaker B: I think it was good because of a marketing requirement.\nSpeaker B: Yeah.\nSpeaker B: The motion function.\nSpeaker B: I guess you should do it.\nSpeaker B: It hides this function.\nSpeaker B: That was also a good thing.\nSpeaker B: Oh yeah.\nSpeaker B: And then matches the operator of the user.\nSpeaker B: That was also a good thing.\nSpeaker B: Okay.\nSpeaker B: So it was a positive question.\nSpeaker B: Yeah.\nSpeaker B: True.\nSpeaker B: By purpose.\nSpeaker B: So yes.\nSpeaker B: So it tells us something.\nSpeaker B: But the picture would be a lot different if we didn't have to drop those things.\nSpeaker B: Those things.\nSpeaker B: Yeah, definitely.\nSpeaker B: Definitely.\nSpeaker F: Because now it's just an average.\nSpeaker F: Yeah.\nSpeaker B: It's remote.\nSpeaker B: It's better than average.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker B: Because of the LCD screen.\nSpeaker F: Yeah.\nSpeaker F: So it looks and stuff.\nSpeaker F: It's still.\nSpeaker F: Yeah.\nSpeaker F: It's not really eye catching except for the color.\nSpeaker B: The color on the screen.\nSpeaker D: Okay.\nSpeaker D: So.\nSpeaker D: This we had.\nSpeaker D: This we had.\nSpeaker D: Pro-typrestation.\nSpeaker D: Pro-typrestation.\nSpeaker D: We dropped.\nSpeaker D: So.\nSpeaker D: The finance.\nSpeaker D: We looked.\nSpeaker D: We have.\nSpeaker D: Redesigned.\nSpeaker D: Not one.\nSpeaker D: That's.\nSpeaker D: Yeah.\nSpeaker B: We have to maybe.\nSpeaker B: We have to redesign it on this case.\nSpeaker B: No, it's right.\nSpeaker D: So you can just make one big LCD screen.\nSpeaker B: Well, some casing around it.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: We're still going to go for the fancy colors.\nSpeaker F: Yeah.\nSpeaker F: Are we going to stick with the green or are we going to do blue?\nSpeaker C: And you can just use this one and then over paint it with the blue color, the green.\nSpeaker D: The only on the screen.\nSpeaker D: The only on the screen.\nSpeaker D: The only on the screen.\nSpeaker D: The only on the screen.\nSpeaker D: One scroll button.\nSpeaker D: Oh, we have to delete it.\nSpeaker D: A single curved.\nSpeaker D: Yeah.\nSpeaker D: Oh, and you might want to add a infrared red.\nSpeaker B: Yeah.\nSpeaker B: Oh.\nSpeaker D: Oh.\nSpeaker D: For what?\nSpeaker D: That doesn't function.\nNone: Okay.\nSpeaker C: So.\nSpeaker B: So that's.\nSpeaker B: We have three other things.\nSpeaker D: Well, I have to make a report for every decision we made so far.\nSpeaker D: All right.\nSpeaker D: And I tried to catch up just before I received the Excel sheet.\nSpeaker D: Yeah.\nSpeaker B: So you made it start, right?\nSpeaker D: Yeah.\nSpeaker D: I'm somewhere.\nSpeaker D: Should you give it some time?\nSpeaker F: Yeah.\nSpeaker F: Yay.\nSpeaker C: Okay.\nSpeaker C: That's how do we make that a score?\nSpeaker C: Yeah.\nSpeaker C: Just one.\nSpeaker C: I'm going to make this away also.\nSpeaker C: And this.\nSpeaker C: And that's the infrared.\nSpeaker C: Oh, yeah.\nSpeaker C: Of course.\nSpeaker F: And pen.\nSpeaker B: Yeah.\nSpeaker B: For what?\nSpeaker B: Shape.\nSpeaker B: Current color.\nSpeaker B: Oh.\nSpeaker B: Oh.\nSpeaker B: Yeah.\nSpeaker F: We did a special color for the line width.\nSpeaker B: No.\nSpeaker F: That's it.\nSpeaker F: Ten.\nSpeaker D: It's going to be one.\nSpeaker D: How many pages?\nSpeaker D: I just took one for every.\nSpeaker D: All right.\nSpeaker B: You have done the first two.\nSpeaker B: Yeah.\nSpeaker F: Because you can, and the look and feel is more natural than this.\nSpeaker D: I think I have to make a.\nSpeaker D: Oh, that's a.\nSpeaker F: Oops.\nSpeaker B: Yeah.\nSpeaker F: Yes.\nSpeaker F: Basically.\nSpeaker F: I do.\nSpeaker C: We have to drop.\nSpeaker C: Yeah.\nSpeaker C: It smells very fashionable anymore.\nSpeaker C: It's okay.\nSpeaker C: It's really ugly.\nSpeaker B: Yeah.\nSpeaker B: Where did we start?\nSpeaker B: Which fries?\nSpeaker B: 26.5.\nSpeaker B: Or.\nNone: Or 26.\nSpeaker B: Something.\nSpeaker B: Like that.\nSpeaker B: Yeah.\nSpeaker C: That looks a little bit more.\nSpeaker C: And it's a special color for it.\nSpeaker C: So we can make it special.\nSpeaker F: What do you mean?\nSpeaker F: Like an olive color and this one.\nSpeaker C: Or spickle, I don't know.\nSpeaker F: Spickle.\nSpeaker F: Yeah.\nSpeaker F: Can we do a print?\nSpeaker C: I'm not sure.\nSpeaker F: I think we have to choose.\nSpeaker D: If you see the up scenes.\nNone: Okay.\nSpeaker F: Special color.\nSpeaker F: We do have special color.\nSpeaker F: Does it mean?\nSpeaker C: It's already a special color.\nSpeaker C: Yeah.\nSpeaker C: It's very special.\nSpeaker E: Shall we just give it some dots to make it look pretty?\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker B: Some big dots.\nSpeaker C: Purple.\nSpeaker C: That's trendy.\nSpeaker C: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nNone: Oh no.\nNone: My room.\nSpeaker A: That's acne.\nSpeaker B: Yeah.\nSpeaker B: Oh.\nSpeaker F: That's why we asked that button.\nSpeaker C: Oh yeah.\nSpeaker F: It's so cute.\nSpeaker C: Oh.\nSpeaker F: What?\nSpeaker B: Whoa.\nSpeaker B: Oh no.\nSpeaker B: Just got.\nSpeaker B: Control Z.\nSpeaker C: Oh no.\nSpeaker F: Is that that?\nSpeaker C: Yeah.\nSpeaker C: Oh no.\nSpeaker B: Yeah.\nSpeaker F: How the...\nSpeaker F: Wait, did we do that?\nSpeaker B: Just dual screen.\nSpeaker C: No.\nSpeaker C: Can we delete it?\nSpeaker C: Yes, we delete it.\nSpeaker F: We can try.\nSpeaker F: It doesn't respond also to undo.\nSpeaker F: It looks like it's...\nSpeaker A: Cresh.\nSpeaker F: No, I guess.\nSpeaker B: Very nice.\nSpeaker F: Why do you want some more dots?\nSpeaker D: Yeah, over here.\nSpeaker D: You can even draw anymore.\nSpeaker D: Even children.\nSpeaker B: You push the button or something.\nSpeaker D: Can you just push the button and then...\nSpeaker C: Oh, there's a select button.\nSpeaker C: It's...\nSpeaker F: Okay.\nSpeaker F: It's not the prettiest.\nSpeaker C: It's not so random.\nSpeaker C: It's okay.\nSpeaker C: I actually think that the R.\nSpeaker B: Yeah, the R and another R.\nSpeaker B: Yeah, okay.\nSpeaker B: It's called the real remote, right?\nSpeaker C: Yeah.\nSpeaker B: Maybe it can say that.\nSpeaker B: The real remote.\nSpeaker D: Yeah, just on the LCD display.\nSpeaker D: Yeah.\nSpeaker E: Welcome.\nSpeaker E: We can make a logo.\nSpeaker E: Yeah.\nSpeaker E: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Put it like this shape.\nSpeaker C: It's designed by...\nSpeaker F: Something like that.\nSpeaker F: I mean, it's not too...\nSpeaker F: It's not their logo, is it?\nSpeaker C: Yeah.\nSpeaker C: Do they have a logo?\nSpeaker C: Oh, here.\nSpeaker C: This is the logo.\nSpeaker C: Yeah.\nSpeaker C: Two R's.\nSpeaker D: One.\nSpeaker D: And then we just reuse that.\nSpeaker B: You can copy and paste it, picture, if you want.\nSpeaker F: So, we do the logo in black or not?\nSpeaker B: Yeah.\nNone: Yeah.\nSpeaker B: What's that?\nSpeaker B: Looks more like campfire.\nSpeaker F: Are you dissing my drawing?\nSpeaker F: This one?\nSpeaker B: Yeah.\nSpeaker F: It looks like a ribbon.\nSpeaker B: Yeah.\nSpeaker C: Okay.\nSpeaker F: So, we want to draw...\nSpeaker F: This is actually quite fun.\nSpeaker F: Do we need to do anything?\nSpeaker F: I hear you people are typing.\nSpeaker B: Typing the report.\nSpeaker F: Okay.\nSpeaker B: I don't see any new messages.\nSpeaker B: Oh, luckily.\nSpeaker B: Luckily.\nSpeaker C: Is the last assignment or?\nSpeaker C: Yeah.\nSpeaker C: It's final.\nSpeaker C: Yeah.\nSpeaker F: It's almost four.\nSpeaker B: What time do we have to deliver the report?\nSpeaker B: Four o'clock?\nSpeaker F: That's four, yeah.\nSpeaker B: Before that.\nNone: Okay.\nSpeaker F: Just compare it.\nSpeaker F: This is really bizarre.\nSpeaker F: It looks like there's a...\nSpeaker D: It looks like a butterfly.\nSpeaker D: It's somewhere...\nSpeaker D: It's inside.\nSpeaker D: It's inside.\nSpeaker B: No, it's on the beamer, I guess.\nSpeaker F: No, but it's not a beamer.\nSpeaker F: It's a normal TV screen.\nSpeaker F: Kind of thing.\nSpeaker C: Yeah, it's somewhere in here.\nSpeaker C: It's somewhere in here.\nSpeaker C: So, you have to...\nSpeaker D: Save copy.\nSpeaker B: Yeah.\nSpeaker B: Strange.\nSpeaker B: This is something that's projection from behind, I guess.\nSpeaker A: There's a little sister.\nSpeaker F: Oh.\nSpeaker F: Yeah, there is some kind of projection, I think.\nSpeaker B: Yeah, it's a beamer.\nSpeaker B: But then we're in a...\nSpeaker B: It's a mirror.\nSpeaker B: Mirror, yeah.\nSpeaker B: So, it looks like a big screen, but in fact it isn't.\nSpeaker F: So, are we going to change anything to this?\nSpeaker F: I mean, as well.\nSpeaker C: Well, it's a single curve.\nSpeaker C: So...\nSpeaker C: This is going to be flat.\nSpeaker C: Yeah, exactly.\nSpeaker C: Okay, it doesn't matter.\nSpeaker B: But it's better to have in the front this kind of shape because it looks nice.\nSpeaker B: I mean, you see more of this than of that.\nSpeaker F: Yeah.\nSpeaker B: Yeah, more like that.\nSpeaker F: Yeah.\nSpeaker F: And this is also going to be...\nSpeaker B: That's very ideal.\nSpeaker F: No.\nSpeaker F: It's sometimes then all of a sudden it does work.\nSpeaker B: Yeah.\nSpeaker B: What's this?\nSpeaker C: That's the detector for the...\nSpeaker C: Okay.\nSpeaker B: I don't see the detector over there.\nSpeaker F: I think you only need two points or not.\nSpeaker B: I thought it was a kind of thing to put it on and then draw right lines or something.\nSpeaker F: It's not working because it's more slanted.\nSpeaker E: Yeah.\nSpeaker B: Or just messed it up.\nSpeaker F: Oh, yeah.\nSpeaker F: Yeah, it matters for the aim of this thing.\nSpeaker B: Yeah, well, it wasn't good.\nSpeaker C: You have to make it 90 degrees.\nSpeaker B: Yeah, it has to touch the corners, I guess.\nSpeaker B: But this one wasn't good because if I was drawing here, I drew a line and then it came over here.\nSpeaker D: No, you probably have to recalibrate.\nSpeaker D: Five minutes to the end of the meeting.\nSpeaker D: Oh, we're out.\nSpeaker D: And the recalibration is done using this icon here.\nSpeaker C: Can we get to that icon?\nSpeaker C: Oh, it's so far.\nSpeaker D: Yeah, well, I did.\nSpeaker B: It's okay.\nSpeaker B: It's working again.\nSpeaker B: It's better than before.\nSpeaker B: Yeah.\nSpeaker C: You do it.\nSpeaker B: Yeah, that's improved pretty much.\nSpeaker F: It's only a bit like to that side, but that is that one, by the way.\nSpeaker B: It's better.\nSpeaker F: It makes the angle either like this.\nSpeaker F: So if I change this, it will go there.\nSpeaker F: If I change that, it will go there.\nSpeaker B: It's better than it was, I guess.\nSpeaker F: I will take this away because it looks silly.\nSpeaker B: Yeah, it works pretty well.\nSpeaker B: Five minutes before the meeting is over and then we have to present.\nSpeaker D: I don't know if you have to present because I didn't receive any information about that.\nSpeaker F: All right.\nSpeaker F: Maybe we will.\nSpeaker B: After these five minutes, you have to.\nSpeaker D: I still have 10 minutes.\nSpeaker B: What's this anyway?\nSpeaker B: All right.\nSpeaker B: After that five minutes, you have to finish it.\nSpeaker B: It looks like candle wax.\nSpeaker B: Yeah.\nSpeaker B: All right.\nSpeaker D: You can just relax with coffee or chill.\nSpeaker D: No, I don't have beer.\nSpeaker D: You can celebrate.\nSpeaker D: You can just, if you finish my presentation.\nNone: Yeah.\nSpeaker D: The presentation is still open.\nSpeaker D: So I should finish that and you'll see.\nSpeaker D: Yeah.\nSpeaker D: Next slide.\nSpeaker B: Oh, yeah.\nSpeaker D: We have to do the project evaluation.\nSpeaker D: Just do that quickly.\nSpeaker D: Yeah.\nSpeaker B: How do you do it?\nSpeaker D: Well, basically what that says, we discuss it.\nSpeaker D: So how are the project process go?\nSpeaker D: Are you really pleased with the process as well?\nSpeaker F: Do you mean the interaction between us or?\nSpeaker D: The interaction and the steps he followed.\nSpeaker F: Yeah, well, at first I was really stressed because it went a bit fast.\nSpeaker F: But then as I knew the tempo that I had to be on the second time, I think I did a bit better.\nSpeaker F: Yeah.\nSpeaker F: In the third time, yeah.\nSpeaker B: And we move more to working together as a team because at first you make your individual contribution and then come here and you have no idea what the others have to make.\nSpeaker B: Yeah.\nSpeaker B: And then finally you have some idea, okay, this is what we're going to make together and okay, I will arrange this and you will arrange that and you have a kind of idea what you can expect on the next half hour when you have to work on your own.\nSpeaker B: So the process, I mean the interaction between us became better and better, I guess.\nSpeaker C: Especially after the first meeting.\nSpeaker B: Yeah, especially if you saw the largest difference from the first to the second meeting, I guess.\nSpeaker D: Okay, and what's that due to my leadership?\nSpeaker C: Yeah, yeah.\nSpeaker F: Well, you did become more assertive the second time round to that.\nSpeaker F: Yeah, yeah.\nSpeaker F: You were more in charge kind of thing.\nSpeaker D: Okay.\nSpeaker D: Was there enough room for creativity?\nSpeaker F: I guess so.\nSpeaker F: Yeah.\nSpeaker C: I mean, only the financial parts.\nSpeaker B: The limited afterwards, but if you don't take that into account there's plenty of room for creativity also, not only the drawing and the making of thing itself, but also in explaining it to the other people by means of the board and the power point and word and stuff like\nSpeaker D: that. So about the board and stuff?\nSpeaker D: Was it helpful?\nSpeaker B: I think in essence the digital pen is better than the smart board because it just works better.\nSpeaker B: I mean, I've made several notes just to test it and just put the pen into it and then it came onto my screen and it looked perfectly so it's better device than the screen.\nSpeaker B: Maybe the screen is useful in essence, but it doesn't work that well.\nSpeaker F: But the pen is more intuitive because we're all used to writing with pen and as I said, I have no idea how power point works.\nSpeaker F: So it will take me half a day to be able to make a normal power point presentation.\nSpeaker F: But once I get to know the program, probably, I mean, it looks better, you know, or something like that, you can give it a kind of a home style like we have the logo and everything.\nSpeaker F: Yeah.\nSpeaker F: I don't know.\nSpeaker D: I don't know.\nSpeaker B: Morning.\nSpeaker B: Finish meeting now.\nSpeaker D: Okay.\nSpeaker D: Are there any new ideas about this?\nSpeaker D: Oh, I think it's really useful.\nSpeaker B: It's especially useful, I guess, to evaluate designs and graphical stuff and the whole screen and stuff like that.\nSpeaker C: I think the power point is too limited.\nSpeaker C: You can draw easy in power point.\nSpeaker C: The drawing board has to be integrated into power points so that you can just easily...\nSpeaker F: Yeah.\nSpeaker F: The problem with power point for me is that you have to put in the pre-made chunks as well.\nSpeaker F: You know, you can draw something on the sketchboard and then take it there or whatever.\nSpeaker B: Yeah.\nSpeaker B: But it's useful to show something to a small audience and then to...\nSpeaker C: It's okay, but...\nSpeaker B: These kind of screens, these sides are useful to discuss things.\nSpeaker B: Yes.\nSpeaker B: Check your email.\nSpeaker D: You should enter our questionnaire.\nSpeaker D: T-shirt.\nSpeaker D: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker B: So we, when we wait quite a minute and then for the heat to reach it, the everyday through\nNone: I understand because there are three questions left. This is number eight.\nSpeaker D: I'm going to go to the island on the coast.\nSpeaker F: And then it's going to be the planet's view.\nSpeaker D: All right.\nSpeaker D: OK.\nSpeaker D: I'm going to go to the island.\nSpeaker F: OK.\nSpeaker B: All right.\nSpeaker B: Yes, boss.\nSpeaker F: I'll just see you in a second.\nSpeaker B: I'll see you soon.\nSpeaker B: Up, up.\nSpeaker D: Yeah.\nSpeaker F: We have to pull it this way.\nSpeaker A: What?\nNone: It's going to hit you.\n"}, {"meeting_id": "ES2014d", "summary": "User Interface and Industrial Designer gave a presentation about the prototype of the remote control based on the team's discussion in the previous meeting. Then the team evaluated the prototype against various criteria they formulated. After that, the team discussed adjusting the design of the remote in order to fit into the budget. Finally, the team evaluated the whole process of the project.", "dialogue": "Speaker B: I'm not seeing that.\nSpeaker D: I have to open it up last year.\nSpeaker B: I already exist.\nSpeaker B: Do you want the places to come up now?\nSpeaker B: I should try to transfer it to my documents, but...\nSpeaker D: I have to close that window because that's the save one.\nSpeaker B: I have to spread it.\nSpeaker B: I'll just redo it.\nSpeaker B: I'll just redo it.\nSpeaker A: I'll put a tight bedding in.\nSpeaker A: It's almost curved, like up to the main display as well.\nSpeaker A: The little line at the bottom in the middle is the panel that you pull down, and the extra flash of the buttons are below that panel.\nSpeaker A: The stick on the bottom is the company no more.\nSpeaker C: I have an apple slash chain at the top.\nSpeaker D: I'll just have my face.\nSpeaker A: It's a bit more fun as me.\nSpeaker A: It's not really a defender.\nSpeaker A: You think Apple, you can completely attack Apple?\nSpeaker D: We might get a copy right here.\nSpeaker A: It's a phone.\nSpeaker C: I'll ask the neat.\nSpeaker A: It might...\nSpeaker A: It'll probably have to be labeled.\nSpeaker A: I couldn't have anything smaller to try.\nSpeaker C: I would just choose simple shapes for all of them.\nSpeaker A: The important ones are the volume ones.\nSpeaker A: We'll move them a bit bigger.\nSpeaker C: We'll separate off and color the volume-related buttons from the channel-based buttons.\nSpeaker C: You've got the volume and volume shown pretty like that.\nSpeaker B: All these things have cost implications.\nSpeaker B: When I've done my thing on cost, I had assumed that the only button that would be a different color would be the red apple button.\nSpeaker B: However, I've never thought I could have said that.\nSpeaker B: Yes, but there would be a cost implication on that.\nSpeaker B: I suppose that whether one of the colors would be open to debate.\nSpeaker B: I suppose.\nSpeaker A: What color would it be?\nSpeaker A: It wouldn't be too important, but we didn't have any white paper.\nSpeaker B: An important consideration.\nSpeaker B: Right, so the second one underneath would be the idea for the side-way.\nSpeaker B: Right, so we've got detailed design meeting.\nSpeaker B: Right.\nSpeaker B: So we've got prototype presentation, which we've just done.\nSpeaker B: Evaluation criteria and finance.\nSpeaker B: I guess we have to evaluate if that meets the various aspects that we're looking for from our previous meeting.\nSpeaker B: So other than the fact that it doesn't have the same layer, but obviously it would.\nSpeaker B: But other than that, we've got the red apple, we've got the buttons.\nSpeaker B: The only thing that's really changed is the coloration of the buttons.\nSpeaker B: But after the evaluation criteria is the finance.\nSpeaker B: And the cost allocation, the only scenario out this is that the cost is probably kind of important.\nSpeaker B: So, and then the production evaluation as to how easy that would be to manufacture.\nSpeaker B: And whether it would.\nSpeaker D: So if you had to do a presentation, we just become a prototype.\nSpeaker B: Right, so as far as the finance of it, whether you consider it, we would make sure that the cost, the production cost, because remember that was one of the first considerations, was to be in under 1250 or two and a half, 12 and a half euros.\nSpeaker B: So, there's no redesign.\nSpeaker B: So that should.\nSpeaker B: Right, so it seems to me that the thing that I have to do is.\nSpeaker C: Is quickly find that.\nSpeaker B: Okay, so by the fact that we've got.\nSpeaker B: Simple chip and the kinetic energy source.\nSpeaker B: We've got a single curved.\nSpeaker B: Case we've got a rubber.\nSpeaker B: Case materials supplements.\nSpeaker B: So we had decided that we're having rubber buttons and.\nSpeaker B: Okay.\nSpeaker B: But the button supplements well originally I thought there would just be one in there because it was the one red apple.\nSpeaker B: But so the so the real.\nSpeaker B: The real question that we come in.\nSpeaker D: Do you make all the buttons.\nSpeaker B: So we've got one special button for which was the apple everything else is going to be a standard.\nSpeaker B: We've got special material rubber wood titanium etc.\nSpeaker B: So I was originally I was thinking what rubber wasn't special but according to this maybe it is.\nSpeaker B: So.\nSpeaker B: And I mean effectively we've got 16 buttons that we're going to have on there.\nSpeaker D: Yeah, I think you just do one.\nSpeaker B: I don't know.\nSpeaker B: Is the sort of answer is that meant to be all 16 buttons and therefore.\nSpeaker B: I mean what's the up the option was maybe not to have rubber buttons.\nSpeaker B: But just to have the one that was.\nSpeaker B: Soft and spongy and therefore.\nSpeaker B: Whereas it would be the special color would be for the so you would only have the one special button that was rubber whereas the rest would be hard plastic.\nSpeaker D: I think the button supplements is just a supplement for all the buttons to be made in a different material.\nSpeaker B: Every design change is.\nSpeaker B: I don't know.\nSpeaker B: Okay.\nSpeaker B: If we just had all the buttons as standard except for the one red apple then that would take care of that I guess.\nSpeaker B: We'd have one special color and one special button for.\nSpeaker B: And when.\nSpeaker B: I plugged that in last time.\nSpeaker B: Remember it has to be under 12 and a half.\nSpeaker B: As far as I know that.\nSpeaker B: That took care of the of the various supplements.\nSpeaker B: And if who had.\nSpeaker D: It's on the bottom.\nSpeaker B: Okay, so but the point would be that if we.\nSpeaker B: If we just did special, sorry you were saying that it would be that one that you would put in one there.\nSpeaker C: Okay.\nSpeaker B: So it was no it's 9.7 I've got.\nSpeaker B: Okay, so that would.\nSpeaker B: That would work out fine if.\nSpeaker B: Assuming your correction or assuming that one change covers all the buttons and that would be fine.\nSpeaker C: And there's nothing else as far as I can see that we would plan to.\nSpeaker C: And if we.\nSpeaker C: Okay, that gives us 12.7.\nSpeaker B: But remember that the idea was to keep it the color of the.\nSpeaker A: So just take out the special color for the apple and.\nSpeaker C: And go for a battery instead that would give you one less.\nSpeaker B: You reckon that I mean the thing is that you wanted to appeal to people and.\nSpeaker B: Not have to replace batteries.\nSpeaker C: Assist this is the whole technology type thing.\nSpeaker C: You're saying in the market that people like to have interesting gadgets and.\nSpeaker C: Whether they were bigger.\nSpeaker C: The sample sensor the sample speaker.\nSpeaker C: Or its recognition to be sort of a worthwhile thing to have.\nSpeaker C: And it's not the batteries or whether they were not.\nSpeaker C: You know what I mean?\nSpeaker C: The problem was the battery running out and losing the.\nSpeaker C: Yeah.\nSpeaker C: Which do you think or which do we think.\nSpeaker B: Is the more important of the options in a sense of the moment we've got a total which we need to reduce down by 1.2.\nSpeaker B: At least remember that was a minimum requirement the other option if we're planning on just going for something cheap and cheerful.\nSpeaker B: Would be to.\nSpeaker D: Make it an originally we're going to make it simple.\nSpeaker D: Show you evaluate the prototype as we've got it now.\nSpeaker D: First.\nSpeaker D: And then to make decisions about what needs to be changed.\nSpeaker D: Makes sense.\nSpeaker D: Have a little thing.\nSpeaker D: So.\nSpeaker B: Sorry.\nSpeaker B: Back up.\nSpeaker D: Presentation.\nSpeaker E: Right.\nSpeaker A: I don't think.\nSpeaker A: Okay.\nSpeaker B: But remember the main the only reason we were planning on having device recognition was so that you could find the remote if it got lost.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: This is about the evaluation criteria that we use for the prototype we've got here.\nSpeaker D: So the method is that design team makes a prototype.\nSpeaker D: We evaluate the prototype against some criteria that we formulated.\nSpeaker D: And those ones are going to be in response to sort of market research.\nSpeaker D: And also finance.\nSpeaker D: I guess.\nSpeaker D: And do that on a scale from true being one.\nSpeaker D: A false being seven.\nSpeaker D: So it's now the June or false.\nSpeaker D: And that's four.\nSpeaker D: So.\nSpeaker D: I've got a set of criteria just based on the marketing that we need to add in a financial one as well.\nSpeaker D: At the end.\nSpeaker D: So.\nSpeaker D: We have to say whether it's true or false that the product looks and feels fancy.\nSpeaker D: Whether the product demonstrates technical innovation.\nSpeaker D: Whether it's easy to use.\nSpeaker D: Whether it's incorporating sort of the fashion elements to try to fire.\nSpeaker D: And whether it's sort of recognizable real reaction product.\nSpeaker D: And if you go up onto the whiteboard and do this apparently.\nNone: So.\nNone: I'll go up here.\nSpeaker D: Right.\nSpeaker D: So the first one is.\nSpeaker D: Does the product look and feel fancy?\nSpeaker D: So if you do this with one.\nNone: So.\nSpeaker D: Yeah.\nSpeaker C: We have a single curve.\nSpeaker C: And maybe like the.\nSpeaker C: Yeah.\nSpeaker C: The other part is.\nSpeaker C: Okay.\nSpeaker C: And we have the rubber kind of sponge.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: At the time.\nSpeaker C: Yeah.\nSpeaker C: So that's some fancy.\nSpeaker A: So I'd really give it there.\nSpeaker D: I think what is one false or is one true.\nSpeaker D: I forgot.\nSpeaker D: One's true.\nSpeaker D: And okay.\nSpeaker D: Seven is full.\nSpeaker D: Four is neutral.\nSpeaker D: Okay.\nSpeaker D: So.\nSpeaker C: Maybe two.\nSpeaker A: Yeah.\nSpeaker A: Because we haven't got the double curve.\nSpeaker A: So we'll call.\nSpeaker A: Go for one.\nSpeaker A: Yeah.\nSpeaker A: Almost everything.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: I've got underneath.\nSpeaker D: So I've got some more space.\nSpeaker D: So false is seven.\nSpeaker D: True is one.\nSpeaker D: So say about two.\nSpeaker D: Yeah.\nSpeaker B: Okay.\nSpeaker B: We're not maybe near three.\nSpeaker D: Okay.\nSpeaker D: Well, it's just that.\nSpeaker B: I'm saying to me.\nSpeaker B: Remember that when you look down, we've got solar power.\nSpeaker B: But at various other things you could have.\nSpeaker B: And we're not going for these options.\nSpeaker D: This is just like the look.\nSpeaker D: Does it look fancy rather than functional?\nSpeaker D: So.\nSpeaker E: Yeah.\nSpeaker D: So should we go for a two on that?\nSpeaker D: Okay.\nSpeaker D: I mean, how much does the product demonstrate technical innovation?\nSpeaker C: We're deciding between the kinetic power or the speech recognition.\nSpeaker C: We had even those for our budget.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker D: So what about the prototype as it is?\nSpeaker D: We've got the speech recognition on it.\nSpeaker C: We're not the kinetic.\nSpeaker D: But not the kinetic power.\nSpeaker B: No, because you can't afford that.\nSpeaker B: We're too bad out to.\nSpeaker C: Yeah.\nSpeaker C: We can't afford both.\nSpeaker D: All right.\nSpeaker E: Didn't know.\nSpeaker D: So it doesn't, it's pretty, the prototype as it is, isn't sort of fulfilling the...\nSpeaker C: No, no.\nSpeaker C: It's made by the choice.\nSpeaker C: It's got something that it has in that.\nSpeaker C: Okay.\nSpeaker B: Well, we're in it.\nSpeaker B: In 13.7, we do have kinetic.\nSpeaker B: The problem is we have to reduce down from there to get it down to 12.5.\nSpeaker B: And one way of doing that would be to take out the kinetic.\nSpeaker B: So it's very much dependent on what you do with your options.\nSpeaker B: And if you're definitely going for the sample sensor and sample speaker, then...\nSpeaker B: Because the other functions we've got in are more...\nSpeaker B: The special material, the rubber, wood titanium, et cetera.\nSpeaker B: If you go for that, that's at the high end because that's.6.\nSpeaker B: Whereas down it, just special color is.2.\nSpeaker B: And you're trying to lose 1.2.\nSpeaker B: So it seems to me that if you're going for the sample sensor speaker, you're basically then having to go for the cheaper options on everything else.\nSpeaker B: And the simple way to do it would be to have a battery, to have your sample sensor speaker.\nSpeaker B: And then you're looking to take out.2, which would be...\nSpeaker B: Come from the button supplements category.\nSpeaker D: I'm just going to check exactly where it's said in the email for...\nSpeaker A: How much of the difference would it make if you made the case in plastic?\nSpeaker A: Because we did say that we don't want to follow the fashion too much.\nSpeaker A: But it's a rubber. That might be strange, you know.\nSpeaker B: So you could have an interface type, well, plastic rather than rubber.\nSpeaker B: That would make the cinema difference.\nSpeaker B: You could either... If you have it in plastic rather than rubber, then that would enable you to get.\nSpeaker B: You could keep kinetic then. You could keep your sample sensor.\nSpeaker B: You were looking to take out.2.\nSpeaker B: Or you could fiddle that down your special form at the bottom or your special colour at the bottom.\nSpeaker B: And that would enable you to do it.\nSpeaker C: So as soon as we make it slightly more fancy, then we use poison innovation.\nSpeaker C: If we make it more innovative, then we use poison.\nSpeaker C: Yeah.\nSpeaker D: I just read the email again and it sort of says, it's evaluated the design sort of as it is I think.\nSpeaker D: So I think we need to think about finance after we've sort of evaluated that design.\nSpeaker D: I don't know whether we've done it in the wrong order or something.\nSpeaker B: Well, okay.\nSpeaker B: But...\nSpeaker A: I suppose it's rubber as it is.\nSpeaker B: It's rubber as it is, yes.\nSpeaker D: So...\nSpeaker D: I think this needs to go up a bit or something because we've got both the...\nSpeaker B: We've got... We've got 13.7 and we've got it in at the moment.\nSpeaker B: And basically we're going to reduce down from that.\nSpeaker B: But the current one you'd see would be fancy, would be two.\nSpeaker D: Yeah.\nSpeaker D: I'm for innovation.\nSpeaker D: So we've got the speech, the speech thing.\nSpeaker D: And...\nSpeaker B: I would have said about a two as well.\nSpeaker D: Okay.\nSpeaker A: I'm happy to do that.\nSpeaker D: Okay.\nSpeaker D: Now the next one is...\nSpeaker D: I'll have to get it back up now.\nSpeaker D: The next one...\nSpeaker B: Well, I can just sing about easy to use.\nSpeaker B: I would have said yes.\nSpeaker B: I would have said yes.\nSpeaker B: I would have go for a one on that at this point and two.\nSpeaker A: Yeah.\nNone: Okay.\nSpeaker A: Yeah, I would say so as well.\nNone: Okay.\nSpeaker B: Yeah, I would say so as well.\nSpeaker B: And incorporates elements of fashion to attract buyer.\nSpeaker B: Well, it certainly has some.\nSpeaker A: Yeah, I thought the chat was...\nSpeaker A: Yeah.\nSpeaker B: Yep.\nSpeaker D: I thought it was three, maybe.\nSpeaker B: I don't know.\nSpeaker C: Yeah, I'd say.\nSpeaker C: Yeah, it was doing it quite well.\nSpeaker C: Yeah.\nSpeaker C: And we can have to do some of these for the moment as it's kind of...\nSpeaker B: Yeah, I would have said two would seem reasonable.\nSpeaker B: The products are recognizable, real...\nSpeaker B: Yeah, reaction product.\nSpeaker C: The business has for using all of this...\nSpeaker C: All of this product, all of this money.\nSpeaker C: It's got a pretty big label on the bottom saying...\nSpeaker C: Yeah.\nSpeaker D:...that's what it comes from.\nSpeaker D: So this is about sort of the corporate image of like new sort of sleek technology.\nSpeaker D: And that sort of thing as well as having the logo on and all that.\nSpeaker D: So it's sort of...\nSpeaker D: Is it sort of a recognizable product?\nSpeaker D: Does it fit in with the art of other products which are sort of coffee makers and spacecraft?\nSpeaker D: Yeah.\nSpeaker A: What's about the same speech?\nSpeaker A: Speech and the copyright issue?\nSpeaker C: As soon as it's kind of a sponge and roberiness is maybe the more kind of comfortable thing can sleep and...\nSpeaker C: Yeah.\nSpeaker C: That's just what you like to do.\nSpeaker B: Okay.\nSpeaker B: So we're going for...\nSpeaker C: Two...\nSpeaker C: Three...\nSpeaker A: Well, the logo would be more recognizable on the actual thing.\nSpeaker A: It's just that the pen wouldn't be right on that.\nSpeaker A: Sure.\nSpeaker A: But I think the logo is definitely a recognizable and it does have...\nSpeaker A: Actually, it's that other product to do.\nSpeaker A: Right.\nSpeaker D: Okay.\nSpeaker E: Three...\nSpeaker B: Two.\nSpeaker B: Right.\nSpeaker B: That's that decision.\nSpeaker B: Right.\nSpeaker B: So...\nSpeaker B: Right.\nSpeaker B: So we're now on to changing it to get it to fit in with the budget requirements.\nSpeaker B: And then...\nSpeaker B: Yeah.\nSpeaker A: What is...\nSpeaker A: What all of them does mean?\nSpeaker A: Do I add them up in the art style or anything?\nSpeaker A: Yeah.\nSpeaker D: I think we sort of...\nSpeaker D: Add them up.\nSpeaker D: So sort of an average is going to be...\nSpeaker B: About a two.\nSpeaker B: Two.\nSpeaker D: One point eight, isn't it?\nSpeaker D: I think anyway.\nSpeaker D: So yeah, pretty close to two.\nSpeaker D: Yeah.\nSpeaker D: So I mean, it's pretty good at the moment, but it's going to get worse, isn't it?\nSpeaker D: Yep.\nSpeaker D: Try and make sure it doesn't get...\nSpeaker B: Two...\nSpeaker C: Two balance.\nSpeaker C: It's a 13.7.\nSpeaker C: It's a 13.7.\nSpeaker C: It's a 13.7.\nSpeaker C: It's a 13.7.\nSpeaker C: Yes.\nSpeaker C: Yep.\nSpeaker C: It's a...\nSpeaker B: Well, you can lose 1.2.\nSpeaker B: And still meet.\nSpeaker B: Do you want to...\nSpeaker B: The requirement...\nSpeaker D: Get into your sort and get up there.\nSpeaker D: Okay.\nSpeaker B: Well, I put it back on.\nSpeaker C: I guess this is the last stage, but it's also about the final.\nSpeaker C: I'm not sure.\nSpeaker D: So we'll probably have to leave it there.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: So we're not thought so.\nSpeaker C: Yeah.\nSpeaker C: Maybe just...\nSpeaker C: See how it lays together.\nSpeaker C: Okay.\nSpeaker B: So what was it?\nSpeaker B: Control...\nSpeaker B: F8, wasn't it?\nSpeaker B: It's a...\nSpeaker B: Oh.\nSpeaker B: Oh.\nSpeaker B: Oh.\nSpeaker B: Okay.\nSpeaker B: Right.\nSpeaker B: Okay.\nSpeaker B: So you can see there that the...\nSpeaker B: If you want to keep kinetic, right, you've got a choice there.\nSpeaker B: Going down to battery, which would save you 1.\nSpeaker B: You've got...\nSpeaker B: The sample sensor and sample speaker, which is your big item.\nSpeaker C: Why should we just have a plastic case that we lose 2?\nSpeaker C: That's right.\nSpeaker C: Which gets us...\nSpeaker C: Which gets you...\nSpeaker C: Like within the budget range.\nSpeaker C: Yes.\nSpeaker A: Yeah.\nSpeaker A: Because that was just a trend.\nSpeaker A: And we do have the rubber buttons anyway.\nSpeaker C: So that's the one.\nSpeaker C: Yeah.\nSpeaker C: And then we can make sure we definitely have all the buttons.\nSpeaker C: All right.\nSpeaker D: Different colours.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I was thinking that because maybe the sort of rubber case is a bit less in sort of the corporate identity than the sort of the sleek, simplistic case.\nSpeaker E: Yeah.\nNone: Okay.\nSpeaker C: Now we're going to have some more time for the innovations.\nSpeaker C: We lose a little bit on...\nSpeaker C: Maybe...\nSpeaker B: So we're going for plastic, yes?\nSpeaker D: We lose a little bit on the fashion, yeah.\nSpeaker D: Yep.\nSpeaker C: And now we know that we have plastic that's within the budget.\nSpeaker C: Do we actually have...\nSpeaker C: Do we just have one special...\nSpeaker C: Only these special forms down here?\nSpeaker B: Well, there was debate as to how you would count them.\nSpeaker B: You got special colour.\nSpeaker B: Well, originally I was assuming we had the red apple and therefore that was the special colour.\nSpeaker A: Yeah, I think we should just imagine white buttons.\nSpeaker C: And we've got old buttons for another...\nSpeaker C: Yeah.\nSpeaker C: We've got 0.8 lives, we've got a lot.\nSpeaker B: Well, we've got special form.\nSpeaker B: Now that would be one button and the question was, was that all buttons or is that just one button?\nSpeaker B: Because our plan really was to have one button only with the special.\nSpeaker B: Yeah.\nSpeaker B: So no matter how you look at that, that would be the same.\nSpeaker B: The other thing would then be special material rubberwood titanium.\nSpeaker C: I think maybe the special colour we've got three now, just because of the volume buttons, the diff and I guess we have one more sharp blue at the moment.\nSpeaker B: Alright.\nSpeaker B: So a special colour you want three in there.\nSpeaker C: Yeah, I think we should...\nSpeaker C: Yeah, that's probably fine for the plan.\nSpeaker D: Okay.\nSpeaker D: You just need two for special colours because it's just two supplements.\nSpeaker D: Yeah, one original colour and then two supplements.\nSpeaker D: I think maybe.\nSpeaker B: That makes sense.\nSpeaker D: Yeah.\nSpeaker D: Yeah, that's probably it.\nSpeaker D: So we only need two for that.\nSpeaker D: Okay.\nSpeaker D: These are molded, the one colour is going to be molded out of a piece of plastic rubberwood.\nSpeaker B: Okay.\nSpeaker B: Yeah, I'd agree with that, I think.\nSpeaker B: So special colours too.\nSpeaker B: We've got special form.\nSpeaker B: Is the one apple, the rest are all standard although you could argue that should maybe be...\nSpeaker B: You were making these buttons down the bottom, I was presuming bigger than the other ones or were you.\nSpeaker B: Was that the idea?\nSpeaker B: Yeah, the only ones.\nSpeaker C: Maybe that'd be a second step.\nSpeaker C: Yeah, but it's a second special form.\nSpeaker C: Uh-huh.\nSpeaker B: I would have thought that's probably about...\nSpeaker B: There you go.\nSpeaker B: Wow.\nNone: Okay, is that...\nSpeaker B: That's just the vehicle.\nSpeaker B: Right.\nNone: 12.2 so we're still...\nNone: We're in a genre.\nSpeaker D: Yep.\nNone: That makes sense.\nSpeaker C: So we'd say we've lost a little bit on the fashion and lost a bit on the fashion.\nSpeaker C: Yep.\nSpeaker C: Yeah.\nSpeaker C: But Jim has been looking at the other rights piece.\nSpeaker B: I would have said so.\nSpeaker B: So you maybe put fashion...\nSpeaker B: Fashion at three rather than...\nSpeaker B: So should we do it?\nSpeaker D: Two.\nSpeaker C: Well, it definitely fits the...\nSpeaker C: The real product.\nSpeaker C: Was it?\nSpeaker C: Was it really?\nSpeaker C: Yeah.\nSpeaker B: Real reaction product.\nSpeaker B: I'm not quite sure what it's happening.\nSpeaker C: Yeah, it still seems that we've set that up so that's like a new age kind of...\nSpeaker C: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker C: In a...\nSpeaker C: In a...\nSpeaker C: In a...\nSpeaker C: In a...\nSpeaker C: In a...\nSpeaker C: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a...\nSpeaker D: In a bit of fashion.\nSpeaker D: Maybe other ones could be...\nSpeaker D: Something else shaped.\nSpeaker D: I don't know, that would be possible.\nSpeaker D: You can see those things as it's cheaper to make them a different series.\nSpeaker D: You know, we've made it a special form so...\nSpeaker D: And that would sort of...\nSpeaker D: Maybe keep us...\nSpeaker D: Close on the sort of fashion type one.\nSpeaker D: And it would sort of...\nSpeaker D: You know, keep it quite fancy as well.\nSpeaker B: I don't know.\nSpeaker B: You could argue you might do it once a year.\nSpeaker B: You would change.\nSpeaker B: Because at the moment you're making it a right apple.\nSpeaker B: right apple so next year you could make next year model the same but have it as a yeah whatever a lemon yeah a lemon lemon or something\nSpeaker D: yeah we got rid of we got rid of the plastic yeah we the main thing we've\nSpeaker B: changed really is the casing isn't it\nSpeaker C: yeah I think we've saved quite a bit so we've just got the pushburn interface which is might not achieve this yeah but maybe in a sense not having that maybe we lost something in the relation side you know and suppose that is like the most standard paper\nSpeaker D: back so we've got we've got rid of the rubber case but we've now we've now sort of got an extra form and an extra color for the buttons so maybe in terms of sort of fanciness and fashionability we're pretty much the same yeah maybe so maybe put two again\nSpeaker C: and everything else is stayed pretty much the same well ease is certainly stayed yeah because of anything the special\nSpeaker D: form is makes it slightly easier yeah yeah and what about the sort of innovation\nSpeaker C: we've still got the kinetic energy and the speech nature and then corporate identity I think we've got just by the same as maybe last as they it's hard to tell how the rubber\nSpeaker D: how we play out yeah whether that's considered to be sort of part of the corporate identity I don't know\nSpeaker C: it's maybe not I think more of the features of the actual control would be more important maybe than yeah either way I think we made it fairly close to what yeah okay\nSpeaker A: yeah do the whole thing in the kind of light but then what color would you like to ask\nSpeaker C: this is that maybe like that design issue\nSpeaker B: but you've got the company logo on there which would effectively be a stick on badge so you're in a sense you're comparing the product without the company logo and then but you've got the space for it to stick it on\nSpeaker D: what's the company color did you get told that the company color is\nSpeaker C: okay so you establish that I don't think so but but in the sense that as you saw\nSpeaker B: with the windows logo badge it doesn't really matter there's virtually the way that you frame you know the windows badge on there it really doesn't matter what color it is as long as our company's logo is framed in the same way as that with a like a black outline in fact they've got black and white or black and silver so basically even if you had a silver the same color of silver on your display because you've got effectively a double edging on the on the logo it means that it splits off what your logo is from the from the product and that's actually quite a sneaky way of doing it\nSpeaker D: because you've got sort of got point to play with if you really wanted to\nSpeaker B: you could put in another another color well in this one you've actually got three colors of buttons yeah we decided that the blue ones were the the standard color okay so you were talking about we are assuming that all the buttons on the second panel the hidden away that would all be standard yeah yeah which\nSpeaker D: may or may not be the case yeah true save the point to the possibility\nSpeaker B: well yeah but but was it was not a button that you were thinking of on the on the other option you know on the second page as it were that you were thinking of maybe having is a different color\nSpeaker A: maybe for the the one that can see all the time\nSpeaker C: they make it look good and um hit the kind of idea of what they want and then for more functional buttons and it works much better it's like it puzzles hidden away anywhere in the theater stuff yeah right it's a sort of thing that\nSpeaker D: I mean you want to pick up the control and just hit a button quickly to change the channel or volume so if it's dead obvious then that's fine but if you're opening the panel and you're looking and you're tuning then you're paying a bit more attention so maybe sort of different color buttons isn't so important all right okay yeah so we just\nSpeaker B: add that to profitability and effect yeah I mean\nSpeaker D: you still drop the cost right so we're meant to finish up in five minutes same sort of function yeah criteria it's all right just made a load of money\nSpeaker A: shouldn't we maybe lose a point of fashion for this thing because we've lost the world because it doesn't seem right about it just hasn't changed at all\nSpeaker D: I think we said that because you got rid of the bubble we put an extra an extra sort of shape on one of the buttons well we could have lemon shaped ones for the volume\nSpeaker C: more of those things\nSpeaker D: did you have to have a rubber case though for rubber buttons or was it the way\nSpeaker C: around?\nSpeaker B: so we're okay this way around until the design team comes in and says get up but you are the design team so what better we on to\nSpeaker A: I can't just check if that's a cherry or an apple we would decide against the apple because of apple because of apple pie and then we'll make it a cherry for sure\nSpeaker C: oh right okay just in case we need that point two for\nSpeaker B:Beats Oh, the rats have botched that all the time.\nSpeaker A: So, what we even put a certain target on that scale?\nSpeaker D: Oh, right, one vertical.\nSpeaker D: But it really has to fit into the budget, so...\nSpeaker D: I guess we just have to adjust things to get it in there.\nSpeaker D: Just try to?\nSpeaker C: And we seem to madly sum from each criteria.\nSpeaker C: Yeah, we haven't really left anything else, so...\nSpeaker C: Yeah, I think.\nSpeaker C: I was nervous, probably, which has to be quite cheap.\nSpeaker C: Yeah.\nSpeaker D: Yeah.\nSpeaker A: So, do I have any help with this guy?\nSpeaker D: I don't know, it's only a chapter.\nSpeaker D: Right.\nSpeaker B: Don't you?\nSpeaker B: No, what's on here?\nNone: Okay.\nSpeaker B: All right, okay.\nSpeaker B: Okay.\nSpeaker B: So, we've done the finance bit and the Excel project, and we've done the redesign, so we're now on to project process, satisfaction with, for example, room for creativity, leadership, teamwork, means accessibility, or a digital plan, etc.\nSpeaker B: So, we're actually now, in a sense, on to the evaluation of the course, rather than the evaluation of the project, as I'm understanding of it.\nSpeaker B: So, what did you...\nSpeaker B: And remembering that nobody's system of the curtain.\nSpeaker B: So...\nSpeaker D: Finish on, please.\nSpeaker C: Which is good.\nSpeaker B: So, as I suppose the easiest way of doing it is to put some notes down, which I will do.\nSpeaker D: Okay.\nSpeaker B: So, I think I have to finish that page.\nSpeaker B: All right, okay.\nSpeaker B: So...\nSpeaker B: So, creativity, do you feel you've got a chance to express yourselves well enough?\nSpeaker D: I've got some of the divisions there between sort of the individual meetings\nSpeaker B: with the arbitrary individual meeting. How do you do India?\nSpeaker D: Well, we were finding out various things in your own.\nSpeaker D: In between the meetings, and the one meeting didn't always follow on for the other one.\nSpeaker D: So, things thrown in the second meeting where, you know, you've looked at the remote controls and seen the Coviness body.\nSpeaker D: And the first one you'd also look to some remote controls and look to the buttons.\nSpeaker D: And I don't understand why it would be a separate meeting.\nSpeaker D: But you'd do that, you know, you'd probably present it at one or something like that.\nSpeaker B: Well, the only thing you find is in a manufacturing process, you would normally go to a meeting, you'd say, right, you do this, you do that, you do that.\nSpeaker B: Then you go away, you find out information, you then come back, you then discuss it, you then go and change things around and then go back.\nSpeaker B: So...\nSpeaker C: Yes, listen to your words, you'd be going out and finding more information this time.\nSpeaker B: Well, as this time you're really getting from a database or so, it's not...\nSpeaker B: Yeah.\nSpeaker B: Anyway, so what do you want to do?\nSpeaker B: I've proved seem dorky, creativity seemed dorky, flow of information on any given subject.\nSpeaker B: Given subject, sometimes disjointed.\nSpeaker C: Yeah.\nSpeaker A: I think it may be kind of been slightly more creative project, but the remote control isn't the most...\nSpeaker A: Yeah, the thing is...\nSpeaker A: No, I was just like, how do you design it?\nSpeaker A: I can't think of a better example, I would...\nSpeaker B: I could have used a different example to increase...\nSpeaker B: Yeah, is this creative?\nSpeaker C: Just a bit of something everyone knows about and then creativity, and then you can at least look at it.\nSpeaker B: You have to do it within a set timeframe as the other thing.\nSpeaker D: Yeah.\nSpeaker D: Well, I always think it's a bit restrictive just to say that you have to design a television remote control.\nSpeaker D: It depends on the business you're in, I guess.\nSpeaker D: I mean, this one seems to have won.\nSpeaker D: On the website it looks quite innovative, but we're coming out with some bookish shop products, don't we?\nSpeaker D: That's the same thing.\nSpeaker C: It's a deep grid.\nSpeaker C: It's not like a range of products.\nSpeaker C: I think...\nSpeaker C: I guess this is going to sign out in the like a piece.\nSpeaker B: But the other thing is that, I mean, I'm guessing that they're trying to use this software to demonstrate how you could do a project.\nSpeaker B: I mean, my wife at the moment, for instance, is acting as a computer for...\nSpeaker B: Normally, you've got a problem, so you go to your tutor and find out information to see how to get it fixed.\nSpeaker B: So what she's doing is she's having to spend a day at the computer terminal at one end so that any student that comes along can ask whatever question.\nSpeaker B: And then you are the idea is that the whichever persons at the other end can point them in the right direction, and should have either give them directly, give them help, or secondly point them in the right direction, either at the library or come back or go and see job logs or whatever.\nSpeaker B: So...\nSpeaker B: And that was a project I suspect similar to this, because they were actually trying to debug the computer software to enable it to work.\nSpeaker B: And of course, you had machine crashing and various things going from...\nSpeaker B: So...\nSpeaker B: All right, so we've got...\nSpeaker D: I mean, if you look at their products on their website here, real reaction, it's all pretty high tech cutting edge.\nSpeaker B: You ideas found? Did we find any? No.\nSpeaker C: So, how do I do with this, the whiteboard, how the non-digital pen is placed?\nSpeaker B: All right.\nSpeaker B: Leadership.\nSpeaker B: Teamwork.\nSpeaker D: So we did find any idea, means, the kinetic remote control.\nSpeaker D: I don't see one of them before.\nSpeaker D: They're all batteries I think.\nSpeaker C: All voice recognition, especially not one with...\nSpeaker C: Yeah.\nSpeaker B: You've got voice recognition computers.\nSpeaker B: Well, it's not remote control.\nSpeaker A: Yeah.\nSpeaker A: What's a different application there?\nSpeaker E: Mm-hmm.\nSpeaker C: Okay.\nSpeaker C: And it's just like the same products, but it's put together in a different one.\nSpeaker B: So, how do you reckon teamwork went?\nSpeaker D: I would have, okay.\nSpeaker C: Yeah, I guess we all had such ideas.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker B: Two, er, go...\nSpeaker B: Reason, I'm playing...\nSpeaker B: Well, okay.\nSpeaker D: I don't think sort of the putchet...\nSpeaker D:...um...\nSpeaker D:...allow just to do anything.\nSpeaker B: But arboring?\nSpeaker D: Well, I mean, I don't think it...\nSpeaker D: I don't think it fitted in with the rest of their products.\nSpeaker D: I've got all these sort of high definition DVD and portable seven inch lightweight computer screens.\nSpeaker D: I think one cheap remote control does really fit in.\nSpeaker B: So, we're meant to comment on leadership and means E.G. Whiteboard, digital pens, et cetera.\nSpeaker B: And new ideas found was the other thing.\nSpeaker B: Yeah.\nSpeaker A: We can't really...\nSpeaker A:...all fail to comment on leadership because we want to project manage our new...\nSpeaker A:...so our experience in leadership wasn't really as much as yours.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Now, I'm much more...\nSpeaker B:...filling like it.\nSpeaker B: But...\nSpeaker B:...I certainly didn't get this computer to work as well as I would have liked.\nSpeaker B: However...\nSpeaker B:...right, means so Whiteboard, so really it's equipment.\nSpeaker B: Yeah.\nSpeaker D: It worked.\nSpeaker D: Yeah, it worked.\nSpeaker D: Yeah, it worked.\nSpeaker D: It could be a bit difficult at times.\nSpeaker D: I mean, I don't...\nSpeaker A:...I'm not sure I see the value in these.\nSpeaker A: I mean, they record what you write in and then what to do with it.\nSpeaker A: Where do you get the recording?\nSpeaker A: Do you...\nSpeaker A:...computers or something?\nSpeaker D: It's a little...\nSpeaker C:...it's not just for us, it's for these families.\nSpeaker D: It's a little chip I think you're just plugging into something else.\nSpeaker A: It's like...\nSpeaker A:...producer.\nSpeaker A:...producer.\nSpeaker A:...producer.\nSpeaker D: I don't know.\nSpeaker D: I've produced the game, so everything you've written or something.\nSpeaker D: No, you have to pay.\nSpeaker B: New ideas found, so one or two?\nSpeaker D: Yeah.\nSpeaker D: Kinetic powered remote control.\nSpeaker D: What was it, the one?\nSpeaker A: Voice.\nSpeaker A: Yeah.\nSpeaker A: What was your thing?\nNone: Right, word.\nNone: And...\nSpeaker B:...voice.\nSpeaker B:...voice.\nSpeaker B: With a voice activated.\nSpeaker D: Voice recognition.\nSpeaker D:...reconation.\nSpeaker D:...almost.\nSpeaker D:...in this...\nSpeaker B:...\nSpeaker B:...in this summary of any celebration?\nSpeaker B: That's good.\nSpeaker D: I think that you've got two questionnaires and a report on the last meeting.\nSpeaker B: So...\nSpeaker B:...we don't...\nSpeaker B:...so really...\nSpeaker B: I tried to save this last time at 12.3 and it doesn't seem to have saved anywhere.\nSpeaker B: Mind you, the figure last time was different, but...\nSpeaker B:...it should be both.\n"}, {"meeting_id": "ES2015d", "summary": "Firstly, User Interface introduced the prototype of remote control based on the previous discussion of its function. The prototype was a pretty simple design in a mango shape with the company logo on it and innovative functions like the plus button. Secondly, Marketing designed an evaluation test under the guidance of both the conceptual as well as functional meetings and the team gave one to seven points to each feature of the product to see if the prototype would meet the original requirements and goals. Thirdly, the team calculated the cost of the prototype and found that the most expensive part was the use of kinetic. But the team decided to use regular chips and cheap materials, which made the budget under control. Lastly, the team gave positive feedback on the project and the process.", "dialogue": "Speaker B: Yeah.\nSpeaker B: There you go.\nSpeaker B: Okay.\nNone: Sorry.\nNone: Oh, yeah.\nSpeaker E: Why do we have two different microcos?\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: What would the difference be?\nSpeaker B: Okay.\nSpeaker B: What would the difference be?\nSpeaker B: Enter thing.\nSpeaker B: I can't see why it was great.\nSpeaker B: All right.\nSpeaker B: Okay.\nSpeaker B: Here you go.\nSpeaker B: All right.\nSpeaker B: The agenda for the.\nSpeaker B: All right.\nSpeaker B: The agenda for this meeting is, will initially have the prototype presentation by our two designers, and then we will evaluate it, given the criteria that we gave it, and talk about our finances, whether we were under or over our budget.\nSpeaker B: I have a spreadsheet where we can calculate our prices for every aspect of what we've made, given our options, and evaluate the product as a group.\nSpeaker B: Okay.\nSpeaker B: And so first we'll have the prototype presentation.\nSpeaker B: Do you need the PowerPoint?\nSpeaker A: Yeah.\nSpeaker A: I'll just do a piece of that.\nSpeaker C: All right.\nSpeaker C: All right.\nNone: Thank you.\nSpeaker C: Yeah.\nSpeaker C: Here we are.\nSpeaker C: This is what we came up with.\nSpeaker C: It's a pretty simple designer based on a mine go.\nSpeaker C: Yeah.\nSpeaker C: And we have the company logo here, and this will be the.\nSpeaker C: Emperage.\nSpeaker C: And just the part point.\nSpeaker C: Oh, okay.\nSpeaker C: Yeah.\nSpeaker C: Sorry.\nSpeaker C: Where's the LED?\nSpeaker C: It's in the middle of the middle of the art.\nSpeaker C: Okay.\nSpeaker C: And then the other one is this part.\nSpeaker C: And we just have a sample design.\nSpeaker C: And we wanted it all to be.\nSpeaker C: Palm hell.\nSpeaker C: Yeah.\nSpeaker C: Palm hell.\nSpeaker B: Yeah.\nSpeaker B: All the.\nSpeaker C: What is our.\nSpeaker C: No, see, I have a number 10 button.\nSpeaker C: Oh, I thought we were going to take it.\nSpeaker A: Right.\nSpeaker E: No, it would be.\nSpeaker C: I wrote.\nSpeaker C: Just got one.\nSpeaker C: Sorry.\nSpeaker C: I was in charge of the number.\nSpeaker C: I'm not.\nSpeaker C: I'm not.\nSpeaker C: I'm just.\nSpeaker A: So, one.\nSpeaker C: Oh, press a plus button.\nSpeaker C: Oh, okay.\nSpeaker B: I've never heard of that kind before.\nSpeaker C: Oh, we just thought we have all the numbers here.\nSpeaker C: So we wanted something.\nSpeaker C: Represented numbers larger than.\nSpeaker C: Yeah.\nSpeaker A: Yeah, because if you on your average.\nSpeaker A: If you press one twice, you just go to.\nSpeaker A: Oh, I say you wanted channel 12.\nSpeaker A: You press one and you go to channel one and then two.\nSpeaker A: Then you just go to channel two instead of.\nSpeaker A: Oh, there's not.\nSpeaker A: So if you just like one plus two.\nSpeaker A: Yeah.\nSpeaker A: You could go to two plus two is channel.\nSpeaker B: Would you have to go zero plus one if you wanted to go to channel one or two?\nSpeaker C: No, no.\nSpeaker C: I have all these numbers.\nSpeaker C: These numbers all work and the pen race up to nine.\nSpeaker B: Yeah, but I mean if you press it, it'll go to that channel right away.\nSpeaker B: Yeah, because you got to press the plus afterwards.\nSpeaker C: Oh no, the plus.\nSpeaker C: I've here going past number nine.\nSpeaker B: Yeah, I know, but if I want to go to say number like sixty five, channel sixty five, I press the six, it'll go to channel six and then I'll press the plus and then we'll go to six and then put the five and we'll go to sixty five.\nSpeaker C: I don't know.\nSpeaker C: It doesn't.\nSpeaker C: We haven't.\nSpeaker C: I wouldn't thought you press the plus.\nSpeaker C: No, I'm a six five, but she says.\nSpeaker C: I don't want to.\nSpeaker A: What do you think?\nSpeaker A: Define that.\nSpeaker A: I wouldn't have thought it would be a problem that it went to channel six.\nSpeaker A: Yeah, like on the way to channel.\nSpeaker A: Yeah, it wouldn't be a problem.\nSpeaker B: I was wondering like as long as we realize that's what it will do.\nSpeaker D: There's a delay on remotes, I think, where you can have it.\nSpeaker D: It's like a five second input time.\nSpeaker D: Yeah, and so as long as we hit them, yeah.\nSpeaker D: Yeah, yeah, that should be fine.\nSpeaker D: Yeah.\nSpeaker D: So it's not a big pause between the two buttons.\nSpeaker D: Yeah.\nSpeaker D: Um, was there, so on the top there is volume and channel, which channel I'm following.\nSpeaker A: Okay, cool.\nSpeaker A: You could just see and feel like that without thinking about it.\nSpeaker B: Yeah, we're, um, where's the power button?\nSpeaker B: It's in the middle of the big air.\nSpeaker A: Oh, okay.\nSpeaker A: So it's just like, yeah.\nSpeaker A: It's all accessible.\nSpeaker C: We thought taking your hand off the remote.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: And we went for like a circular design for the numbers because we thought that was kind of a more natural movement than just going like that with your thumb.\nSpeaker A: Oh, okay.\nSpeaker C: Yeah, it could actually help with the repair of stress.\nSpeaker C: It could cause another type of repression.\nSpeaker A: Okay.\nSpeaker A: But yeah, no, I mean, it's a different movement.\nSpeaker A: Yeah.\nSpeaker A: Um, and the feel of it.\nSpeaker A: I mean, we've made this out of Plato, which is representing the, you know, the rubber.\nSpeaker A: Yeah.\nSpeaker A: You know, the spongy, the rubberness.\nSpeaker A: Um, because it was said before in the material specification that this, this anti RSI, um, materials often used in stress books.\nSpeaker A: So this has got a, you know, a bit of give to it.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: The verstraspar field.\nSpeaker A: Yeah.\nSpeaker A: Feel it.\nSpeaker B: Yes.\nSpeaker B: How fit time you were.\nSpeaker B: My goodness.\nSpeaker B: There you go.\nSpeaker B: Thank you.\nSpeaker B: Ready?\nSpeaker D: Yeah.\nSpeaker D: Oh, it's nice.\nSpeaker D: Oh, I think I killed the fire.\nSpeaker D: I did.\nSpeaker D: Oh, God.\nSpeaker A: Okay.\nSpeaker A: As for the colors, we will presented with a limited range of colors for this prototype.\nSpeaker A: But we're thinking that seeing as we're having an interchangeable casing anyway, that this is not necessarily a representation of the true colors that we would necessarily use or the combination.\nSpeaker A: Um, and we're thinking to carry on with the fruit and vegetable theme, the color, um, combination just to be named after different fruit, like banana, with black and yellow water, and green, you know, vanilla might be the most popular if it's just a blending.\nSpeaker A: It's kind of subtle and more settled cream.\nSpeaker A: Yeah.\nSpeaker A: It looks more a bit garish.\nSpeaker B: I think like vanilla and banana would.\nSpeaker A: But an is more representative of our color scheme, like the company, the yellow and the black.\nSpeaker A: Yeah.\nSpeaker A: Oh, okay.\nSpeaker A: So that for corporate identity, that would probably be the most strange.\nSpeaker A: I mean, watermelon, you know, probably appealing to the best.\nSpeaker A: Kind of Christmas, you know.\nSpeaker A: Yeah.\nSpeaker A: Seasonal.\nSpeaker A: Apple, green, brown, more kind of trendy, you know, kind of neat.\nSpeaker C: The polygonics kind of yellow and funky kind of things.\nSpeaker C: Cool.\nSpeaker C: Then you can know it's more sophisticated cast around here.\nSpeaker C: But that's almost all decor.\nNone: Okay.\nSpeaker A: Um, yeah, we thought of the components.\nSpeaker A: It was definitely a focus of ergonomics and just a single hand hand.\nSpeaker A: Hand held device.\nSpeaker A: And you don't need to use both hands.\nSpeaker A: One hand to hold this and type in with the other.\nSpeaker A: You can just use your thumb.\nSpeaker A: Yeah.\nSpeaker A: Um, as I said, the rubber is probably used for comfort and anti-hour size.\nSpeaker A: That's about it.\nSpeaker B: All right.\nSpeaker B: Thank you very much.\nSpeaker B: Good work, everyone.\nSpeaker B: All right.\nSpeaker B: And so now that we have, we have a prototype and we need to go over the finances and seeing if this prototype matches what our budget can handle.\nSpeaker B: So I have something I'm going to.\nSpeaker D: Wait, do you need to do a presentation first?\nSpeaker D: I don't know what order it goes.\nSpeaker B: Yeah.\nSpeaker B: I'm going to check that out for a second.\nSpeaker B: Oh, yeah.\nNone: Sorry.\nSpeaker B: You're right.\nSpeaker B: I'm going to go over the evaluation criteria.\nSpeaker C: Hello.\nNone: Hello.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Sorry.\nSpeaker D: My computer is going to be technical difficulties.\nSpeaker B: Just press the button again.\nSpeaker B: I'll still do that.\nSpeaker B: Okay.\nSpeaker D: And then again, I think.\nNone: One more time.\nSpeaker B: Yeah.\nSpeaker D: Okay.\nSpeaker D: Now I think for this one, I could, would you guys prefer to use the whiteboard or maybe I'll just do it on right on the screen where you can see it.\nSpeaker D: We're going to be doing an evaluation report together based on the prototype that we've just seen.\nSpeaker D: And looking, looking back at my notes from our, both our conceptual and our functional meetings, I made a list of what our original requirements and goals were back to our kickoff meeting this morning.\nSpeaker D: And we'll evaluate as to whether we've done what we set up to do.\nSpeaker D: And we're going to do it on a one to seven scale where one is true and seven is false.\nSpeaker D: So basically the lower, lower the points, the better.\nSpeaker D: Okay.\nSpeaker D: So question number one.\nSpeaker D: Does the remote, whoops, sorry.\nSpeaker D: Oh, I'm not going to, I'll do it on the whiteboard.\nSpeaker D: I can't change it.\nSpeaker D: So I'm going to ask you to push it down once.\nSpeaker D: Okay.\nSpeaker D: I'll write down our scores up on the, okay.\nSpeaker D: So number one, do we have a fancy look and feel?\nSpeaker E: Feel, I think.\nSpeaker A: I think quite successful with the rubber, the look is a little bit more playful.\nSpeaker C: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: It depends on your definition of fancy, but it's definitely different.\nSpeaker B: Oh, definitely.\nSpeaker D: Yeah.\nSpeaker A: I think the color has a lot to do with it.\nSpeaker A: I mean, the colors we were given for making the photos like the colors that I think we would have necessarily chosen is not the kind of tool.\nSpeaker A: It's all red and black and yellow and orange.\nSpeaker A: Yes.\nSpeaker A: I'm sorry.\nSpeaker A: I'm sorry.\nSpeaker C: I don't want to go for black because most people are going to go for black and gray.\nSpeaker C: Yeah.\nSpeaker A: But if you can imagine that in like maybe kind of pale metallic finish or something.\nSpeaker A: Well, I know it's for rubber.\nSpeaker A: I can visualize this in nice colors.\nSpeaker A: I think it would look great.\nSpeaker A: Yeah.\nSpeaker D: I kind of like the potato look.\nSpeaker D: That's very cool.\nSpeaker D: Oh, well.\nSpeaker D: That's why I was bringing the vegetables.\nSpeaker D: I'm sorry, the mango.\nSpeaker D: Yeah, the fruit or vegetable depends on your mood.\nSpeaker B: So I myself would say a one or two.\nSpeaker C: I would say two.\nSpeaker C: Okay.\nSpeaker D: One being true.\nSpeaker A: I add two.\nSpeaker D: Okay.\nSpeaker D: That's this might get confusing.\nSpeaker D: So question number two was is it technologically innovative?\nSpeaker D: So I know we have the kinetic energy, which is very innovative.\nSpeaker D: Yeah.\nSpeaker A: And the use of the rubber.\nSpeaker A: The use of the rubber.\nSpeaker A: The use of the LED.\nSpeaker A: The LED isn't used to take Killy innovative.\nSpeaker A: And we don't have any scroll buttons.\nSpeaker A: There's all push buttons.\nSpeaker A: There's no LCD control.\nSpeaker A: So if you're thinking about the rest of the market.\nSpeaker A: Yeah.\nSpeaker A: It's probably halfway in seven aspects.\nSpeaker A: Yeah.\nSpeaker C: Three.\nSpeaker D: And I think it's tough to say because we didn't want it to be any more innovative than this.\nSpeaker D: Yeah, we want it to be a purpose.\nSpeaker D: So I mean, we'll put three, but I think we actually reached our goal.\nSpeaker D: Yeah.\nSpeaker D: We didn't want it any more than that.\nSpeaker D: Okay.\nSpeaker D: Question number three will be easy to use.\nSpeaker B: I think so.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I think it's, yeah.\nSpeaker D: Can't really get confused with that.\nSpeaker D: I mean, we had to work out the number, the plus system.\nSpeaker D: But once that's figured out, it's tough to turn it on.\nSpeaker D: Number four, is this a good looking remote?\nSpeaker D: Remember that 75% of users find most remote controls ugly.\nSpeaker A: Again, I think the color comes into this.\nSpeaker B: Yeah.\nSpeaker B: Color will definitely be effective.\nSpeaker B: I think that the logo could be smaller.\nSpeaker B: Okay.\nSpeaker C: Yeah.\nSpeaker B: And maybe an odd, it's a prominent way.\nSpeaker B: But maybe like at the bottom, kind of.\nSpeaker C: The management said that it had to be taller than it.\nSpeaker B: Whoops.\nSpeaker B: It just had to be on there, I guess.\nSpeaker B: Just not touch it.\nSpeaker D: It's time for the three I killed.\nSpeaker D: I was just wondering if it should be like flatter.\nSpeaker B: I suppose I've got quite the appeal of it being like a big glob.\nSpeaker B: Yeah.\nSpeaker C: But you know what I've just thought of then, right?\nSpeaker C: Where's the connie set in your living room?\nSpeaker C: Maybe it's the bottom is just sort of flat.\nSpeaker C: Yeah, the bottom could be like chopped a bit.\nSpeaker C: But then around.\nSpeaker C: That is comfortable.\nSpeaker C: That's true.\nSpeaker B: It would still be comfortable.\nSpeaker B: Maybe it could be on the bottom.\nSpeaker D: So you wouldn't lose, like if it's flat here, so it sits up.\nSpeaker D: Yeah.\nSpeaker D: Like that.\nSpeaker B: Oh, that was the last fall over all the time.\nSpeaker B: Yeah.\nSpeaker B: It's less.\nSpeaker D: If it's weighted, maybe.\nSpeaker D: I mean, I would definitely go for that rather than like your average.\nSpeaker D: Yeah.\nSpeaker D: Like that.\nSpeaker D: But definitely in another color, I'm not happy with that.\nSpeaker C: Okay.\nSpeaker C: So it's a good looking remote.\nSpeaker C: Do you want to show it off to our friends?\nSpeaker B: You would though, because it's more interesting than all the other.\nSpeaker B: Yeah.\nSpeaker B: I think it was another color and it was like, I think it would look okay.\nNone: Maybe you too.\nSpeaker A: I mean, I mean, I would definitely go for that rather than like your average.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: We're like that.\nSpeaker A: But definitely in another color.\nSpeaker A: I'm not happy with that.\nSpeaker D: Okay.\nSpeaker D: So should we say to?\nSpeaker D: For that?\nSpeaker D: Sure.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Question number five was, will people be willing to spend 25 euros on this product?\nSpeaker D: Remember that 80% of users were willing to spend more money when remote control looked fancy?\nSpeaker C: I think we have to market it in the right way.\nSpeaker C: Yeah.\nSpeaker C: To say that is really simple.\nSpeaker C: So people didn't just say it and think this is so simple.\nSpeaker C: I do want to spend 25 euros.\nSpeaker C: Yeah.\nSpeaker D: The marketing of lots of units.\nSpeaker A: The kinetic energy she kissed early.\nSpeaker A: Oh, no.\nSpeaker A: She kicked her butt.\nSpeaker A: She kicked her butt.\nSpeaker A: The video.\nSpeaker A: There's a beef video.\nSpeaker C: I don't know what the plot is.\nSpeaker D: No, I guess I don't know much about the remote control industry.\nSpeaker D: How much your average sells for.\nSpeaker D: But you're a marketing expert.\nSpeaker D: I know I am right here.\nSpeaker C: I think they're quite tan.\nSpeaker C: I think they're tan.\nSpeaker C: I think they're tan.\nSpeaker C: I think they're tan.\nSpeaker C: But you don't have to buy batteries.\nSpeaker D: So in the long term, this can actually save you money.\nSpeaker B: So we'll market it that way.\nSpeaker B: Exactly.\nSpeaker D: Yeah.\nSpeaker D: So yeah, I think with the good marketing scheme and the personalization options.\nSpeaker D: So yeah, I would give it a two.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nNone: Okay.\nSpeaker D: Number six.\nSpeaker D: Can someone read it out?\nSpeaker A: Does this product type match the operating behavior of the output?\nSpeaker A: She's there.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: So that was mainly the.\nSpeaker A: Yeah.\nSpeaker A: I think it does.\nSpeaker A: Yeah.\nSpeaker A: Because the most accessible buttons are the volume and the channel changing.\nSpeaker A: And it's just you won't have to think about it.\nSpeaker A: You don't have to look down to find them.\nSpeaker A: They're clearly there.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker D: I guess the key word there is average because there were some people that use the video input and sound and stuff.\nSpeaker D: Yeah.\nSpeaker D: But they're not you and I really.\nSpeaker D: So okay.\nSpeaker D: So one.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Number seven.\nSpeaker D: Heather, can you push it down?\nSpeaker D: Well, this remote control be easy to find when lost.\nSpeaker D: Remember that 50% of users lose their remote regularly.\nSpeaker D: We had the alarm system.\nSpeaker D: Is the alarm system still?\nSpeaker C: Was it implemented?\nSpeaker C: Yeah.\nSpeaker C: When you press the alarm system, the light.\nSpeaker C: Yeah.\nSpeaker A: The light's on the end or flash the bulb.\nSpeaker A: I mean, it's not obviously obvious from outside that's going to happen.\nSpeaker A: Because you can't particularly see the alarm.\nSpeaker A: I have the light on the inside.\nSpeaker C: Yeah.\nSpeaker C: The light it will.\nSpeaker C: Oh yeah.\nSpeaker A: But when the alarm is not.\nSpeaker A: Yeah.\nSpeaker B: If you must button makes a noise, right?\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker C: It turns into a duck and starts cracking.\nSpeaker D: Yeah.\nSpeaker A: You're right.\nSpeaker A: Awesome.\nSpeaker A: I think if it was had an alarm system, I mean, when it like suffered, I mean, it could light up when the alarm went.\nSpeaker A: Yeah.\nSpeaker A: If it was hidden underneath the question or something, it wouldn't be any point.\nSpeaker A: So you can't see the alarm, but it would like.\nSpeaker B: I thought it was going to make a noise.\nSpeaker B: Yeah.\nSpeaker A: Yeah.\nSpeaker A: You can't see an alarm inside the alarm system itself.\nSpeaker A: Okay.\nSpeaker D: You just explained why it's not on the prototype.\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker D: It'll be there.\nSpeaker D: Oh, okay.\nSpeaker D: Sorry.\nSpeaker D: We could give it a one because compared to every other remote ever made, this one will be easier to find.\nSpeaker A: Yeah.\nSpeaker A: Sorry, Heather.\nSpeaker D: No problem.\nSpeaker D: Question number eight.\nSpeaker D: Will it be easy to learn how to use this remote when brand new?\nSpeaker D: Remember that 34% of users usually found it difficult.\nSpeaker D: Totally.\nSpeaker D: It has to be.\nSpeaker D: Yeah.\nSpeaker C: It's not the one to do.\nSpeaker C: The plus thing is we walked on.\nSpeaker C: What's that written on the page?\nSpeaker C: That.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: I think just because it's, we've decided to reduce it down for the basic buttons.\nSpeaker A: I think that in itself makes it.\nSpeaker D: Does it make more sense for the middle one to be just an enter button?\nSpeaker D: So then you would have to push two buttons every time at least.\nSpeaker D: Yeah.\nSpeaker B: So it's just like channel six.\nSpeaker B: That kind of noise.\nSpeaker B: And then zero sexy.\nSpeaker C: That's the first.\nSpeaker C: We don't have to press zero.\nSpeaker C: You could just press six.\nSpeaker D: Enter.\nSpeaker D: Oh, okay.\nSpeaker A: Oh, okay.\nSpeaker A: I know.\nSpeaker A: One, two, three, four, five.\nSpeaker A: Yeah.\nSpeaker A: I think that's probably the most straightforward.\nSpeaker A: Yeah.\nSpeaker D: Good.\nSpeaker D: Okay.\nSpeaker D: So we can.\nSpeaker D: We'll say yes.\nSpeaker D: It's one.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Question number nine.\nSpeaker D: Well, minimize the effects of RSI, which was repeated strain injury.\nSpeaker D: Which affected over a quarter of users.\nSpeaker B: I think so.\nSpeaker B: But if your thumb might get a little.\nSpeaker A: Yeah.\nSpeaker A: I don't think it will.\nSpeaker A: We may have to do some more research in the other strain injuries that we don't know about.\nSpeaker B: But it is soft.\nSpeaker B: And that's kind of what the.\nSpeaker B: And people PowerPoint slide thing said would be good for.\nSpeaker B: Yeah.\nSpeaker B: I don't know what other option we get.\nSpeaker D: I mean, you can hold your left hand and.\nSpeaker D: Use your index figure.\nSpeaker D: But there's not really any other options.\nSpeaker D: It's like a keyboard.\nSpeaker D: So I think we did as much as you can with remote control.\nSpeaker D: Yeah.\nSpeaker D: So one or two, do you think?\nSpeaker A: I think.\nSpeaker D: Yeah.\nSpeaker D: Two.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: Number 10.\nSpeaker D: Did we somehow incorporate the company color and logo?\nSpeaker D: Yes.\nSpeaker D: No.\nSpeaker D: Is the color going to be there even if it's like vanilla is the yellow.\nSpeaker C: We can't really do that.\nSpeaker C: For example, on the banana thing, we can't have everything yellow.\nSpeaker C: It will sound like.\nSpeaker C: So it's not always going to be the same color.\nSpeaker D: It sounds like the color is something that we.\nSpeaker B: Yeah.\nSpeaker B: The colors are yellow and gray.\nSpeaker B: Gray.\nSpeaker C: So it could be gray on the banana one.\nSpeaker C: Yeah.\nSpeaker B: Could be gray.\nSpeaker D: Yeah.\nSpeaker D: I mean, the yellow, yeah, the yellow is ugly though, depending on that.\nSpeaker D: Yeah.\nSpeaker D: So I think we'll have to talk to our executive managers and see if we can get away with just the R.\nSpeaker A: Yeah.\nSpeaker A: And this is the.\nSpeaker A: The fact that we need to be successful.\nSpeaker A: Okay.\nSpeaker A: Perhaps.\nSpeaker A: That's great.\nSpeaker C: And then the buttons in the middle.\nNone: Yeah.\nSpeaker D: So do you think that's more of a three then?\nSpeaker D: Four.\nSpeaker D: Four.\nSpeaker D: I think four.\nSpeaker D: Well, I don't.\nSpeaker D: Well, we have good reasons for it.\nSpeaker D: So we can still put a.\nSpeaker D: Four.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: And final question.\nSpeaker D: Do we stay true to our motto and put the fashion into electronics?\nSpeaker D: Current trends of fruits and veggies desire for sponginess.\nSpeaker B: I would say so.\nSpeaker B: But maybe more like two because there's no like pictures of fruit.\nSpeaker B: It's just sort of naming it by a fruit.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker D: So you can have banana and kiwi and.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: So that's something that's kind of in.\nSpeaker D: The making to like maybe it'll become more.\nSpeaker D: The fashionable in itself.\nSpeaker D: Haven't you change a ball place?\nSpeaker B: Mm hmm.\nSpeaker B: I think.\nSpeaker B: Yeah.\nSpeaker C: Well, that was a brief.\nSpeaker C: Well, we have.\nSpeaker C: I think if somebody saw that and he said, what was that inspired from?\nSpeaker C: I don't know.\nSpeaker C: I think it's a.\nSpeaker C: I think it's a.\nSpeaker A: Mind go.\nSpeaker A: Maybe if it was centered.\nSpeaker B: Oh, yeah.\nSpeaker B: There we go.\nSpeaker B: That was great.\nSpeaker B: We have money for that.\nSpeaker B: All right.\nSpeaker B: So based on.\nSpeaker B: This evaluation, do we average the notes?\nSpeaker B: Yes, we do.\nSpeaker B: So what was I going to put for that?\nSpeaker B: To for fashion?\nSpeaker B: I would say to.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: So that's a good question.\nSpeaker D: I would say to.\nSpeaker D: Okay.\nSpeaker D: So our average there.\nSpeaker D: Five, six, eight, 10, 11, 12, 13, 15, 19, 21 divided by 11.\nSpeaker D: One point nine.\nSpeaker B: Yeah.\nSpeaker B: I don't know.\nSpeaker B: It's a one and two between one and two.\nSpeaker B: Okay.\nSpeaker A: That's great.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker D: Yeah, that's very good because the highest could have got is one.\nNone: All right.\nSpeaker D: That seemed right.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker A: I think like it should be more around to.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker C: So you have an online campaign.\nSpeaker A: And attempting to do that.\nSpeaker B: Okay.\nSpeaker B: Yeah, it is one point nine.\nSpeaker B: Oh, wow.\nSpeaker A: All that cause.\nSpeaker B: Yeah.\nNone: All right.\nNone: Now with that.\nSpeaker B: Over and done with our next step is to see if we are under budget.\nSpeaker B: And my computer is frozen.\nSpeaker B: And now it's not.\nSpeaker B: Okay.\nSpeaker B: So in our shared folder, if everyone could go there right now.\nSpeaker B: Sorry.\nSpeaker B: I'm going to.\nSpeaker B: Oh, you're going to steal a game.\nSpeaker B: Okay.\nSpeaker B: It's, it's a.\nSpeaker B: It's an Excel file.\nSpeaker B: Oh.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: Production costs.\nSpeaker B: And I have to access that as well.\nSpeaker B: One moment.\nSpeaker B: Okay.\nSpeaker B: So far I've added what I think.\nSpeaker B: What is going on?\nSpeaker B: I'm going to open up the second one.\nSpeaker B: It's locked for editing.\nSpeaker B: I have the original in my.\nSpeaker B: But you know I can't.\nSpeaker B: Or is it locked?\nSpeaker D: Cause I'm in it.\nSpeaker A: I don't think it just means that we can add any more to it.\nSpeaker A: Now.\nSpeaker A: Have you, have you completed it?\nSpeaker A: No.\nSpeaker B: No, I was hoping that you guys could.\nSpeaker B: Okay.\nSpeaker B: There we go.\nSpeaker B: Okay.\nSpeaker B: So we need to tally up how much our product will be costing.\nSpeaker B: So you can look up at the screen.\nSpeaker B: The live screen.\nSpeaker B: I guess looking at your own too and telling you which one you think.\nSpeaker B: Okay.\nSpeaker B: We're using kinetic, which is quite.\nSpeaker B: A large expense of three euros.\nSpeaker B: Yeah.\nSpeaker B: We're using a regular chip.\nSpeaker B: Yep.\nSpeaker B: It's curve.\nSpeaker B: It's double curve.\nSpeaker B: So it's curve all around.\nSpeaker B: It's another three.\nSpeaker B: We're already at five.\nSpeaker B: We're using plastic and rubber.\nSpeaker B: Yeah.\nSpeaker B: So.\nSpeaker B: The thing with plastic is free.\nSpeaker C: Eight.\nSpeaker B: Yeah.\nSpeaker B: I guess we should do it just for one kind.\nSpeaker B: Yeah.\nSpeaker B: So it's like special color.\nSpeaker B: We'll have two colors, right?\nSpeaker B: Well, one color for the case, one color for the buttons.\nSpeaker B: So we can.\nSpeaker B: We have push button interface.\nSpeaker B: That's inexpensive.\nSpeaker B: And you have a special color for the button.\nSpeaker B: And we also have a special form.\nSpeaker B: And special material.\nSpeaker B: Which puts us just barely under budget.\nSpeaker A: Hey, congratulations.\nSpeaker B: Okay.\nSpeaker B: So, um, our operating cost.\nSpeaker B: Is.\nSpeaker B: 12, 20 euros.\nSpeaker B: Awesome.\nSpeaker B: And back to our PowerPoint.\nSpeaker B: Okay.\nSpeaker B: Yes, we are.\nSpeaker B: So we need to do a project evaluation.\nSpeaker B: Again, which is probably, um.\nSpeaker B: I don't know.\nSpeaker B: A different extension of the.\nSpeaker A: Of the actual project rather than the project.\nSpeaker A: Yeah.\nSpeaker A: A project.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: All right.\nSpeaker B: So, um.\nSpeaker B: Do you guys feel like there was real creativity?\nSpeaker A: I think we were pushed.\nSpeaker A: For creativity.\nSpeaker A: We went.\nSpeaker A: Really given.\nSpeaker A: Yeah.\nSpeaker A: To go about.\nSpeaker A: So I think we could have done it a bit more.\nSpeaker A: Okay.\nSpeaker A: Okay.\nNone: Okay.\nSpeaker B: So it'd be like.\nSpeaker B: But you were allowed.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: I think so, but you were supposed to.\nSpeaker B: Yeah.\nSpeaker B: And the conceptual and functional.\nSpeaker D: Like we were very creative and coming up with an idea.\nSpeaker D: Yeah.\nSpeaker D: What we just limited by resources.\nSpeaker A: Okay.\nSpeaker A: If we had decided to use the L.T.\nSpeaker A: E screen and like solar power backup and everything.\nSpeaker A: And we would have been able to afford that.\nSpeaker A: So that did.\nSpeaker A: Great.\nSpeaker A: Yeah.\nSpeaker A: Leadership.\nSpeaker B: Does this mean you like guys?\nSpeaker B: Do you like me?\nSpeaker D: Good leadership.\nSpeaker D: I think we stayed on task.\nSpeaker A: Yeah.\nSpeaker A: We did.\nSpeaker A: We've come out with what we intended a product within the budget.\nSpeaker A: I think that's the leadership.\nSpeaker A: I'm also a personal coach.\nSpeaker A: Helps along the way.\nSpeaker D: And the timing was good.\nSpeaker D: We never were pushed for time or sat around doing nothing.\nSpeaker D: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Special Coke.\nSpeaker B: I'm still working on the teamwork tool.\nSpeaker B: Yeah.\nSpeaker D: I think we were well suited to our roles.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nNone: All right.\nNone: How are our means?\nSpeaker B: We needed more Play-Doh colors.\nSpeaker B: Yeah.\nSpeaker A: And more Play-Doh because that was all the red we had.\nSpeaker A: So if we wanted to make a bigger.\nSpeaker A: Oh really?\nSpeaker A: We wouldn't have been able to.\nSpeaker A: Okay.\nSpeaker B: I don't think we could.\nSpeaker B: I don't think anything else was satisfactory.\nSpeaker B: Is that?\nSpeaker D: Yeah.\nSpeaker A: The computer programs are good.\nSpeaker A: Yeah.\nSpeaker A: It could be really straightforward.\nSpeaker A: Yeah.\nSpeaker A: I think the only thing was having to remember to, you know, take the okay.\nSpeaker A: Yeah.\nSpeaker A: I think I was the only one who struggled with that.\nSpeaker C: I'm actually not sure if I'm staying with my presentation.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker D: I'll probably still be there.\nSpeaker B: No ideas found.\nSpeaker B: I don't really know what that means.\nSpeaker A: I think we've all learned stuff from each other.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: About each different.\nSpeaker A: Got any ideas for each other?\nSpeaker A: I'm not sure.\nSpeaker D: No ideas found.\nSpeaker D: Yeah.\nSpeaker D: Well, I guess we really, we bounced off of each other, which was cool.\nSpeaker D: Yeah.\nSpeaker D: Like based on marketing stuff and then you'd say something about interface and.\nSpeaker D: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nNone: Yeah.\nSpeaker D: Oh, does it have smart materials by the way?\nSpeaker D: Sorry.\nSpeaker D: Does it have smart materials?\nSpeaker A: Oh yeah.\nSpeaker A: Well, did it come with it?\nSpeaker A: I don't know if we counted that in the cost.\nNone: Okay.\nNone: Okay.\nSpeaker B: Well, that achieved our last slide is our closing slide.\nSpeaker B: Yes, our costs are within budget.\nSpeaker B: Yeah.\nSpeaker B: It's evaluated generally, positively.\nSpeaker B: Yeah.\nSpeaker B: And don't forget to complete final question.\nSpeaker B: Meeting summary, then we celebrate such a way that I have no idea.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker A: Thank you very much.\nSpeaker A: Okay.\nSpeaker B: Bye.\nSpeaker B: Okay.\n"}, {"meeting_id": "TS3007d", "summary": "The group mainly talked about the interface design and price setting of the product. Firstly, the user interface designer gave a presentation on the prototype of the product. They provided the prototypes with a lot of details on the appearances and those details were then further discussed by the group.After that, the marketing expert used a seven-point scale to evaluate the prototype according to the past user requirements and trends analysis. They Cleared some misconceptions and reached a consensus on the targeting group and how they could adapt the product better to suit their needs. Then they discussed the price setting of the product and how many functions needed to be retained or removed. As they were trying very hard to remain the price at twelve and a half Euro, they squabbled over the plus and minus of the functions. In the end, they finalized on the costs and planned a celebration for the closing down of the project.", "dialogue": "None: Mr. Big.\nNone: You still have some minutes left, don't we?\nNone: Yeah.\nNone: Or a minute?\nNone: Sorry.\nNone: We still have some time left for the meeting.\nNone: Yeah.\nSpeaker D: One minute.\nSpeaker D: Like a minute.\nSpeaker D: Yeah.\nSpeaker E: Do whatever you want.\nSpeaker E: Oh, what a lot of other.\nSpeaker E: Yeah.\nNone: Oh, maybe you can display it on the screen.\nNone: I'm just doing this, sorry.\nSpeaker C: Nah, this is just about...\nSpeaker C: Okay.\nSpeaker C: I have some shok over here.\nNone: It tastes funny.\nNone: What if you don't?\nSpeaker C: It's like mixed chocolate milk and coffee or something.\nNone: Oh, that's good.\nNone: No, Steve.\nNone: Picture's mixed after each other in the mouth.\nNone: Shana pot.\nNone: Yeah, that would be better.\nSpeaker E: Maybe easier with open and...\nSpeaker E: Oh, see.\nSpeaker E: Okay.\nNone: Did you put it in the chat for a while?\nNone: Yeah.\nNone: Oh, we can begin.\nNone: We have to the meeting room and bring your letter.\nNone: Thank you.\nSpeaker C: That's tight too.\nNone: Welcome to the...\nNone: Sorry.\nNone: Yeah.\nNone: Alright.\nSpeaker E: Maybe good for a presentation if I just...\nNone: Not sure if yet.\nNone: Oh, I see.\nNone: Okay.\nNone: But welcome to the fourth meeting, the details design meeting, where we have to reach the session on the details of the remote control.\nSpeaker D: Presented by your...\nSpeaker D:...pornomally...\nSpeaker D:...protic manager.\nSpeaker D: What we're going to do?\nSpeaker D: Once again, I'm going to take minutes, so no presentation for me.\nSpeaker D: First, we have a prototype presentation by G and G.\nSpeaker D: After what some evaluation criteria...\nSpeaker D: Evaluation criteria.\nSpeaker D: Sorry.\nSpeaker D: Evaluation criteria.\nSpeaker D: In combination with finance, I received an Excel file, which we have to fill in later on.\nSpeaker D: You see?\nSpeaker D: And then we must see if we stay under the 12 and a half euro.\nSpeaker D: So that's...\nSpeaker A: That's a bit...\nSpeaker D: So let's...\nSpeaker D:...proble.\nSpeaker D: We have...\nSpeaker D: We must have some time for that because it will be quite a lot of mathematics.\nSpeaker D: After that evaluation of the process, how we have done it here with the Smartboard, with our laptops, with all this.\nSpeaker D: And afterwards, we're closing.\nSpeaker D: Once again, 40 minutes.\nSpeaker D: Let's start.\nSpeaker D: I will give the word to G and G for the prototype.\nSpeaker D: Presentation.\nSpeaker E: Presentation.\nSpeaker E: And then...\nSpeaker D: Oh, well, sure.\nSpeaker D: J and J.\nSpeaker C: J and J.\nSpeaker C: J and J.\nSpeaker C: Okay, guys.\nSpeaker C: Take it away.\nSpeaker C: Take it away.\nNone: Tsk.\nSpeaker E: This was our first concept.\nSpeaker E: We decided to use a single touchscreen.\nSpeaker E: So we've worked out these concepts, how to hold it, where to put it, what's this and stuff.\nSpeaker E: And, well, we began with a form of shape that is easy to hold in one hand, left or right-handed.\nSpeaker E: So we made it a little bit less thick and have some artistic meaning.\nSpeaker E: Oh, yes, it's not that easy.\nSpeaker E: Yeah, that's better.\nSpeaker E: Well, during the meeting, I showed you the concept of placing the buttons on top, usable with your thumb, and the menu structure, if necessary, with your other hand.\nSpeaker E: So it's going to hold it easily, and it has to be accessible with your other hand too, of course.\nSpeaker E: So we began working our concepts.\nSpeaker B: Yeah, well, as usual, we would just have to basically mode with the panel LCD screen.\nSpeaker B: Well, these would be the main buttons.\nSpeaker B: You could change them later on in your own profile if you want to, but, well, the standard will be delivered with this kind of setup, with the more advanced menu sitting right here, with the sub-menus and stuff.\nSpeaker B: We made a top or front view, just like you want a back view.\nSpeaker B: As you can see, this is two weird bumps in it.\nSpeaker B: This is for the added effect of our youth and dynamic, and this is for artistic effect.\nSpeaker B: Well, what we figured is, we'll show you a picture later on, you'll have a better idea after that, but the idea is to stay in balance with these two.\nSpeaker B: So when you put it on the table, it will just lay down, it won't roll around and stuff, but it will lie more in your hand like an old telephone, maybe, or like these old phones.\nSpeaker B: You may get the idea.\nSpeaker B: So this is about how we figured it should be.\nSpeaker B: You would hide with some more rubber layers, like we discussed earlier.\nSpeaker B: You wouldn't see a straight panel, but more fluidly and round.\nSpeaker E: The panel just, of course, goes like this.\nSpeaker E: But the overlaying layer is a little bit cursed and stuff.\nSpeaker B: Okay. And in these bumps, you could actually put some electronics out of which you could make a more thinner design, and that would actually look very nice.\nSpeaker B: And about the colour.\nSpeaker E: Oh, yeah, that this can be held with your hands.\nSpeaker E: So this maximum is one and a half centimeters.\nSpeaker E: So you have room here for your battery, and maybe even other electronic chips.\nSpeaker E: So you can just be the layer of the touchscreen and some wires underneath it to make it as clean as possible in the middle for a good, great.\nSpeaker B: As cutters, do you have a picture in there?\nSpeaker B: Yeah.\nSpeaker B: Oh, this is the idea about the bumps.\nSpeaker B: You can see there's a very youthful dynamic exterior.\nSpeaker B: You just want to hold it if you were young and dynamic like us.\nSpeaker C: It's like an Easter egg.\nSpeaker B: Yeah, it's like, but this is for children, we want a more adult version, but this is like a remote control for children.\nSpeaker B: It's called a we-mode.\nSpeaker B: We-mode.\nSpeaker B: A we-mode, yeah.\nSpeaker B: It's actually brilliant. Mark Tingston.\nSpeaker B: Yeah.\nSpeaker C: Wait, what I got in mind.\nSpeaker B: So this is actually basically the idea.\nSpeaker B: We just want to build a more adult version of this.\nSpeaker B: Yeah, I can imagine that.\nSpeaker B: And for colors, we figured starting with basic colors like white or metallic gray.\nSpeaker B: Those are the technological colors actually.\nSpeaker E: It would be best to appeal to a broad public and make the conference exchangeable, so the young people will buy an orange and a red and a blue one.\nSpeaker E: Sure, but when older people go in the shop and see an orange remote control, it would be less appealing than a white one.\nSpeaker E: And young people we think are a little bit more flexible.\nSpeaker E: I'll buy for a couple of euros.\nSpeaker E: It's a nice hit.\nSpeaker C: Maybe it's an idea to sell it without a cover.\nSpeaker C: So that you can pick a cover in the shop.\nSpeaker E: I think a cover is necessary because I'll just have the LCD screen.\nSpeaker E: There must be some cheap standard cover, maybe white or something that comes with it and you can buy it so we can make it.\nSpeaker D: Yeah, but you mustn't forget that our target aim is younger people.\nSpeaker D: We had decided to put some flashy fruit-free colors in it.\nSpeaker D: And in the surfy from Milan and Paris, it came out that the older people are more willing to spend money on extra futures.\nSpeaker D: So I think it will be a better idea to have some flashy-free colors as standard for the people who really want a more sophisticated, more traditional look.\nSpeaker D: They are willing to pay that.\nSpeaker D: They want more luxury stuff, but they have the money to do it and they want to buy that.\nSpeaker D: So maybe it's an idea to put that as an extra.\nSpeaker B: That's a standard.\nSpeaker B: Yeah, perhaps, all right.\nSpeaker B: I would agree with this sound logical.\nSpeaker C: Another idea.\nSpeaker C: Maybe we could develop a cover with wood style.\nSpeaker C: They would please the elder users as well.\nSpeaker E: Well, yeah, a color of wood style, a white and a couple of pretty colors.\nSpeaker E: And the leverage standard with a pretty color but not too much.\nSpeaker E: Banana and mango, not purple or orange.\nSpeaker D: But I think the standard must be some kind of attractive, flashy color.\nSpeaker D: Not too button, a little.\nSpeaker B: Yeah, like this.\nSpeaker B: This isn't too much, is it?\nSpeaker D: Well, the buttons.\nSpeaker C: It could be a standard model.\nSpeaker B: Yeah, something like this would be nice.\nSpeaker C: Okay, that's for most.\nSpeaker C: Okay, it's my time now.\nSpeaker D: The marketing expert.\nSpeaker C: Okay.\nSpeaker C: During the design lifecycle, we made a lot of requirements and trend analysis and stuff.\nSpeaker C: Now is the time to evaluate our prototype concept to the past requirements.\nSpeaker C: So we are going to evaluate the design according to the past user requirements and trends analysis.\nSpeaker C: We're going to do that with a seven point scale.\nSpeaker C: Opening a word document now.\nSpeaker C: Okay.\nSpeaker C: One.\nSpeaker C: I have to explain something.\nSpeaker C: We have to be consensus about things.\nSpeaker C: So it has to be a group decision.\nSpeaker D: Okay.\nSpeaker D: So we're going to evaluate the.\nSpeaker C: Yeah, the thing we need.\nSpeaker C: Okay. One.\nSpeaker C: The remote controls designed for people with age below 40.\nSpeaker D: Yeah.\nSpeaker D: Seven.\nSpeaker D: Seven is fourth.\nSpeaker B: True.\nSpeaker B: Yeah.\nSpeaker B: One.\nSpeaker B: Yeah, one.\nSpeaker B: I think most true.\nSpeaker C: Yeah, it's not just designed for people under the age of 40.\nSpeaker B: It's also designed for people above 40.\nSpeaker B: Yeah, it's not just the question is aimed at design for people with age below 40.\nSpeaker B: But it's also designed for people above 40.\nSpeaker B: Exactly.\nSpeaker B: Also, it's about three.\nSpeaker E: Yeah, it will be primarily appealing to minus 40.\nSpeaker E: Yeah.\nSpeaker C: Also, yeah.\nSpeaker C: Okay.\nSpeaker C: The second.\nSpeaker C: The remote controls.\nSpeaker C: Beautiful.\nSpeaker C: Yeah.\nSpeaker C: Wow.\nSpeaker C: Yeah.\nSpeaker C: According to us, it's one.\nSpeaker C: Yeah.\nSpeaker C: It's the marketing.\nSpeaker E: Yeah.\nSpeaker E: We have a one.\nSpeaker C: Of course, you have to be very positive and enthusiastic about your own problem.\nSpeaker C: That's also three.\nSpeaker C: The remote control looks fancy.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: One.\nSpeaker C: Yeah.\nSpeaker C: Perfect.\nSpeaker C: Good.\nSpeaker C: For the remote control, it's big, clear, channel switching buttons.\nSpeaker E: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: These two user faces.\nSpeaker E: User interface experts.\nSpeaker C: Data text buttons and volume buttons.\nSpeaker C: No, data text.\nSpeaker D: But data text is in the menu.\nSpeaker B: False?\nSpeaker B: False.\nSpeaker C: And volume.\nSpeaker C: Volume is true.\nSpeaker C: Big and clear.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Big and clear.\nSpeaker E: Yeah.\nSpeaker E: Big and clear.\nSpeaker E: But you could make a teletext button six.\nSpeaker E: Otherwise people would read this.\nSpeaker E: I'm going to think we have no teletext button.\nSpeaker B: Yeah.\nSpeaker B: It's not in your show.\nSpeaker B: It isn't entirely unclear.\nSpeaker B: So I wouldn't give it a seven.\nSpeaker B: I would give it a five or a six.\nSpeaker B: Five?\nSpeaker B: Okay.\nSpeaker B: I don't know.\nSpeaker B: Yeah.\nSpeaker D: It's a project.\nSpeaker D: Well, I agree.\nSpeaker D: I was thinking very black and white.\nSpeaker B: Yeah.\nSpeaker E: Looking at it.\nSpeaker D: Thank you, Jenny.\nSpeaker D: Brad.\nSpeaker C: Okay.\nSpeaker C: I forget to say to volume.\nSpeaker C: The remote control is easy to be found.\nSpeaker E: Well, we put fancy colors here.\nSpeaker B: And it has all these fruity colors and it has a strange shape.\nSpeaker B: So if you have trouble finding it.\nSpeaker E: It's not making any sounds.\nSpeaker E: Okay.\nSpeaker C: So, look.\nSpeaker C: If you put your normal remote control under your bed, or you throw the S remote control under your bed, it's a better finding.\nSpeaker E: It will make a difference.\nSpeaker E: We have the better remote.\nSpeaker E: I don't know.\nSpeaker E: I think so.\nSpeaker E: My remote control is black.\nSpeaker E: A little bit maybe.\nSpeaker C: A little bit.\nSpeaker C: Yeah.\nSpeaker D: We can do it glow in the dark.\nSpeaker D: So even the dark place you feel.\nSpeaker C: I think five.\nSpeaker C: It doesn't really make a lot of.\nSpeaker B: Well, then I'll go for four.\nSpeaker B: Four?\nSpeaker B: Because four is between three and between two.\nSpeaker E: Okay, you're right.\nSpeaker D: I think five is between four and six.\nSpeaker D: According to the other remote controls, there may be there in your TV room.\nSpeaker D: There's one we'll stand out.\nSpeaker D: Yeah, that's a better question.\nSpeaker D: I think that's one more spot.\nSpeaker D: If this line on your couch, you think, what's that for?\nSpeaker C: Sure, sure.\nSpeaker C: The survey under users was that they really lost it.\nSpeaker E: They're stupid.\nSpeaker C: Like not seeing it, but lost it in the house or something.\nSpeaker B: Okay.\nSpeaker B: If you see a strange shape line somewhere, then you'd recognize it as, whoa, that's strange.\nSpeaker E: That's a remote control.\nSpeaker C: What is that?\nSpeaker C: I agree, I agree.\nSpeaker C: Okay, most of the time, you lose it your shirt.\nSpeaker E: So your remote controls are fresh through the collar.\nSpeaker E: I would call it a choose two because we decided not to make two fresh collars.\nSpeaker C: Not too fresh.\nSpeaker C: The remote controls are made of soft material.\nSpeaker C: Yeah, rubber is kind of soft.\nSpeaker B: Yeah, but not too soft.\nSpeaker B: Not too soft.\nSpeaker D: Three?\nSpeaker D: Three.\nSpeaker D: Yeah.\nSpeaker D: Yeah, easy to use.\nSpeaker D: Easy to use?\nSpeaker D: Yeah, can it be zero?\nSpeaker B: Well, I don't, yeah, it's just kind of easy to use.\nSpeaker C: It's not the most easy to use.\nSpeaker E: No, it can do two because...\nSpeaker E: It can be easier.\nSpeaker E: It can be easier.\nSpeaker E: Just with ten buttons.\nSpeaker E: Yeah, but then you'll lose our functionality.\nSpeaker E: Yeah, functionality and our fancy look.\nSpeaker C: Yeah, but the most easy to use.\nSpeaker C: It's just with one button.\nSpeaker E: The other easy to use because you have primary buttons always.\nSpeaker C: Yeah, okay, but not the most easy to use.\nSpeaker B: No, actually, I'll go for two.\nSpeaker B: My vote's on two.\nSpeaker B: Yeah, mine too.\nSpeaker C: We also have to compare it to the remote controls on the market nowadays.\nSpeaker E: Yeah, wait, it's just a minute.\nSpeaker E: Inspiration.\nSpeaker D: What's the time we also have to do the evaluation and the production.\nSpeaker C: Regular remote stuff.\nSpeaker C: Yeah, I'm hurrying.\nSpeaker C: Okay, eleven.\nSpeaker C: The remote controls and all of this.\nSpeaker C: Yes, yes.\nSpeaker E: Two, one.\nSpeaker E: You agree, Tim?\nSpeaker D: Of course.\nSpeaker D: You have to see the more innovative thing in there.\nSpeaker C: The remote controls, remove both of them.\nSpeaker E: Yes, one.\nSpeaker E: Very multifunctional.\nSpeaker E: No.\nSpeaker C: The remote control.\nSpeaker C: It has speech recognition.\nSpeaker E: This is used with speech recognition.\nSpeaker C: The remote control has building games.\nSpeaker C: Yes.\nSpeaker E: But maybe make it too because the games are in a sub-menu and not, it's not the one.\nSpeaker B: Yeah, but they are built in.\nSpeaker E: So it's one.\nSpeaker E: Oh, no, I'll build in.\nSpeaker B: Yeah, okay.\nSpeaker B: Okay.\nSpeaker C: The lost.\nSpeaker B: But, I thought it was going to be a bit less.\nSpeaker B: You really like the parental advice.\nSpeaker B: Yes.\nSpeaker E: Did you make this or the safe house?\nSpeaker B: Yeah, he made it.\nSpeaker E: He changed it, man.\nSpeaker C: Okay.\nSpeaker C: Oh, yeah.\nSpeaker C: The moss.\nSpeaker C: Oh, dear.\nSpeaker D: How is your turn?\nSpeaker D: Okay.\nNone: Thank you.\nSpeaker A: Let's see.\nSpeaker A: Hmm.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker D: We have now to calculate the production cost.\nSpeaker D: If it's under 12 and a half euro, then it's a.\nSpeaker D: Okay.\nSpeaker D: But it is, if it is, but no, this is right.\nSpeaker D: No.\nSpeaker D: Redesign.\nSpeaker D: If you're under 50.\nSpeaker D: Yeah.\nSpeaker D: No.\nSpeaker D: Oh, yeah.\nSpeaker D: Yeah.\nSpeaker D: Yes.\nSpeaker A: Yes.\nSpeaker D: Sorry.\nNone: Yeah.\nSpeaker D: It's a car.\nSpeaker D: It's under 12 and a half euro.\nSpeaker D: Then we can move on to the project evaluation as we have experienced it.\nSpeaker D: Otherwise we have to do a little redesign thingy.\nSpeaker D: So we have to fill in the numbers of the components.\nSpeaker D: We have to want to do it in and see if we stay under the 12 and a half euro.\nSpeaker D: So do we have a hand dynamic?\nSpeaker D: No.\nSpeaker D: That's zero.\nSpeaker D: Yes.\nSpeaker D: Battery.\nSpeaker D: One.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Simple chip on print?\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: Simple sensor.\nSpeaker D: Simple speaker.\nSpeaker C: No.\nSpeaker C: It's fast chip.\nNone: It's three.\nSpeaker C: Three.\nSpeaker C: Yes.\nSpeaker C: One.\nSpeaker C: One.\nSpeaker C: One piece.\nSpeaker D: Three euro.\nSpeaker D: No.\nSpeaker B: What's a sample sensor?\nSpeaker B: Well, that's speech recognition.\nSpeaker B: Yeah.\nSpeaker D: It's speech recognition.\nSpeaker D: Okay.\nSpeaker D: One.\nSpeaker D: Zero.\nSpeaker D: Uncurfed flat.\nSpeaker D: No.\nSpeaker D: Zero.\nSpeaker D: No.\nSpeaker D: But it's not made from a single uncurfed thingy and then...\nSpeaker A: No.\nSpeaker D: Thingy.\nSpeaker D: And then...\nSpeaker D: No.\nSpeaker D: Okay.\nSpeaker D: So it's only once double curved.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker E: Yeah.\nSpeaker E: Because the layer around it fits around the bubbles on the...\nSpeaker E: Okay.\nSpeaker D: We're now in a problem.\nSpeaker D: Because we have...\nSpeaker D: We have 11 euro.\nSpeaker D: Yeah.\nSpeaker E: We don't have anything else.\nSpeaker E: Okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: We'll see how much we are over budget.\nSpeaker E: Okay.\nSpeaker E: You step a little to the wrong.\nSpeaker E: Yeah.\nSpeaker E: Sorry.\nSpeaker D: Two.\nSpeaker D: Rubber.\nSpeaker D: You don't...\nSpeaker D: Ditani, I'm no.\nSpeaker E: Special.\nSpeaker E: It's the special color.\nSpeaker B: I don't think so.\nSpeaker B: No.\nSpeaker B: It's a standard color.\nSpeaker A: Yes.\nSpeaker B: Yes.\nSpeaker B: This is a special color.\nSpeaker D: We want to make the wood color.\nSpeaker D: So that's...\nSpeaker E: Yeah.\nSpeaker E: If you're honest with you all the time, you have a special color.\nSpeaker C: That's an adult.\nSpeaker C: Yeah.\nSpeaker C: That's another brand.\nSpeaker C: That's another article.\nSpeaker C: The cell.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But we're going to...\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: That's true.\nSpeaker D: But yeah.\nSpeaker D: It doesn't account for this.\nSpeaker E: Maybe we'll finish this.\nSpeaker D: And then look back.\nSpeaker D: Okay.\nSpeaker E: No.\nSpeaker E: Scoville.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: Yes.\nSpeaker D: One.\nSpeaker D: One.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: We don't have...\nSpeaker B: No.\nSpeaker B: We're only four of our budget.\nSpeaker D: Okay.\nSpeaker D: So, what's the thing we can change?\nSpeaker B: Well, are there a case?\nSpeaker B: We can make it single-curved or uncurved.\nSpeaker A: Can I...\nSpeaker A: I say something?\nSpeaker A: Yeah.\nSpeaker A: Of course.\nSpeaker D: Can I say something?\nSpeaker D: Asparatic management.\nSpeaker D: Just...\nSpeaker D: The kinetic thing.\nSpeaker D: Yeah.\nSpeaker D: Can we just skip it?\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: You have to shake it, but that's not really enough.\nSpeaker E: Yeah.\nSpeaker E: We just put it in battery and it's mobile phone nowadays.\nSpeaker C: Daniel.\nSpeaker C: Daniel.\nSpeaker C: Daniel.\nSpeaker C: Yeah.\nSpeaker C: What do you think about putting a battery in it?\nSpeaker C: Yeah.\nSpeaker C: But also selling like the coffee or a docking station.\nSpeaker C: Just apart from the thing.\nSpeaker C: Mm-hmm.\nSpeaker C: And then you can put rechargeable batteries in it.\nSpeaker C: Okay.\nSpeaker E: Just...\nSpeaker E: What you can use rechargeable batteries anyway.\nSpeaker E: Just...\nSpeaker E: Yeah.\nSpeaker D: Not really enough.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: But we...\nSpeaker D: You forget about the kinetic.\nSpeaker D: Yeah.\nSpeaker D: Sure.\nSpeaker D: Well, if we do that...\nSpeaker D: Which are you?\nSpeaker B: Thirteen to health.\nSpeaker B: Okay.\nSpeaker B: Well, you can go from double-curved to single-curved.\nSpeaker C: Yeah.\nSpeaker B: Of course.\nSpeaker B: And that would solve the budget problem.\nSpeaker B: So we have to break the...\nSpeaker D: But the single-curved is just...\nSpeaker D: It's just...\nSpeaker D: Oh.\nSpeaker E: Yeah.\nSpeaker E: Well, the single-curved...\nSpeaker E: One-curved and not the back-curved I think.\nSpeaker D: Okay.\nSpeaker D: Yeah.\nSpeaker D: Exactly.\nSpeaker D: That's one option.\nSpeaker D: And then, yeah.\nSpeaker D: Then we could have it.\nSpeaker D: Yeah.\nSpeaker D: It's...\nSpeaker D: It's a main point of the...\nSpeaker D: Yeah.\nSpeaker D: But we'll look.\nSpeaker E: Then we have to cut out.\nSpeaker E: We don't fall.\nSpeaker E: It's a little bit of problem.\nSpeaker E: No, no, no.\nSpeaker D: That can be done.\nSpeaker D: So...\nSpeaker E: Oh, though, can we make it with the back- Okay.\nSpeaker D: A little less...\nSpeaker D: Conversation.\nSpeaker D: Hey.\nSpeaker C: Those are...\nSpeaker C: Art.\nSpeaker C: Art.\nSpeaker C: What are they for?\nSpeaker C: Some...\nSpeaker C: Blue...\nSpeaker D: Blue...\nSpeaker D: Fill in the mirror.\nSpeaker D: Just...\nSpeaker D: Okay.\nSpeaker D: Explanation.\nSpeaker C: So 50?\nSpeaker D: I can delete it for you if you want.\nSpeaker C: Oh, no.\nSpeaker D: So if we do this, we are on 12.\nSpeaker D: And a half euro.\nSpeaker D: And we're done?\nSpeaker D: Yeah.\nSpeaker E: But does it fit with our design?\nSpeaker E: Maybe actually...\nSpeaker D: Well, the only thing that don't...\nSpeaker D: The single curve.\nSpeaker D: Yeah, single curve.\nSpeaker D: But there's a curve in it.\nSpeaker D: So...\nSpeaker E: We just make the bubbles cut off the back and then...\nSpeaker E: Yeah, we just make it flat.\nSpeaker B: Okay.\nSpeaker B: But you do look...\nSpeaker C: Look, what is the...\nSpeaker C: If you make it double curve.\nSpeaker C: It costs one euro more.\nSpeaker C: Yeah.\nSpeaker C: You make it optional.\nSpeaker C: No, but...\nSpeaker C: Does it have a lot of extra...\nSpeaker C: Function?\nSpeaker B: Actually...\nSpeaker C: Function more.\nSpeaker C: What does it have added with?\nSpeaker C: A static value, but not function.\nSpeaker D: Yeah, it's really a static value.\nSpeaker C: A static.\nSpeaker C: Yeah.\nSpeaker C: I mean, you make like 11 and a half euros a profit.\nSpeaker C: Instead of 12 and a half, but...\nSpeaker C: I don't know if 12 and a half is a fixed price.\nSpeaker C: Now, well, that's a true price.\nSpeaker B: We should assume that it is.\nSpeaker B: Yeah.\nSpeaker B: I think the kinetic would be a marketing promotion.\nSpeaker B: If you promote kinetic...\nSpeaker B: I...\nSpeaker B: Kinetic remote control.\nSpeaker B: I mean, that would sell better than a normal remote control.\nSpeaker D: Yeah.\nSpeaker D: Do you think?\nSpeaker D: Yeah.\nSpeaker D: Well, now you can shake your remote control.\nSpeaker B: No, well, I mean, you can count to your neighbor and tell them, Ha!\nSpeaker B: My remote control is kinetic.\nSpeaker B: You have standard old battery control remote control.\nSpeaker C: Yeah.\nSpeaker C: What about all the environments freaks?\nSpeaker C: Yeah, but it does fit in a cost profile.\nSpeaker C: No.\nSpeaker C: I think it's...\nSpeaker E: It's a good light there, as well.\nSpeaker E: You can make it in a specialized extra gold version.\nSpeaker D: Because if you want to go to kinetic, you're on turning a half and you must go to flat.\nSpeaker D: And I think now it's more of a compromise.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And if you make the single curve, just a big curve, then it's a...\nSpeaker D: Yeah, one big...\nSpeaker A: One big...\nSpeaker D: Good curve.\nSpeaker D: I was coming to a...\nSpeaker D: Say nasty words, but I don't see.\nSpeaker D: This is strange, by the way.\nSpeaker D: Wood is cheaper than rubber.\nSpeaker D: We thought that wood would be more expensive.\nSpeaker D: It's American figures.\nSpeaker E: You just cut down some trees.\nSpeaker E: Yeah, maybe.\nSpeaker D: But this is it?\nSpeaker D: Yeah, this is it.\nSpeaker D: Okay, this is it.\nSpeaker D: I'm going to save it.\nSpeaker E: I'm going to save it.\nSpeaker E: Titanium.\nSpeaker E: Yeah.\nSpeaker B: It is possible.\nSpeaker B: But you can't use double curves for titanium.\nSpeaker B: That's one of the functional biddity.\nSpeaker D: Okay.\nSpeaker D: Well, considering we have...\nSpeaker C: Oh, no.\nSpeaker C: I have to do all those hours again.\nSpeaker C: One back.\nSpeaker C: Gas on.\nSpeaker C: No, for redesign.\nSpeaker D: Well, we were above.\nSpeaker D: Yeah, we did a little redesign.\nSpeaker E: Yeah, it's that rule.\nSpeaker D: Okay.\nSpeaker D: Now, it's about time to talk about this project.\nSpeaker D: Some things.\nSpeaker D: Was there room for creativity?\nSpeaker D: In our meetings, or in your individual meetings?\nSpeaker C: I didn't think so.\nSpeaker C: There was a lot of room for it.\nSpeaker C: But that's mainly because of the information that was delivered to us.\nSpeaker C: Yeah.\nSpeaker C: It was just fixed information and get your stuff from there.\nSpeaker C: And I couldn't go on the internet and share my own stuff.\nSpeaker D: Yeah.\nSpeaker D: That's true.\nSpeaker D: I agree with that.\nSpeaker B: Well, I think you too.\nSpeaker B: Especially you and the...\nSpeaker B: I don't know, you both had the less creative roles in the project.\nSpeaker D: That's true.\nSpeaker E: For us, I think...\nSpeaker E: Also, I think we could just sign up and remove it if you like.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker B: That's true.\nSpeaker B: I think you're run and I.\nSpeaker B: We had more design.\nSpeaker B: We could have more...\nSpeaker B: We had more room for creativity than you too.\nSpeaker C: Yeah.\nSpeaker C: Yeah, that's true.\nSpeaker D: Okay.\nSpeaker D: How about leadership?\nSpeaker D: Leadership was crappy.\nSpeaker C: The leadership wasn't crappy.\nSpeaker B: It was the leader that was crappy.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Thank you.\nSpeaker B: Leadership was okay.\nSpeaker B: We're done.\nSpeaker E: Example of crappy leadership.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker E: Okay.\nSpeaker C: I thought the first meeting was a little bit of unstructured meeting.\nSpeaker C: Yeah.\nSpeaker C: You could have...\nSpeaker C: But it was your first?\nSpeaker C: No disrespect or something.\nSpeaker C: But you could have structured a little bit more.\nSpeaker C: Yeah.\nSpeaker C: So, like, I was talking most of the time, the first meeting.\nSpeaker C: You could say shut up.\nSpeaker E: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: But I know the city too.\nSpeaker D: I was also very unhappy, very unsatisfied about...\nSpeaker D: About me.\nSpeaker D: About the first meeting.\nSpeaker D: So, I hope...\nSpeaker D: That's a little fun on the stage.\nSpeaker B: And we will never do it again.\nSpeaker D: Get better.\nSpeaker D: And I think the last two meetings also, we reached some good decision.\nSpeaker D: Yeah.\nSpeaker D: More consensus.\nSpeaker E: Yeah.\nSpeaker E: Everybody was really...\nSpeaker D: Much more constructive.\nSpeaker D: Okay.\nSpeaker D: So, that's cool.\nSpeaker D: Teamwork?\nSpeaker D: Well, maybe that's only...\nSpeaker D: Yeah, well, it's for us.\nSpeaker E: Well, we worked together on the project, but everybody has his own tasks.\nSpeaker E: Yeah.\nSpeaker D: And it's more like presentation.\nSpeaker D: And some points were discussed.\nSpeaker D: Yeah.\nSpeaker D: Really teamwork where you two...\nSpeaker D: So, I just went to crazy.\nSpeaker B: Yeah, it went well.\nSpeaker B: It's just...\nSpeaker B: It was hard to feel.\nSpeaker B: Stupid then, but...\nSpeaker B: Yeah, we had some trouble with the pen.\nSpeaker D: Now you must push a little while.\nSpeaker E: Yeah, but it was something difficult.\nSpeaker E: Just write your name right now.\nSpeaker E: Try to write your name.\nSpeaker E: Okay.\nSpeaker E: In writing letters, of course.\nSpeaker E: Yeah, normally...\nSpeaker E: This is a block letter sign there.\nSpeaker E: Just write your name in one line.\nSpeaker E: Just a little bit small, ankylric, quicker now.\nSpeaker D: You can be...\nSpeaker D: You can go quicker, because then it won't notice it.\nSpeaker E: Oh, he knows how to work.\nSpeaker D: I followed him out to class for the smartboard.\nSpeaker D: I think that's the main issue.\nSpeaker D: Okay.\nSpeaker D: So, about this one you were...\nSpeaker D: You were dealing with?\nSpeaker D: Yeah.\nSpeaker D: The digital pen?\nSpeaker E: The idea is great, but it works properly.\nSpeaker C: The digital pen, I thought, the first time I did individual work, I used it.\nSpeaker C: Yeah.\nSpeaker C: And the first two meetings, I brought it with me, but I didn't use it at all after the first meeting.\nSpeaker C: No, I haven't proven it.\nSpeaker B: You know, I was not real used.\nSpeaker B: No, I don't have that much added value to it.\nSpeaker E: Yeah.\nSpeaker E: As I said a few moments ago, I would like myself to write it with a normal pen, because it must...\nSpeaker E: Yeah, it's almost the same concept, but you can just more simply put it on our scanner.\nSpeaker E: It's the same concept.\nSpeaker E: It must depend.\nSpeaker E: You have to download software, very slow.\nSpeaker D: And it is still your own handwriting, up and up.\nSpeaker E: It doesn't give any added value.\nSpeaker D: No, it's really not.\nSpeaker D: And the smartboard is useful, but the pen is not user-friendly, I think.\nSpeaker D: Yeah, not user-friendly.\nSpeaker D: It takes a lot of time to draw things and write things.\nSpeaker C: It's not very precise.\nSpeaker C: We try to do this.\nSpeaker E: Yeah, and try to write your name in a normal size.\nSpeaker E: Smaller.\nSpeaker E: Smaller?\nSpeaker C: Yeah, smaller.\nSpeaker E: Just when you're writing on a letter.\nSpeaker D: Yeah, but that's not...\nSpeaker D: When you send an added foreign audience, you don't gonna write...\nSpeaker E: As you saw on this drawing, it's open this one or that one.\nSpeaker E: Yeah, we have more problems even here.\nSpeaker E: When we try to draw this button, it's almost impossible for clear roundings when you're...\nSpeaker D: Oh.\nSpeaker D: But maybe there's some function with...\nSpeaker D: No, it isn't.\nSpeaker E: And the eraser was not a problem.\nSpeaker E: It's this large.\nSpeaker E: I mean, you try to erase this line.\nSpeaker D: Yeah.\nSpeaker D: Well, I'm gonna erase my name.\nSpeaker E: I'm gonna erase my name.\nSpeaker D: Yeah, it's a big eraser.\nSpeaker C: Okay, new ideas.\nSpeaker E: Erm, what kind of new ideas?\nSpeaker D: Well, the idea of the touch screen is...\nSpeaker D: Yeah, two years.\nSpeaker D: Go on.\nSpeaker D: I'm just...\nSpeaker E: New ideas about the working of this software, about the project, about remote controls.\nSpeaker D: I don't know what I mean.\nNone: I don't know what I mean.\nSpeaker D: Did you heard what you said?\nSpeaker D: I don't...\nSpeaker C: I don't know what I mean.\nSpeaker C: Oh, I have some figure.\nSpeaker C: Here.\nSpeaker C: The evaluation.\nSpeaker C: Oh, yeah.\nSpeaker C: The mean number is 1.8, 1.86.\nSpeaker C: That's interesting.\nSpeaker C: So that's fairly good, I think.\nSpeaker C: Okay.\nSpeaker D: Because what does it mean?\nSpeaker C: So true.\nSpeaker C: That...\nSpeaker C: All the requirements are between 1 and 2.\nSpeaker D: True or varied.\nSpeaker D: True.\nSpeaker D: Oh, okay.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Thank you.\nSpeaker D: Expert.\nSpeaker D: But the new ideas found with working with this software?\nSpeaker E: Not really.\nSpeaker E: They have to improve.\nSpeaker E: The concept is okay.\nSpeaker E: Yeah, I think.\nSpeaker E: But it has to be quicker.\nSpeaker E: It's still opening my program.\nSpeaker E: How much of my entire computer is locked up?\nSpeaker E: Yeah.\nSpeaker E: It just takes too many time.\nSpeaker E: People will still feel the need to write it quickly on a page and not download it and save it.\nSpeaker D: You're expected to be more user friendly.\nSpeaker E: When you use a pen, you can just draw like you do normally.\nSpeaker C: Maybe the idea you proposed is a screen here.\nSpeaker C: You can draw it and it's placed over there.\nSpeaker E: Yeah.\nSpeaker E: Or at least when you don't have to adapt to the technologies, you can write in a way you normally write.\nSpeaker E: Now you have to keep constantly in mind that you're drawing on this screen and that's a very bad concept.\nSpeaker E: Yeah, that's true.\nSpeaker C: Very bad.\nSpeaker E: No, okay.\nSpeaker C: It's my opinion that this is better than regular flipovers.\nSpeaker E: Yeah, it can be safety easier.\nSpeaker E: But if you're in normal flipover, a lot of people write text.\nSpeaker E: There's not a text option.\nSpeaker E: And writing text, you've got to really do your best.\nSpeaker D: And maybe some functions for a circle or a square.\nSpeaker D: You have to draw it yourself now.\nSpeaker E: Maybe even insert picture if you have some presentation.\nSpeaker B: Or text function.\nSpeaker B: Just type text.\nSpeaker B: That would be excellent.\nSpeaker C: Yeah, but insert image isn't available.\nSpeaker C: Because then you could picture from scanner clip-on.\nSpeaker C: That can be done already.\nSpeaker D: But not the hyperlink predefined squares.\nSpeaker C: What if you do like hyperlink?\nSpeaker E: It was www.google.com.\nSpeaker C: Real reaction.\nSpeaker A: No.\nSpeaker C: That's okay.\nSpeaker E: You'll just make a link in.\nSpeaker E: Well, that's nice.\nSpeaker E: There's one way to double click it.\nSpeaker E: Maybe if you're not using the eraser.\nSpeaker C: Erasing.\nSpeaker C: Something else.\nSpeaker C: Select.\nSpeaker C: Double click.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Well, you have, as you saw, you have a little.\nSpeaker D: Yeah, thank you.\nSpeaker D: You can go.\nSpeaker D: Okay.\nNone: There.\nSpeaker D: So there are the functionality is there.\nSpeaker D: It's not ideal.\nSpeaker D: It costs a lot of time to use.\nSpeaker E: To use.\nSpeaker D: And that's a pity.\nSpeaker D: If you have 30, 40 minutes.\nSpeaker D: Yeah.\nSpeaker D: For this kind of thing.\nSpeaker D: And we are now with four people.\nSpeaker D: Well, imagine you are here with ten people.\nSpeaker D: Everyone.\nSpeaker E: You're here with management.\nSpeaker E: You get two minutes to make your case.\nSpeaker E: Two minutes of drawing.\nSpeaker E: You better use power points and work it out in a font.\nSpeaker E: And one or two things you have to draw when you're there.\nSpeaker C: What I really miss also is a.\nSpeaker C: It's a decision system like.\nSpeaker C: You have to.\nSpeaker C: You have to.\nSpeaker C: Both like what do you want?\nSpeaker C: One or two?\nSpeaker C: Yeah.\nSpeaker C: Maybe a little application like.\nSpeaker C: Give your number and click one, two, three, four, five, six, seven.\nSpeaker D: Yeah.\nSpeaker D: Just like you said with the screen which you can write.\nSpeaker D: Also kind of voting.\nSpeaker C: Yeah.\nSpeaker C: Voting application.\nSpeaker C: Just a little group.\nSpeaker C: Group decision application.\nSpeaker B: But the problem is you can't discuss anything.\nSpeaker B: Well, you can.\nSpeaker B: You can.\nSpeaker B: But you will discuss a lot less than like we did now.\nSpeaker B: Yeah.\nSpeaker B: I mean, one person maybe said free.\nSpeaker B: But we said no.\nSpeaker B: I think too because this and this and then you can react on it.\nSpeaker B: But if you put a free on it.\nSpeaker B: Just figure out everybody knows what I'm knowing.\nSpeaker B: So they all just put a two on.\nSpeaker B: Yeah, of course.\nSpeaker C: But you can still discuss about it.\nSpeaker C: Yeah.\nSpeaker C: But click it in an application.\nSpeaker C: That's a lot easier to process.\nSpeaker B: Yeah.\nSpeaker B: For process.\nSpeaker B: Yeah.\nSpeaker B: But then I think the idea of one person entering it and discussing it.\nSpeaker B: Isn't that bad idea actually.\nSpeaker A: Yeah.\nSpeaker B: Not that your opinion isn't value, but still.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker E: So.\nSpeaker E: That's it.\nSpeaker D: Well, just about.\nSpeaker E: Because when are we going to produce it?\nSpeaker D: Yeah.\nSpeaker D: Tomorrow.\nSpeaker D: Celebrate.\nSpeaker D: Because we're within the budget.\nSpeaker D: Yeah.\nSpeaker D: The project is a fairly good.\nSpeaker D: But before we don't celebrate.\nSpeaker D: I have a little question which you can't answer because there must be some kind of entry port.\nSpeaker D: I'm busy with the entry port right now.\nSpeaker D: You might think what the hell is he doing?\nSpeaker C: What is an entry port?\nSpeaker D: About all the meetings, what we have decided.\nSpeaker D: Yeah.\nSpeaker D: A report of this day.\nSpeaker D: Yeah.\nSpeaker D: That must be made.\nSpeaker D: But I don't know.\nSpeaker D: Here is standing.\nSpeaker D: Well, we can celebrate now.\nSpeaker C: Oh, yeah.\nSpeaker C: You have ten minutes left.\nSpeaker C: I read.\nSpeaker C: You have now ten minutes left to finish up the entry.\nSpeaker D: Okay.\nSpeaker D: Well, that can be done.\nSpeaker D: Maybe we can do it together.\nSpeaker D: You can see when I've.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: I will put it on a storyboard.\nSpeaker D: You can see it.\nSpeaker D: Because I think it will.\nSpeaker D: It must be.\nSpeaker C: You're all made better.\nSpeaker C: Yeah.\nSpeaker D: It's a three.\nSpeaker D: 75 pages.\nSpeaker D: Yeah.\nSpeaker D: 75 pages.\nSpeaker D: Well, just a moment.\nSpeaker D: Okay.\nSpeaker D: Daniel.\nSpeaker C: You want a chair?\nSpeaker C: Maybe.\nSpeaker C: A chairman.\nSpeaker C: No, no, no.\nSpeaker D: I'm just.\nSpeaker D: Oh, you can read it.\nSpeaker D: Here it is.\nSpeaker D: You finish it actually.\nSpeaker B: So we just have to read it.\nSpeaker B: And so yes or no?\nSpeaker D: Well, it's not.\nSpeaker D: Read only.\nSpeaker D: But it's not.\nSpeaker D: It's not.\nSpeaker C: Five minutes.\nSpeaker D: Yep.\nSpeaker D: For finishing.\nSpeaker D: This is about the functional design.\nSpeaker D: The management expert here.\nSpeaker C: Yeah.\nSpeaker D: Change that.\nSpeaker D: Yeah.\nSpeaker D: When I said it, I remember.\nSpeaker C: I said read only version.\nSpeaker C: Yeah, but you can save it under another name.\nSpeaker D: Okay.\nSpeaker D: One thing actually.\nSpeaker D: Okay.\nSpeaker D: About three functions which are most used and which must immediately be visible on our remote control.\nSpeaker D: It must be simple to use.\nSpeaker D: Very clear what to do and the younger people.\nSpeaker D: So this is really about what kind of things must be in it.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Maybe.\nSpeaker C: Oh.\nNone: Yeah.\nNone: Yeah.\nSpeaker C: I understand you.\nSpeaker D: I can talk a little bit.\nSpeaker C: You have to put switch channels at the top because that's the most function.\nSpeaker C: Okay.\nSpeaker C: Data text at the second.\nSpeaker D: I really didn't.\nSpeaker C: All right.\nSpeaker C: Volume changing.\nSpeaker D: So this one switch.\nSpeaker C: Yes.\nSpeaker D: You go there and you go.\nSpeaker D: Yeah.\nSpeaker A: Okay.\nSpeaker D: Cool.\nSpeaker D: Well, maybe I can do it.\nSpeaker D: One, two, three.\nSpeaker D: Yes.\nSpeaker D: I've got it.\nSpeaker D: Good.\nSpeaker D: If the order is in.\nSpeaker D: Yeah.\nSpeaker D: Important.\nSpeaker D: Yeah.\nSpeaker D: That's the word for the order.\nSpeaker D: Then the conceptual design.\nSpeaker D: Well, all the things we have discussed the energy.\nSpeaker D: Which turn out to be a battery.\nSpeaker D: So that's working.\nSpeaker E: Maybe you can add it later.\nSpeaker D: Yeah.\nSpeaker D: We decided because of the cost.\nSpeaker D: It is still double curve.\nSpeaker D: The flash.\nSpeaker D: The fruity.\nSpeaker D: The removable.\nSpeaker D: It's not double anymore.\nSpeaker D: Not double anymore.\nSpeaker D: Now, but this is what.\nSpeaker D: Was on the shield.\nSpeaker C: Yeah.\nSpeaker D: And the added function flag.\nSpeaker D: The parent control and touchscreen.\nSpeaker D: Just a summary of what we had discussed.\nSpeaker D: One small thing.\nSpeaker B: The added functions.\nSpeaker B: Was it included in the cost?\nSpeaker B: I don't think so.\nSpeaker E: It's very cheap.\nSpeaker E: It's very cheap.\nSpeaker D: Maybe you come at the.\nSpeaker C: No, it's not very cheap but it's very necessary.\nSpeaker C: No, but it's.\nSpeaker C: Development inside corporation.\nSpeaker C: Like we don't have to buy parental control.\nSpeaker C: Our own people can.\nSpeaker C: Yeah, it has some cost.\nSpeaker D: Okay.\nSpeaker A: So we can discuss that.\nSpeaker D: Yeah.\nSpeaker D: We can discuss that kind of things I think with.\nSpeaker D: With the board of.\nSpeaker D: Of direction.\nSpeaker D: What's the company call?\nSpeaker D: I just keep forgetting it.\nSpeaker D: Real reaction.\nSpeaker D: Real reaction.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: So anybody miss something here about.\nSpeaker D: Yeah, the end conclusion.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker C: That's what I'm going to write.\nSpeaker C: Still the end conclusion.\nSpeaker D: That's all.\nSpeaker D: I think.\nSpeaker D: In here nothing.\nSpeaker E: The decision to make the buttons on the top and the menu on the.\nSpeaker E: On the bottom.\nSpeaker E: Yeah.\nSpeaker E: Clearly touch screen.\nSpeaker E: You've mentioned.\nSpeaker D: Yeah.\nSpeaker C: Touch screen.\nSpeaker C: Yeah.\nSpeaker C: The decisions are put in the conclusion.\nSpeaker C: Okay.\nSpeaker C: Why we decided to.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker D: LCD.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Well, I got a.\nSpeaker D: Redeciner something now.\nNone: Yeah.\nSpeaker D: Because I think it will five minutes from.\nSpeaker E: Oh, for you change anything.\nSpeaker E: Save the list.\nSpeaker E: You can select file exports and then there's the impact files.\nSpeaker C: Yeah, we're done.\nSpeaker C: You've had the same thing.\nSpeaker E: Eight.\nSpeaker E: We have another blank.\nSpeaker E: Sample of children.\nSpeaker C: You change anything?\nSpeaker C: Oh, wait.\nSpeaker C: Wait.\nNone: Wow.\nNone: Yeah.\nSpeaker A: A new way.\nSpeaker E: Commercial logo.\nSpeaker E: That's a petite.\nSpeaker E: Don't save it.\nSpeaker D: Oh, that's cool, Tim.\nSpeaker D: Finish meeting now.\nNone: Why are we only the first five smartboard five?\nSpeaker D: I'm going to finish Mario and before.\nSpeaker D: Okay.\nSpeaker B: Here, by meeting is finished.\nSpeaker B: You declare.\nSpeaker D: I am the one who can say that.\nSpeaker D: Yeah.\nSpeaker D: Here, by the meeting is finished.\nNone: Yeah.\nNone: Okay.\nNone: Okay.\nNone: How about the N34?\nNone: And the other one is while you are finished, before the other one is, I know it's a match.\nNone: Next, the science day.\nNone: It only takes five minutes.\nNone: And I do explain to you later, it's just five minutes because we already did.\nNone: Okay.\nNone: And you'll still be in time.\nNone: Okay.\nNone: You have a lot of cool things.\nNone: Okay.\nNone: Who do you lose?\nNone: Question.\nNone: What the fuck?\nSpeaker B: I thought there's quite an irritating sound.\nSpeaker D: Some questionery.\n"}, {"meeting_id": "IS1003d", "summary": "This was a meeting on the detailed design. First, the user interface designer presented a prototype to show the basic colour and shape of the product. Next, the marketing suggested doing some product evaluation instead of just talking about it, so the group started to judge the design in three aspects: how fancy, technologically innovative, and easy-to-use the remote control was. It got three, two, and four in separate aspects, which meant the design was acceptable. Later, the group discussed the quotation and made some decisions on details of industrial design to make sure the cost was within the budget. Furthermore, they came up with some new ideas about future products, such as a fruit collection of electronic devices or a customizable TV that followed the trend and people's moods. Finally, the group named the product as pineapple remote control.", "dialogue": "Speaker C: I'm sorry, not really.\nSpeaker C: Thanks for being here.\nSpeaker B: Your judgment is biased.\nSpeaker C: I need to do my control and stay for her time.\nSpeaker B: So, let's go for our detailed design meeting.\nSpeaker B: So, I will still play the role of the secretary and we'll have first the presentation by our user interface designer, David Jordan, and our industrial designer, Baba.\nSpeaker B: So, we'll have to evaluate your proposed remote control and have an idea of the price that this thing will cost.\nSpeaker B: In case if we all agree on the fabric of building a C3 remote control, we'll evaluate the production.\nSpeaker B: So, maybe I will let our two designers talk about some...\nSpeaker B: I have slides.\nSpeaker B: You have slides.\nSpeaker C: This is our product prototype.\nSpeaker C: This is made by Klin.\nSpeaker C: It looks strange.\nSpeaker C: The basic color is yellow and red. Yellow is our company color. I read it. It's more attractive.\nSpeaker C: So, we use two basic colors.\nSpeaker C: At the ship, there's two basic ships. The first is the cycle.\nSpeaker C: And the second is the triangle space.\nSpeaker C: We call it mushroom design. It looks like the mushroom design.\nSpeaker C: This is the introduction of our product.\nSpeaker B: Genetically modified mushroom, I will say.\nSpeaker C: Okay.\nSpeaker C: Yeah, the next slide. So, there are several key features of our prototype.\nSpeaker C: The first is that it is fancy. I'm sure this would be the unique design in the market.\nSpeaker B: Maybe I hope so.\nSpeaker C: Yeah, so it's a fancy design and a unique design.\nSpeaker C: The second key feature is that the circle channel selection.\nSpeaker C: In the traditional controller, you use button to select the channel.\nSpeaker C: But now we have a suck circle. So, we can turn this ball to select channel.\nSpeaker C: It's quite convenient for users to use it.\nSpeaker B: Yeah, but don't touch. Don't destroy your first time.\nSpeaker C: Okay. The third feature is the stable triangle base.\nSpeaker C: This triangle base is very stable. So, it's unlikely you cannot find it.\nSpeaker C: So, you can put it in the table. So, you can turn the ball to select the channel.\nSpeaker C: And there's some cute button. You can see the shape of the button is a mushroom.\nSpeaker B: Everything's mushroom. So, we can call our remote control that mushroom.\nSpeaker C: It was not really mushroom because you have lemon shape.\nSpeaker C: Yeah, that's why if you put it in the table, careful somebody will eat it.\nSpeaker C: I hope nobody will eat it anyway. You're trying to get the food aspect.\nSpeaker D: Mushroom was not in the trance.\nSpeaker D: For its vegetables.\nSpeaker D: It's vegetable.\nSpeaker C: It's vegetable.\nSpeaker B: It's something eatable.\nSpeaker C: We integrated the different color.\nSpeaker D: But this is not a mushroom anyway. So, it's fine.\nSpeaker C: I think we take into account what you said about fruit and vegetable biasing.\nSpeaker D: These colors.\nSpeaker C: And spider color.\nSpeaker C: And very sophisticated material.\nSpeaker C: So, next slide.\nSpeaker C: So, what I can add is that he talks about what is outside.\nSpeaker C: So, what is inside is what we talked about.\nSpeaker C: The chip is a low level chip.\nSpeaker C: So, we cut it.\nSpeaker C: The low level chips inside.\nSpeaker C: And the LCD button.\nSpeaker A: So, where's the battery?\nSpeaker C: The battery is under the base.\nSpeaker B: In the base.\nSpeaker D: And where is the solar set?\nSpeaker C: This is a kind of new revolutionary receptor that we can put outside.\nSpeaker B: Do you think it won't be, it won't Christ increase the price?\nSpeaker C: I don't think so.\nSpeaker B: Okay, we'll see after.\nSpeaker B: We'll have the first.\nSpeaker B: Okay.\nSpeaker B: So, Mr. Monet, what's your opinion, according to this remote control?\nSpeaker D: I mean, we're going to try to measure how good it is instead of just talking about.\nSpeaker D: We had three key points for this remote control design.\nSpeaker D: And first one was fancy look and feel.\nSpeaker C: Yeah.\nSpeaker D: So, we tried to judge this feature with a one to seven scale.\nSpeaker D: One being no.\nSpeaker D: I think.\nSpeaker D: Just let me check.\nSpeaker C: So, 4.3.5, it means it's acceptable.\nSpeaker D: One being true and seven being false.\nSpeaker D: Okay.\nSpeaker D: Oops.\nSpeaker D: So, do we have a fancy look and feel?\nSpeaker C: I think you have nice colors.\nSpeaker C: The shape, the ball shape.\nSpeaker C: Yeah, the shape is unique.\nSpeaker C: The colors.\nSpeaker D: I agree, it's unique, but is it really?\nSpeaker C: So, is it really fancy?\nSpeaker C: Do we define fancy?\nSpeaker D: Yeah, I mean, fancy was defined by fruit and vegetable look.\nSpeaker C: You have the lemon aspect of this.\nSpeaker C: This is it.\nSpeaker A: Yeah.\nSpeaker B: Maybe if you change, if you take the buttons out.\nSpeaker D: Maybe do you think like that?\nSpeaker D: I don't know where the lemon is, but I mean, it's not obvious.\nSpeaker C: This shape is the lemon line.\nSpeaker B: It would be more like a lemon.\nSpeaker D: If.\nSpeaker D: Maybe improving the texture, like having less.\nSpeaker D: Less buttons.\nSpeaker D: Oh.\nSpeaker C: So, my mush.\nSpeaker B: Looks like more fruit.\nSpeaker B: Maybe a pineapple.\nSpeaker D: Yeah, I don't know.\nSpeaker B: And you know, you have the finger here with the buttons?\nSpeaker C: Yeah, it's a good idea.\nSpeaker B: It looks more like a pineapple.\nSpeaker B: Sorry.\nSpeaker B: What's the use for that?\nSpeaker B: I don't know.\nSpeaker C: If you want to tame the ball, I have no idea.\nSpeaker C: It's very convenient for you to tame the ball, to change the channel.\nSpeaker B: Okay.\nSpeaker C: It's embedded.\nSpeaker C: Yeah, that's the second point.\nSpeaker D: The second point is we have to judge the.\nSpeaker B: If it is fancy or not.\nSpeaker B: Fancy looking feel.\nSpeaker B: Is it better like that?\nSpeaker B: Okay, let's say it's a pineapple now.\nSpeaker C: The color is the color acceptable.\nSpeaker D: No, the color is okay.\nSpeaker D: That's fine.\nSpeaker C: So, the shame.\nSpeaker D: It's like.\nSpeaker D: I'd say there is more too much red.\nSpeaker C: Not too much red.\nSpeaker B: In the basement.\nSpeaker B: It looks like a pineapple with cherry on top.\nSpeaker B: Okay.\nSpeaker D: Right.\nSpeaker D: So, from one to seven.\nSpeaker B: I'll give.\nSpeaker B: Seven is the maximum?\nSpeaker D: No, seven is false.\nSpeaker D: And one is true.\nSpeaker C: I'll give two or three.\nSpeaker C: Three, I will say three.\nSpeaker C: Three is fine for me.\nSpeaker C: Okay.\nSpeaker D: Three.\nSpeaker B: Is it better like that?\nSpeaker D: Yeah.\nSpeaker D: Then, let me.\nSpeaker D: What are the other criterias?\nSpeaker D: Is it easier to use technically, technologically innovative?\nSpeaker B: You said previously that there is some microphone inside.\nSpeaker C: Yeah, this is microphone ring.\nSpeaker C: It's a microphone.\nSpeaker B: Okay.\nSpeaker B: And you have the microphone ring.\nSpeaker B: There's the technology inside that recognizes simple vocal comments.\nSpeaker B: And you can turn it so maybe it's techno.\nSpeaker B: Yeah.\nSpeaker B: So, you can.\nSpeaker C: I have a lot of things around remote control.\nSpeaker D: Yeah, but when you say technology, it's more in the core of the single.\nSpeaker B: We have tactile buttons.\nSpeaker D: Mm-hmm.\nSpeaker C: I said you have microphone ring, but you have...\nSpeaker C: No, that's good.\nSpeaker C: Yeah.\nSpeaker C: That's good.\nSpeaker C: And you have this solar receptor.\nSpeaker D: Yeah.\nSpeaker D: That's another really good point.\nSpeaker D: Yeah.\nSpeaker C: I think that's not going to do it acceptable.\nSpeaker B: Maybe two.\nSpeaker B: So using the same scale?\nSpeaker B: Two?\nSpeaker B: Two, yeah.\nSpeaker C: Two.\nSpeaker C: You agree?\nSpeaker B: Yeah.\nSpeaker D: Okay.\nSpeaker D: Now, maybe the most critical one.\nSpeaker D: Okay.\nSpeaker D: Last one I would like to judge is...\nSpeaker D: Is it easy to use?\nSpeaker C: The local common, yes.\nSpeaker C: It might be easy.\nSpeaker C: Yeah, but it's stunning.\nSpeaker D: Can you just explain me the...\nSpeaker D: Can you just explain me the...\nSpeaker C: It's a print saying.\nSpeaker C: Yeah.\nSpeaker C: Please.\nSpeaker C: This is the base.\nSpeaker C: Yeah.\nSpeaker C: So you can tend to change the channel.\nSpeaker D: Yeah, but how intuitive is it to turn things to change channels?\nSpeaker D: I think maybe if you...\nSpeaker C: Like if you want to go from...\nSpeaker C: Oh, I understand.\nSpeaker B: If you take...\nSpeaker B: Take from the remote.\nSpeaker B: So...\nSpeaker B: And you can turn like that to change the channel to...\nSpeaker B: Yeah, but imagine...\nSpeaker B: I think it's quite easy to do zapping, but maybe it will be too fast.\nSpeaker D: Yeah, I'm not definitely convinced.\nSpeaker D: It's the best way to...\nSpeaker D: If you want to jump from, I don't know, 1 to 20.\nSpeaker D: Yeah, that's different.\nSpeaker D: How can you go directly to 20, for example?\nSpeaker C: No, no.\nSpeaker C: It depends on the angle.\nSpeaker C: Yeah.\nSpeaker C: I agree.\nSpeaker C: I agree.\nSpeaker C: But I mean...\nSpeaker C: You need to know...\nSpeaker C: Is it the channel wide?\nSpeaker C: So it could be channel...\nSpeaker C: I think something that's three, channel four, channel five.\nSpeaker B: Yeah, you have a click click click click click click click.\nSpeaker D: Just imagine you have 50 channels.\nSpeaker C: Oh, yeah.\nSpeaker C: 50 channels you buy the number by 330 degree.\nSpeaker C: Yeah.\nSpeaker C: So you got how many degree you...\nSpeaker B: Yeah, but you have to go through all the channels.\nSpeaker C: No, no, no, no, no, no, no, no, no, no.\nSpeaker C: When you stop...\nSpeaker C: When you stop the...\nSpeaker C: The tone...\nSpeaker C: Mm-hmm.\nSpeaker C: Then the angle you stop, it's angle you, it's the channel you...\nSpeaker D: How do you know this angle is the correct one?\nSpeaker C: It's very easy because you know how many channels are there...\nSpeaker D: So you count at one degree, two degrees, no?\nSpeaker C: Yeah, yeah, you can do it.\nSpeaker D: I don't think so.\nSpeaker D: Yeah, that's great.\nSpeaker C: That's great.\nSpeaker C: I think so, you can...\nSpeaker C: Yes.\nSpeaker C: I think...\nSpeaker C: But the vocal common is easy to...\nSpeaker C: You can say 50.\nSpeaker C: Yeah, vocal common is okay.\nSpeaker C: 50, it's okay.\nSpeaker D: So we've said previously that maybe it's not going to be...\nSpeaker B: Yeah, but when you're zapping, you're changing from one channel to the other, so you're passing through all the channels.\nSpeaker B: So when you say I want to go to the channel number 20, that's you've decided to go to channel 20, so you can say channel 20.\nSpeaker B: All right, yeah.\nSpeaker B: Or channel 4, because you really want to go on this channel, but if you really want to...\nSpeaker B: And this would be more for browsing.\nSpeaker B: To do zapping, you don't really know what you want to do, you can turn it.\nSpeaker B: Yeah.\nSpeaker B: Just go through all the channels and maybe stop if there is something interesting.\nSpeaker D: Mm-hmm.\nSpeaker B: Seems to be good.\nSpeaker B: Good choice, Mr. David Jordan.\nSpeaker B: Yeah, I mean, you're famous.\nSpeaker B: And whatever you do.\nSpeaker B: Yeah, what's this?\nSpeaker B: What's this?\nSpeaker C: Cherry?\nSpeaker C: Oh, you can...\nSpeaker A: Also, the turner.\nSpeaker C: Turner for the tune.\nSpeaker C: Turner for the tune.\nSpeaker C: If you want to skip from channel 1 to channel 2, you skip this.\nSpeaker C: If you want to skip from channel 1 to channel 10, you turn this.\nSpeaker C: It's like fine from cost to fine.\nSpeaker C: This is cost.\nSpeaker C: This is fine.\nSpeaker C: Yeah, yeah, yeah.\nSpeaker C: It's cost to fine.\nSpeaker C: It's great technology.\nSpeaker C: Yeah, this looks better.\nSpeaker C: Yeah, you see, this is from one channel to...\nSpeaker C: I didn't see where it is.\nSpeaker C: It's the turner.\nSpeaker C: Turner for the tuneer.\nSpeaker C: Turner for the tuneer.\nSpeaker C: Actually, it's one of the...\nSpeaker B: Yeah, you have the vocal comment.\nSpeaker C: Uh, it's...\nSpeaker C: On off.\nSpeaker D: On off, yeah.\nSpeaker D: Yeah, but it has to be on to...\nSpeaker B: Most of the time you have them.\nSpeaker B: Yeah, it's sleeping remote control.\nSpeaker B: Yeah, sleeping.\nSpeaker D: That's not the ecological...\nSpeaker D: That's true.\nSpeaker B: That's why we have the solar...\nSpeaker C: To come back to the end.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker D: So, which number?\nSpeaker C: Three.\nSpeaker C: It is, it's very relative, but three, it's fine, I think.\nSpeaker C: It's rather than a...\nSpeaker C: Do you agree?\nSpeaker D: Yeah, I agree.\nSpeaker D: So, the reason about this is four is one.\nSpeaker C: Three.\nSpeaker C: Three for me.\nSpeaker C: It's...\nSpeaker C: It's okay.\nSpeaker C: It's...\nSpeaker C: So...\nSpeaker C: Four.\nSpeaker C: Three.\nSpeaker C: Maybe you can vote to see how many everybody gives and just take the...\nSpeaker B: Or what's your opinion?\nSpeaker D: Uh, we can even say, I mean...\nSpeaker D: Those are sort of agree, but this one would be more five to me.\nSpeaker D: Five?\nSpeaker D: I would like...\nSpeaker D: I mean, this is just a prototype.\nSpeaker D: I'm not really convinced it's so easy to use.\nSpeaker D: Yeah, so...\nSpeaker C: I don't know.\nSpeaker C: What do you...\nSpeaker C: What...\nSpeaker C: You compare with traditional...\nSpeaker C: Yeah.\nSpeaker C: To...\nSpeaker C: Traditional controller.\nSpeaker C: I think it's easier than traditional controller.\nSpeaker C: If you use traditional controller, you have to put the button.\nSpeaker C: And now you don't have to put the button.\nSpeaker C: You have...\nSpeaker C: You just tend the...\nSpeaker C: Yeah, but you know...\nSpeaker C: That's who kind of...\nSpeaker C: The smaller...\nSpeaker C: Yeah.\nSpeaker C: So you can control the scale.\nSpeaker C: But in the traditional controller, how do...\nSpeaker C: How can you control the scale?\nSpeaker C: By pushing zero...\nSpeaker C: You just put two buttons.\nSpeaker C: Zero.\nSpeaker C: That's the first one.\nSpeaker C: And one.\nSpeaker C: Yeah, yeah, you can do it.\nSpeaker C: Yeah.\nSpeaker C: But you can also do it in the...\nSpeaker C: But if you do not want to browsing all the channels, you can just...\nSpeaker C: There's also a button here.\nSpeaker C: And there's some buttons, okay.\nSpeaker C: Yeah.\nSpeaker C: This function is just for your broadened.\nSpeaker C: From one channel to the next one.\nSpeaker C: The next...\nSpeaker C: That's not what you say previously.\nSpeaker B: Yeah, if you're changing your mind...\nSpeaker B: This was the fine...\nSpeaker B: Fine to course.\nSpeaker B: Fine.\nSpeaker B: Of course.\nSpeaker B: And from 10 to 10 channels.\nSpeaker C: Why do 10, 10 to 12...\nSpeaker C: 10 to 20?\nSpeaker C: And this one, one, two, three, four, five, six.\nSpeaker C: Like this.\nSpeaker C: Okay.\nSpeaker C: So there's different scale.\nSpeaker C: So you can...\nSpeaker C: Okay.\nSpeaker C: You can choose.\nSpeaker D: But this...\nSpeaker D: This has to stay on the table, right?\nSpeaker D: Mm-hmm.\nSpeaker D: This has to stay on the table.\nSpeaker C: Alright.\nSpeaker C: You can fuck on a flat.\nSpeaker C: This is just a bit.\nSpeaker C: You can just...\nNone: You can...\nSpeaker C: Yeah.\nSpeaker D: But then when you turn it...\nSpeaker C: Oh, yeah.\nSpeaker B: Just right?\nSpeaker B: No, no.\nSpeaker B: You can't put it out.\nSpeaker B: That's just your turn in from the base.\nSpeaker B: You need to have everything in hand.\nSpeaker B: If you want to turn, you can't use it and turn.\nSpeaker B: You need to put it on and turn.\nSpeaker D: You know, that's the weak point because with the traditional one, you just have one hand.\nSpeaker B: Yeah, but nobody would be able to take it in the pocket and bring it in the kitchen.\nSpeaker B: Say, I've lost the remote control.\nSpeaker B: Nobody would take it.\nSpeaker D: Yeah, you would never lose it on the...\nSpeaker C: So this is the next generation.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: Maybe.\nSpeaker C: It's the next prototype.\nSpeaker B: Maybe we can change from...\nSpeaker B: Maybe for it's okay.\nSpeaker B: For...\nSpeaker B: Easy to use for.\nSpeaker B: You can save it.\nSpeaker D: Yeah, for it's more.\nSpeaker D: It's even easier to...\nSpeaker D: Maybe.\nSpeaker C: You can...\nSpeaker C: You can error this.\nSpeaker C: Yeah.\nSpeaker D: I don't know.\nNone: Yeah.\nNone: Yeah.\nNone: Yeah.\nSpeaker C: Yeah, for it's more than I would say.\nSpeaker C: So average...\nSpeaker C: Nine over three?\nSpeaker C: Three?\nSpeaker C: Three.\nSpeaker C: Three.\nSpeaker C: It's average, it's more than I.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker C: It need maybe some further work, but...\nSpeaker D: Yeah, especially on the easy to use.\nSpeaker D: Yeah.\nSpeaker D: Absolutely.\nSpeaker D: Target.\nSpeaker B: So the project is accepted.\nSpeaker B: Yeah.\nSpeaker C: Yeah.\nSpeaker B: I think...\nSpeaker B: I think that it will be good to do some more work to transform this into a pineapple.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: If you really want to have a fruitful, remote country.\nSpeaker C: Okay.\nSpeaker B: So...\nSpeaker B: Is that fine?\nSpeaker B: Okay.\nSpeaker B: We had a prototype presentation with the evaluation.\nSpeaker B: So as we all agree to accept...\nSpeaker B: Yeah.\nSpeaker B: Under certain conditions, the prototype will have looked to the financial view.\nSpeaker B: So we need to calculate the projection cost.\nSpeaker B: As I said in the first meeting, when to have remote control, that would cost more than 12 and more than 50 euros.\nSpeaker B: And if not, if it's not the case, you would have to redesign it.\nSpeaker B: Yeah.\nSpeaker B: So just have a look at the...\nSpeaker B: Okay.\nSpeaker B: Here is the...\nSpeaker B: Excel...\nSpeaker B: Shit.\nSpeaker B: So...\nSpeaker B: So as the energy source, we have Hundinamo.\nSpeaker B: No.\nSpeaker B: We don't use that.\nSpeaker B: We have battery, right?\nSpeaker B: Yeah.\nSpeaker B: Not battery.\nSpeaker B: Kinetic.\nSpeaker B: We don't have it, I suppose.\nSpeaker B: But we have solar cells.\nSpeaker B: Yeah.\nSpeaker B: How many do you need solar cells?\nSpeaker B: Do you think one would be enough for them?\nSpeaker B: Such as...\nSpeaker C: I think it's about...\nSpeaker B: You have three.\nSpeaker C: Three.\nSpeaker C: Three.\nSpeaker C: Three batteries, three solar cells.\nSpeaker B: So...\nSpeaker B: Electronic.\nSpeaker B: Single, simple chip on print.\nSpeaker B: Just one would be necessary.\nSpeaker B: Simple, yeah.\nSpeaker C: One.\nSpeaker B: Zero for the others.\nSpeaker B: And sample sensors and pull speaker.\nNone: One.\nSpeaker B: One.\nSpeaker D: One.\nSpeaker B: As we have voice recognition, I think.\nSpeaker C: Yeah.\nNone: Yeah.\nSpeaker B: So the case...\nSpeaker C: It would be...\nSpeaker D: So we are all already 19.\nSpeaker C: Wooden.\nSpeaker C: Yeah.\nSpeaker B: Okay.\nSpeaker B: Just...\nSpeaker B: Keep on going.\nSpeaker B: Just have an idea.\nSpeaker B: Yes, yes.\nSpeaker C: Wooden...\nSpeaker C: There's no good plastic just on anything.\nSpeaker B: Yeah, but what about case?\nSpeaker B: Uncurved, flat, single curved, double curved.\nSpeaker B: I think it's more like double curved.\nSpeaker B: That's a little bit of a double curved.\nSpeaker B: Yeah, double curved.\nSpeaker B: Yeah, double curved.\nSpeaker B: One.\nSpeaker C: One, yeah.\nSpeaker B: One, yeah.\nSpeaker B: Uh, wood?\nSpeaker C: But it's...\nSpeaker C: A rubber...\nSpeaker B: Do you want it's special color?\nSpeaker B: You may be too.\nSpeaker A: Yeah, yeah.\nSpeaker B: Two special colors.\nSpeaker B: Yeah, yeah.\nSpeaker B: Push button.\nSpeaker C: Yeah, we have four.\nNone: Five.\nSpeaker C: Five.\nSpeaker C: Okay.\nSpeaker C: Okay.\nNone: Okay.\nSpeaker B: Scroll wheel?\nSpeaker B: No.\nSpeaker B: It's more like integrated.\nSpeaker B: I think this will be like a scroll wheel.\nSpeaker C: Yeah, you turn it.\nSpeaker C: Sorry.\nSpeaker C: Yeah.\nSpeaker B: Maybe two scroll wheel eyes.\nSpeaker B: Yeah, true.\nSpeaker B: Yeah.\nSpeaker B: We have the course to find a scroll wheel.\nSpeaker B: Okay.\nSpeaker B: So, no button supplements?\nSpeaker B: No.\nSpeaker D: Okay.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker B: Okay.\nSpeaker B: I think we have problem.\nSpeaker C: As you display maybe.\nSpeaker C: Interphase.\nSpeaker B: I think with...\nSpeaker B: If we keep on adding things.\nSpeaker B: It's okay.\nSpeaker B: That's fine.\nSpeaker B: So, we have to...\nSpeaker D: Yeah.\nSpeaker D: One of the key points is solar cells.\nSpeaker B: Maybe if...\nSpeaker B: If suppose if we change and we put just one.\nSpeaker D: Yeah.\nSpeaker B: Has it changed?\nSpeaker D: Yeah, it was actually...\nSpeaker D: It did change, but...\nSpeaker D: Okay.\nSpeaker C: We have to switch to electronics.\nSpeaker C: From the samples of the cell to regular tip.\nSpeaker C: Oh.\nSpeaker C: We have to delete the samples.\nSpeaker B: Yeah, but we have the voice recognition.\nSpeaker B: No.\nSpeaker D: Yeah, but this one, one of the features we were not able to...\nSpeaker C: Yeah, it's the one...\nSpeaker C: Maybe we have tumor and the first part in the basic version.\nSpeaker C: I think you can transform the wood into plastic maybe.\nSpeaker B: It seems that this can be...\nSpeaker C: Oh, it would be better.\nSpeaker C: Yeah, wood into plastic.\nSpeaker C: It should be fine.\nSpeaker C: Plastic is free.\nSpeaker D: Yeah.\nSpeaker B: So, do we need special color?\nSpeaker D: Yeah, that's one of...\nSpeaker D: Because we have red.\nSpeaker B: Red and yellow.\nSpeaker B: Red and yellow.\nSpeaker D: We can turn everything in either yellow or black.\nSpeaker D: Black then is a regular color.\nSpeaker D: I think so.\nSpeaker B: If we try to have a kind of...\nSpeaker B: Yeah.\nSpeaker A: Pineapple B.\nSpeaker B: Okay.\nSpeaker D: So, push button then it's the next...\nSpeaker D: Yeah.\nSpeaker B: And we have integrated scroll wheel with push button.\nSpeaker B: But now I think this is one.\nSpeaker B: Integrated scroll wheel push button.\nSpeaker B: So, we have only one.\nSpeaker B: And...\nSpeaker B: Push button.\nSpeaker B: Close to.\nSpeaker B: So, if we have all integrated in the scroll wheel and push button.\nSpeaker B: It's...\nSpeaker C: Okay, remove maybe...\nSpeaker C: You have five push buttons so we can just...\nSpeaker C: Try to modify some of them.\nSpeaker D: Okay, so what's the...\nSpeaker D: Like...\nSpeaker C: How about we change our color?\nSpeaker C: Double curve.\nSpeaker C: We can transform the double curve into single...\nSpeaker B: Something flat.\nSpeaker D: Yeah, but flat.\nSpeaker B: Uncurved.\nSpeaker B: Okay.\nSpeaker B: Yeah, maybe not.\nSpeaker B: Single curve better.\nSpeaker C: Yeah.\nSpeaker C: Single curve should be fine.\nSpeaker A: Ooh, quick.\nSpeaker A: Oh, we have only one.\nSpeaker D: Really?\nSpeaker D: One you were left.\nSpeaker A: Um...\nSpeaker B: Maybe don't...\nSpeaker B: No battery on me, so myself.\nSpeaker D: I think somebody gets...\nSpeaker B: I think there's a problem with the push button.\nSpeaker B: We only need...\nSpeaker A: Just one.\nSpeaker A: One push.\nSpeaker A: Just one.\nSpeaker A: Yeah.\nSpeaker A: That's fine.\nSpeaker A: Okay.\nSpeaker C: So, actually...\nSpeaker D: So, we have one button.\nSpeaker D: Yeah, we have...\nSpeaker B: So, one button and...\nSpeaker C: One scroll wheel.\nSpeaker C: Scroll wheel with push button on it.\nSpeaker C: And the vocal cover, it's fine.\nSpeaker C: It's fine.\nSpeaker B: So...\nSpeaker D: It's good they're not charging anything for them.\nSpeaker D: So, yeah.\nSpeaker B: I think we've done a good job as the cost is...\nSpeaker B: Could we have looked at...\nSpeaker C: Could we have looked at...\nSpeaker C: Yeah, yeah.\nSpeaker C: Actually, it's wrong.\nSpeaker D: We're not under...\nSpeaker D: The cost?\nSpeaker D: Yeah, but it's under our equal.\nSpeaker B: It's not rated.\nSpeaker B: Sometimes it's under our equal.\nSpeaker B: It's under our equal.\nSpeaker C: It's fine.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: So, let's see...\nSpeaker C: Twelve.\nSpeaker B: Twelve.\nSpeaker C: Fifty.\nSpeaker C: Yeah.\nSpeaker C: It's fine, right?\nSpeaker C: Which part is the most defensive part?\nSpeaker C: Solar cells.\nSpeaker C: Solar cells, right?\nSpeaker C: Yeah.\nSpeaker C: Is it?\nSpeaker C: Is it?\nSpeaker B: I think...\nSpeaker B: Yeah.\nSpeaker B: But it would be interesting for our marketing team to make lots of advertisements...\nSpeaker B: Yeah, to be able to sell to the same.\nSpeaker B:...and concerning these solar cells to be...\nSpeaker D: That's a nice argument.\nSpeaker D: Yeah.\nSpeaker C: It's still for our...\nSpeaker C: It's really, really, really, really, very expensive.\nSpeaker C: Yeah.\nSpeaker C: Maybe...\nSpeaker B: Yeah, but it would be technologically innovative.\nSpeaker D: Yeah, but we just have one button.\nSpeaker B: So, it's easy to use in Powerful, as the remote controller has only one button?\nSpeaker D: Easy, you know, about Powerful.\nSpeaker D: Yeah.\nSpeaker C: Easy Power.\nSpeaker C: Easy to use.\nSpeaker B: So, I don't think we need to redesign the product.\nSpeaker D: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Or that's what we've just done.\nNone: We've done it.\nSpeaker D: Yeah.\nSpeaker D: It's under the...\nSpeaker D: If it was low, high and so...\nNone: So...\nSpeaker C: Okay.\nSpeaker C: Now...\nSpeaker C: So, what are we going to do with this project evaluation?\nSpeaker C: So...\nSpeaker D: I think we have to just have to discuss it.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker B: So, did you enjoy your clay modeling?\nSpeaker B: Yes.\nSpeaker B: Of course.\nSpeaker B: This is my job.\nSpeaker B: Yeah.\nSpeaker B: Was it a nice way to create your remote control?\nSpeaker C: Yeah.\nSpeaker C: It's a...\nSpeaker C: It's good to create a control instead of a computer.\nSpeaker C: Yeah.\nSpeaker C: I think I found it really funny and amazing and interesting to go through all the process from the beginning to the end and designing, looking at the chips, solar cells, and...\nSpeaker C: It is very informative for me.\nSpeaker B: And for the marketing guy?\nSpeaker D: That was good, but we should have more brainstorming like meetings maybe.\nSpeaker D: Because we're just presenting, one is presenting his stuff, next one his stuff, and then we try to combine afterwards.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker B: And new ideas about new products maybe, which would be fashion and yellow.\nSpeaker B: Yellow.\nSpeaker C: I think you can think about yellow TV now after a food TV or I don't know.\nSpeaker C: It can be interesting.\nSpeaker B: Yes, just lemon.\nSpeaker C: Yeah, just lemon TV.\nSpeaker C: You know the traditional TV is flat.\nSpeaker C: Yeah, it's flat.\nSpeaker C: It's square.\nSpeaker C: Yeah, the chip is very boring.\nSpeaker C: It's really boring.\nSpeaker A: It's really boring.\nSpeaker C: Come up with new TV.\nSpeaker C: Yeah.\nSpeaker C: With such as this kind of TV.\nSpeaker C: So you can...\nSpeaker C: You can have this, triangle base, so you have the TV.\nSpeaker B: The lemon TV with the pineapple remot control.\nSpeaker B: Yeah, that's...\nSpeaker C: Yeah.\nSpeaker C: That is really interesting.\nSpeaker C: Yeah.\nSpeaker C: Because the TV...\nSpeaker B: Oh, that's interesting.\nSpeaker B: We could do a kind of fruit collection of...\nSpeaker B: Oh, you like this.\nSpeaker B:...electronic device.\nSpeaker B: Yeah.\nSpeaker D: Yes.\nSpeaker D: You just don't trust too much the trends.\nSpeaker D: Maybe.\nSpeaker D: Because fruit and vegetables last for 10 years.\nSpeaker D: Yeah, maybe the message will be insects.\nSpeaker C: Yeah.\nSpeaker C: But I think it's good to follow the flow and make it now and after, you know, if the people have changed their mind, you changed their mind.\nSpeaker D: Yeah, but this is good because it's not a long life product.\nSpeaker D: Yeah.\nSpeaker D: TV is more like 15 years maybe so.\nSpeaker D: Yeah, that's a lemon TV for...\nSpeaker C: Yeah.\nSpeaker C: We can think about TV with, you know, where you can change, you know, the aspect.\nSpeaker C: So, like for the cell phones.\nSpeaker C: Yeah.\nSpeaker B: You customize it every time.\nSpeaker C: So every time you go to a channel, you just change the appearance.\nSpeaker B: Oh, so, yeah.\nSpeaker B: You can do it all the time.\nSpeaker C: You can make it.\nSpeaker C: Yeah.\nSpeaker C: And keep the number up here.\nSpeaker C: Yeah, and following...\nSpeaker C: The mood of person, the fashion.\nSpeaker C: Yeah.\nSpeaker C: It's interesting.\nSpeaker B: Maybe we can create a line of TV.\nSpeaker C: Yeah.\nSpeaker C: TV.\nSpeaker C: Yeah.\nSpeaker C: But TV for autumn and a TV for winter, you know.\nSpeaker C: So, that's...\nSpeaker C: So, what is it?\nSpeaker C: Okay.\nSpeaker B: So, I think the costs are within the project.\nSpeaker B: We're just...\nSpeaker B: Yeah, that's actually...\nSpeaker B: 12.50 euros.\nSpeaker B: So...\nSpeaker B: Do you think you can celebrate your own?\nSpeaker B: Yeah.\nSpeaker B: You can celebrate your creation.\nSpeaker C: Yeah, you can celebrate your leadership.\nSpeaker B: Oh, thanks a lot.\nSpeaker B: Thank you.\nSpeaker B: Well, I say yes.\nSpeaker B: David Jordan.\nSpeaker C: It's really a celebrating object.\nSpeaker C: So, it's, you know, very...\nSpeaker C: Yeah.\nSpeaker C: It's very happy.\nSpeaker C: Yeah.\nSpeaker C: It's a party.\nSpeaker D: It's a very much control.\nSpeaker B: Yeah.\nSpeaker B: The thing now is to...\nSpeaker B: to sell it.\nSpeaker C: Yeah, it's your job.\nSpeaker C: Yeah.\nSpeaker B: You sell it.\nSpeaker C: You sell it.\nSpeaker A: Good luck.\nSpeaker A: Yeah.\nSpeaker C: Okay, and the...\nSpeaker C: 12.50...\nSpeaker C: 12.25 euros.\nSpeaker C: 25 euros.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I think it's...\nSpeaker C: 25 euros.\nSpeaker C: 25 euros.\nSpeaker C: Yeah, I think it's...\nSpeaker C: 25 euros.\nSpeaker C: So, maybe a little bit expensive.\nSpeaker A: Maybe a little bit expensive.\nSpeaker A: No, I'm...\nSpeaker A: It's not so easy.\nSpeaker D: I'm not so happy about the fruit chip, you know.\nSpeaker C: What, really?\nSpeaker C: It should be fine, you know, actually.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: Maybe you eat it.\nSpeaker C: Maybe you eat it.\nSpeaker A: Maybe you eat it.\nSpeaker B: Maybe you can do a test, put it outside and you become...\nSpeaker B: I mean, you become...\nSpeaker B: I eat...\nSpeaker A: really fruit.\nSpeaker A: I think...\nSpeaker A: But don't put sugar in it.\nSpeaker A: The color I'm making is...\nSpeaker C: The color I'm making.\nSpeaker C: So, actually...\nSpeaker C: No, the colors are...\nSpeaker D: It's perfect, yeah.\nSpeaker D: It's perfect.\nSpeaker D: And two.\nSpeaker D: Yeah, another thing is the logo is missing still.\nSpeaker B: Yeah, but the color, I think the color is more...\nSpeaker B: The yellow, more...\nSpeaker B: The yellow, more...\nSpeaker B: I don't think you have ever seen something like that before for remote control.\nSpeaker D: Still, that was one of the requirements we had.\nSpeaker D: Yeah, but I don't think it's such a problem.\nSpeaker D: Just putting the logo somewhere.\nSpeaker C: We can put some...\nSpeaker B: Yeah, we decided to have something yellow and red for the costs.\nSpeaker B: Yeah.\nSpeaker B: So, maybe we can just integrate it on one side.\nSpeaker B: The double R.\nSpeaker C: Okay, yeah.\nSpeaker C: That's a really good idea, so...\nSpeaker B: So, we have logo, we have the color, and we have the fashion in electronics.\nSpeaker B: So, we have the slogan too?\nSpeaker C: So, we have give him...\nSpeaker C: Give him...\nSpeaker C: Give him...\nSpeaker C: A cute name.\nSpeaker B: A cute man.\nSpeaker B: Yes, mushroom control.\nSpeaker B: No, it's the pineapple control.\nSpeaker B: You cannot say mushroom because it's not the same.\nSpeaker B: It's the pineapple now.\nSpeaker B: It's changed.\nSpeaker C: It's pineapple.\nSpeaker C: Pine apple.\nSpeaker C: Pine apple.\nSpeaker C: Pine apple.\nSpeaker C: Yeah, just flying saucer.\nSpeaker C: Let's carry him top.\nSpeaker C: Okay.\nSpeaker C: Oh, that's a flying saucer.\nSpeaker C: And he didn't fire him once in one.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker C: That's pineapple remote control.\nSpeaker C: I think it's fine.\nSpeaker C: Would you buy one?\nSpeaker B: Yeah, I'd try to.\nSpeaker B: I tried to.\nSpeaker B: 25.\nSpeaker C: That's beautiful.\nSpeaker C: I can't...\nSpeaker C: I would try to work and see how easy it is.\nSpeaker C: Yeah, you taught me you lost your control.\nSpeaker C: Yeah, I already saw that.\nSpeaker C: So...\nSpeaker C: To need buy buy.\nSpeaker C: So, if it is a universal, I can use it to my TV.\nSpeaker C: Okay.\nSpeaker B: One thing I like is the shape because, you know, it's not like the...\nSpeaker B: The remote controls you can put in your pockets, in your jacket.\nSpeaker C: This is why, this is why.\nSpeaker C: What do you choose?\nSpeaker C: I prefer the laser remote control.\nSpeaker C: What do you choose?\nSpeaker C: I think I will choose this one because of the color.\nSpeaker C: It's okay.\nSpeaker C: It's you light your house and your home and your TV.\nSpeaker C: Okay.\nSpeaker B: Oh, maybe next if we decide to do something, we can put light inside.\nSpeaker B: Yeah.\nSpeaker B: That would be a nice one.\nSpeaker D: That's gonna be expensive, you know.\nSpeaker D: We have some problems.\nSpeaker D: I think...\nSpeaker B: Going to have euros and...\nSpeaker B: It would be interesting to just to see if we can...\nSpeaker B: If people will buy this one and maybe add some features to it.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker D: If it's a new trend.\nNone: Okay.\nSpeaker D: So...\nSpeaker D: That's fine.\nSpeaker C: So, we can celebrate now.\nSpeaker C: It's our new product.\nSpeaker C: Champagne, see you later.\nSpeaker B: Mr. Baba.\nSpeaker C: It's fine.\nSpeaker C: I like it.\nSpeaker C: So bye-bye.\nNone: Next time.\nSpeaker C: We're eight by one.\nSpeaker A: Yeah.\nSpeaker C: So...\nSpeaker C: That's fine.\nSpeaker C: Okay.\nNone: I will close this.\nSpeaker B: Yeah.\nSpeaker B: Go.\nNone: So, I think we have...\nNone: Finished.\nNone: Yeah.\nNone: Okay.\nNone: Okay.\nSpeaker C: I will close this.\nSpeaker B: Yeah.\nSpeaker B: Go.\nSpeaker B: So, I think we have...\nSpeaker B: Finished.\nSpeaker C: Okay.\nSpeaker C: Okay.\nNone: Okay.\nNone: Okay.\nSpeaker B: So, I think we have...\nSpeaker B: Finished.\nSpeaker B: Okay.\nSpeaker B: The designing and the evaluation of our remote control.\nSpeaker D: Okay.\nSpeaker B: Yeah.\nSpeaker B: And we have a nearly final product.\nSpeaker D: Oh, final prototype.\nSpeaker B: Final prototype, right?\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker B: Okay.\nSpeaker B: Thank you very much.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker C: Very predictive.\nSpeaker C: Okay.\nSpeaker C: Thanks.\nNone: Okay.\nSpeaker C: So what is that storage in the remote?\nNone: Mm.\nNone: Yeah.\nNone: A memory upgrade is closing so that the grid is in there.\nNone: So, that should consume.\nNone: Okay.\nNone: Okay.\nNone: Mm.\n"}, {"meeting_id": "TS3011d", "summary": "This is the fourth meeting of the design group. The group firstly introduced a prototype which contained the characteristics determined at the last meeting. Then Project Manager pointed out that since the production costs were over budget, they should reduce costs by leaving out solar cells, not using voice recognition and changing the button material into plastics. Then the group went through the evaluation criteria of the product. They thought the remote control should be fancy, innovative, easy to use, easy to find and spongy and evaluated the prototype with these criteria. When evaluating the project, the group considered that it was tempered by the choice of components and the price. Eventually, the group decided to redesign the product.", "dialogue": "None: you you you you you you you N as It's the same as on the top of it with the round like this one.\nSpeaker B: But we do think it's what it is for.\nSpeaker B: Which prefers the customer.\nSpeaker D: I think this is the device which has a learning curve.\nSpeaker D: Not for users.\nSpeaker D: Users device as normal users use remote control.\nSpeaker D: And after a while they start to develop some skills in the voice recognition functions.\nSpeaker D: And then they will not use this dial as often.\nSpeaker D: But other users who are new to this device need something like that.\nSpeaker D: They need to understand how to change channels and change the volume.\nSpeaker D: So it's easier for them.\nSpeaker F: Could I see this crowbar as a sort of shortcut?\nSpeaker C: Yeah maybe so.\nSpeaker F: And voice recognition as well.\nSpeaker F: Maybe you could.\nSpeaker D: Wow it's another approach.\nSpeaker D: It's more that are as I feel back in the room.\nSpeaker D: There are many ways of doing things on such a device.\nSpeaker D: Yeah.\nSpeaker F: It's quite easy.\nSpeaker F: Okay good.\nSpeaker F: And the case is rubber.\nSpeaker D: Yes rubber.\nSpeaker F: And the buttons.\nSpeaker D: Plastic or rubber.\nSpeaker D: Plastic or rubber.\nSpeaker F: Oh yeah.\nSpeaker F: Okay.\nSpeaker F: And the coloring yellow with gray or black.\nSpeaker D: Or something like that.\nSpeaker D: Whatever cost the least.\nSpeaker F: Okay we'll come to that later.\nSpeaker F: Okay.\nSpeaker F: Anything else to add or?\nSpeaker D: No.\nSpeaker D: Maybe we should think about these buttons.\nSpeaker D: Yeah.\nSpeaker E: Or you could use two of them to program the channels on the two channel button.\nSpeaker D: But these are tasks that are only executed once I think.\nSpeaker F: Yeah.\nSpeaker F: But maybe you do want to program buttons to for example activate the voice recognition or train the voice recognition.\nSpeaker D: Oh okay.\nSpeaker B: That's right.\nSpeaker B: Or something on the button for disabling the voice recognition.\nSpeaker F: Yeah although by pressing the button for I don't know two or three seconds you could also say you disable it with a little beep.\nSpeaker F: Yeah.\nSpeaker F: Okay that's not really really important.\nSpeaker B: This is the basic idea.\nSpeaker B: The basic.\nSpeaker B: Okay.\nSpeaker F: For prototype.\nSpeaker F: Okay.\nSpeaker F: You thought of some evaluation criteria?\nSpeaker F: Yes.\nSpeaker C: Okay.\nNone: My name is not the name.\nSpeaker E: You are nameless.\nSpeaker F: You are nameless.\nSpeaker E: Well I use the earlier documents.\nSpeaker E: And these are the most important criteria.\nSpeaker E: That's how the fashion guy is stated.\nSpeaker E: Fancy look and feel.\nSpeaker B: So just walk through it step by step.\nSpeaker B: I mean it's fancy.\nSpeaker B: Everything I believe.\nSpeaker B: I believe it's fancy.\nSpeaker E: I believe it's fancy too.\nSpeaker E: But apparently we shouldn't evaluate it yet.\nSpeaker E: Oh.\nSpeaker F: Okay.\nSpeaker F: So these are the criteria.\nSpeaker E: I think these are the most important criteria.\nSpeaker E: Okay.\nSpeaker E: Well.\nSpeaker F: So that's about it.\nSpeaker F: Then we'll switch to my presentation.\nSpeaker F: The prediction costs.\nSpeaker F: The costs are not under.\nSpeaker F: Can I?\nSpeaker F: Aww.\nSpeaker F: This is the.\nSpeaker E: 22.\nSpeaker F: Yeah.\nSpeaker F: It's way above the 12 year of 50.\nSpeaker F: And what makes it very expensive is for example the solar cell.\nSpeaker F: So I guess we should skip that because it's not that important.\nSpeaker B: Okay.\nSpeaker B: What is the price?\nSpeaker B: And the price.\nSpeaker B: Yeah.\nSpeaker F: The number of items and the sum.\nSpeaker F: Well this is what I would call our luxury model.\nSpeaker F: If you look at the what we could do to make it more to make it just between the 12 year of 50 then I did the following changes.\nSpeaker F: 12 euro 40 cents I came up with by leaving out solar cells by not using the voice recognition feature because it's a four euro addition to the price.\nSpeaker F: Yeah.\nSpeaker F: I believe.\nSpeaker F: Push button.\nNone: Well.\nSpeaker F: It makes it 13.\nNone: Yeah.\nSpeaker F: Push buttons are not the most expensive but do at extra cost.\nSpeaker F: So yeah this design is not within our price model.\nSpeaker D: Okay.\nSpeaker D: But I'm afraid it's not complete because we use special materials.\nSpeaker D: The last item.\nSpeaker D: You have not added one item there.\nSpeaker D: Yeah.\nSpeaker D: So it's way too expensive.\nSpeaker F: It's still too expensive.\nSpeaker F: Yeah.\nSpeaker D: But that's only for the buttons.\nSpeaker F: I guess if we leave the if we leave this one out.\nSpeaker F: And maybe not use a special form.\nSpeaker B: And a plus and just plus button.\nSpeaker B: So instead of rubber.\nSpeaker F: It becomes a very dull remote control.\nSpeaker F: I know.\nSpeaker F: But it's the board decision.\nSpeaker F: And yeah.\nSpeaker D: Well basically when this is our only option we should even consider changing the casing because I think there's very little added value in an enhanced case with these dual functions.\nSpeaker F: Yeah.\nSpeaker F: I know maybe we should look at a focus on another type of movement.\nSpeaker F: Yeah.\nSpeaker F: Not all that fancy but just way way more easy.\nSpeaker F: Basic.\nSpeaker F: Yes.\nSpeaker F: And maximize the profit.\nSpeaker D: And that's maybe that's better.\nSpeaker D: Yeah.\nSpeaker D: Maybe we could we should go for straight and simple.\nSpeaker D: Yeah.\nSpeaker D: But that's not what has been asked.\nSpeaker F: I know.\nSpeaker D: I know.\nSpeaker D: So we should kick the boards.\nSpeaker D: Well.\nSpeaker F: Yeah.\nSpeaker F: Although I think we could still make a remote control that applies more to young people by giving it another color already.\nSpeaker F: So it is possible to make a device that attracts a little bit more to young people.\nSpeaker F: Yes.\nSpeaker F: But I agree it's not a fancy high tech device.\nSpeaker F: Definitely not.\nSpeaker F: It's not that innovative or however you pronounce that.\nSpeaker F: So, okay.\nSpeaker F: Oh, this is the wrong one.\nSpeaker F: So that means redesign.\nSpeaker F: We do not have the time now to redesign the product.\nSpeaker F: But we can evaluate the process and the satisfaction on how things went.\nSpeaker F: I'm not sure if we need to evaluate the device first.\nSpeaker F: I guess.\nSpeaker E: Well, since we're not going to manufacture it anyway.\nSpeaker F: Yeah.\nSpeaker F: That's true.\nSpeaker D: Well, maybe it's good to do it anyway.\nSpeaker D: Because if we can also determine if our objectives are good.\nSpeaker D: Yeah.\nSpeaker F: So.\nSpeaker F: I agree.\nSpeaker F: Well.\nSpeaker D: Is it fancy?\nSpeaker D: It is.\nSpeaker F: It is.\nSpeaker F: Yeah.\nSpeaker E: I think it's a rubber.\nSpeaker F: You like the rubber at all.\nSpeaker D: I'm into it.\nSpeaker D: So, one.\nSpeaker D: Yeah.\nSpeaker D: That's not that fancy.\nSpeaker F: No, I'll give it a two.\nSpeaker D: I think it would have been more fancy if we used the titanium housing.\nSpeaker D: The casing.\nSpeaker D: It would be.\nSpeaker D: You like that.\nSpeaker D: I'm into it.\nSpeaker F: That's a flavor as well.\nSpeaker D: It has flavor.\nSpeaker D: Yes, that's right.\nSpeaker D: But you taste it.\nSpeaker E: Right.\nSpeaker F: Yeah, I know.\nSpeaker F: But that's fancy in the way.\nSpeaker F: I mean, fancy has a lot of.\nSpeaker D: It has to do with fashion, I guess.\nSpeaker B: Yeah.\nSpeaker B: So, do we.\nSpeaker E: It's trendy.\nSpeaker E: Yeah.\nSpeaker E: What they wanted was colors and soft materials.\nSpeaker E: So, in that way.\nSpeaker F: It applies.\nSpeaker B: Yeah.\nSpeaker B: It's fancy.\nSpeaker B: I'll just give it a two.\nSpeaker C: Yeah.\nSpeaker D: Right.\nSpeaker C: I think it's not the ultimate thing.\nSpeaker D: It's not the ultimate thing.\nSpeaker D: It's not the ultimate thing.\nSpeaker D: It's not the ultimate thing.\nSpeaker D: But I would have been more fancy by using the double curved case.\nSpeaker D: It would have been even more fancy.\nSpeaker D: But we decided not to because if we use double.\nSpeaker D: Yeah, but that's the case.\nSpeaker D: We couldn't use solar.\nSpeaker B: So looking at the user needs, we only don't have the double curved case.\nSpeaker B: We do have the rubber.\nSpeaker B: We do have the colors.\nSpeaker B: There's two out of three.\nSpeaker B: Yes.\nSpeaker D: So, I believe we're close to two.\nSpeaker D: Yeah, I agree.\nSpeaker D: Yes, I agree.\nSpeaker D: It's okay.\nSpeaker D: We did good.\nSpeaker D: Yes, we did good.\nSpeaker E: Okay.\nSpeaker E: And what is it?\nSpeaker E: Inevitive.\nSpeaker F: Well, with the voice recognition feature.\nSpeaker D: And it's over.\nSpeaker D: Or can we?\nSpeaker D: No, we are evaluating this design now.\nSpeaker F: This prototype.\nSpeaker D: I think it is.\nSpeaker D: I think it's innovative.\nSpeaker D: Yeah.\nSpeaker D: In the scroll wheel, the solar, not many remotes have the solar, I think.\nSpeaker D: It could have been a little bit more innovative by using the kinetic energy source.\nSpeaker D: But it's what it would have been a thrill.\nSpeaker E: Yes.\nSpeaker E: Also, I think it's a two.\nSpeaker E: Is it easy to use?\nSpeaker F: I'm not sure.\nSpeaker F: I'm not sure.\nSpeaker B: Well, yeah, the voice recognition, of course, is hard to learn, I think.\nSpeaker B: Well, hard.\nSpeaker B: It's not for the elderly.\nSpeaker D: They won't use it.\nSpeaker D: There are two parts in this remote control.\nSpeaker D: What you see here is the basic part.\nSpeaker D: Everybody can use it.\nSpeaker D: So that's easy to use.\nSpeaker D: That's for another user.\nSpeaker D: When you have a more advanced level of users, well, such a user really would like to explore all these additional functions.\nSpeaker D: In that way, it is advanced.\nSpeaker D: I think it's easy to use for both types of users.\nSpeaker D: I think it's not very easy for you.\nSpeaker B: I think it's free.\nSpeaker F: Wouldn't give it more.\nSpeaker F: I'm doubting as well.\nSpeaker E: The most important function is easy to use.\nSpeaker E: The zapping, channel switching, volume.\nSpeaker E: But the more advanced functions are probably a bit harder.\nSpeaker F: Okay.\nSpeaker F: Two or three, what would be your guess?\nSpeaker F: I mean, ease of use does not only apply to the most basic functions.\nSpeaker F: It's overall.\nSpeaker F: It's the device easy to use.\nSpeaker D: That's right.\nSpeaker D: You're right in that.\nSpeaker D: But I guess an advanced user will find the voice recognition function easy to use.\nSpeaker D: Because he is already, she is already an advanced user.\nSpeaker D: Personally, I would give it two.\nSpeaker F: Okay.\nSpeaker F: At all, three.\nNone: Yeah.\nNone: Good question.\nSpeaker E: I'll go for the two.\nSpeaker C: Okay.\nSpeaker D: Two, two, and three.\nSpeaker D: Two-free.\nSpeaker D: So I could make it easy.\nSpeaker D: So that will be three in general.\nSpeaker D: No, two and a half.\nSpeaker D: If you make a four.\nSpeaker D: No, six and four.\nSpeaker D: Six and four is ten.\nSpeaker D: And by four is two and a half.\nSpeaker D: So darn.\nSpeaker D: May.\nSpeaker D: Yes.\nSpeaker D: But all.\nSpeaker D: Seven, three.\nSpeaker D: No.\nSpeaker D: Four and a three together.\nSpeaker D: Four.\nSpeaker B: Yeah.\nSpeaker B: You have a two.\nSpeaker B: Two, three.\nSpeaker B: Three.\nSpeaker B: No, I haven't said anything yet.\nSpeaker B: May.\nSpeaker F: I know.\nSpeaker F: Okay.\nSpeaker F: But if I would say a three, then it's six and four is ten.\nSpeaker B: Divided by four.\nSpeaker F: Divided by four is two.\nSpeaker B: Yeah.\nSpeaker B: If you want to have the conclusion as a three, then you would make a four.\nSpeaker B: If you fill out a four.\nSpeaker F: But I'm filling in a three, so it will be a 2.5.\nSpeaker D: But that's not possible to fill in.\nSpeaker D: Yes, it is.\nSpeaker F: So we have to run.\nSpeaker F: I have a veto.\nSpeaker F: Exactly.\nSpeaker F: It's not about the content.\nSpeaker F: It's about...\nSpeaker F: Okay.\nSpeaker E: Is it easy to find?\nSpeaker F: Yeah, definitely.\nSpeaker B: Yes.\nSpeaker B: Yeah, we haven't really worked it out, but it most doesn't...\nSpeaker B: You can just say find and he repeats find.\nSpeaker F: Yeah, or beeps or...\nSpeaker B: Yeah, but that's the basic idea of the speaker.\nSpeaker B: I think.\nSpeaker D: I'm here.\nSpeaker D: I'm here.\nSpeaker D: Something like that.\nSpeaker D: Maybe you have to program it once to let it respond to a certain word or a sentence and something like...\nSpeaker D: Even without it.\nSpeaker D: Where are you?\nSpeaker D: And then it will sing, I'm here.\nSpeaker D: Yeah.\nSpeaker D: So something like that.\nSpeaker F: We should not stay too long on this subject because of the time.\nSpeaker F: Yeah.\nSpeaker F: But I personally give it a one.\nSpeaker F: Sebastian?\nSpeaker D: Yeah, me too.\nSpeaker D: Me too.\nSpeaker F: Yeah, I think.\nSpeaker F: Right.\nSpeaker F: Well, the feel of remote control is spongy.\nSpeaker F: Well, it can't be more spongy.\nSpeaker D: Well, it can be.\nSpeaker D: There are cases in which the outside casing can be...\nSpeaker D: Is this one of our...\nSpeaker D: One of our options.\nSpeaker D: No, it's not one of our options.\nSpeaker D: No, okay.\nSpeaker D: This is when you're looking at the market.\nSpeaker F: For your options given, it's the most points you want.\nSpeaker D: Yes, but that's not what they're talking about, I think, because we compare all these characteristics with markets, with the real market.\nSpeaker D: So there are remote controls out there, which are a lot more spongy.\nSpeaker D: They're out there.\nSpeaker D: In this case, we've done the best we could.\nSpeaker D: Yes.\nSpeaker D: But it's not good enough, so it's a tool.\nSpeaker F: I'll give it a one.\nSpeaker D: I'll take one.\nSpeaker D: You'll take one?\nSpeaker D: What do you give it?\nSpeaker E: Oh, yeah, it depends, because it's the most spongy we can...\nSpeaker F: Yeah, I know, but you have to name it, because we need to go on for the time.\nSpeaker E: Well, if I give it a one, it will be a one- No, we want health calculations.\nSpeaker E: So I'll just give it a two and make this a 1.5.\nSpeaker D: No, no, no.\nSpeaker D: I'll change it.\nSpeaker D: I'll make it...\nSpeaker D: My mark will be a four.\nSpeaker F: You are...\nSpeaker F: Okay, the remote control offers enough features.\nSpeaker F: Well, Root, what do you think about it?\nSpeaker E: Well, the basic layout doesn't offer much, but the voice recognition could add a lot.\nNone: So...\nSpeaker D: Basically, it's completely programmable.\nSpeaker D: You can add very much functionality by using the voice recognition mode.\nSpeaker C: Yeah, I know.\nSpeaker D: It's quite a sense.\nSpeaker D: What we didn't talk about is...\nSpeaker B: Yeah, but it doesn't have the digits.\nSpeaker B: I believe it's...\nSpeaker B: If you ask yourself, it offers enough features, I don't think it has all the features a normal remote has.\nSpeaker F: I think it has.\nSpeaker E: It depends on what you implement.\nSpeaker F: Plus we didn't talk about it, but you do have remote controls that are able to adapt another signal.\nSpeaker F: So you place a regular remote control in front of the other one, hit the one or the two or the three, whatever, and it records the series...\nSpeaker F: The signals.\nSpeaker F: The signals.\nSpeaker F: So you could enter any comment you like.\nSpeaker F: As long as our device is able to reproduce the infrared signal.\nSpeaker F: So I think this is remote control with very high level features.\nSpeaker F: Absolutely.\nSpeaker F: Although there are a few buttons, but the inside is quite...\nSpeaker D: That's it.\nSpeaker D: Power, I guess.\nSpeaker D: Because a regular programmable remote control contains, well, really a lot of buttons, at least 40 buttons.\nSpeaker D: So it's quite complicated to get used to.\nSpeaker D: And this is quite simple.\nSpeaker D: You can use your voice to program it.\nSpeaker F: Okay.\nSpeaker F: Let's give it a number.\nSpeaker F: I'll give it a one.\nSpeaker F: For this type of market, I think it's a one.\nSpeaker B: I'll give it a two.\nSpeaker D: I'll give it a one.\nSpeaker E: I think a one, because with the voice recognition you could add anything you want.\nSpeaker E: That's like...\nSpeaker D: Yes.\nSpeaker D: I think we've succeeded in developing a product that's actually quite good, but not for this kind of market and not for this kind of price.\nSpeaker F: Yeah.\nSpeaker F: So high quality, low acceptance.\nSpeaker F: The product has high quality and is advanced, but whether or not our clients are willing to pay 25 euros for this kind of device is not sure.\nSpeaker F: Do you agree?\nSpeaker D: Yes, I agree.\nSpeaker E: I mean, even because it doesn't look advanced.\nNone: Yeah.\nSpeaker F: Yeah.\nSpeaker F: Maybe we should have a radar function.\nSpeaker B: Couldn't...\nSpeaker B: What's the selling price?\nSpeaker B: 50?\nSpeaker B: 25 euros.\nSpeaker B: 25.\nSpeaker B: And costs were 12.50.\nSpeaker B: But even now, if only our production costs were exceeded, the double, I think, production costs were 22.\nSpeaker B: So selling price would be about 50 euros.\nSpeaker D: Yeah.\nSpeaker D: That's quite expensive.\nSpeaker D: It's not very expensive for remote control that...\nSpeaker D: No.\nSpeaker D: It has this functional remote control of any TV kind, a Philips remote control.\nSpeaker D: You pay more than 50 euros.\nSpeaker D: It's quite expensive.\nSpeaker D: Yeah.\nSpeaker B: And I know from a few years ago, it was a hundred kilos.\nSpeaker F: Well, yeah, I know.\nSpeaker F: But you're paying for the brand because there are remote controls which control your stereo, television, DVD, CD player.\nSpeaker D: Yes.\nSpeaker D: 25 euros.\nSpeaker D: Yes, but you can learn this thing, all these functions.\nSpeaker D: And it's easier to use because those remote consoles don't offer voice recognition.\nSpeaker D: This one does.\nSpeaker D: So I think it's worth its price.\nSpeaker F: Okay.\nSpeaker F: You had an overall rating.\nSpeaker E: Yeah, but...\nSpeaker E: That's counting.\nSpeaker E: These ratings.\nSpeaker D: Yeah.\nSpeaker D: Well, it's about...\nSpeaker D: It should be about 1.15.\nSpeaker D: So something like that.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker F: Okay.\nSpeaker F: We'll go further on with the rest of the evaluation.\nSpeaker F: About the project itself, not about the product.\nSpeaker F: What did you think about the process, the project process?\nSpeaker C: Ruth.\nNone: Yeah.\nSpeaker E: Try to translate that.\nSpeaker A: Any other...\nSpeaker F: I think...\nSpeaker B: Okay.\nSpeaker B: The process was good.\nSpeaker B: But we weren't aware of the prices of the cost.\nSpeaker B: And that was in the big deal.\nSpeaker B: If we knew that before, actually we could have made the choice between...\nSpeaker B: Better decisions.\nSpeaker D: Yeah.\nSpeaker D: We had too little information actually.\nSpeaker D: And while looking at room for creativity, the choice of components was way too narrow.\nSpeaker D: So that was not really a process of...\nSpeaker F: So we could have been creative, but it was tempered by the choice of components and...\nSpeaker F: Yes, the price.\nSpeaker D: The price.\nSpeaker D: Well, in the first meeting, we already were very creative.\nSpeaker D: We thought of possibilities who are not possible with the occurrence or of manufacturing components.\nSpeaker D: So...\nSpeaker C: Yeah.\nSpeaker D: We're tempered by that, yes.\nSpeaker F: Okay, rule.\nSpeaker F: Any other thoughts on that?\nSpeaker B: No.\nSpeaker F: No.\nSpeaker F: Ruth.\nSpeaker F: I agree.\nSpeaker F: You agree.\nSpeaker F: Okay.\nSpeaker F: Leadership.\nSpeaker D: Fantastic.\nSpeaker D: Okay.\nSpeaker F: Rules on for his promotion.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker C: Is it okay?\nSpeaker D: Okay.\nSpeaker D: I think we're a good team.\nSpeaker F: I think so too.\nSpeaker F: It's, of course, a laboratory environment.\nSpeaker F: I missed it to be able to contact you in between.\nSpeaker F: And say, I'll...\nSpeaker F: I tried once, but it was not allowed.\nSpeaker D: Yeah.\nSpeaker F: So...\nSpeaker F: Yeah, but...\nSpeaker F: When taking in account the situation, I think we performed pretty well.\nSpeaker F: I think so too.\nSpeaker F: Yeah.\nSpeaker F: The means.\nSpeaker F: The smart board, the digital pen, did you like them?\nSpeaker D: The digital pen was okay, but the smart board was really bad.\nSpeaker D: Yeah.\nSpeaker D: Because of the response or...\nSpeaker D: The response is very slow and the possibilities are very limited.\nSpeaker D: It's not accurate.\nSpeaker C: Okay.\nSpeaker B: It has... yeah, yeah.\nSpeaker B: It's not accurate.\nSpeaker B: The pointing of the pen is not the place where it writes...\nSpeaker B: Where it draws?\nSpeaker B: Yeah, where it draws.\nSpeaker D: It's the drawing on the board is right from the pen.\nSpeaker D: Okay.\nSpeaker F: So it had to be...\nSpeaker F: You have to take that line.\nSpeaker D: There are lines.\nSpeaker D: Or what's the word?\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker F: Maybe it needs to be calibrated.\nSpeaker F: It was calibrated just before this means.\nSpeaker F: The one before the third meeting.\nSpeaker F: Okay.\nSpeaker F: So it's not the calibration.\nSpeaker F: It's the thing itself.\nSpeaker F: Okay.\nSpeaker F: I think.\nSpeaker F: Ruit, did you use the pen a lot?\nSpeaker F: Not at all.\nSpeaker F: Not at all.\nSpeaker F: Not really?\nSpeaker F: Okay.\nSpeaker F: I thought it was quite a handy thing.\nSpeaker F: I think so too.\nSpeaker F: Although I would like to see OCR.\nSpeaker F: Yeah.\nSpeaker B: Yes.\nSpeaker B: If it has OCR, I think I would use...\nSpeaker B: I just took notes for myself and that's it.\nSpeaker B: It was necessary for me to...\nSpeaker F: To digitize them.\nSpeaker B: Yeah.\nSpeaker B: Because if I want something on the computer, I just typed it.\nSpeaker B: I typed faster than the other right.\nSpeaker D: But I think it's a great solution for a gnome problem.\nSpeaker D: Writing down some notes, some information.\nSpeaker D: And then forgetting your notebook somewhere.\nSpeaker D: And losing all that information.\nSpeaker D: Because you have everything in one place.\nSpeaker D: And it's quite easy, quite easy.\nSpeaker D: It's possible to make this information digital.\nSpeaker D: I share it with others.\nSpeaker D: Yeah.\nSpeaker D: Quite easy way.\nSpeaker D: I think it's a good product.\nSpeaker D: I only think the shape of the pen is too big.\nSpeaker D: Yeah.\nSpeaker D: It's not quite an anomis.\nSpeaker D: Economic.\nSpeaker D: Er-genomis.\nSpeaker C: I know.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Erm...\nNone: What...\nSpeaker F: Ruit, what did you think about the smart boards?\nSpeaker E: Oh, I only use it to draw a rabbit.\nSpeaker E: Yeah.\nSpeaker E: Okay.\nSpeaker E: You can't really decide.\nSpeaker F: I missed a feature to easily select a slide and distribute it to the laptops.\nSpeaker F: I think that would be very easy if you could say, okay, I want to use this for my own work or my own presentation further on or...\nSpeaker E: Or the other way around.\nSpeaker F: Or the other way around that you could show.\nSpeaker F: But quite what PowerPoint does.\nSpeaker B: I know.\nSpeaker B: If you save this image, you can open it in your shared work folder.\nSpeaker B: I know.\nSpeaker B: So it's almost...\nSpeaker F: I know.\nSpeaker F: But we couldn't use that feature.\nSpeaker F: So I missed it.\nSpeaker C: Yeah.\nSpeaker F: We weren't able to do that.\nSpeaker F: At least I wasn't explained how to do such.\nSpeaker B: You know, and the function of filling an oval or an object.\nSpeaker D: The drawing...\nSpeaker C: It's not...\nSpeaker C: Capabilities are very limited.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: And when you're using windows, you're used to a certain interface and certain buttons which you can use for drawing.\nSpeaker D: Yeah.\nSpeaker D: And a lot of these buttons don't appear here.\nSpeaker D: So...\nSpeaker D: That looks like a paint.\nSpeaker D: Okay.\nSpeaker D: So it's not even as advanced as paint.\nSpeaker D: Not...\nSpeaker D: Not...\nSpeaker D: It's quite limited.\nNone: Okay.\nNone: No.\nSpeaker F: Yeah.\nSpeaker F: The project is evaluated.\nSpeaker F: But, well, we need to redesign the product.\nSpeaker D: Oh, very good celebration.\nSpeaker D: So...\nSpeaker D: It's a...\nSpeaker B: It's a privilege for working with you.\nSpeaker F: You're dismissed.\nSpeaker F: No, I think we are ready.\nSpeaker D: Okay.\nSpeaker D: Two private rooms?\nSpeaker D: I see some action over there.\nSpeaker F: Private room, at all.\nSpeaker F: That sounds quite scary.\nNone: Okay.\nSpeaker F: Now, let's find out the way to...\nSpeaker F: They don't want to do this.\nNone: There.\nNone: That was...\nSpeaker C: We're done.\nSpeaker D: We're finished.\nSpeaker D: I believe.\nSpeaker D: So...\nSpeaker D: Are there any more cycles in this process?\nSpeaker D: I think not.\nSpeaker D: I don't believe so.\nSpeaker F: Well, maybe we get an email.\nSpeaker F: Thank you for your...\nSpeaker D: But how much time did we get for this meeting?\nSpeaker D: Fourteen minutes.\nSpeaker D: And how much time is left?\nSpeaker F: A minute or ten, maybe?\nSpeaker F: Ten minutes.\nSpeaker F: Yeah, ten or five.\nSpeaker F: Okay.\nSpeaker D: Okay.\nSpeaker D: Okay.\nSpeaker F: So we can redesign our...\nSpeaker F: I would like to...\nSpeaker D: Well, I think we all know what the redesign should be.\nSpeaker D: A simple, built, one-colored box.\nSpeaker E: No added value.\nSpeaker D: No, it's just the same product that is already on the market.\nSpeaker B: You see the problem?\nSpeaker B: Yeah.\nSpeaker B: You can continue your line.\nSpeaker B: Well, it's fluffy, all right?\nSpeaker B: Spongy.\nSpeaker D: What is that?\nSpeaker D: Giraffe.\nSpeaker D: Giraffe eating leaves from a tree.\nSpeaker F: It's blue tongue.\nSpeaker E: Interesting design.\nSpeaker F: Yeah.\nSpeaker F: This is a new model.\nSpeaker D: So you're actually promoting Bluetooth or blue tongue?\nSpeaker F: Blue tongue.\nSpeaker E: But it does have a natural feeling.\nSpeaker E: So...\nSpeaker F: It's spongy.\nSpeaker C: Yeah.\nSpeaker F: Opa.\nSpeaker F: That is...\nSpeaker F: It's a new feature next to Bluetooth to disable all Bluetooth devices.\nSpeaker F: Okay.\nSpeaker D: Right.\nSpeaker D: Let's wrap it up.\nSpeaker F: Yeah.\nSpeaker F: We're done here.\nSpeaker F: Gentlemen, thank you for your cooperation.\nSpeaker D: Thank you, Mr. Manager.\nSpeaker D: Now, let's have a bottle of champagne.\nSpeaker B: All right. Let's give it here.\nSpeaker F: Yes.\nSpeaker B: That's it, mate.\nSpeaker D: No, it's not.\nSpeaker F: It's very good.\nSpeaker F: Mate, it's a microphone.\nSpeaker F: It's an all-tinger.\nNone: It's a nice one.\n"}, {"meeting_id": "ES2003d", "summary": "Industrial Designer and User Interface presented a prototype of the new remote control according to the team's previous meetings. However, due to the budget limit, the team had to give up the spongy rubber material and the double-curved design. Instead, the remote control would be made of plastic and have only one curve. There would not be any location function, either. The team decided to make the remote control conspicuous by designing a bright yellow banana shape in case it got lost easily in a room. In the product evaluation, the team was satisfied with its success in reducing the number of unused buttons. The user interface was considered to be user-friendly enough. However, the team also admitted that there was still room for improvement on the location function, technological innovation, the material, as well as the fashion style of the remote control. At the end of the meeting, all team members expressed themselves about the teamwork sincerely.", "dialogue": "Speaker B: should hopefully do a trick.\nSpeaker B: Sorry about this small delay.\nSpeaker B: I'm going a little bit behind schedule.\nSpeaker B: Okay, so just to try and roughly go over what we agreed in the last one.\nSpeaker B: We're going to go for something.\nSpeaker B: How was it the new black of a leaf?\nSpeaker B: Something that looks good because that seems to be in preference to actual functionality in the end, but we should never avoid functionality of course.\nSpeaker B: Many of our components are going to be standard off the shelf, but it seemed like we were going to require at least an advanced chip and we were still very much for the idea of using an LCD display.\nSpeaker B: Other things where we were hoping to use rubber, most likely going to be double curved, etc.\nSpeaker B: Okay, so due to your hard work we might as well let the two designers go first and show us the prototype.\nSpeaker B: Okay, that's right.\nSpeaker B: Why, how the best way to do this is I'm not sure, but...\nSpeaker A: Thank you, both.\nSpeaker A: Step up and outline our ideas.\nSpeaker A: Okay.\nSpeaker A: Now, doing the prototype gave us a bit more insight into the ergonomics of the design.\nSpeaker A: For one thing, it turned out that the only point at which it needs to be articulated for handedness is down here for the LED as it turned out.\nSpeaker A: The whole thing transfers from the right to left hand fairly well from the point of view of operating the function buttons in joystick.\nSpeaker A: So it might be an idea to be able to adjust the position as with the base of the joystick just a little bit.\nSpeaker A: So, perhaps I just thought you could simply have a slightly over-oiled shaped joystick that could then just be turned on a twisted round so that the stickier...\nSpeaker A: so that the bit that sticks out a bit more is on one side or the other.\nSpeaker A: But as you see with the holding it on the left hand, the LCD is nowhere useful, so that would need to be articulated.\nSpeaker A: If we're going to retain ergonomic design.\nSpeaker A: I got your note about keeping the cost down.\nSpeaker A: I'll say, yeah.\nSpeaker B: We'll go into that a bit more, but it's not.\nSpeaker A: This design could be done with plastic casing, though I would recommend around the grip part here in the middle, having to be just a rubber grip over that, which would allow for a slightly more bio-market form and a bit more ergonomic as well.\nSpeaker A: As for the single curve, well, this edge and this edge, like I said, would be nice to have some curvature to it, but it's not absolutely necessary.\nSpeaker A: Really the curve that's mostly needed is the underside so that the joystick rests over the edge of the hand like this.\nSpeaker A: Then you have the transmitter here.\nSpeaker A: No mean speaker for the remote control finder.\nSpeaker A: Any further comment?\nSpeaker D: Obviously it's going to be bulkier than it looks because it's going to be flat on one side, so the LCD will stick down like this one.\nSpeaker D: Because we can't get a curve.\nSpeaker D: Yeah, I mean that it costs and it's plastic as well, so it'll be as comfortable on the hand.\nSpeaker A: And with the rubber design, it could pretty much mould very much to the user's hand.\nSpeaker A: One nice feature if we could still do the rubber.\nSpeaker A: I thought it was to have the rubber extend beyond the end of the rigid substructure.\nSpeaker A: It has a sort of tail that you just drape over your wrist so it stays in position nicely.\nSpeaker A: Absolutely.\nSpeaker D: Yeah, that's it.\nSpeaker B: Okay, yeah, great.\nSpeaker C: Yeah, I've got a few load up my evaluation.\nSpeaker C: Yeah, okay.\nSpeaker B: We'll go for it.\nSpeaker E: Excellent work.\nSpeaker C: So, basically, we'll have a list of criteria that we need to rate their prototype by.\nSpeaker C: Then we will, it's a seven step kind of evaluation process.\nSpeaker C: So, not seven steps, seven scale.\nSpeaker C: So, after we finish doing all the ratings for each criteria, we average that and that would give us some type of confidence in our prototype.\nSpeaker C: And the criteria were based on real reactions, kind of goals and policies, marketing strategies, and also those are put together from the user requirements.\nSpeaker C: Okay, if you put the.\nSpeaker C: So, those are the criteria.\nSpeaker C: Perhaps I could put them a bit better, but you notice a few things that we've totally abandoned, which means that the product was scored very badly on some of those points.\nSpeaker C: So, we have them.\nSpeaker C: True.\nSpeaker C: Seven, eight.\nSpeaker C: Okay, so we have to go through each point.\nSpeaker C: We imagine it's actually straight and just give it a school.\nSpeaker C: So, how well would you say the prototype is how well we realize the dream of being able to stop remotes from being lost or to able to find them once they lost. I mean, is their home and things still the locator that's still.\nSpeaker C: Yeah, that's still part of the design.\nSpeaker C: Sure, and Adam, we can keep that in.\nSpeaker B: Yeah, I believe so.\nSpeaker B: So, I mean, I don't think anybody could actually stop a remote being lost because that would mean doing something a bit human.\nSpeaker B: But I'd like to think that we've done something about finding a damn thing once we have.\nSpeaker A: Sure.\nSpeaker A: And making it a bright color helps.\nSpeaker A: Personally, I would have gone for purple.\nSpeaker C: Bright color.\nSpeaker C: So, we still have that noise being done.\nSpeaker C: On a scale of one to seven, how would you guys rate it for finding it once it's lost?\nSpeaker A: I'd say number one.\nSpeaker C: Number one?\nSpeaker C: Okay.\nSpeaker C: Number one for the first criteria.\nSpeaker A: I think if it was just the sounder, then there's something I found with, what was he trying to find, the cardless phone on mobile.\nSpeaker A: You can hear it, but you can't quite pin it down where it is.\nSpeaker C: Yeah, you can tell what the mobile is.\nSpeaker D: What about the volume on the TV's turned up massively?\nSpeaker D: And this one I turned down the volume you can't find the remote.\nSpeaker D: Suppose you have to go to the TV and do a manual.\nSpeaker E: Yeah.\nSpeaker D: You wouldn't hear that it's speaking.\nSpeaker B: Just before we go through all of the steps.\nSpeaker B: You want to see something here?\nSpeaker B: Well, what we'll do is, if we can look at the criteria you're going to evaluate and then we'll come back to the product evaluation if that's all right.\nSpeaker B: That's fine.\nSpeaker B: Yeah.\nSpeaker C: That's the first one.\nSpeaker B: So is there anything here that you wanted to cover as a criteria that you've covered?\nSpeaker B: And then we'll come back pretty much properly to this.\nSpeaker B: Where's the main...\nSpeaker B: Is there any explaining or is there anything that you thought that really would stand out compared to the others?\nSpeaker C: A few.\nSpeaker C: Some have got neglected from my initial research is that rural reactions has a goal strategy that all of their products be inspired by interior fashion and clothing fashion.\nSpeaker C: That is why food and veg being popular in the home and in clothing was important.\nSpeaker C: And they want all their products to be somehow inspired by current trends and fashion.\nSpeaker C: So they say we put the fashion and electronics where they really mean it.\nSpeaker C: They're very big on fashion.\nSpeaker C: So, okay.\nSpeaker C: That's this bit right here.\nSpeaker C: And this bit is this one easy to use for visitors or for anybody.\nSpeaker C: I guess it's just the same as saying it's easy to use interface.\nSpeaker C: So we're kind of condensed into one.\nSpeaker C: And we can come back to it.\nSpeaker C: You said so.\nSpeaker B: No.\nSpeaker B: And which we will do very, very shortly.\nSpeaker B: Okay.\nSpeaker B: Slate problem we had was that we have an amazing four euros over budget for what we were hoping to do.\nSpeaker B: Most of it stems from the use of the LCD, which I think in the end accounted for about half of our extended trip because of course we required a chip as well.\nSpeaker B: The only way to get this down was either to ditch the LCD, at which point we removed a large part of how we were going to interface with the requirement buttons, etc.\nSpeaker B: Or what we did was that we, as an eyes of quickly going over it was altering the actual structure.\nSpeaker B: Changing it to plastic and a solid unit with a single curved design would allow us to come back into the proposed costs.\nSpeaker B: And we're just scraping it in.\nSpeaker B: We've got point two of a euro left over there.\nSpeaker B: So we're just managing it really.\nSpeaker B: Even then as well.\nSpeaker B: There was no criteria technically defined for a joystick.\nSpeaker B: So I've used what I think is appropriate with any luck.\nSpeaker B: That won't mean that we've incurred more cost than we can actually afford to.\nSpeaker B: It blows a lot of our really good ideas kind of slightly to one side.\nSpeaker B: For example, the possibility of having a USB connection is definitely not all valuable now.\nSpeaker B: Different languages.\nSpeaker B: That should still be viable.\nSpeaker B: We've got an advanced chip.\nSpeaker B: We've got the use of the LCD.\nSpeaker B: So being able to communicate in multiple languages is still very much a possibility.\nSpeaker B: But what's something we need to decide on is how we're going to go from here.\nSpeaker B: We do need to try and come up with an idea which could be continued with other people if need be.\nSpeaker B: Looking now can bring the Excel sheet up and show you if you wish.\nSpeaker B: I really think as much as it pains me is that we might have to go with plastic and some kind of solid design possibly meaning that the LCD wouldn't be in this perfect place.\nSpeaker B: It might be slightly between what would be good for left handed and what would be good for the right handed person.\nSpeaker A: I suppose one thing that could be done is have it circular and so that the pink is actually goes a bit over the pinky finger.\nSpeaker A: Okay.\nSpeaker B: So it very much is about making concessions unfortunately.\nSpeaker C: Do you have any data on how much different prints cost? I mean can you get the entire thing printed with a design?\nSpeaker B: I have as much information as I do.\nSpeaker B: So as you can see here for example the battery really not very little choice in that one.\nSpeaker B: We've gone from one of the cheaper options as well. Unfortunately we require the advanced chip if we're going to do what we're needing to.\nSpeaker B: I've said single curved. We really do need it to be that way for the ergonomics of it.\nSpeaker B: Plastic for some reason occurs no cost which I've had to very much make advantage of despite the fact that robbers only got a value to euros per unit.\nSpeaker B: Problem comes here as you can see in the interface.\nSpeaker B: If I've read this incorrectly then you can save 0.5 of a euro here and it's not per push button.\nSpeaker B: That might make sense because then a numeric keypad would come in at 4.5 euros which is an awful lot so that could well be wrong.\nSpeaker B: Even if we save 0.5 there it would just mean it was most likely placing it and actually just gaining a color for the unit which we set to be put to some one side.\nSpeaker B: As you can see the use of an LC display advanced chip and what would determine a scroll wheel here as well because it's an integrated school school push button.\nSpeaker B: It wasn't quite what I think they had in mind really joystick.\nSpeaker C: Why would that be more expensive than an individual push button? A scroll wheel together?\nSpeaker C: That's quite significantly expensive.\nSpeaker B: That's something I have to take up with the bean counters.\nSpeaker B: As you can see that's taken up well over half the price.\nSpeaker B: I'm very much open to suggestions of where we go but because we need to shed what was for euros off of the price of what we really desired.\nSpeaker B: This one comes in on the price as you can see but this was the one that sacrificed the curio for the case and for the actual case design.\nSpeaker C: We don't even have speakers here.\nSpeaker C: What about speakers and transmitters and stuff like that? I really like to tell them.\nSpeaker B: I know we haven't.\nSpeaker C: We have to receive our speakers plus the extra device itself that's going to be on the TV.\nSpeaker B: It literally would be a button.\nSpeaker D: We might have to express them.\nSpeaker B: It looks like almost nothing.\nSpeaker B: No good call.\nSpeaker B: I mean it's not on here but I'm doing a very valid point.\nSpeaker C: Do we have to use an advanced chip for the LCD?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: That's...\nSpeaker B: So if we're going to go with the LCD display then that's...\nSpeaker C: What's that hand down?\nSpeaker C: Dynamo.\nSpeaker C: You have to wind it up?\nSpeaker B: I feel like so yeah.\nSpeaker B: That would probably not be in keeping with the fashion statement.\nSpeaker B: Yeah.\nSpeaker D: This is basically the only new thing is the LCD on the...\nSpeaker B: Manipulated by the joystick here.\nSpeaker D: I'm just...\nSpeaker B: Which I'm defining as scroll wheel.\nSpeaker C: And we couldn't replace the joystick quite because we would need to pull extra buttons to replace it up down left and right and that would be more expensive than it.\nSpeaker C: But is the scroll wheel not just back and forth?\nSpeaker B: Yeah, it's just because there was no actual definition for what a joystick might be that...\nSpeaker B: That's what I've labeled it for, purposes of this evaluation.\nSpeaker D: The LCD basically is the big selling point of...\nSpeaker B: If we remove the LCD display we could save ourselves a fair amount.\nSpeaker D: Which you can't sort of make original but...\nSpeaker A: Yeah.\nSpeaker A: If we remove the LCD display then there would absolutely no point to any of these meetings.\nSpeaker A: But we just put our branding on any other remote control.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker B: It's a shame we should possibly...\nSpeaker B: If we could have increased the price we could have manufactured that and then we could have got some...\nSpeaker B: Part closer to what we were.\nSpeaker B: Does this open to you?\nSpeaker C: Does this bear in mind that...\nSpeaker C: I mean it's a bit ridiculous that they're going to charge us.\nSpeaker C: Was it like this much money for three million if we're going to buy three million components?\nSpeaker B: Again, you'll have to argue the accountants on that one.\nSpeaker B: But for the purposes of this meeting we're going to have to stick with these figures.\nSpeaker B: So...\nSpeaker B: I would say that it would seem like the general opinion is we're going to keep the LCD display because it's about what really separates us.\nSpeaker B: It's like the cost it's going to incur.\nSpeaker B: Our people maybe not happy with but are willing to go ahead with this and go in for a plastic solid case to keep the LCD.\nSpeaker A: I'm one thing I...\nSpeaker A: How much extra would it be to keep the articulation?\nSpeaker B: It's hard to tell. I would say that you're at least going to take double curved and even then I'm not quite sure if that's incorporating the idea of articulation.\nSpeaker B: Yeah.\nSpeaker A: Well, I think that it didn't require it to be double curved.\nSpeaker D: Yeah, it could still be single curved but it's just a little bit of articulation.\nSpeaker A: It's just that the case would come into it and would be made in two parts and then join together with an articulation.\nSpeaker C: Could we not get rid of the curved base and focus on the menu being the best interface?\nSpeaker C: Do we have restrictions on software?\nSpeaker D: That's what you need for the joystick.\nSpeaker A: Oh, but it has to be.\nSpeaker A: I mean, if you look closer at the prototype here, the lines here along grip are actually quite straight.\nSpeaker A: But the curve's all over the hand, is it?\nSpeaker A: Yeah, on the LCD, I mean, although we've done it with the curve, it could just as easily be done without curves.\nSpeaker A: The curve that's really needed is up here to keep the joystick in a good ergonomic position without to have it rest on the top of the hand.\nSpeaker A: Okay.\nSpeaker C: Sure.\nSpeaker C: Okay, my bad.\nSpeaker B: We wouldn't actually save a lot by reducing it anyway.\nSpeaker B: So, I mean, for the purposes of this meeting, maybe we can state that single curve still allows articulation.\nSpeaker B: Unless we hear otherwise, we could go ahead with that proposal.\nSpeaker C: Sure.\nSpeaker C: The product is not going to perform so well for my criteria.\nSpeaker B: Which is what we can get on to now.\nSpeaker B: As long as, so we're going to say, we'll have to keep an eye on the time as well, but we're going to say, um, Oh, wait a minute.\nSpeaker C: Single curve design sample speaker.\nSpeaker C: What is the sample speaker?\nSpeaker C: Is that somewhat similar to what we want?\nSpeaker C: It could well be the end of the course.\nSpeaker A: That voice response thing.\nSpeaker A: I forgot the email about.\nSpeaker A: I thought it was just completely pointless.\nSpeaker C: You got an email about voice response?\nSpeaker C: Yeah.\nSpeaker A: I did not.\nSpeaker A: All right.\nSpeaker A: But basically, it's saying that your labs had come up with a chip that you could say hello to and it would say hello back in a friendly female.\nSpeaker A: Okay.\nSpeaker B: We'll definitely won't go with that one.\nSpeaker B: We won't go with that one.\nSpeaker C: I mean, that's voice recognition.\nSpeaker B: Okay.\nSpeaker C: Okay.\nSpeaker C: Um, so, okay.\nSpeaker B: Battery definitely.\nSpeaker B: So it looks like we're going to get rid of the whole locate locator.\nSpeaker B: It looks like unless we can manage to put it in under point two years.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Maybe even slick.\nSpeaker B: Well, no, you have pretty much point two euros, I'd say.\nSpeaker B: So we'll leave that one for now.\nSpeaker D: Also, we don't for a special color at all.\nSpeaker B: It's a case of, um, I'm slightly unsure.\nSpeaker B: One point five of a euro for one push button doesn't sound quite right.\nSpeaker B: So maybe it's a case of a push button is maybe one or more.\nSpeaker B: Um, well, at which point for a case.\nSpeaker D: Oh, you already incorporated that.\nSpeaker C: Oh, special color for the case.\nSpeaker B: Well, you got point five there.\nSpeaker B: It's literally a case of whether or not this is correct.\nSpeaker B: I'm not quite sure if they're, I don't think they mean point five euros per button.\nSpeaker A: Okay.\nSpeaker A: Well, let's say that and then we can have our special color case and then we at least have.\nSpeaker A: Make it a little harder to lose.\nSpeaker A: There we go.\nSpeaker A: Because most, most remotes are a fairly dingy color that gets camouflaged under any pile of crap.\nSpeaker C: What's the default color, white or black?\nSpeaker B: Black probably normal color you'd see gray.\nSpeaker B: I quite like that color that you're fetching now.\nSpeaker B: It's definitely make it glow in the dark even better.\nSpeaker B: Yeah.\nSpeaker B: So, will we go with that then?\nSpeaker E: Okay.\nSpeaker B: It's not and we can see, we'll come back.\nSpeaker B: You're evaluation in which you're probably now going to pan us.\nSpeaker C: Okay.\nSpeaker C: Okay.\nSpeaker B: So, uh, just to give you an idea, um, you'll want to go maybe a bit quickly as well.\nSpeaker B: I'm not sure how much time we've not hit the five minute mark.\nSpeaker D: We're working yet, but okay.\nSpeaker D: Okay.\nSpeaker C: Okay.\nSpeaker C: Ability to start remote some ring gloss.\nSpeaker C: Let's find them once they lost.\nSpeaker C: Um, okay.\nSpeaker C: Special color.\nSpeaker B: Uh, four, four, three, three, we're being generous.\nSpeaker A: I think we can do three, three.\nSpeaker A: We're being generous here with a special color.\nSpeaker A: What I think make a difference.\nSpeaker A: It makes it stand out from, you know, it's lost in a big pile of crap.\nSpeaker A: It stands out from the rest of the crap.\nSpeaker A: Okay.\nSpeaker C: Okay.\nSpeaker C: We'd use the number of one use buttons.\nSpeaker C: We're down to two buttons.\nSpeaker C: Two buttons.\nSpeaker C: Okay.\nSpeaker C: So that's a one, you know, we're, that's.\nSpeaker B: I'd say we're doing well there.\nSpeaker C: Okay.\nSpeaker C: That was good.\nSpeaker C: Easy to use interface buttons.\nSpeaker C: Many, many is cool.\nSpeaker C: That's good.\nSpeaker C: Okay.\nSpeaker C: That's when I've done so badly.\nSpeaker C: Easy to use.\nSpeaker C: Oh, okay.\nSpeaker C: Let's forget that one.\nSpeaker C: Fancy looking.\nSpeaker A: It doesn't get much.\nSpeaker C: Sure.\nSpeaker C: We could do whatever we like with the LCD.\nSpeaker C: Yeah.\nSpeaker C: Let's just assume it's a good LCD display.\nSpeaker C: Maybe I was panicking for no reason.\nSpeaker D: I'll tell you go to because like the fans used to work the door.\nSpeaker D: Yeah.\nSpeaker C: Maybe a bit.\nSpeaker C: Yeah.\nSpeaker C: With the articulators.\nSpeaker C: There we go.\nSpeaker C: That's better.\nSpeaker C: To more accurate numbers.\nSpeaker C: No logically innovative.\nSpeaker C: Well, we're getting rid of the locator thing, which is a shame.\nSpeaker C: Which, yeah.\nSpeaker A: I don't need for this for that.\nSpeaker A: Free.\nSpeaker A: Yeah.\nSpeaker A: And the menus thing is something you don't normally see on a remote.\nSpeaker A: You see it in a lot of other places.\nSpeaker C: Yeah.\nSpeaker C: Mobile phones.\nSpeaker C: And what you're doing is moving the menu from the television to the remote control source.\nSpeaker C: You say free?\nSpeaker C: I might go as far as two on that.\nSpeaker C: Free.\nSpeaker A: I'm going to 10, 2, 3.\nNone: OK.\nSpeaker B: OK.\nSpeaker B: And the next one anyway.\nSpeaker C: There are some people find pleasing.\nSpeaker C: Spungyness is what they really would have wanted.\nSpeaker B: It is.\nSpeaker B: Yep.\nSpeaker B: Don't blame them.\nSpeaker B: Because of the way that we've minimized the number of buttons and such.\nSpeaker B: Plastic.\nSpeaker B: It sucks.\nSpeaker B: But it's no worse than any of the other remote controls they have.\nSpeaker C: That's true.\nSpeaker C: It's not a step backwards.\nSpeaker D: Yeah.\nSpeaker A: Five.\nSpeaker A: I give it a six.\nSpeaker A: Yeah.\nSpeaker A: OK.\nSpeaker C: Let's give it a six.\nSpeaker C: OK.\nSpeaker C: That's totally from everything with balance.\nSpeaker C: Inspired by the latest interior and clothing fashion.\nSpeaker C: We could.\nSpeaker C: What color will we going to make it?\nSpeaker A: Leopard print on it.\nSpeaker A: I would say give us a selection of colors.\nSpeaker A: I don't know.\nSpeaker A: Yellow for the prototype.\nSpeaker A: If I were buying one, I'd go for the particle.\nSpeaker C: Leopard print would be cool.\nSpeaker D: But I think it's more OK.\nSpeaker D: Yeah.\nSpeaker D: The color of the border there.\nSpeaker D: It's fine.\nSpeaker D: That's not standard.\nSpeaker B: That yellow.\nSpeaker B: It would also help keep the product placement local.\nSpeaker B: But they're referring to the fruit and veg thing.\nSpeaker C: Is this like a banana type color?\nSpeaker C: Could we stretch?\nSpeaker C: No, there's no shape like a banana.\nSpeaker A: It's kind of dope.\nSpeaker A: It won't be when it's been.\nSpeaker C: Oh, like because it's flat.\nSpeaker A: What is this?\nSpeaker C: Fruit.\nSpeaker C: It's just flat.\nSpeaker A: I think this is not particularly fruit and veg.\nSpeaker A: Yeah.\nSpeaker C: We might have to suffer badly for this one as well.\nSpeaker C: Yeah.\nSpeaker B: Well, I mean, it's probably more fruit and veg than most other things that are fruit and veg.\nSpeaker B: So, what?\nSpeaker C: Four?\nSpeaker C: Four.\nSpeaker C: All right.\nSpeaker A: It's very ambitious.\nSpeaker A: Yeah.\nSpeaker A: I don't think fruit and veg is the sole criteria.\nSpeaker A: The sole criteria for being.\nSpeaker A: Inspired by the way.\nSpeaker A: Action for inspired by current passions.\nSpeaker A: I can't wait to start with that.\nSpeaker D: This is their motto.\nSpeaker D: Yeah.\nSpeaker D: I mean, this is their strategy.\nSpeaker C: I imagine we actually had some money invested in this and the amount that we invested is going to be proportional to the marks.\nSpeaker C: Might want to be a bit more skeptical about this one.\nSpeaker B: What would you think, yourself?\nSpeaker C: I would say, I mean, it's not at all.\nSpeaker C: Right.\nSpeaker C: In any way of shape or form.\nSpeaker B: Well, it's kind of curved and we can make it yellow.\nSpeaker B: And that's pretty much banana-like.\nSpeaker C: Okay.\nSpeaker C: Yellow banana light fingers.\nSpeaker C: I think it's okay.\nSpeaker B: It's got a curve to it.\nSpeaker C: Five.\nSpeaker C: Is that so unreadable?\nSpeaker C: Yeah.\nSpeaker B: I'm not stretching the use of a banana.\nSpeaker B: Yeah.\nSpeaker C: Okay.\nSpeaker C: So we have one, two, three, four, five, six, seven.\nSpeaker C: So five, seven, ten, sixteen, twenty one.\nSpeaker C: This gives us an average of three.\nSpeaker C: Well, this would be in the middle.\nSpeaker C: So it's not bad.\nSpeaker B: It's in a good section.\nSpeaker B: It's not bad and considering the...\nSpeaker B: Don't break the fan.\nSpeaker C: Oops.\nSpeaker C: Sorry.\nSpeaker C: I'm sorry.\nSpeaker B: What are you not badly scared of?\nSpeaker B: It's bad to design that.\nSpeaker B: Considering the price we have to get this in, we have a positive, you know, even based on the four of us being heavily biased.\nSpeaker B: Yeah.\nSpeaker B: It was going to be quite hard to get anything standing out, I'd say, possibly, based on the cost teachers.\nSpeaker C: Even if we were to increase this entire thing by seven, we were to go down a grade to four, we would have to do...\nSpeaker C: I mean, we weren't that kind of optimistic, too overly optimistic, you know, like we didn't add, we didn't subtract our whole seven points from the use of things.\nSpeaker C: So I think we're definitely on the good bit.\nSpeaker C: Even if we gave this one seven and this one seven, that's still only three extra points over seven.\nSpeaker C: You know, it's...\nSpeaker C: Yeah, we did. It was okay. It was good.\nSpeaker A: Personally, I think...\nSpeaker A: Given that the product only replaces a single remote control that you already got, I think we're really going to show out 25 euros for something that's only marginally good.\nSpeaker D: Well, depends here, what the target people are, like you'd say.\nSpeaker D: Maybe fashion conscious.\nSpeaker B: Maybe women target.\nSpeaker D: Oh, no, look at that. It's cool. It looks like a...\nSpeaker D: It's yellow. It looks like a banana. It's cool. It's got a look good in the sitting room.\nSpeaker D: And rather than LCD, whereas more technical, like more people in the divideds technology, like as good as going on LCD screens.\nSpeaker D: Only got two buttons and a joystick.\nSpeaker D: So which kind of people would be more likely to buy?\nSpeaker B: Probably the people technologically do. Usually it wants to buy pointless stuff.\nSpeaker C: I mean, my mom still has not learned how to use text messaging on her phone, and she's had it for a long time. She uses it to make phone calls and that's it.\nSpeaker C: So I think if my mom's a remote control like this, we're only two buttons and a joystick.\nSpeaker C: I mean, that'll probably be the first one she decides not to buy.\nSpeaker C: She'd be like, is this a remote control? I don't...\nSpeaker C: How do you use it and stuff like that?\nSpeaker C: Even if it is really user friendly to us, but we're used to using menus all the time.\nSpeaker A: That's one thing is that because it's technically innovative, for someone who's sort of technophobic, in fact, it simply looks unfamiliar or be daunting.\nSpeaker C: Yeah. I think it's totally radical to have a remote control, no numbered buttons.\nSpeaker C: But like radical, good.\nSpeaker B: Okay. I don't know how much longer we've got. At least five minutes, I think.\nSpeaker B: Quickly, we'll pop onto project evaluation.\nSpeaker B: So...\nSpeaker B: We've got these four criteria here for satisfaction.\nSpeaker B: Does anybody want to...\nSpeaker B: Do you have any opinions on any of them? For example, or backwards, I suppose, the ability to work in this project using the technology we've been presented with.\nSpeaker B: These people made good use of the pen and paper.\nSpeaker C: I would say...\nSpeaker C: I wrote Mary A. Page, but not.\nSpeaker B: I'm quite sure what the advantage for us using a digital pen might be.\nSpeaker A: I think it's a thing with digital pens, most of it.\nSpeaker A: I mean, the...\nSpeaker A: It goes into their corpus.\nSpeaker A: That would have been nice to be able to transfer the paper notes onto the computer.\nSpeaker B: Yeah, that would have been pretty good.\nSpeaker B: It does seem like paper's still a heavy consideration for taking notes.\nSpeaker B: So maybe this is literally just the way I owned it.\nSpeaker B: I heard people satisfied with the teamwork we've managed to display today.\nSpeaker C: I like it.\nSpeaker B: Leadership.\nSpeaker B: As much as can be leader than this thing.\nSpeaker B: I like that.\nSpeaker B: Last one was about a group of creativity.\nSpeaker A: Until accounts came along.\nSpeaker C: We're not lacking in ideas.\nSpeaker C: That was not the problem.\nSpeaker B: Yeah, I think in the end ideas can be used and sadly is meaning not so much that we weren't full of ideas but of ones that are going to allow us to actually build the thing.\nSpeaker B: It's a bit of a pity.\nSpeaker B: I would have to agree in that I think we needed a larger budget.\nSpeaker B: If you're going to, in your product, maybe a technological kind of sector, then you can afford to maybe chat to price up slightly from what it is because they will pay out.\nSpeaker B: They just cash the bevel.\nSpeaker A: I think retaining the modern biomarker for the new articulation would gain more profit in sales than it would lose in added expense.\nSpeaker D: The price was twice the assembly cost.\nSpeaker D: And would it have to be twice the, it could have had the assembly cost.\nSpeaker D: It could still set up for 25.\nSpeaker B: It's true.\nSpeaker B: I suppose these are all, will have to be taken up with a different group, I guess, as to who the costs involved.\nSpeaker B: But we've got a prototype.\nSpeaker B: So I think it's gone okay today considering the information that we've had at this point.\nSpeaker C: Maybe the accounts would have been better if we had a list to begin with.\nSpeaker B: It probably would have, you could have come up with a lot more solid design in the end.\nSpeaker B: I would have to agree.\nSpeaker B: It is very much a pity to get so far into the stage and then find out that maybe some of your ideas are just a bit too extensive.\nSpeaker B: Always hard to tell until you know the costs.\nSpeaker B: Okay.\nSpeaker B: So we're at a cost-within-budget, we are now that we have our slightly less than capable product.\nSpeaker B: We've evaluated it and we can see we came out with a value of three.\nSpeaker A: One thing I would say, I mean something that could perhaps be part of the product, the product testing and market research process would be to produce more cups of both versions and seeing just how much of a difference the over going over budget would make the sales.\nSpeaker C: Yeah, and like response from consumers.\nSpeaker A: We could even market two versions.\nSpeaker A: We keep you in a nice fine market.\nSpeaker B: And then the final one where you get to call it how.\nSpeaker B: Right.\nSpeaker B: Is there anything else that anybody would like to add?\nSpeaker B: Anything that's not been covered before I quickly write up a final report?\nSpeaker B: I don't know, I mean we've got a product, we maybe aren't as happy with it as we like to be, but we've got something we think we can maybe stick on to the market and sell.\nSpeaker B: And of course something we have been avoiding talking about because we've no information is selling them directly to the manufacturers.\nSpeaker B: There's a huge market.\nSpeaker B: I mean we briefly touched in it with no more knowledge than there's a little we can say on that.\nSpeaker B: Yeah.\nSpeaker B: So unless anybody's got anything they'd like to add, we can maybe round this up slightly earlier than we need to and then we can finish up the writing in such.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Thank you for your participation.\nNone: Thank you.\nSpeaker C: I was actually kind of upset at the budget and that we had to cut a lot of stuff.\nSpeaker C: But man, we can't have the located thing.\nSpeaker C: Yeah, that's a fact.\nSpeaker C: If it may be were the prices were made.\nSpeaker B: That's a question you've been asking about.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "Bro010", "summary": "The meeting was about various approaches that the team could pursue for further research. The team initially discussed the delay between the frequency bands which was causing problems in the model deployment. They then discussed how spectral subtraction could help various tasks. The meeting ended with more ideas on improving the model, like directionality of voice and looking at smaller units of sound in recordings.", "dialogue": "Speaker E: Okay, we're on.\nSpeaker F: Okay, what are we talking about today?\nSpeaker B: I don't know.\nSpeaker B: We have news from the conference talk.\nSpeaker B: This was program for yesterday.\nSpeaker F: Oh, I'm sorry.\nSpeaker F: Now I know what you're talking about.\nSpeaker F: No one told me anything.\nSpeaker E: This was the talk where they were supposed to try to decide.\nSpeaker E: To decide what to do.\nSpeaker F: That would have been a good thing to find out for this meeting.\nSpeaker F: I have no idea.\nSpeaker F: So, let's assume for right now we're just kind of plugging on ahead because even if they tell us that the worlds are different, we're still interested in doing what we're doing.\nSpeaker F: So what are you doing?\nSpeaker B: Well, we've a little bit worked on trying to see what were the bugs and the problems with the latencies.\nSpeaker B: We took the LDA filters and redesigned new filters using recursive filters.\nSpeaker F: So when you say we did something similar as that.\nSpeaker F: I'm sorry.\nSpeaker B: Who is doing that?\nSpeaker B: So we took the filters, the fear filters, and we designed our filters that have the same frequency response.\nSpeaker B: Well, similar.\nSpeaker B: But that have shorter delays.\nSpeaker B: So they had two filters, one for the low frequency bands and another for the high frequency bands.\nSpeaker B: So we redesigned two filters and the low frequency bands has 64 milliseconds of delay and the high frequency band filter has something like 11 milliseconds compared to the 200 milliseconds of the high-high filters.\nSpeaker B: But it's not yet tested.\nSpeaker B: So we have the filters but we still have to implement the routine that does recursive filtering.\nSpeaker F: You had a discussion with Sonia about this though?\nSpeaker F: No.\nSpeaker F: You should talk with them.\nSpeaker F: No, the whole problem happened before was coordination.\nSpeaker F: So you need to discuss with them what we're doing because they could be doing the same thing.\nSpeaker B: I don't know if that's what they were trying to do.\nSpeaker B: I'm trying to do something different like taking while using filter that takes only a past.\nSpeaker B: This is just a little bit different.\nSpeaker B: But I will send him an email and tell him exactly what we are doing.\nSpeaker F: Yeah.\nSpeaker F: We just have to be in contact more.\nSpeaker F: I think that the fact that we did that was, had that thing with the latency.\nSpeaker F: It was a decade of the fact that there wasn't enough communication.\nSpeaker B: Yeah.\nSpeaker B: Well, there is one remark about these filters that they don't have a linear phase.\nSpeaker B: So, but I don't know.\nSpeaker B: Rapid.\nSpeaker B: Rapid doesn't hurt because the phase is almost linear.\nSpeaker B: And so, yeah, for the delay, I give you air.\nSpeaker B: It's a computing done.\nSpeaker B: The five-hertz modulation frequency, which is the most important for speech.\nSpeaker B: This is the first thing.\nSpeaker F: So that would be a reduction of 136 milliseconds.\nSpeaker F: Which, what was the total we ended up with?\nSpeaker B: 330.\nSpeaker B: So that would be within.\nSpeaker B: Yeah, but there are other points actually, which will perhaps add some more delay.\nSpeaker B: Is that some other stuff in the process where perhaps not very correct.\nSpeaker B: Like the down sampling, which has simply dropping frames.\nSpeaker B: So we will try also to add a nice down sampling having a filter.\nSpeaker B: That, that, well, a low pass filter at 25 hertz.\nSpeaker B: Because when we look at the LDA filters, well, they are basically low pass, but they leave a lot of what's above 25 hertz.\nSpeaker B: And so, yeah, this would be another filter, which would add 10 milliseconds again.\nSpeaker B: Yeah, and then the third thing is that basically the way online normalization was done is just using this recursion on the feature stream.\nSpeaker B: But this is a filter, so it has also a delay.\nSpeaker B: And when we look at this filter actually has a delay of 85 milliseconds.\nSpeaker B: So if we want to be very correct, so if we want to the estimation of the mean to be, well, the right estimation of the mean, we have to take 85 milliseconds in the future.\nSpeaker F: That's a little bit of a problem.\nSpeaker B: But, well, when we add up everything, it would be at 65 plus 10 plus for the down sampling plus 85 for the online normalization.\nSpeaker B: So it's, yeah, plus 80 for the neural net and PCA.\nSpeaker B: So it would be around 240.\nSpeaker B: What's the allowable?\nSpeaker F: 250.\nSpeaker F: Which there's some discussion of.\nSpeaker E: What were they thinking of changing it to?\nSpeaker F: Well, the people who we had very low latency wanted to be low, very narrow latency bound.\nSpeaker F: And people have longer latency down.\nSpeaker F: So fortunately, we're the main ones with long latency.\nSpeaker B: But yeah, and basically the best proposed on that something like 30 or 40 milliseconds.\nSpeaker F: So they were basically, I mean, they were more or less trading computation for performance.\nSpeaker F: And we were trading latency for performance.\nSpeaker F: And they were dealing with noise explicitly and we weren't.\nSpeaker F: So I think of it as complimentary.\nSpeaker F: I think the best systems.\nSpeaker F: So everything that we did in a way was it was just adamantly insisting on going in with the brain damage system.\nSpeaker F: Which is something actually we've done a lot over the last 15 years.\nSpeaker F: We say, well, this is the way we do it.\nSpeaker F: And then we do it.\nSpeaker F: And then someone else does something that's straightforward.\nSpeaker F: So this was a test that largely had as with noise.\nSpeaker F: And we did absolutely nothing explicitly to handle the added noise.\nSpeaker F: We just trained up systems to be more discriminant.\nSpeaker F: And we did this Rastelike filtering, which just on the log domain, and it's attending to handle convolutional noise.\nSpeaker F: We actually did nothing about that.\nSpeaker F: So the spectral subtraction schemes, a couple of places, seemed to do a nice job.\nSpeaker F: And so we're talking about putting some of that in, still keeping some of our stuff.\nSpeaker F: I think you should be able to end up with a system that's better than both.\nSpeaker F: But clearly, the way that we're operating for this other stuff does involve some latency.\nSpeaker F: To get rid of most of that latency, you get down to 40 or 50 milliseconds.\nSpeaker F: We'd have to throw out most of what we're doing.\nSpeaker F: And I don't think there's any good reason for it.\nSpeaker F: And the application actually, I mean, you're speaking to a recognizer.\nSpeaker F: I'm a remote server.\nSpeaker F: And having a quarter second versus some processing to clean it up doesn't seem like it's that.\nSpeaker F: You could deal.\nSpeaker F: These aren't large vocabulary things, so the decoder should take a really long time.\nSpeaker E: And I don't think anybody's going to notice the difference between a quarter of a second of latency and 30 milliseconds of latency.\nSpeaker F: No.\nSpeaker F: What does- What was your experience when you were doing stuff with the surgical?\nSpeaker F: Microscopes and so forth.\nSpeaker F: How long was it from when somebody finished an utterance to when something started happening?\nSpeaker E: We had a silence detector, so we would look for the end of an utterance based on the silence detector.\nSpeaker E: And I can't remember now if the top of my head, how many frames of silence we had to detect before we would declare it to be the end of an utterance.\nSpeaker E: But it was, I would say it was probably around the order of 250 milliseconds.\nSpeaker F: Yeah.\nSpeaker F: And that's when you'd start doing things.\nSpeaker E: Yeah, we did the backtrace at that point to get the answer.\nSpeaker E: Of course, I didn't take too long at that point.\nSpeaker E: It was pretty quick.\nSpeaker F: So you had a quarter second delay before plus some little processing time.\nSpeaker F: And then the microscope would start moving or something.\nSpeaker F: Right.\nSpeaker F: And there's physical inertia there, so probably the motion itself.\nSpeaker E: And it felt to the users that it was instantaneous.\nSpeaker E: I mean, as fast as talking to a person, I don't think anybody ever complained about the delay.\nSpeaker F: So you would think as long as it's under half a second or something.\nSpeaker E: Yeah.\nSpeaker E: I don't remember the exact numbers, but something like that.\nSpeaker E: I don't think you can really tell a person, I don't think a person can tell the difference between a quarter of a second and a hundred milliseconds.\nSpeaker E: Yeah.\nSpeaker E: I'm not even sure if you can tell the difference between a quarter of a second and a half a second.\nSpeaker E: Yeah.\nSpeaker E: I mean, it just feels so quick.\nSpeaker F: Yeah, if you said, what's the shortest route to the opera?\nSpeaker F: And it took half a second to get back to you.\nSpeaker F: Yeah.\nSpeaker F: I mean, it might be too abrupt.\nSpeaker F: You might have to put it in.\nSpeaker E: It may feel different than talking to a person because when we talk to each other, we tend to step on each other's utterances.\nSpeaker E: Yeah.\nSpeaker E: So like if I'm asking you a question, you may start answering before I'm even done.\nSpeaker E: Right.\nSpeaker F: So it would probably feel different, but I don't think it would feel slow.\nSpeaker F: All right.\nSpeaker F: Not anyway.\nSpeaker F: I mean, I think we could cut, we know else we could cut down on the neural net time by, by, by, uh, playing around a little bit going more into the past or something like that.\nSpeaker F: We talked about that.\nSpeaker E: How far away do you think the neural net is going to relate and see from the neural net caused by how far ahead you're looking?\nSpeaker F: And there's also, well, there's neural net and there's also this multi-frame, uh, KLT.\nSpeaker E: Was it in the recurrent neural nets where they weren't looking ahead at all?\nSpeaker F: They weren't looking ahead much.\nSpeaker E: They looked ahead a little bit.\nSpeaker F: Yeah, I mean, you could do this with a recurrent net.\nSpeaker F: But you also could just, I mean, we have an experiment with this, but I imagine you could predict a label from more in the past than in the future.\nSpeaker F: We've done some stuff with that before.\nSpeaker F: I think it works okay.\nSpeaker E: We've always had, usually we use the symmetric windows, but yeah, but we played a little bit with asymmetric.\nSpeaker F: You can do it.\nSpeaker F: So, that's what you're busy with.\nSpeaker F: Oh, yeah.\nSpeaker A: Also, we were thinking to apply the spectral saturation from Ericsson and to change the contextual RKLT for LDA.\nSpeaker E: Change the way?\nSpeaker A: The contextual RKLT.\nSpeaker E: I'm missing that last word.\nSpeaker A: KLT.\nSpeaker A: KLT, oh, KLT.\nSpeaker A: I'm just using LDA, this community.\nSpeaker B: Yeah.\nSpeaker A: How do you mean?\nSpeaker E: What is the advantage of that?\nSpeaker B: Well, it's that, well, for a moment we have something that's discriminant and nonlinear.\nSpeaker B: And the other is linear, but it's not discriminant.\nSpeaker B: I thought, well, it's a linear transformation.\nSpeaker F: So, at least just to understand maybe what the difference was between how much you were getting from just putting the frames together and how much you're getting from the discriminative, what the nonlinear already does for you or doesn't do for you.\nSpeaker F: Just to understand it a little better, I guess.\nSpeaker B: What, yeah.\nSpeaker B: Actually, what we want to do, perhaps it's to replace 212, something that's discriminant, but linear.\nSpeaker B: And to see if it's fitting proof of work or the non-discriminating of transformation.\nSpeaker B: And if the new run net is better than this.\nSpeaker F: Yeah, well, that's why I'm honest.\nSpeaker F: To see whether, whether having a neural net really buys you anything.\nSpeaker F: I mean, it did look like it buys you something over just the KLT, but maybe it's just discrimination.\nSpeaker F: And maybe, yeah, maybe the nonlinear discrimination isn't necessary.\nSpeaker F: Good to know.\nSpeaker F: But the other part you're saying was the spectrosotraction.\nSpeaker F: So, just kind of, at what stage do you do that?\nSpeaker F: Do you, doing that?\nSpeaker B: So, it would be on the...\nSpeaker B: An undemanding.\nSpeaker A: Yeah.\nSpeaker A: Okay, so just do that on the phone before.\nSpeaker A: We were thinking to do before after the video.\nSpeaker A: Or we don't know exactly when we spent it before after the video.\nSpeaker F: So, you know that the way that there's one thing that would be no good to find out about from this conference call is that what they were talking about, what they're proposing doing, was having a third party run a good VAD and determine boundaries.\nSpeaker F: Yeah.\nSpeaker F: And then given those boundaries, then everybody do the recognition.\nSpeaker F: The reason for that was that if someone, one group put in the VAD and another didn't, or one had a better VAD than the other, since they're not viewing that as being part of the task and that any manufacturer would put a bunch of effort on having some kind of good speech-science detection still wouldn't be perfect.\nSpeaker F: But I mean, the argument was let's not have that be part of this task.\nSpeaker F: Let's separate that out.\nSpeaker F: And so, I guess they argued about that yesterday.\nSpeaker F: And I'm sorry, I don't know the answer, but we should find out.\nSpeaker F: I'm sure we'll find out soon what they decided.\nSpeaker F: So, yeah, so there's a question of the VAD, but otherwise it's on the Mel filter bank energy, I guess.\nSpeaker F: You're doing the thing.\nSpeaker F: And you're just attracting in the power domain, or magnitude domain, probably power domain.\nSpeaker B: I guess it's power domain.\nSpeaker B: I don't remember exactly.\nSpeaker B: So it's before everything else.\nSpeaker F: If you look at the theory, it should be in the power domain.\nSpeaker F: But I've seen implementations for people to do the magnitude domain.\nSpeaker F: They ask people why and they strike their shoulders and say, oh, it works.\nSpeaker F: And I guess this is mysterious.\nSpeaker F: I mean, people who do this a lot, I guess, have developed little tricks of the trade.\nSpeaker F: I mean, there's this, you don't just subtract the estimate of the noise spectrum, you subtract it at times or less.\nSpeaker B: Good.\nSpeaker B: Yeah.\nSpeaker B: And generally, this, so you have the estimation of the power spectrum of the noise.\nSpeaker B: You multiply this by a factor, which is dependent on the SNR.\nSpeaker B: So when the signal level is more important, compared to this noise level, the coefficient is small around one.\nSpeaker B: But when the signal level is a small compared to the noise level, the coefficient is more important.\nSpeaker B: And this reduce, actually, the musical noise, which is more important during silence portions.\nSpeaker B: The energy is small.\nSpeaker B: So there are tricks like this.\nSpeaker B: Yeah.\nSpeaker E: Is the estimate of the noise spectrum a running estimate?\nSpeaker F: Yeah.\nSpeaker F: Well, that's, I mean, that's what differs from different tasks and different spectral subtraction methods.\nSpeaker F: I mean, if you have a fair assurance that the noise is quite stationary, then the smartest thing to do is use as much data as possible to estimate the noise, get a much better estimate and subtract it off.\nSpeaker F: But if it's varying at all, which is going to be the case from almost any real situation, you have to do it online with some forgetting factors.\nSpeaker E: So do you, is there some long window that extends into the past over which you calculate the average?\nSpeaker F: Well, there's a lot of different ways of computing the noise spectrum.\nSpeaker F: So one of the things that Hans Gunter Hirsch did, and other people, actually, he's, isn't anyone, I guess, was to take some period of speech and in each band to develop a histogram.\nSpeaker F: So to get a decent histogram of these energies takes at least a few seconds, really.\nSpeaker F: But you can do it with a smaller amount, but it's pretty rough.\nSpeaker F: And in fact, I think the NIST standard method of determining signal noise ratio is based on this.\nSpeaker F: So a couple of seconds?\nSpeaker F: No, it's based on this kind of method, this histogram method.\nSpeaker F: So you have a histogram.\nSpeaker F: Now, if you have signal and you have noise, you basically have these two bumps in histogram.\nSpeaker F: She could approximate its two Gaussian's.\nSpeaker E: But don't they overlap sometimes?\nSpeaker F: Oh, yeah.\nSpeaker F: So you have a mixture of two Gaussian's.\nSpeaker F: Yeah.\nSpeaker F: Right, you can use EM to figure out what it is.\nSpeaker F: So basically now you have this mixture of two Gaussian's.\nSpeaker F: You know what they are, and I mean, sorry, you estimate what they are.\nSpeaker F: And so this gives you what the signal is and what the noise energy is in that band and the spectrum.\nSpeaker F: And then you look over the whole thing and then you have a noise spectrum.\nSpeaker F: So Hans-Gunther Hirsch and others have used that kind of method.\nSpeaker F: And the other thing to do is which is sort of more trivial and obvious is to determine through magical means that there's no speech in some period and see what the spectrum is.\nSpeaker F: And you know, that's strictly due, it has mistakes.\nSpeaker F: And if you've got enough time, this other method appears to be somewhat more reliable.\nSpeaker F: A variant on that for just determining signals and noise ratio is to just, you can do an iterative thing, yeah, I'm like thing to determine means only, I guess it is EM still, but just determine the means only.\nSpeaker F: Don't worry about the variances.\nSpeaker E: And then just use those mean values as being the noise ratio in that band.\nSpeaker E: Depending on where the window was that you used to calculate the signal and noise ratio.\nSpeaker F: Yeah, sure.\nSpeaker F: But, that's certainly because if you start looking to the future, right?\nSpeaker E: Okay, well that was my question.\nSpeaker F: Yeah, I mean, if you just, if you, at the beginning you have some gas.\nSpeaker F: Yes, some gas.\nSpeaker F: Yeah, interesting question. I wonder how they did.\nSpeaker B: Actually, it's if you want to have good estimation on non-stationary noise, you have to look in the future.\nSpeaker B: I mean, if you take your window and build your Instagram on this window, what you can expect is to have an estimation of the noise in the middle of the window.\nSpeaker B: Not at the end, so.\nSpeaker B: Well, yeah, but what does, what does our hotel do?\nSpeaker B: They just look in the past, I guess it works because the noise are almost stationary.\nSpeaker F: Well, the thing, I mean, you talk about non-stationary noise, but I think that sexual subtraction is rarely, is not going to work really well for non-stationary noise.\nSpeaker B: Well, if you have a good estimation of the noise, yeah, because, what do you, let's talk about it.\nSpeaker B: That's hard to do.\nSpeaker F: Yeah, that's hard to do.\nSpeaker F: So, I think that what is, what's more common is that you're going to be helped with slowly varying or stationary noise.\nSpeaker F: That's what spectral subtraction will help with, practically speaking.\nSpeaker F: If it varies a lot, to get a good estimate, you need a few seconds of speech, even if it's centered.\nSpeaker F: If you need a few seconds to get a decent estimate, but it's changed a lot in a few seconds, then, you know, it's kind of a problem.\nSpeaker F: I mean, imagine five hertz is the middle of the speech modulation spectrum, right?\nSpeaker F: So imagine a jackhammer growing in five hertz.\nSpeaker F: I mean, good luck, so.\nSpeaker B: So in this case, yeah, sure, you can.\nSpeaker B: But I think, uh, earth does experiment with windows of, like, between 500 milliseconds in one second.\nSpeaker B: And, well, 500 was not so bad.\nSpeaker B: And he worked on non-stationary noise.\nSpeaker B: He's like, noise modulated with, well, with, with, with, and between modulation, so things like that.\nSpeaker E: So, what is, uh, windows centered around that?\nSpeaker B: Yeah, well, I think, yeah, when the paper we show that, actually, the estimation of the noise is delayed while it's, there is, you have to center the window.\nSpeaker F: Yeah.\nSpeaker F: No, I understand it's better to do, but I just think that, uh, for real noises, what, what's most likely to happen is that there'll be some things that are relatively stationary where you can use one or another spectral subtraction thing.\nSpeaker F: And, uh, there's a lot of things where it's not so stationary.\nSpeaker F: I mean, you can always pick something that, that falls between your methods.\nSpeaker F: But I don't know if, you know, sinusoidally, amplitude modulated noise is sort of a big problem in practice, I think.\nSpeaker E: Yeah.\nSpeaker E: We could probably get a really good estimate of the noise if we just went to the noise files and built the averages from them.\nSpeaker B: What do you, he's actually, but if the noise is stationary, perhaps we don't even need some kind of noise estimation algorithm, just take the beginning of the utterance.\nSpeaker B: Oh, yeah, sure.\nSpeaker B: I know, I don't know if people try this for, well, everybody seems to use some kind of adaptive, but, but, but, but, it's game, but, you know, stationary useful.\nSpeaker F: Very useful. Very slow adaptation.\nSpeaker F: Right. The word stationary is a very precise statistical meaning, but, you know, in, in signal processing, really, what we're talking about, I think, is things that change slowly compared with our processing techniques.\nSpeaker F: So, if you're driving along in a car, I, I would think that most of the time the nature of the noise is going to change relatively slowly.\nSpeaker F: It's not going to stay absolutely the same if you, if you check it out five minutes later, you may be in a different part of the road or whatever, but it's, it's using the local characteristics.\nSpeaker F: Time is probably going to work pretty well, but you could get hurt a lot if you just took something from the beginning of all the speech of, you know, an hour speech and that lighter.\nSpeaker F: So, they may be, you know, maybe overly complicated for, for this test, but, but, but, I don't know.\nSpeaker F: But what you're saying, you know, makes sense though. I mean, if possible, you should, you should make it center of the center of the window, but we're already having problems with these delay.\nSpeaker F: Yeah, it's a bit of a waste without it.\nSpeaker E: If they're going to provide a voice activity detector that will tell you the boundaries of the speech, then couldn't you just go outside those boundaries and do your estimate there?\nSpeaker E: Oh yeah.\nSpeaker F: Yeah, so I imagine that's what they're doing, right?\nSpeaker F: They're probably looking in non-speech sections and getting some, yeah, they have some kind of threshold on the previous estimate.\nSpeaker B: So, yeah, I think, I think, Ericsson used this kind of threshold.\nSpeaker B: Yeah, so they have an estimate of the noise level and they put a threshold like 6 or 10 dB above.\nSpeaker B: What's under this threshold is used to update the estimate.\nSpeaker B: Isn't that right?\nSpeaker A: I think so.\nSpeaker A: So, it's, yeah.\nSpeaker B: It's like saying what's under the threshold is silence.\nSpeaker F: Does trans talk harm do the same thing?\nSpeaker B: I, you know, perhaps?\nSpeaker A: No, I have no idea what that is.\nSpeaker F: Okay, we're done with that.\nSpeaker F: Let's see.\nSpeaker F: Maybe we talked about a couple of other things, briefly, just things that we've been chatting about.\nSpeaker F: And these meetings yet, so you're coming up with your Quiles proposal.\nSpeaker F: And when I just give a two, three minute summary of what you're about to be doing.\nSpeaker C: Two, three, you can be sure of that.\nSpeaker C: Well, I've talked to somebody already, but I'm looking into extending the work done by Larry Saw and John Allen and Mazinari.\nSpeaker C: They have a system that's a multi-band system, but the multi-band is a little different than the way that we've been doing multi-band in the past, where we've been taking subband features and training up these new formats and on phonetic targets and then combining them somehow down the line.\nSpeaker C: They're taking subband features and training up a detector that detects for these phonetic features.\nSpeaker C: For example, it presents a detector to detect sonorous.\nSpeaker C: And so what it basically is, there's at the lowest level, it's an AND gate.\nSpeaker C: So on each subband, you have several independent tests to test whether there's the existence of sonorous in a subband.\nSpeaker C: And then it's combined by a soft AND gate.\nSpeaker C: And then at the higher level, for every, if the higher level, there's a soft or gate.\nSpeaker C: If this detector detects the presence of sonorous in any of the subband, then the or gate at the top says, okay, well, this frame has evidence of sonorous.\nSpeaker E: What are some of the low-level detectors that they use?\nSpeaker C: Well, the low-level detectors are logistic regressions.\nSpeaker F: And the one other way, basically, is one of the units in our neural networks.\nSpeaker F: That's all it is.\nSpeaker F: It's a sigmite with weighted sum of TMP.\nSpeaker F: Is it trained by gradient?\nSpeaker C: Yes.\nSpeaker C: So he uses an EM algorithm to train up these parameters for logistic regression?\nSpeaker F: Well, actually, so he's using EM to get the targets.\nSpeaker F: So you have this AND gate, calling an AND gate, but it's a product, product rule thing at the output.\nSpeaker F: And then he uses, and then feeding into that, or, I'm sorry, it's nor at the output, isn't it?\nSpeaker F: Yeah, that's the product.\nSpeaker F: And then he has each of these AND things.\nSpeaker F: So they're little neural units, and they have to have targets.\nSpeaker F: So the targets come from the EM.\nSpeaker E: And so are each of these low-level detectors?\nSpeaker E: Are these something that you decide ahead of time, like I'm going to look for this particular feature, and look at this frequency, or what are they looking at?\nSpeaker E: What are their input?\nSpeaker C: Right, so the, okay, so, for each subband, there are basically several measures of SNR and correlation.\nSpeaker C: Oh, okay.\nSpeaker C: And he said there's like 20 of these per subband.\nSpeaker C: And for every subband, you just take ahead of time, I'm going to have like five independent logistic tests.\nSpeaker C: And you initialize these parameters in some way, and use the EM to come up with your training targets for the low-level detectors.\nSpeaker C: And then once you get that done, you train the whole thing on maximum likelihood.\nSpeaker C: And he showed that using this method to detect sonarances is very robust compared to two typical full band Gaussian mixture estimations of sonnets.\nSpeaker C: And so that's just one detector.\nSpeaker C: And you can imagine building many of these detectors on different features, you get enough of these detectors together, then you have enough information to do higher-level discrimination, example discriminating between phones, and then you keep working your way up and telling you to build a full recognizer.\nSpeaker C: So that's the direction I'm thinking about going.\nSpeaker C: Cool.\nSpeaker F: It has a number of properties that I really like.\nSpeaker F: I mean, one is going towards using narrow band information for phonetic features of some sort rather immediately going for the typical sound units.\nSpeaker F: Another thing I like about it is that this thing is going to be explicitly trained for a product of errors rule, which is what Alan keep pointing out that Fletcher is observed in the 20s for people listening to narrow band stuff.\nSpeaker F: That's Friday's time.\nSpeaker F: And the third thing I like about it is, and we've played around with this in a different kind of way a little bit, but it hasn't been a dominant way of evaporating, I think.\nSpeaker F: This issue of where the targets come from.\nSpeaker F: I mean, in our case, when we've been training at multi-band things, the way we get the targets for the individual bands is that we get the phonetic label of the sound.\nSpeaker F: But this is saying, that's maybe what our ultimate goal is, or not ultimate, but an ultimate goal is getting these small sound units. But along the way, what should we be training these intermediate things for?\nSpeaker F: I mean, because we don't know that this is a particularly good feature. There's no way someone in the audience yesterday was asking, well, couldn't you have people go through a market for the individual bands and say where they're in the first place on earth?\nSpeaker F: But I think having a bunch of people listening to critical band-wide, I think it would be possible. It's all going to sound like sign waves to you more or less. Well, narrow band.\nSpeaker F: I think it's very hard for someone to make that determination. So we don't really know how those should be labeled. It could be that you should not be paying that much attention to certain bands for certain sounds, in order to get the results.\nSpeaker F: So what we have been doing there to sort of mixing it all together is certainly much, much cruder than that. We train these things up on the final label. Now we have, I guess, done experiments. You've probably done stuff where you have done separate the therapies on the different force alignment on the subband labels.\nSpeaker F: Yeah, you've done that. Did that help at all? It helps for one iteration, but anything after that, it doesn't help. So that may or may not be helpful because in the sense that's the same sort of thing. You're taking global information and determining how you should.\nSpeaker F: But this is, I think, a little more direct. How do they measure the performance of their detector? Well, he's just actually looking at the confusions between sound and non-sound.\nSpeaker F: So he hasn't applied it to recognition, or if he did, he didn't talk about it. And one of the concerns in the audience actually was that he did a comparison to an old foil that asked the old standard recognizing it with, well, no filter bank at the front and HM admins and so forth.\nSpeaker F: And it didn't do nearly as well, especially in noise. But the good questions in the audience was, well, yeah, but that was great for that. I mean, the use of a very smooth, a spectral envelope is something that has evolved generally a good thing for speech recognition.\nSpeaker F: But if you knew that what you were going to do is detect sounder, it's not. So sounder, it's almost like voice down voice to accept, I guess, that the voice stops are also called up joints.\nSpeaker F: So it's, it's, but with the exception of the stops, I guess it's pretty much the same as voice down voice, right? So, so if you knew you were doing that, if you were doing something safe for a vocoder, you wouldn't use the same kind of features. You would use something that was sensitive to the periodicity and not just the envelope. And so in that sense, it was an unfair test.\nSpeaker F: So I think the question was right. It was, in that sense, an unfair test. Nonetheless, it was one that was interesting because this is what we are actually using for speech recognition, these sort of envelopes.\nSpeaker F: This says that perhaps even trying to use them the best way we can that we ordinarily do with Gaussian mixers and, did you mention so forth, you don't actually do that well on determining whether something is sounder or not, which means you're going to make errors between similar sounds that are sounder or instrument.\nSpeaker E: Didn't they also do some kind of an oracle experiment where they said if we could detect sounder and perfectly and then show how it would improve speech recognition, I thought I remember hearing about an experiment like that.\nSpeaker F: These same people? I don't remember that. That's, you're right. That's exactly the question to follow up this discussion. I suppose you did that. Got that right.\nSpeaker B: Yeah. What could be the other low level detectors and for other kind of features in addition to detecting sounder and sounder?\nSpeaker C: That's what you want to go for. So, what? Oh, build other other, other, other different different features.\nSpeaker C: Let's see. Yeah, I don't know. Easiest thing would be to go do some voicing stuff, but that's very similar to sounder.\nSpeaker E: When we talked with John O'Halla the other day, we made a list of some of the things like forcation, abrupt closure, our coloring, naizality, voicing.\nSpeaker F: Yeah, so is it half doesn't like that better? Now this was coming at it from a different angle, but maybe it's a good way to start with these things, which John felt that a human annotator would be able to reliably mark.\nSpeaker F: So the sort of things he felt would be difficult for a human annotator to reliably mark would be tongue position.\nSpeaker E: Please, thanks. Yeah. Yeah. There's also things like stress.\nSpeaker F: But stress doesn't fit in this thing of coming up with features that will distinguish words from one another.\nSpeaker F: It's a good thing to mark and will probably help us ultimately.\nSpeaker E: Yeah, there's a few cases where it can like permit, permit, but that's not very common in English. In other languages, it's more important.\nSpeaker F: Well, yeah, but either case you'd write P, R, I, T, you get the word right.\nSpeaker E: No, I'm saying, I thought you were saying that stress doesn't help you distinguish between words.\nSpeaker E: Oh, I see what you're saying. As long as you get the secret. We're talking about transcriptionists.\nSpeaker E: Yeah, yeah, yeah. Yeah. Yeah. Right. So where it could help is maybe at a higher level. Right. Yeah. Understandings. Yeah. Yeah. Yeah.\nSpeaker F: Exactly. But that's this afternoon's meeting. Yeah. That's what I understand. That's what I'm saying.\nSpeaker F: Yeah. So that's, yeah, that's, you know, the neat thing.\nSpeaker C: So, oh, how that's going to help you with these transcription of the leading data?\nSpeaker E: Well, I don't know. We, we sort of didn't get that far. We just talked about some possible features that could be marked by humans.\nSpeaker E: And because of having maybe some extra transcribe or time, we thought we could go through and mark some portion of the data for that.\nSpeaker F: Yeah. That's not an immediate problem. We have a lot of extra transcribe.\nSpeaker F: Right. But I'm long term, I guess, Chuck is going to continue to dialogue with John.\nSpeaker E: Well, we'll end up doing some. I'm definitely interested in this area too.\nSpeaker E: A acoustic feature. Okay. Stuff.\nSpeaker F: Yeah. I think it's an interesting, interesting way to go.\nSpeaker F: I said, it's an amazing thing. There's never good things.\nSpeaker F: So, we're going to talk maybe two, three minutes about what we've been talking about, right?\nSpeaker D: Yeah. Okay. So, we're interested in methods for far-makes, speech recognition, mainly methods to deal with the reverberation in the far-makes signals.\nSpeaker D: So, one approach would be, say, MSG and PLP, like was used in Aurora 1, and there are other approaches which actually attempt to remove the reverberation instead of being robust to it like MSG.\nSpeaker D: And so, we're interested in comparing the performance of a robust approach like MSG with these speech enhancement or do reverberation approaches.\nSpeaker D: And it looks like we're going to use the meeting recorded digits that after that.\nSpeaker B: And the dear reverberation algorithm. Can you give some more details on this? Use one microphone.\nSpeaker D: Several microphones. Okay. Well, there was something that was done by a guy named Carla, so I forget his last name. He worked with Hinat, who Evan Donnell. It was like Rasta in the sense of it was deconvolution by filtering, except he used a longer time window like a second maybe. And the reason for that is Rasta's time window is too short to include the whole reverberation.\nSpeaker D: I don't know what you call the reverberation response. If you see what I mean. The reverberation filter for my mouth to that mic is like, it's too long in the time domain for the Rasta filtering to take care of it.\nSpeaker D: And then there are a couple of other speech enhancement approaches which haven't been tried for speech recognition yet, but have just been tried for enhancement, which have the assumption that you can do LPC analysis of the signal you get at the far microphone.\nSpeaker D: And the all-pull filter that you get out of that should be good. It's just the excitation signal that is going to be distorted by the reverberation.\nSpeaker D: And so you can try and reconstruct a better excitation signal and feed that through the all-pull filter and get enhanced speech with reverberation produced.\nSpeaker F: There's also this echo cancellation stuff that's sort of chasing. So when we're seeing these digits now, we do have a close microphone signal. And then there's the distant microphone signal.\nSpeaker F: And you could, as a kind of baseline, say, okay, given that we have both of these, we should be able to do a cancellation so that we essentially identify the system in between the linear time and the varying system between the two microphones and re-cancel it out to some reasonable approximation.\nSpeaker F: That's not a practical thing. If you don't have a close mic, we thought that by making a good baseline. It still won't be perfect because there's no noise.\nSpeaker F: But there are single microphone methods that I think people have done for this kind of due reverberation. Do you know any references to any? Because I was, I literally thought that I guess when people are working with single microphones, they are more trying to do, well, not very,\nSpeaker B: well, there is the event and work, but also trying to find the deconvolution filter, but not in the time domain, but in the stream of features. But there's someone working on this in mons, so perhaps, yeah, we should try to, it's working on this on trying to, on reverberation.\nSpeaker F: The first paper on this is going to have great references. I can tell you what I mean. I always go to have references, especially when reviewers read it or one of the authors, and you're okay, you cited me.\nSpeaker B: Well, it did deconvolution, and it did some fancier things like training different network and different reverberation conditions, and then trying to find the best one.\nSpeaker F: The other thing that Dave was talking about earlier was multiple mic things, where they're all distant. So, I mean, there's all this work in the race, but the other thing is, what can we do that's clever, that can take some advantage of only two mics, particularly if there's an obstruction between them.\nSpeaker F: It creates a shadow, which is helpful, it's part of why you have such good direction. We've two ears, even though they're not several feet apart, for most people, since.\nSpeaker F: That could help though. That's what it's for. It's basically heads for, separate ears.\nSpeaker E: Okay, I think that's all we have this week, and I think it's digit time. Actually, for some reason, the digit forms are blank. I think that maybe due to the fact that Adam ran out of digits, and didn't have time to regenerate it.\nSpeaker E: Well, this is no real reason to write our names on here. If you want to put your credit card numbers, or do we need the names for the other stuff? Yeah, I do need your names and the time and all that, because we put that into the key files.\nSpeaker F: Okay, that's why we have the forms. Okay, yeah, I didn't notice this. I was sitting here. I was in a box reading.\nSpeaker F: So, I guess we're done. I'll do my credit card later.\n"}, {"meeting_id": "Bro004", "summary": "The participants discussed results from the experiments that had been conducted. Switching between tasks in the same language had smaller errors than multilingual models. The professor thought that increasing the parameters of the net for larger multi-lingual models would be helpful. The team decided that they should experiment further with different linguistic features. They also discussed how they could speed up their work by relying on greater computational resources.", "dialogue": "Speaker G: I don't know.\nSpeaker G: I can't do the crash.\nSpeaker G: I think I'm going to say.\nSpeaker G: Wow.\nSpeaker B: I'm going to say that.\nSpeaker B: Hello.\nSpeaker B: Hello.\nSpeaker B: Hi.\nSpeaker B: Well, maybe it's a turning off.\nSpeaker D: It's a turning on.\nSpeaker D: I'm not turning.\nSpeaker D: I'm going to turn it over.\nSpeaker D: So I can't figure out how to tell.\nSpeaker D: If it's the car or the sentence in illegal,\nNone: I don't know how to tell it. I think it's you.\nNone: Yeah, I'll have a good mind.\nSpeaker B: Oh.\nNone: Okay.\nNone: Okay.\nSpeaker E: Okay.\nSpeaker E: Okay.\nSpeaker E: So, I guess we are going to do the digits at the end.\nSpeaker C: General tree.\nSpeaker C: Okay.\nSpeaker A: 75.\nSpeaker A: Yes, sir.\nSpeaker E: Yeah, that's the mic over there.\nNone: It's a written.\nSpeaker E: Mike, the guy who's here.\nSpeaker E: The channel.\nSpeaker B: The channel 4.\nSpeaker A: Yes.\nSpeaker A: Oh.\nSpeaker E: Yes.\nSpeaker E: And I'm channel 2.\nSpeaker E: I think we're channel 1.\nSpeaker E: I think I'm channel 1.\nSpeaker E: Oh, I'm channel 1.\nSpeaker E: Channel 1?\nSpeaker E: Yes.\nNone: Okay.\nSpeaker E: Okay.\nSpeaker E: So, I also copied the results that we all got in the mail.\nSpeaker E: I think from OGI.\nSpeaker E: Go through them also.\nSpeaker E: So, where are we on our runs?\nSpeaker C: So, we, so as I was already said, we mainly focused on four kind of features.\nSpeaker C: The PLP, the PLP with Jerasda, the MSG and the MFCC from the business.\nSpeaker C: Over.\nSpeaker C: And we focused for the test part on the English and the Italian.\nSpeaker C: We've trained several neural networks on the DIA digits English.\nSpeaker C: And on the Italian data and also on the broad English French and Spanish databases.\nSpeaker C: So, there is a result tables here for the tandem approach.\nSpeaker C: And actually what we've observed is that if the network is trained on the task data, it works pretty well.\nSpeaker E: Okay.\nSpeaker E: I'm sorry.\nSpeaker E: There's a pausing for a photo from the front of the room.\nSpeaker E: Yeah.\nSpeaker E: It's longer.\nSpeaker E: We're pausing for a photo opportunity here.\nSpeaker D: So, wait, wait, wait, wait.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker E: Let me give you a black screen.\nSpeaker E: Facing this one.\nSpeaker E: Okay, this could be a good section for our silencing section.\nSpeaker E: Musical chairs, everybody.\nSpeaker E: So, you were saying about the training data.\nSpeaker C: Yeah.\nSpeaker C: So, if the network is trained on the task data, tandem works pretty well.\nSpeaker C: And actually we have results are similar.\nSpeaker G: Do you mean if it's trained only on the data from just that task, that language?\nSpeaker C: Yes, but actually we didn't train network on both types of data.\nSpeaker C: I mean phonetically balanced data and task data.\nSpeaker C: We only did either task data or wrote data.\nSpeaker C: Yeah.\nSpeaker E: So, how, I mean, clearly it's going to be good then, but the question is how much worse is it if you have broad data?\nSpeaker E: I mean, from what I saw from the early results, I guess last week, was that if you trained on one language and tested on another, say that the results were relatively poor.\nSpeaker E: But the question is if you train on one language, but you have a broad coverage and then test on another, because that improves things in comparison.\nSpeaker C: So, you use the same language, you mean?\nSpeaker E: No, no, definitely.\nSpeaker E: So, if you train on TI digits and test on Italian digits, you do poorly.\nSpeaker E: Say, I don't know if the numbers in front of me, so I'm just imagining.\nSpeaker E: Yeah, but I did not do that.\nSpeaker E: So, you did not?\nSpeaker E: We train on 10 minutes and test on Italian digits.\nSpeaker C: No, we did, for kind of testing, actually, the first testing is with task data.\nSpeaker C: So, with net strain on task data, so for Italian, on the Italian speech data curve, the second test is trained on a single language, but the broad database, but the same language as the task data.\nSpeaker C: But for Italian, we choose Spanish, which we assume is close to Italian.\nSpeaker C: The third test is by using the tree language database.\nSpeaker C: And the fourth is...\nSpeaker E: In S3 languages, that's including the...\nSpeaker E: This includes the one that it's...\nSpeaker E: Yeah.\nSpeaker G: But not digits, I mean.\nSpeaker G: The three languages is not digits, it's the broad data.\nSpeaker G: Yeah, data, okay.\nSpeaker C: And the fourth test is excluding from these three languages, the language that is the task language.\nSpeaker E: Oh, okay, yeah, so that's what I wanted to know.\nSpeaker E: And just wasn't saying it very well.\nSpeaker C: Yeah.\nSpeaker C: So, for the digits, for example, when we go from the digits training to the limit training, we lose around 10%.\nSpeaker C: The error rate increases...\nSpeaker C: Right.\nSpeaker C:...of 10% relative.\nSpeaker C: So this is not so bad.\nSpeaker C: And then when we jump to the multilingual data, it's...\nSpeaker C: It becomes worse and...\nSpeaker E: Well, how much?\nSpeaker C: Around, let's say, 20% further.\nSpeaker C: So...\nSpeaker C: Yeah.\nSpeaker C: 12% further.\nSpeaker C: So 30% further, yeah.\nSpeaker G: And so remind me that multilingual stuff is just the broad data, right?\nSpeaker G: Yeah, it's not the digits.\nSpeaker G: So it's the combination of two things there.\nSpeaker G: It's removing the task-specific training and it's adding other languages.\nSpeaker C: Yeah, okay.\nSpeaker C: But the first step is already a regular task is specific.\nSpeaker C: So the building.\nSpeaker C: Yeah, okay.\nSpeaker C: So basically, when it's trained on the multilingual broad data or number, so the ratio of error rates with baseline error rate is around 1.1.\nSpeaker E: Yeah.\nSpeaker E: And it's something like 1.3 of the...\nSpeaker E: If you compare everything to the first case, it's the baseline.\nSpeaker E: You get something like 1.1 for the using the same language with a different task.\nSpeaker E: Something like 1.3 for three languages, a lot of stuff.\nSpeaker C: Same language we are for English at 0.8.\nSpeaker C: So it improves compared to the baseline.\nSpeaker C: But...\nSpeaker C: That's good.\nSpeaker E: I meant something different by baseline.\nSpeaker E: So let me...\nSpeaker E: Okay, fine.\nSpeaker E: Let's use the conventional meaning of baseline.\nSpeaker E: By baseline here, I meant using the task-specific data.\nSpeaker E: Okay.\nSpeaker E: But because that's what you were just doing with this 10%.\nSpeaker E: Yes.\nSpeaker E: I just trying to understand that.\nSpeaker E: So if we call a factor just 1, just normalize to 1, the word error rate that you have for using TI digits as training and TI digits as test.\nSpeaker E: Different words, I'm sure, but the same task and so on.\nSpeaker E: If you call that one, then what you're saying is that the word error rate for the same language but using different training data and you're testing on TI digit and so forth, it's 1.1.\nSpeaker E: Yeah, it's around 1.1.\nSpeaker E: Right. And if you do go to three languages including the English, it's something like 1.3.\nSpeaker E: That's what you were just saying, I think.\nSpeaker C: More, actually.\nSpeaker C: 1.4?\nSpeaker C: Yeah.\nSpeaker C: So it's an additional 30%.\nSpeaker C: What would you say?\nSpeaker C: Around 1.4.\nSpeaker E: Okay. And if you exclude English from this combination of that.\nSpeaker C: If we exclude English, there is not much difference with the data with English.\nSpeaker E: So, that's interesting.\nSpeaker E: You see because, so that's important. So what it's saying here is just that yes, there is a reduction in performance when you don't have task data.\nSpeaker E: Wait a minute.\nSpeaker E: Wait a minute.\nSpeaker E: Actually, it's interesting. So when you go to a different task, there's actually not so different.\nSpeaker E: So what's the difference between 2 and 3? Between the 1.1 case and the 1.4 case, I'm confused.\nSpeaker D: It's multilingual.\nSpeaker C: Yeah, the only difference is that it's multilingual.\nSpeaker E: Because in both of those cases, you don't have the same task.\nSpeaker E: So is the training data for this 1.4 case? Does it include the training data for the 1.1 case?\nSpeaker C: Yeah.\nSpeaker C: Yeah, I'm proud to fit you.\nSpeaker C: How much bigger is it?\nSpeaker C: It's 2 times, actually.\nSpeaker C: The multilingual databases are 2 times the broad English data.\nSpeaker C: We just wanted to keep it.\nSpeaker E: So it's 2 times.\nSpeaker E: So it includes the broad English data.\nSpeaker E: So that's timet, basically.\nSpeaker E: So it's band limited timet.\nSpeaker E: This is all Angular sampling.\nSpeaker E: So your band limited timet gave you almost as good a result as using TI digits on a TI digits test.\nSpeaker E: But when you add in more training data, it keeps the neural net the same size.\nSpeaker E: It performs worse on the TI digits.\nSpeaker E: Now all of this is noisy TI digits, I assume?\nSpeaker E: Yeah, both training and test.\nSpeaker E: Yeah, okay.\nSpeaker E: Well, we made this need to...\nSpeaker E: So it's interesting that going to a different task didn't seem to hurt us that much.\nSpeaker E: Going to a different language...\nSpeaker E: It doesn't seem to matter.\nSpeaker E: The difference between 3 and 4 is not particularly great.\nSpeaker E: So that means that whether you have the language in or not is not such a big deal.\nSpeaker E: It sounds like we may need to have more of things that are similar to a target language.\nSpeaker E: I mean, you have the same number of parameters in the neural net.\nSpeaker E: You haven't increased the size of the neural net.\nSpeaker E: And maybe there's just not enough complexity to it to represent the very increased variability in the training set.\nSpeaker E: That could be.\nSpeaker E: So what about... So these are results that you're describing now that are pretty similar for the different features?\nSpeaker C: Let me check.\nSpeaker C: So this was for the PLP.\nSpeaker C: For the PLP with Geras, that we...\nSpeaker C: This is quite the same tendency with the slight increase of the error rate.\nSpeaker C: If we go to the team it, and then it gets worse with the mid-deling.\nSpeaker C: Yeah, there is a difference actually between PLP and Geras, that Geras seems to perform better with the I-limit-matched condition, but slightly worse for the well-matched condition.\nSpeaker E: I have a suggestion actually, you know, to lay us slightly.\nSpeaker E: Would you mind running in the other room and making copies of this?\nSpeaker E: Because we're all sort of... If we could look at it while we're talking, I think, I'll sing a song or dance or something like that.\nSpeaker E: So go ahead.\nSpeaker G: What you're going to ask someone my question.\nSpeaker E: This way and to slide it to the left, yeah.\nSpeaker G: What was this number 40?\nSpeaker G: It was roughly the same as this one, he said.\nSpeaker G: You had the two language versus the three language.\nSpeaker G: That's what he was saying.\nSpeaker E: Or he removed English.\nSpeaker E: Sometimes actually depends on what features you're using.\nSpeaker E: Yeah.\nSpeaker E: But it sounds like... I mean, that's interesting because it seems like what it's saying is not so much that you got hurt because you didn't have so much representation of English because in the other case, you don't get hurt anymore, at least when.\nSpeaker E: It seems like it might simply be a case that you have something that is just much more diverse but you have the same number of parameters representing it.\nSpeaker G: I wonder were all three of these nets using the same output, this multi-language...\nSpeaker G: labeling.\nSpeaker E: It's using 64 phonemes from sample.\nNone: Okay.\nSpeaker G: So this would, from this you would say, well, it doesn't really matter if we could finish into the training of the neural net if there's going to be, you know, finish in the test data, right?\nSpeaker E: Well, it sounds...\nSpeaker E: We have to be careful because we haven't done a good result yet, comparing different bad results.\nSpeaker E: I think it does suggest that it's not so much cross language as cross type of speech.\nSpeaker E: It's...\nSpeaker E: But we did, oh yeah, the other thing I was asking though is that I think that in the case...\nSpeaker E: Yeah, you do have to be careful because of compounded results.\nSpeaker E: I think they got some earlier results in which you trained on one language and tested on another and you didn't have three, you just had one language, so you trained on one type of digits and tested on another.\nSpeaker E: Wasn't there something of that where you say trained on Spanish and tested on TI digits or the other way around?\nSpeaker E: No.\nSpeaker E: Not there was something like that that he showed me last week.\nSpeaker E: The way to where you could...\nSpeaker E: Yeah, that would be interesting.\nSpeaker E: This may have been what I was asking before stuff, but...\nSpeaker E: Wasn't there something that you did where you trained on one language and tested on another?\nSpeaker E: No mixture, but just...\nSpeaker C: Hello.\nSpeaker E: We've never just trained on one language.\nSpeaker C: Training on the single language and testing on one language.\nSpeaker C: Yeah.\nSpeaker C: Right, so the only test that's similar to this is training on two languages.\nSpeaker E: But we've done a bunch of things where we just trained on one language, right?\nSpeaker E: I mean, you haven't done all your tests on multiple languages.\nSpeaker C: No, either this is test with the same language, but from the broad data or its test with different languages.\nSpeaker C: The list of different languages.\nSpeaker G: Did you do different languages from digits?\nSpeaker C: No, you mean training digits on one language and using the net to recognize digits on another language?\nSpeaker G: No.\nSpeaker E: See, I thought you showed me something like that last week.\nSpeaker E: You had a little...\nSpeaker C: No, I didn't think so.\nNone: What?\nSpeaker D: He's almost sorry.\nSpeaker E: So, I mean, what's this table that we're looking at is...\nSpeaker E: Is all testing for TI digits?\nSpeaker C: So, you have basically two parts, the upper part is for TI digits.\nSpeaker C: And it's divided into three rows of four rows each.\nSpeaker C: Yeah.\nSpeaker C: The first four rows is well matched, then the second group of four rows is mismatched.\nSpeaker C: Finally, I'm mismatched.\nSpeaker C: And then the lower part is for Italian, and it's the same thing.\nSpeaker G: So, the upper part is training TI digits?\nSpeaker C: It's the HTK results.\nSpeaker C: I mean, so it's HTK training testings with different kinds of features and what appears in the left column is the networks that are used for doing this.\nSpeaker C: So...\nSpeaker E: What was it that you had done last week when you showed your number?\nSpeaker E: Why?\nSpeaker E: When you showed me the table last week?\nSpeaker C: It was part of these results.\nSpeaker G: So, where is the baseline for the TI digits located in here?\nSpeaker C: You mean the HTK or a baseline?\nSpeaker C: Yeah.\nSpeaker C: It's the 100 number.\nSpeaker C: All these numbers are in the ratio with respect to the baseline.\nSpeaker E: So, this is where it aerates, so a high number is bad.\nSpeaker C: Yeah, this is a world aerarade ratio.\nSpeaker G: Okay.\nSpeaker C: So, 70.2 means that we reduce the aerarade to 30%.\nSpeaker C: Okay.\nSpeaker E: Okay.\nSpeaker E: So, if we take...\nSpeaker E: Let's see, POP with online normalization and delta-delts.\nSpeaker E: So, that's the thing you have circled here in the second column.\nSpeaker E: And multi-english refers to what?\nSpeaker C: To demit.\nSpeaker C: Then you have MF, MF and ME, which are from French, Spanish and English.\nSpeaker C: Actually, I forgot to say that the multilingual net are trained on features without the derivatives.\nSpeaker C: But with increased frame numbers.\nSpeaker C: And we can see on the first line of the table that it's less worse when we don't use delta, but it's not that much.\nSpeaker E: So, I'm sorry, Mr. Watts-MF, does an ME?\nSpeaker C: Multi-french with this Spanish.\nSpeaker E: So, it's a broader vocabulary.\nSpeaker E: Okay.\nSpeaker E: So, I think what I saw in the smaller chart that I was thinking of was...\nSpeaker E: There were some numbers I saw, I think, that included these multiple languages.\nSpeaker E: And I was seeing that it got worse.\nSpeaker E: I think that was almost.\nSpeaker E: You had some very limited results at that point, which showed having in these other languages.\nSpeaker E: In fact, it might have been just this last category having two languages broad that were where English was removed.\nSpeaker E: So, that was cross-language, and the result was quite poor.\nSpeaker E: What I hadn't seen yet was that if you had it in the English, it's still poor.\nSpeaker E: Yeah.\nSpeaker E: Now, what's the noise condition of the training data?\nSpeaker E: Well, I think this is what you were explaining. Noise condition is the same. It's the same Aurora noises in all these cases.\nSpeaker E: You have the training.\nSpeaker E: So, there's not a statistically strong, statistically different noise characteristic between...\nSpeaker E: No, these are the training attest.\nSpeaker E: And yet, we're seeing some kind of data.\nSpeaker C: At least for the first...\nSpeaker C: Well matched.\nSpeaker E: So, there's some kind of an effect from having this broader coverage.\nSpeaker E: Now, I guess what we should do by doing with this is try testing these on this same sort of thing.\nSpeaker E: You probably must have this lined up to do to try the same...\nSpeaker E: With the exact same training, do testing on the other languages.\nSpeaker E: Oh well, you have it here for the Italian.\nSpeaker E: That's right.\nSpeaker C: So, for the Italian, the results are stranger.\nSpeaker C: So, what appears is that perhaps Spanish is not very close to Italian, because when using the network training on Spanish, your rate is almost twice the baseline error rate.\nSpeaker E: Well, I mean, let's see.\nSpeaker E: Is there any difference in...\nSpeaker E: So, you're saying that when you train on English and test on...\nSpeaker E: No, you don't have training on English.\nSpeaker C: There is another difference is that the noises are different.\nSpeaker C: For the Italian part, I mean the networks are trained with noise from...\nSpeaker C: or the ATGETs.\nSpeaker C: And the noise...\nSpeaker C: Perhaps the noise are quite different from the noises in the speech that Italian.\nSpeaker E: Do we have any test sets in any other language that have the same noise as in...\nSpeaker E: You are...\nSpeaker G: Can I add something real quick?\nSpeaker G: In the upper part, the English stuff, it looks like the very best number is 60.9.\nSpeaker G: And that's in the third section in the upper part under PLP, J. Rosta, for the middle column.\nSpeaker G: Yeah.\nSpeaker G: Is that a noisy condition?\nSpeaker F: Yeah.\nSpeaker G: So, that's matched training. Is that what that is?\nSpeaker C: It's not a third part, so it's an I-limy smashed.\nSpeaker C: So, training...\nSpeaker G: So, why do you get your best number in...\nSpeaker G: Wouldn't you get your best number in the clean case?\nSpeaker D: It's relative to the baseline mismatching.\nSpeaker G: Oh, okay. So, these are not...\nSpeaker G: Okay. All right. I see.\nSpeaker G: Yeah.\nSpeaker G: Okay.\nSpeaker G: And then, so, in the...\nSpeaker G: In the non-mismatched clean case, your best one was under MFCC.\nSpeaker G: That's 61.4.\nSpeaker C: Yeah. But it's not a clean case.\nSpeaker C: It's a noisy case, but training and test noises are the same.\nSpeaker G: Oh, so this upper third?\nSpeaker G: Yeah.\nSpeaker G: That's still noisy?\nSpeaker G: Yeah.\nSpeaker G: Okay.\nSpeaker C: So, it's always noisy, basically.\nSpeaker C: What?\nSpeaker C: Nice.\nSpeaker E: Okay.\nSpeaker E: So, I think this will take some...\nSpeaker E: Looking at, thinking about it.\nSpeaker E: What is currently running that's...\nSpeaker E: That just filling in holes here?\nSpeaker C: No, we don't plan to fill the holes, but...\nSpeaker C: Actually, there is something important.\nSpeaker C: Is that we made a lot of assumption concerning the online normalization.\nSpeaker C: And we just noticed recently that the approach that we were using was not leading to very good results when we use the straight features to HDK.\nSpeaker C: So, basically, if you look at the left of the table, the first row with 86, 143 and 75, these are the results we obtained for Italian with straight PLP features using online normalization.\nSpeaker C: And the...\nSpeaker C: What's in the table just at the left of the PLP 12 online normalization column.\nSpeaker C: So, the number 79, 54 and 42 are the results obtained by PDIBA with...\nSpeaker C:...is online normalization.\nSpeaker C: Where is that 79?\nSpeaker E: It's just sort of sitting right on the column line.\nSpeaker E: Oh, I see.\nSpeaker C: Yeah. So, these are the results of OGI with online normalization and straight features to HDK.\nSpeaker C: And the previous result, 86 and so on, yes.\nSpeaker C: With our features, straight to HDK.\nSpeaker C: So, what we see there is that the way we were doing this was not correct, but still the networks are very good.\nSpeaker C: When we use the networks, our numbers are better.\nSpeaker E: So, do you know what was wrong with the online normalization?\nSpeaker C: Yeah, there were different things.\nSpeaker C: Basically, my fourth thing is the alpha values, so the recursion part.\nSpeaker C: I used 0.5%, which was the default value in the programs here.\nSpeaker C: And the pretty values are 5%.\nSpeaker C: So, it adapts more quickly.\nSpeaker C: But, yeah, I assume that this was not important because previous results from Dan and show that basically both values give the same results.\nSpeaker C: It was true on TI digits, but not true on Italian.\nSpeaker C: Second thing is the initialization of the stuff.\nSpeaker C: Actually, what we were doing is to start the recursion from the beginning of the iterations and using initial values that are the global mean and variance.\nSpeaker C: Measure that across the world database.\nSpeaker C: And pretty bad, it's something different is that she initialized the values of the mean and variance by computing this on the 25 first frames of each other.\nSpeaker C: There were other minor differences, the fact that she used 15 DCT instead of 13 and that she used C0 instead of log energy.\nSpeaker C: But the main difference is concerns the recursion.\nSpeaker C: So, I changed the code and now we have a baseline that's similar to the OGI baseline.\nSpeaker C: It's slightly different because I don't exactly initialize the same way she does.\nSpeaker C: Actually, I don't wait to have 25 frames before computing the mean and variance to start the recursion.\nSpeaker C: I use our line scheme and only start the recursion after the 25th frame.\nSpeaker C: But it's similar.\nSpeaker C: I retrained the networks with these, well, the networks are retraining with these new features.\nSpeaker C: So, basically what I expect is that these numbers will a little bit go down but perhaps not so much because I think the neural networks learn perhaps to, even if the feature sound is normalized, it will learn how to normalize.\nSpeaker E: I think that given the pressure of time, we probably want to draw some conclusions from this, do some reductions in what we're looking at and make some strong decisions for what we're going to do testing on for next week.\nSpeaker E: Did you have something going on on the side with multi-band?\nSpeaker C: No, we plan to start this. So, actually, we have discussed what we could do more as a research.\nSpeaker C: We were thinking perhaps that the way we use the tandem is not, well, there is basically perhaps a flow in the stuff because we train the networks.\nSpeaker C: What we ask is the network is to put the decision boundary somewhere in the space and ask the network to put one side of the particular phoneme at one side of the boundary decision boundary and one for another phoneme at the other side.\nSpeaker C: So, there is kind of reduction of the information there that's not correct because if we change task and if the phonemes are not in the same context and the new task, obviously the decision boundaries should not be at the same place.\nSpeaker C: But the way the network gives the features is that it removes completely information from the features by placing the decision boundaries that optimal places for one kind of data.\nSpeaker C: But this is not the case for another kind of data. So, what we were thinking about is perhaps one way to solve this problem is to increase the number of outputs of the neural networks.\nSpeaker C: Doing something like phonemes within contexts, well, basically context dependent phonemes.\nSpeaker E: Maybe. I mean, I think you could make the same argument, be justice legitimate for hybrid systems as well. But we know that things get better with context dependent versions.\nSpeaker E: Yeah, but here it's something different. We want to have features. Yeah. But it's still true that what you're doing is you're ignoring, you're coming up with something to represent whether it's distribution, probably distribution or features.\nSpeaker E: If you're coming up with a set of variables that are representing things that vary over context and you're putting it all together, ignoring the differences in context. That's true for the hybrid system, the street for a tandem system.\nSpeaker E: So for that reason, when you in the hybrid system, when you incorporate context one way or another, you do get better scores. Yeah. Okay. But it's a big deal to get that.\nSpeaker E: I'm sort of. And once you the other thing is that once you represent start representing more and more context, it is much more specific to a particular task in language.\nSpeaker E: So the acoustics essentially particular context, for instance, you may have some kinds of context that will never occur in one language and will occur frequently in the others.\nSpeaker E: The issue of getting enough training for a particular kind of context becomes harder. We already actually don't have a huge amount of training data.\nSpeaker C: Yeah, but. I mean, the way we do it now is that we have a neural network and basically the network is trained almost to give binary decisions.\nSpeaker C: Right. And binary decisions about phonemes. Almost. But I mean, it does give a distribution. Yeah.\nSpeaker E: And it is true that if there's two phones that are very similar that they may prefer one that it will give a reasonably high value to the other two. Yeah. Sure. But.\nSpeaker C: Basically, it's almost binary decisions and the idea of using more classes is to get something that's less binary decision. Oh, no, but it would still be even more of a binary decision.\nSpeaker E: It would be more of one because then you would say that in that this phone in this context is a one, but the same phone in a slightly different context is a zero. That would be even even more distinct for binary decision. I have to think you'd want to go the other way and have fewer classes.\nSpeaker E: I mean, for instance, the thing I was arguing for before, but again, which I don't think we have time to try is something in which you would modify the code so you could train to have several outputs on and use articulatory features.\nSpeaker E: Because then that would go that would be much broader and cover many different situations. If you got a very, very fine category. Yeah, but I think.\nSpeaker C: Yeah, perhaps you're right, but you have more classes. So you have more information in your features. So you have more information in the posterials vector, which means that.\nSpeaker C: But still information is relevant because it's information that absolutely discriminates if it's posterior to discriminate amongst phonemes in context.\nSpeaker E: Well, it's an interesting. So I mean, we could disagree about it length, but the real thing is if you're interested in it, you'll probably try it and we'll see.\nSpeaker E: But, but what I'm more concerned with now and this operational level is, you know, what do we do in four or five days? And so we have to be concerned with, are we going to look at any combinations of things?\nSpeaker E: You know, once the nets get retrained, so you have this problem out of it. Are we going to look at multi-band, are we going to look at combinations of things?\nSpeaker E: What questions are we going to ask? Now that we should probably turn shortly to this. So, gee, I know how are we going to combine with what they've been focusing on?\nSpeaker E: We haven't been doing any of the LDA roster sort of thing. And they, although they don't talk about it in this note, there's the issue of the.\nSpeaker E: New law business versus the logarithm. So, so what is going on right now? What's right? You've got nets retraining.\nSpeaker A: Is there any HTK training? I'm trying the HTK with PLP 12 online delta delta MSG feature together.\nSpeaker E: Combination, I see. But the combination is so. MSG and PLP. And is this with the revised online normalization?\nSpeaker E: With the old one. So it's using all the nets for that. But again, we have the hope that we have the hope that it maybe is not making too much difference.\nSpeaker C: So there is this combination, yeah, working on combination, obviously. I will start to work on multi-band. And we plan to work also on the idea of using both features and net outputs.\nSpeaker C: And we think that with this approach, perhaps we could reduce the number of outputs of neural network. So get simple networks because we still have the features.\nSpeaker C: So we have come up with different kind of broad phonetic categories. And we have basically we have three types of broad phonetic classes.\nSpeaker C: Well, something using a base of articulation, which leads to nine, I think, broad classes. Another which is based on manner, which is something also like nine classes.\nSpeaker C: And then something that combined both. And we have 25, 27, broad classes.\nSpeaker E: So like back forwards, front forwards. So what should you do? So you have two nets or three nets? How many nets do you have?\nSpeaker C: For a moment, we don't have nets. I mean, it's just we're just changing the labels to retrain nets with fewer outputs. Right. And then I didn't understand.\nSpeaker E: The software currently just has allows for I think the one one hot output. So you're having multiple nets and combining them. Or how are you how are you coming up with if you say, if you have a place characteristic and a manner characteristic, how do you?\nSpeaker C: It's the single net one. Oh, it's just one net. It's one net with 27 outputs if we have 27 classes. So it's basically a standard net with fewer classes.\nSpeaker E: So you're sort of going the other way of what you were saying a bit ago. Yeah, but I think yeah, including the features. Yeah, I don't think this will work alone.\nSpeaker C: I think it will get worse because well, I believe the effect that of to reducing too much information is basically what happens.\nSpeaker C: And I think if you include that plus the other feature. Yeah, because there's perhaps one important thing that the net brings and what I show show that is the distinction between speech and silence because these nets are trained on well controlled condition.\nSpeaker C: I mean, the labels are obtained on clean speech. We had no is after. So this is one thing. But perhaps something intermediary using also some broad classes could could bring some much more information.\nSpeaker E: So again, we have these broad classes and well, somewhat, I mean 27 is 64 basically. And you have the original features with your POP or something. Then just to remind me, all of that goes into that all of that is transformed by KL or something.\nSpeaker C: So we'll probably be one single KL to transform everything.\nSpeaker E: Well, no, I do something that you know, so there's a question of whether you would write whether you would transform together or just one. Yeah, I want to try it both ways. That's interesting. So that's something that you haven't trained yet, but are preparing to train. Yeah.\nSpeaker E: So I think you know, we need to choose the choose the experiments carefully.\nSpeaker E: We get key questions answered before then and leave other ones aside, even if it's in complete tables someplace. It's really time to choose. Let me pass this out, by the way.\nSpeaker E: Did I interrupt you with or other things that you want?\nSpeaker E: Something I asked, so they're doing the VAD, I guess they mean voice activity section. So against the silence. So they've just trained up in that, which has two outputs, I believe.\nSpeaker E: I asked he and I asked he and I, whether they compared that to just taking the nets we already had and summing up the probabilities to get the speech voice activity detection or else just using the silence if there's anyone silence up.\nSpeaker E: And he didn't think they had, but on the other hand, maybe they can get by with a smaller net and maybe sometimes you don't run the other. Maybe there's a computational advantage to having a separate net anyway.\nSpeaker E: So the results look pretty good. I mean, not uniformly. I mean, there's an example or two that you can find where it made it slightly worse, but in all but a couple examples.\nSpeaker A: They have a question of the result. How are trying the LDA filter? How are trying the LDA filter?\nSpeaker A: I'm sorry, don't understand the question. The LDA filter needs some trying set to obtain the filter. Maybe I don't know exactly how.\nSpeaker A: Training. Training with the training test of each understanding.\nSpeaker A: For example, LDA filter needs a set of trying to obtain the filter. Maybe for the Italian, for the EU and for Finnish. These filter are obtained with the own training set.\nSpeaker E: Yes, I don't know. That's a very good question. Where does the LDA come from? In earlier experiments, they had taken LDA from a completely different database.\nSpeaker A: Yeah, because maybe it's the same situation that the neural network is running with the own.\nSpeaker E: So that's a good question. Where does it come from? Yeah, I don't know. But to tell you the truth, I wasn't actually looking at the LDA so much when I was looking at it. I was mostly thinking about the VAD.\nSpeaker A: What is ASP? Oh, that's the features. I don't understand what you're saying. What is the difference between ASP and baseline over?\nSpeaker E: Because there's baseline error above it. And this is mostly better than baseline, although in some cases it's a little less.\nSpeaker E: So it's basically ASP is 23 ml, minus 13. Yeah, it says what it is, but I don't know how that's different from the baseline.\nSpeaker C: I think this is the same point we were at when we were at the CZ row, using CZ row instead of luck energy. It should be that.\nSpeaker G: They say in here that the VAD is not used as an additional feature. Does anybody know how they're using it?\nSpeaker E: So what they're doing here is if you look down at the block diagram, they get an estimate of whether it's speech or silence, then they have a median filter of it.\nSpeaker E: So basically they're trying to find stretches. The median filter is enforcing and having some continuity. You find stretches where the combination of the frame wise VAD and the median filters say that the stretch of silence. And then it's going through and destroying the data away.\nSpeaker E: So it's throwing out frames. The median filter is enforcing that it's not going to be single cases of frames.\nSpeaker E: So it's throwing out frames. And the thing is what I don't understand is how they're doing this with HTK. That's what I was just going to ask. How can this throw out frames?\nSpeaker E: You can, right? It stretches again. For single frames I think it'd be pretty hard. If you say speech starts here, speech ends there.\nSpeaker C: Yeah, you can basically remove the frames from the feature files.\nSpeaker E: Yeah, so I mean in the decoding, you're saying we're going to decode from here to here. I think they're treating it like, well, it's not isolated words, but connected.\nSpeaker G: In the text they say that this is a tentative block background of a possible configuration we could think of.\nSpeaker G: So that sort of sounds like they're not doing that yet.\nSpeaker E: Well, no, they have numbers though, right? So I think they're doing something like that. I think that they're, I think what I mean by that is they're trying to come up with a block diagram that's plausible for the standard.\nSpeaker E: From the point of view of reducing the number of bits you have to transmit, it's not a bad idea to do that.\nSpeaker G: I'm just wondering what exactly did they do up in this table if it wasn't this?\nSpeaker E: But it's the thing is that I certainly would be tricky about it in transmitting voice for listening to is that these kinds of things cut speech off a lot.\nSpeaker G: Plus it's going to introduce delays.\nSpeaker E: It does introduce delays, but they're claiming that it's within the boundaries of it. And the LDA introduces delays. And what he's suggesting this here is a parallel path so that it doesn't introduce any more delay.\nSpeaker E: It introduces 200 milliseconds of delay, but at the same time the LDA down here.\nSpeaker E: What's the difference between TLDA and SLDA?\nSpeaker E: It's a temporal spectrum. Oh, thank you. You wouldn't know that.\nSpeaker E: So the temporal LDA does in fact include the same.\nSpeaker E: So I think by saying this is a tentative block diagram I think means if you construct it this way this delay would work out in that way.\nSpeaker E: It clearly did actually remove silent sections because they got these word error results.\nSpeaker E: So I think that it's nice to do that in this because in fact it's going to give a better word error result.\nSpeaker E: And therefore what helps with an evaluation was to whether this would actually be in a final standard.\nSpeaker E: I don't know. As you know part of the problem with evaluation right now is that the word models are pretty bad and nobody wants has approached improving them.\nSpeaker E: So it's possible that a lot of the problems with so many insertions and so forth would go away if there were better word models.\nSpeaker E: So this might just be a temporary thing. But on the other hand, maybe it's a decent idea.\nSpeaker E: The question that we're going to want to go through next week when Henrik shows up I guess is given that we've been, look at what we've been trying. We're looking at by then I guess combinations of features and multi-band.\nSpeaker E: And we've been looking at cross language, cross task issues.\nSpeaker E: And they've been not so much looking at the cross task multiple language issues.\nSpeaker E: But they've been looking at these issues, the online normalization and the voice activity detection.\nSpeaker E: And I guess when he comes here we're going to have to start deciding about what do we choose from what we've looked at to blend with some group of things, what they've looked at.\nSpeaker E: And once we choose that, how do we split up the effort? Because we still have, even once we choose, we've still got another month or so.\nSpeaker E: I mean there's holidays in the way but I think the evaluation data comes January 31st. So there's still a fair amount of time to do things together.\nSpeaker E: It's just that there probably should be somewhat more coherent between the two sites.\nSpeaker G: When they remove the silence range, do they insert some kind of a marker so that the recognizer knows when it's time to backtrace it?\nSpeaker E: Well see there, I think they're, I don't know the specifics of how they're doing it.\nSpeaker E: They're getting around the way the recognizer works because they're not allowed to change the scripts.\nSpeaker E: So the recognizer I believe so. Maybe they're just inserting some dummy frames or something?\nSpeaker E: You know that's what I had thought but I don't think they are.\nSpeaker E: I mean that's sort of what I had imagined would happen is that on the other side yeah you put some low level noise or something.\nSpeaker E: Probably don't want all zeros, most of the guys just don't like zeros but you know put some epsilon in or some random variable in or something.\nSpeaker G: Some constant vector. I mean not a constant.\nSpeaker E: Or something divide by the variance of that but I mean it's...\nSpeaker G: But something that what I mean is something that is very distinguishable from speech so that the silence model in HDK will always pick it up.\nSpeaker E: Yeah so that's what I thought they would do or else. Maybe there is some indicator until it's starting to stop.\nSpeaker E: I don't know but whatever they did I mean they have to play within the roles of the specific variation.\nSpeaker E: We can find out.\nSpeaker G: You got to do something otherwise if it's just a bunch of speech stuck together.\nSpeaker G: No they're...\nSpeaker E: Yeah.\nSpeaker E: They would do badly and they did badly right?\nSpeaker E: Yeah.\nNone: Yeah.\nSpeaker E: So okay so I think this brings me up to date a bit.\nSpeaker E: I have play things other people have to date a bit.\nSpeaker E: I think I want to look at these numbers offline a little bit and take a bit and talk to everybody outside of this meeting.\nSpeaker E: But I mean it sounds like...\nSpeaker E: I mean they're the usual number of little problems and bugs and so forth but it sounds like they're getting ironed out.\nSpeaker E: Now we seem to be kind of in a position to actually look at stuff and compare things.\nSpeaker E: That's pretty good.\nSpeaker E: I don't know what the...\nSpeaker E: One of the things I wonder about coming back to the first results you talked about is...\nSpeaker E: how much things could be helped by more parameters and how many more parameters we can afford to have...\nSpeaker E: in terms of the computational limits.\nSpeaker E: Because when we go to twice as much data and have the same number of parameters, particularly when it's twice as much data and it's quite diverse, I wonder if having twice as many parameters would help.\nSpeaker E: It's kind of a bigger hidden layer.\nSpeaker E: But I doubt it would help by 40%.\nSpeaker E: But it's curious.\nSpeaker E: How are we doing on resources to ask them?\nSpeaker C: I think we're all right.\nSpeaker C: There's much problems we can.\nSpeaker C: Computation?\nSpeaker C: Well, this table took more than five days to get...\nSpeaker E: Were you folks using Jin?\nSpeaker E: That just died, you know.\nSpeaker C: No, you were using Jin, perhaps.\nSpeaker E: That's good.\nSpeaker E: We're going to get a replacement server that will be a faster server.\nSpeaker E: 750 megahertz.\nSpeaker E: But it won't be installed for a little while.\nSpeaker B: Do we have that big UIVM machine?\nSpeaker E: We have the little tiny IBM machine that might someday grow up to be a big IBM machine.\nSpeaker E: It's got slots for 8.\nSpeaker E: IBM was donating 5.\nSpeaker E: I think we only got 2 so far processors.\nSpeaker E: We originally hope we were getting 800 megahertz processors.\nSpeaker E: They end up in 550.\nSpeaker E: So instead of having 8 processors that were 800 megahertz, we end up with 2 that are 5 or 15 megahertz.\nSpeaker E: And more are supposed to come soon.\nSpeaker E: And there's only a moderate amount of memory.\nSpeaker E: So I don't think anybody has been sufficiently excited by it to spend much time with it.\nSpeaker E: But hopefully they'll get us some more parts soon.\nSpeaker E: Yeah, I think that'll be, once we get a populated, that'll be a nice machine.\nSpeaker E: I mean, we will ultimately get 8 processors in there.\nSpeaker E: And a nice amount of memory.\nSpeaker E: So it will be a pretty fast Linux machine.\nSpeaker B: And if we can do things, not Linux, some of the machines we have going already, like, sweet, it seems pretty fast.\nSpeaker B: I think 5 is pretty fast, too.\nSpeaker E: Yeah, I mean, you can check with Dave Johnson.\nSpeaker E: I think the machine is just sitting there.\nSpeaker E: And it does have 2 processors.\nSpeaker E: Somebody could do, you know, check out the multi-threading libraries.\nSpeaker E: I mean, it's possible that the, I guess, a prudent thing to do would be for somebody to do the work on getting our code running on that machine with 2 processors, and then, you know, the right 5 rate, there's going to be debugging hassles.\nSpeaker E: Then we'd be set for when we did have 5 or 8.\nSpeaker E: They have it really be useful.\nSpeaker E: Notice how I said somebody in my head, you direction.\nSpeaker E: That's one thing you don't get in these recordings.\nSpeaker E: You don't get the visuals.\nSpeaker B: Mostly the, you know, that work training that are, slow down or the HDK runs, that are slow down.\nSpeaker E: I think, yes.\nSpeaker E: You're right. I mean, I think you're sort of held up by both, right?\nSpeaker E: If the neural net trainings were 100 times faster, you still wouldn't be anything running through you as 100 times faster, because you'd be stuck by the HDK training.\nSpeaker E: But if the HDK, I mean, I think they're both, it sounded like they were roughly equal.\nSpeaker E: So that's not right?\nSpeaker B: Because I think that'll be running Linux and sweet, sweet, and fudge already running Linux.\nSpeaker B: So I could try to get the neural network trainings with the HDK stuff running under Linux, to start with, I'm wondering which one I should pick first.\nSpeaker E: Probably the neural net, this is probably, it's, it's, well, I don't know, they both, HDK we use for this Aurora stuff.\nSpeaker E: I think it's not clear yet what we're going to use for trainings.\nSpeaker E: Well, is the trainings, is it the training that takes the time or the decoding?\nSpeaker E: Is it about equal?\nSpeaker E: Training the two for HDK?\nSpeaker C: For, yeah, for the Aurora.\nSpeaker C: Training is longer.\nSpeaker E: Okay.\nSpeaker E: I don't know how we can, I don't know how to, do we have HDK source?\nSpeaker E: Yeah.\nSpeaker E: You would think that would fairly trivially, the training would anyway.\nSpeaker E: The testing, I don't think would parallelize all that well, but I think that you could certainly do distributed sort of, and though it's the each individual sentence, it's going to be tricky to parallelize, but you could split up the sentences and tell them.\nSpeaker G: They have a thing for doing that, they have for a while in HDK, and you can parallelize the training in a hundred and several machines, and it just basically keeps counts, and then there's something, a final thing that you run, and it accumulates all the counts together.\nSpeaker G: I see.\nSpeaker G: I don't know what their scripts are set up to do for their Aurora stuff,\nSpeaker E: but something we haven't really settled on yet is other than this Aurora stuff, what do we do, large vocabulary training slash testing for tandem systems, because we haven't really done much for tandem systems for larger stuff.\nSpeaker E: We had this one collaboration with CMU and we used Sphinx.\nSpeaker E: We're also going to be collaborating with SRI, and we have theirs.\nSpeaker E: So, I don't know.\nSpeaker E: So, I think the advantage of going with the neural net thing is that we're going to use the neural net training no matter what, for a lot of things we're doing, whereas exactly which HMM, the SRI mixture based HMM thing we use is going to depend.\nSpeaker E: So, with that, maybe we should go to R,\nNone: digit recitation. Task.\nSpeaker E: It's about 1150.\nNone: And...\nSpeaker E: I can start over here.\nSpeaker E: 2011-2030.\nSpeaker E: 0690601423051081.\nSpeaker E: 4004722617428789759.\nSpeaker E: 0103010322430101556924063703.\nSpeaker A: And, transcript number 19912010908276193342055305163274891090737312374743.\nSpeaker A: 25316616005679196951.\nSpeaker G: Transcript 2071-209012084641536603789290900581.\nSpeaker G: 1153564275603654045556779194086031091218100.\nSpeaker D: Transcript 2051-207000220432213469567808837702861051020219270502.\nSpeaker D: 263983405725610882849647400.\nNone: Transcript 1971-199091010101.\nSpeaker E: 315-074-560338367.\nSpeaker B: 35074560338163359187.\nSpeaker C: Transcript 2031-20501012122663273497905606002.\nSpeaker C: 81294791650834053120607305264881.\nSpeaker C: 786748619490.\nSpeaker B: 810101063531.\nSpeaker B: 810101010106.\nSpeaker E: 8naden1010107.\nSpeaker E: Does everyone sign the consent form before in previous meetings?\nSpeaker E: So you don't have to do it again each time?\nSpeaker E: The government could only give you two Donna, it's 18, her eyes, and her eyes cannot show it again, I think it's an unrisk campaign\n"}, {"meeting_id": "Bro005", "summary": "The participants went over the results of the most recent experiments, discussing future directions to take. Adding in more languages was still decreasing model performance. The best network consisted of the TIMIT and noise. The team thought that they should try to reduce their dependence on noise in the data. The Professor outlined how they could continue their experimentation.", "dialogue": "Speaker H: Hey, Mike, Mike one.\nNone: Uh?\nNone: Yes, please.\nSpeaker D: I mean, we're testing oyster busting, but listen, you get to wait.\nSpeaker C: Okay, so, uh, we've got some, uh, Zerax things to pass out.\nSpeaker H: Yeah.\nSpeaker H: I'm sorry for the table, but...\nNone: There's a bruise.\nSpeaker D: This one's nice though.\nSpeaker D: This is the last column we use for imagination.\nNone: Okay.\nSpeaker D: Uh, this one's nice though.\nSpeaker D: This has nice big font.\nSpeaker J: Yeah.\nSpeaker J: Yeah.\nSpeaker D: Yeah.\nSpeaker D: You get older, you have these different perspectives.\nSpeaker D: I mean, lowering the word area to spine, but having big font.\nSpeaker H: Extending the width.\nSpeaker H: Put the cut off for some.\nSpeaker E: Yeah, it's a nice little big line.\nSpeaker H: Okay.\nSpeaker H: Okay.\nSpeaker H: So there is kind of summary of what has been done.\nSpeaker H: Oh, summary of experiments since, well, since last week and also since the...\nSpeaker H: We've started working this.\nSpeaker H: So since last week, we've started to fill the column with, uh, features with net strain on PLP with our nine normalization, but with delta also, because the column must not completely, what?\nSpeaker H: It's still not completely filled, but we have more results to compare with network using without BLP. And finally, delta seems very important.\nSpeaker H: I don't know if you take, um, let's say, anyway, or a 2B.\nSpeaker H: So the next, the second part of the table, uh, when we use the large training set using French Spanish and English, you have 106 without delta and 89 with the delta.\nSpeaker D: And again, all of these numbers are with 100% being the baseline performance, but with the no cap straw system going straight into the edge.\nSpeaker H: So now we see that the gap between the different training set is much, uh, much smaller.\nSpeaker H: But actually, for English training on timid is still better than the other languages.\nSpeaker H: And, yeah.\nSpeaker H: And also for Italian, actually. If you take the second set of experiment for Italian, so the mismatched condition, um, when we use the training on timid, so it's multi-english, we have 91 number, and training with other languages is a little bit worse.\nSpeaker D: Oh, I see it down near the bottom of this sheet.\nSpeaker H: So, yeah. And, yeah. And here, the gap is still more important between using delta and not using delta.\nSpeaker H: If I take the, the large training set, it's, we have 172, and 104 when we use delta.\nSpeaker H: Even if the context is quite the same because without delta, we use 17 frames.\nSpeaker H: Yeah. So, the second point is that we have no single cross-language experiments that we did not have last week.\nSpeaker H: So, this is training the net on French only, or on English only, and testing on Italian.\nSpeaker H: And training the net on French only, and Spanish only, and testing on the attitudes.\nSpeaker H: And, yeah. What we see is that these nets are not as good, except for the multi-english, which is always one of the best.\nSpeaker H: Yeah. Then, we've started to work on a large data base containing sentences from the French, from the Spanish, from the Tibet, from Spain, from English digits, and from Italian digits.\nSpeaker H: So, this is another line, another set of lines in the table, and left to the same as what it's fine.\nSpeaker H: And, actually, we did this before knowing the result about the delta. So, we have to redo the experiment training the net with PLP, but with delta.\nSpeaker H: But, this net performed quite well. It's better than the net using French, Spanish, and English.\nSpeaker H: So, we have also started feature combination experiments, mainly experiments using features and net outputs together.\nSpeaker H: And, this is the results on the other document.\nSpeaker H: We can discuss this after a problem, just to know if we need to.\nSpeaker H: Yeah. So, basically, there are four kind of systems. The first one is combine two feature streams using each feature stream as its own MLP, so it's similar to the tandem that was proposed for the first multi-stream tandem for the first proposal.\nSpeaker H: The second is using features and PLT transform MLP outputs, and the third one is to use a single PLT transform features as well as MLP outputs.\nSpeaker H: Yeah. You can command this result.\nSpeaker I: Yes, I would like to say that, for example, if we doesn't use the delta delta, we have an improve when we use some combination.\nSpeaker H: But, we... Just to clear the numbers here are recognition accuracy. So, it's another one.\nSpeaker I: Again, we switch to another one.\nSpeaker I: Yes, and the base line is 82.\nSpeaker C: The base line is 82.\nSpeaker C: Yeah.\nSpeaker H: So, it's experiment only on the Italian mismatched for the moment for this.\nSpeaker I: This is Italian mismatched. Okay, I move.\nSpeaker I: This is MLP, and it's obviously that the English MLP is dependent. For the rest of the experiment, I use multi-English, only multi-English.\nSpeaker I: The result is that the MSG-3 feature doesn't work for the Italian database, because it never helps to increase the algorithms.\nSpeaker H: Actually, if we look at the table, the huge table, we see that for TI digits, MLG performs as well as the PLP. But this is not a case for Italian.\nSpeaker H: Where the error rate is almost twice the error rate of PLP.\nSpeaker H: So, I don't think this is a bug, but this is something in probably the MLG process that...\nSpeaker H: I don't know what exactly perhaps the fact that there's no low pass filter, well, or low-prim phase filter, and that there is some VC offset in the Italian.\nSpeaker H: Well, something simple like that, but that we need to sort out if we want to get improvement by combining PLP and MSG, because on the moment the MSG doesn't bring much information.\nSpeaker H: And as Karman said, if we combine the two, we have the result basically of PLP.\nSpeaker D: The baseline system, when you said the baseline system was 82%, that was trained on what and tested on what.\nSpeaker D: That was Italian mismatched digits, is the testing, and the training is Italian digits.\nSpeaker D: So, the mismatch just refers to the noise and microphone, so forth. So, would that then correspond to the first line here of where the training is?\nSpeaker D: Is the Italian digits...\nSpeaker I: The training of the HTG?\nSpeaker I: Yes.\nSpeaker D: Training of the net, yeah. So, what that says is that in a matched condition, we end up with a fair amount worse putting in the PLP.\nSpeaker D: Now, do we have a number I suppose for the matched? I don't mean matched, but use of Italian training on Italian digits for PLP only.\nSpeaker H: So, basically this is in the table. So, the number is 52%.\nSpeaker H: 52% of 82% of 18%. So, it's error rate.\nSpeaker H: So, this is accuracy. So, we have 90%.\nSpeaker H: Which is what we have also if we use PLP and MSG together, 89.7%.\nSpeaker D: Okay, so, even just PLP, it is not in the matched condition. I wonder if it's the difference between PLP and Mellkebstra or whether it's that the net of some reason is...\nSpeaker H:...it's not PLP and Mellkebstra give the same result. Well, we have this result. It's not...\nSpeaker H: So, PLP is a result with PLP alone feeling HTK. That's what you mean. Just PLP.\nSpeaker D: So, PLP. So, adding MSG, well, but that's without the neural net, right?\nSpeaker H: Yeah, that's without the neural net. And that's the result, basically, that OGI as well. So, with the MFCC with online normalization.\nSpeaker D: But, she had said 82%.\nSpeaker H: But, this is without online normalization.\nSpeaker H: Oh, this is the 82.\nSpeaker H: 82 is the Aurora baseline. So, MFCC, then we can use...\nSpeaker H: Well, OGI, they use MFCC, the baseline MFCC plus online normalization.\nSpeaker D: I keep getting confused because this is accuracy.\nSpeaker D: Alright, alright. So, this is... I was thinking, this was worse. Okay, so, this is all better because...\nSpeaker D: I'm bigger than 82. Okay, I'm all better now. Okay.\nSpeaker H: So, what happens is that when we apply online normalization, we jump to almost 90%.\nSpeaker H: When we apply a neural network is the same. We jump to 90%.\nSpeaker H: And whatever the normalization actually, if we use a neural network, even if the features are not correctly normalized, we jump to 90%.\nSpeaker D: So, we go from 88.6 to 290.\nSpeaker H: No, I mean 90, it's around 89.\nSpeaker H: 89.\nSpeaker D: And then, adding MFCC does nothing, basically.\nSpeaker H: No.\nSpeaker H: Okay.\nSpeaker H: For Italian.\nSpeaker H: For this case, right?\nSpeaker D: Alright. So, actually, the answer for experiments with one is that adding MSG, if you does not help in that case...\nSpeaker G: But don't, it's really good.\nSpeaker D: And the multi-English does...\nSpeaker D: So, if we think of this as an error rate, we start off with 18% error rate, roughly.\nSpeaker D: And we almost cut that in half by putting in the online normalization and the neural net.\nSpeaker D: And the MSG doesn't, however, particularly affect things.\nSpeaker D: And we cut off, I guess, about 25% of the error.\nSpeaker D: No, not quite that. Is it 2.6?\nSpeaker D: Not 18, but 16% or something of the error.\nSpeaker D: If we use multi-English instead of the magic condition.\nSpeaker D: But the Italian track.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker I: We select these stats because it's the more difficult.\nSpeaker D: Yes. Good.\nSpeaker D: Okay. So, then, you're assuming multi-English is closer to the kind of thing that you could use, since you're not going to have magic data for the other languages and so forth.\nSpeaker D: One thing is that, I think I asked you this before, but I want to double check.\nSpeaker D: When you say ME in these other tests, that's the multi-English.\nSpeaker D: But it is not all of the multi-English, right?\nSpeaker D: It's a part of it.\nSpeaker H: It's a part of it.\nSpeaker H: Or one million frames.\nSpeaker H: And the multi-English is how much?\nSpeaker H: It's one million and the half.\nSpeaker D: Yeah.\nSpeaker D: Oh, so you use almost all these two thirds of it.\nSpeaker D: So, it still, it hurts you.\nSpeaker D: It seems to hurt you a fair amount to add in this French and Spanish.\nSpeaker D: Yep.\nSpeaker C: That's why it's too wide.\nSpeaker B: Well, Stefan was saying that they weren't hand-label.\nSpeaker B: Yeah.\nSpeaker B: Yeah, he's French and Spanish.\nSpeaker B: It's a bit...\nSpeaker I: Maybe for that.\nSpeaker D: Still.\nSpeaker D: Okay.\nSpeaker D: All right, go ahead.\nSpeaker I: And then...\nSpeaker I: With the experiment type 2, I first tried to combine some feature from the MLP and other feature.\nSpeaker I: Another feature.\nSpeaker I: And we can...\nSpeaker I: First, the feature are without Delta Delta Delta and we can see that in this situation, the MSG3, the same help, nothing.\nSpeaker I: And then I do the same, but with the Delta Delta Delta, PLP Delta Delta Delta.\nSpeaker I: And the output of the MLP is without Delta Delta Delta.\nSpeaker I: And we have a little bit less result than the...\nSpeaker I: The baseline PLP with Delta Delta Delta.\nSpeaker I: Maybe when we have the new...\nSpeaker I: The new neural network, China with PLP Delta Delta Delta, maybe the final result must be better.\nSpeaker I: I don't know.\nSpeaker H: Actually, it's just to be so more...\nSpeaker H: This number, this 87.1 number has to be compared with...\nSpeaker D: Yes, yeah, I mean, you can't compare with the other because this is a with multi-English training.\nSpeaker D: She has to compare it with the one over that you've got in a box, which is that 84.6.\nSpeaker H: So...\nSpeaker H: Yeah, but I mean, in this case, for the 87.1, we use MLP outputs for the PLP net and straight features with Delta Delta.\nSpeaker H: And straight features with Delta Delta gives you what's on the first G, 88.6.\nSpeaker D: No, no, no, no, not trained with multi-English.\nSpeaker H: Yeah, but this is the second configuration.\nSpeaker H: So we use feature or net outputs together with features.\nSpeaker H: So, yeah, this is not perhaps not clear in, but in this table, the first column is for MLP and the second is for the features.\nSpeaker H: So, you're saying, so, so, in the question, what, what has adding the MLP down?\nSpeaker H: Yeah, so, actually, it, it decrease the accuracy because we have 88.6 and even the MLP alone, what gives the MLP alone?\nSpeaker H: Multi-English PLP, oh no, it gives 83.6.\nSpeaker H: So, we have 83.6 and 88.6 that gives 87.1.\nSpeaker D: I thought it was 80.\nSpeaker D: Okay, 83.6 and 88.6.\nSpeaker F: Okay, is that right?\nSpeaker I: Yeah.\nSpeaker I: But, you don't know, but maybe if we have the neural network, it will be the PLP delta Delta Delta, maybe.\nSpeaker I: Perhaps, yeah.\nSpeaker D: But that's one thing. But see, the other thing is that, I mean, it's good to take the difficult case, but let's consider what that means.\nSpeaker D: What, what we're saying is that, one of the things that, I mean, my interpretation of your, your, your, your suggestion is something like this, as motivation.\nSpeaker D: When we train on data, that is in one sense or another similar to the testing data, then we get a wind by having a instrument in training.\nSpeaker D: When we train on something that's quite different, we have a potential to have some problems.\nSpeaker D: And, if we get something that helps us when it's somewhat similar and doesn't hurt us too much, when it, when it's quite different, that's maybe not so bad.\nSpeaker D: So, the question is, if you took the same combination and you tried it out on, on, on, on, say digits, you know, was that experiment done?\nSpeaker D: No, okay.\nSpeaker D: Then, does that, you know, maybe with similar noise conditions and so forth, does it, does it then look much better?\nSpeaker D: And so, what is the range over these different kinds of tests? So, anyway, okay.\nSpeaker D: Yes.\nSpeaker I: And, with this type of configuration, what I do, I experiment using the new neural net with, named BloodClath 27, and I have found more let the same result.\nSpeaker H: So, it's slightly better.\nSpeaker H: It's slightly better.\nSpeaker I: Yeah, it's better, yes, it's better.\nSpeaker D: And, and, you know, again, maybe if you use the delta there.\nSpeaker D: Yeah, maybe.\nSpeaker D: You bring it up to where it was, you know, at least about the same, for a difficult case.\nSpeaker H: Yeah, so, well, so perhaps let's, let's jump at the last experiment.\nSpeaker H: Yes.\nSpeaker H: It's either less information from the neural network, if we use only the silence output, it's again better.\nSpeaker H: So, it's 89.\nSpeaker I: 40, 40, 40, 40, because in the situation, we prefer one kind of answer.\nSpeaker I: Yeah.\nSpeaker I: And then, with the first configuration, I am found that work, that's not work, work, work, but it's better the second configuration.\nSpeaker I: Because, for the, in the PLP delta and delta delta, we have 85.3 agglance, and with the second configuration, we have 87.1.\nSpeaker D: And, by the way, there was another suggestion that would apply to the second configuration, which was made by, and that was that if you have a feed two streams into HTK, and you change the variances, you scale the variances associated with these streams, you can effectively scale the streams.\nSpeaker D: Right, so, you know, without changing the scripts, for HTK, which is the rule here, you can still change the variances, which would effectively change the scale of these two streams that come in.\nSpeaker D: And so, if you do that, for instance, it may be the case that the MLP should not be considered as strongly.\nSpeaker D: And so, this is just setting them to be equal weight. Maybe it shouldn't be equal weight. I'm sorry to say that gives more experiments if we want to look at that.\nSpeaker D: Yeah, on the other hand, it's just experiments at the level of the HTK recognition. It's not even HTK. Well, I guess you have to do the HTK training also. Do you?\nSpeaker D: Yeah, you have to change the... No, you can just do it. Once you've done the training, the training is just coming up with variances, so I guess you can scale them all.\nSpeaker D: Scaled variances. But, is it... I mean, HTK models are diagonal variances, so... That's exactly the point I think that if you change what they are, it's diagonal variance matrices, but you say what those variances are.\nSpeaker D: So, that, you know, it's diagonal means that then you're going to... It's going to generally multiply it.\nSpeaker D: It's implicitly expenetuated to get probabilities, and so it's going to affect a range of things if you change the variances of some of the features. So, it's precisely given that model, you can very simply affect the strength of the two by the features.\nSpeaker D: So, that was... I already suggested. Yeah. So... So, it could just be that repeating them equally, treating two streams equally is just not the right thing to do.\nSpeaker D: Of course, it's potentially opening a can of worms because maybe it should be a different number of free queries. That's it.\nSpeaker D: Okay. So, I guess the other thing is to take... Yeah, if you want more to take a couple of the most successful of these.\nSpeaker H: Yeah, test, press everything. Yeah, try out these different tests.\nSpeaker H: So, the next point here, we've had some discussion with Steven Chauen about the artillery stuff. So, we'll perhaps start something next week.\nSpeaker H: The discussion with Inek, Sunil and Pratyba for trying to plug in our networks within their block diagram, where to plug in the network after the feature before or as a plug-in or as another path.\nSpeaker H: The discussion about multi-bound traps actually in Inek would like to see, perhaps if you remember the block diagram, there is a temporal LDA followed by a spectral LDA for each critical bound.\nSpeaker H: And it would like to replace these by a network which would make the system look like a trap. Well, basically it would be a trap system.\nSpeaker H: Basically, this is a trap system, kind of trap system, I mean, but where the neural network are replaced by LDA.\nSpeaker H: Yeah, and about multi-bound. I started multi-bound MLP training. Actually, I prefer to do exactly what I did when I was in Belgium.\nSpeaker H: So, I take exactly the same configuration, seven bands with nine frames of context, which is trained on the T-MIT and on the large database. So, with Spine and everything.\nSpeaker H: I started to train all the networks with larger contexts. So, this would be something between traps and multi-bound because we still have quite large bands, but with a lot of context also.\nSpeaker H: Yeah, we still have to work on finish. Basically, to make a decision on which MLP can be the best across the different languages for the moment, it's the T-MIT network and perhaps the network trained on everything.\nSpeaker H: So, we can test these two networks with delta and large networks. Test them also on finish and see which one is the best.\nSpeaker H: Well, the next part of the document is basically a kind of summary of everything that has been done. So, we have 79 MLPs trained on 1, 2, 3, 4, 5, 10 on 10 different databases.\nSpeaker H: The number of frames is also, so we have 1 million and half for some, 3 million for other and 6 million for the last one.\nSpeaker H: Yeah, as we mentioned, T-MIT is the only that's unlabeled. And perhaps this is what makes the difference. Yeah, the other are just Viter B-Lite.\nSpeaker H: So, these 79 MLPs differ on different things. First, with respect to the online normalization, there are that use bad online normalization and other good online normalization.\nSpeaker H: With respect to the features, with respect to the use of delta or no, with respect to the hidden layer size and to the targets. But of course, we don't have all the combination of these different parameters.\nSpeaker H: What's this? We only have 288 trained tests, not 2000. I was impressed by 2000.\nSpeaker H: Yeah, basically the observation is what we discussed already, the MLG problem. The fact that the MLP trained on target task decreased the error rate.\nSpeaker H: But when the MLP trained on the target task increased the error rate compared to using straight features. Except if the features are bad, actually except if the feature are not correctly online normalized.\nSpeaker H: In this case, the T-M is still better even if it's trained on not on the target digits. Yeah, so it sounds like the net corrects some of the problems with some poor normalization. But if you can do good normalization, it's...\nSpeaker H: So the fourth point is, yeah, the T-Mit plus noise seems to be the training set that gives the best network.\nSpeaker D: Before you go on to the possible issue, so on the MSG problem, I think that in the short time solution, that is trying to figure out what we can proceed forward with to make the greatest progress.\nSpeaker D: As I said with J-Rasta, I really like J-Rasta and I really like MSG. I think it's kind of in the category that it may be complicated.\nSpeaker D: And it might be if someone's interested in it, certain encourage anybody to look into it in a longer term.\nSpeaker H: So we get out of this particular rush for results, but in the short term, unless you have some strong idea of what's wrong.\nSpeaker H: No bypass filter. Yeah, there's supposed to be an MSG. Supposed to have an online normalization now. There is an edge kind of a GC.\nSpeaker D: Yeah, but also there's an online norm. Besides the a GC, there's an online normalization that you're supposed to be.\nSpeaker D: Yeah, taking out means and variances and so forth. In fact, the online normalization that we're using came from the MSG design.\nSpeaker H: Yeah, but this was the bad online normalization. Actually.\nSpeaker I: Are you a result of still with the bad online two?\nSpeaker H: Online two is good.\nSpeaker D: Yeah, so yeah, I agree. It's probably something simple. If someone, you know, let's play with it for a little bit. I mean, you're going to do what you're going to do, but my guess would be that it's something that is a simple thing that could take a while to find.\nSpeaker D: Yeah, and the other results observations two and three is, yeah, it's pretty much what we've seen that what we were concerned about is that if it's not on the target task, if it's on the target task, then it helps to have the MLP transforming it.\nSpeaker D: If it's not on the target task, then depending on how different it is, you can get a reduction in performance.\nSpeaker D: And the question is now how to get one and not the other or how to ameliorate the problems.\nSpeaker D: Because it certainly does is nice to have in there when there is something like that.\nSpeaker H: Yeah, so that's what you say.\nSpeaker H: The reason is that perhaps the target, the task dependency, the language dependency, and the noise dependency.\nSpeaker H: Well, but this is still not clear because I don't think we have enough result to talk about the language dependency.\nSpeaker H: Well, the team network is still the best, but there is also the other difference, the fact that it's unlabeled.\nSpeaker E: Hey.\nSpeaker E: I don't have a link.\nSpeaker D: I'm just sitting here.\nSpeaker D: I don't think we want to mess with the microphones, but just have a seat.\nSpeaker D: The summary of the first 45 minutes is that some stuff works and some stuff doesn't.\nSpeaker H: Okay.\nSpeaker H: We still have one of these.\nSpeaker D: Yeah, I guess we can do a little better than that.\nSpeaker D: I think if you start off with the other one, it sort of has it in words, and that has the associated result.\nNone: Okay.\nSpeaker D: So you're saying that although we see, yes, there's what you would expect in terms of a language dependency and a noise dependency.\nSpeaker D: That is when the neural net is trained on one of those and tested on something different.\nSpeaker D: You don't do as well as in the target thing, but you're saying that it is, although that general thing is observable so far, there's something you're not completely convinced about.\nSpeaker D: And what is that?\nSpeaker H: When you say not clear yet, what do you mean?\nSpeaker H: I mean, the fact that for the IDGITs, the timet net is the best, which is the English net, but the other has likely worse.\nSpeaker H: But you have to do affect the effect of changing language and the effect of training on something that's bitterly aligned instead of end and level.\nSpeaker H: Yeah.\nSpeaker D: Do you think the alignments are bad?\nSpeaker D: I mean, have you looked at the alignments at all? What do the terribly alignments do?\nSpeaker H: I don't know.\nSpeaker H: Did you look at the Spanish alignments, Carmen?\nSpeaker D: No.\nSpeaker D: It's interesting to look at it, because I mean, that is just looking, but it's not clearly necessarily what you do so badly from a terribly aligned net.\nSpeaker D: Depends on whether the ranking measure is that the engine is doing the alignment.\nSpeaker F: Yeah.\nSpeaker H: But perhaps it's not really the alignment that's bad, but just the phoneme string that's used for the alignment.\nSpeaker H: The pronunciation.\nSpeaker H: I mean, it's single pronunciation.\nSpeaker H: French phoneme strings were corrected manually, so we ask people to listen to the sentence, and we give the phoneme string and they kind of correct them.\nSpeaker H: But still, there might be errors just in the string of phonemes.\nSpeaker H: Yeah, so this is not really the real deal, I mean.\nSpeaker H: The third issue is the noise dependency, perhaps, but this is not clear yet, because all the nets are trained on the same noises.\nSpeaker D: I thought some of the nets were trained with the spine and so forth, so they're not that nice.\nSpeaker H: But the results are only coming for the nets.\nSpeaker D: Okay, yeah, just don't need more results there with that.\nSpeaker H: So, from these results, we have some questions with answers.\nSpeaker H: What should be the network input?\nSpeaker H: PLP work as well as MFCC, I mean.\nSpeaker H: But it seems important to use the delta.\nSpeaker H: We respect to the network size. There's one experiment that's still running, and we should have the result today, comparing networks with 500 and 1000 units.\nSpeaker H: Still no answer actually.\nSpeaker H: The training set, well, some kind of answer.\nSpeaker H: We can tell which training set gives the best result, but we don't know exactly why.\nSpeaker D: Right, I mean, the multi-English so far is the best.\nSpeaker D: Multi-English just means Tim.\nSpeaker D: So, yeah, so when you add other things into broad nets, it gets worse.\nSpeaker H: So, some questions with answers. Training set.\nSpeaker H: Training targets.\nSpeaker D: I like that the training set is both questions with answers and without answers.\nSpeaker D: It's multi-purpose.\nSpeaker H: Training, so yeah, the training targets actually.\nSpeaker H: The two of the main issues perhaps are still the language dependency and the noise dependency.\nSpeaker H: And perhaps to try to reduce the language dependency, we should focus on finding some other kind of training targets.\nSpeaker H: And labeling seems important because of the limit results.\nSpeaker H: For a moment, we use phonetic targets, but we could also use articulatory targets, soft targets, perhaps even use networks that doesn't do classification, but just regression.\nSpeaker H: So, trying to have neural networks that...\nSpeaker H: There's a regression and, well, basically, compute features without noise.\nSpeaker H: I mean, transform the noisy features in other features that are not noisy, but continuous features, not art targets.\nSpeaker D: Yeah, that seems like a good thing to do, probably not again a short term sort of thing.\nSpeaker D: I mean, one of the things about that is that...\nSpeaker D: I guess the major risk you have there being is being dependent on the kind of noise.\nSpeaker H: Yeah, but yeah, so this is...\nSpeaker H: Well, this one thing could help perhaps to reduce language dependency and for the noise part, we could combine this with other approaches like, the clenchmit approach, so the idea of putting all the noise that we can find inside the database, I think clenchmit works using more than 50 different noises to train this network.\nSpeaker H: So, this is one approach, and the other is multi-band.\nSpeaker H: I think it's more robust to noise changes, so perhaps doing something like multi-band, train on a lot of noises with feature space targets.\nSpeaker D: Yeah, if you could...\nSpeaker D: It's interesting though, maybe if you just trained up, I mean, one fancy would be you have something like articulate-right targets, and you have some reasonable database, which is copied over many times with a range of different noises.\nSpeaker D: And because what you're trying to do is come up with a core reasonable feature set, which is then going to be used by the HMF system.\nSpeaker H: So, yeah, the future work is... well, try to connect to the... to plug in the system to the OGI system.\nSpeaker H: There are still open questions there, where to put the MLB, basically.\nSpeaker D: And I guess, you know, the real open question, I mean, there's lots of open questions, but one of the core core open questions for that is...\nSpeaker D: If we take the best ones here, maybe not just the best ones, but the best few, or something, the most promising group from these other experiments, how well do they do over a range of these different tests, not just the Italian?\nSpeaker D: And then, then, then, see, again, how we know that there's a loss of performance from the neural net is trained on conditions that are different than we're going to test on.\nSpeaker D: But if you look over a range of these different tests, how well do these different ways combining the straight features with the MLB features stand out over that range?\nSpeaker D: That seems like the real question, if you know that.\nSpeaker D: So, if you just take PLP with double deltons, assume that's the feature.\nSpeaker D: Look at these different ways of combining it, and take... let's say, just take multi-English, because that works pretty well for the training.\nSpeaker D: Just look, take that case, and then look over all the different things. How does that compare between these?\nSpeaker D: All the tests set. All different test sets, and for the couple different ways that you have of combining them.\nSpeaker D: How well do they stand out over that?\nSpeaker H: And perhaps doing this for changing the variance of the streams, and so on.\nSpeaker H: That's another possibility.\nSpeaker H: Different times.\nSpeaker H: Yeah, so this would be more working on the MLP as an additional path instead of an insert to their diagram.\nSpeaker H: Perhaps the insert idea is kind of strange because they make an LDA, and then we will, again, add a network that is discriminant and that...\nSpeaker D: Yeah, this is a little strange, but on the other hand, they did before.\nSpeaker H: And because of so perhaps we know that when we have very good features, the MLP doesn't help.\nSpeaker D: So, I don't know.\nSpeaker D: The other thing though is that...\nSpeaker D: So, we want to get their path running here, right?\nSpeaker D: So we can add this other stuff as an additional path, right?\nSpeaker H: Yeah, the way we...\nSpeaker D: Because what are doing LDA Rasta?\nSpeaker D: What?\nSpeaker D: They're doing LDA Rasta.\nSpeaker H: Yeah, the way we want to do it, perhaps, is just to get the VAD labels and the final features.\nSpeaker H: So they will send us the...\nSpeaker H: Well, I see.\nSpeaker H: So, the feature files and with the VAD binary labels so that we can get our MLP features and filter them with the VAD and then combine them with their feature stream.\nSpeaker D: So, the first thing we're going to do there is to make sure that when we get those labels to the final features that we get the same result system without putting in a second path.\nSpeaker H: Yeah.\nSpeaker H: You mean...\nSpeaker H: Yeah, just retraining...\nSpeaker D: Yeah, just to make sure that we have...\nSpeaker D: We understand properly what things are.\nSpeaker D: The very first thing to do is to double check that we get the exact same results as the MLP and the K.\nSpeaker D: I mean, I don't know that we need to...\nNone:...training?\nSpeaker D: I mean, we can just take their training files also, but just for the testing, just to make sure that we get the same results.\nSpeaker D: We can duplicate it before we add in another...\nSpeaker D:...because otherwise, yeah, we won't be able to do anything.\nSpeaker H: Yeah, so LDA Rasta, I don't know if we want to...\nSpeaker H: We can try networks with LDA Rasta, filter with features.\nNone: I'm sorry.\nNone: Yeah, well...\nNone: Yeah.\nNone: Oh, you know, the other thing is when you say...\nSpeaker F: I'm sorry, I'm sorry, I'm sorry, I'm sorry, I'm sorry.\nSpeaker D: When you're talking about combining MLP features, so, suppose we said, okay, we've got these different features and so forth, but the PLP seems pretty good.\nSpeaker D: If we take the approach that Mike did, and have...\nSpeaker D: I mean, one of the situations we have is we have these different conditions, different languages, we have different noises.\nSpeaker D: If we have some drastically different conditions, then we're just trying to have different MLPs.\nSpeaker D: And put them together. What Mike found for the reverberation case, at least, I mean, who knows if we'll work for these other ones, that you did have nice, interpretative effects.\nSpeaker D: That is that, yes, if you knew what the reverberation condition was going to be, and you trained for that, then you got the best results.\nSpeaker D: But if you had, say, a heavily reverberation case, and I know reverberation case, and then you fed the thing, something that was modest on a reverberation, then you get some results in between the two, so sort of behaviorism.\nSpeaker D: That affair?\nSpeaker E: Yeah, that's fair.\nSpeaker E: It also seems like if you try to train a little bit, so much noise, you'll just like to put a little help in the right case,\nNone: and you have space, so it's better. Yeah, so you think it's better to have several MLPs?\nSpeaker F: Of course, it's easier.\nSpeaker E: I don't know how you interpret this on.\nSpeaker E: Yeah, I think I won't feel much for the work, but it's a lot of making it happen, which is a lot of pictures.\nSpeaker D: It works better if what?\nSpeaker E: Well, you were doing something that was, so maybe the analogy isn't quite right, you were doing something that was in a way a little better behaved.\nSpeaker D: But, you know, it's a very simple variable, which is reverberation.\nSpeaker D: Here the problem seems to be is that we don't have a really huge net with a really huge amount of training data, but we have, for this kind of task, I would think, sort of a modest amount, many million frames actually isn't that much.\nSpeaker D: A modest amount of training data from a couple different conditions.\nSpeaker D: So, the, that we anticipate in the test set in terms of language, and noise type, and channel characteristic, sort of all of them at a bunch of different dimensions.\nSpeaker D: And so, I'm just concerned that we don't really have the data to train up.\nSpeaker D: I mean, one of the things that we were seeing is that when we added in, we don't have a good explanation for this, we are seeing that we're adding in a few different databases, and the performance is getting worse.\nSpeaker D: And we just take one of those databases, a pretty good one. It actually is better.\nSpeaker D: And that says to me, yes, there might be some problems with the pronunciation models, some of the databases writing in or something like that, but one way or another, we don't have, seemingly, the ability to represent in the neural net of the size that we have, all of the variability that we're going to be covering.\nSpeaker D: So, I'm hoping that this is another take on the efficiency arc we're making, which I'm hoping it would moderate size neural nets, that if we look at more constrained conditions, then we'll have enough parameters to really represent them.\nSpeaker E: That also has a huge, and it's very online.\nSpeaker E: I feel that I feel that the online conversation that we've over-bad, online conversation, has to do some extrusion.\nSpeaker E: And take away the online conversation, online conversation, so that we can talk about that.\nSpeaker G: So doing both is not right.\nSpeaker E: That's not right, but if you fill in online conversation, then it might not be necessary to use the word that's going to be.\nSpeaker D: I just sort of have a feeling, but yeah. I mean, I think it's true that the OGI-folk found that using LDA Rasta, which is a kind of log-rasta, it's just that they have, I mean, it's done the log domain as Eric Allen, and it's just that they've just trained up, right?\nSpeaker D: That that benefited from online normalization.\nSpeaker D: So they did, at least in their case, it did seem to be somewhat complimentary.\nSpeaker D: So will it be in our case where we're using the neural net? I mean, they were not using the neural net.\nSpeaker D: I don't know.\nSpeaker D: Okay, so the other things you have here are trying to improve results from a single, yeah, make stuff better.\nSpeaker D: Basically, yeah, and CPU memory issues, yeah, we've just sort of ignoring that.\nSpeaker H: Yeah, so I don't know.\nSpeaker H: Yeah, but I have to address this problem of CPU memory?\nSpeaker D: Well, I think my impression, you folks have been looking at this more than me, but my impression was that there was a strict constraint on the delay.\nSpeaker D: But beyond that, it was kind of that using less memory was better, using less CPU was better, something like that.\nSpeaker F: Yeah, but yeah.\nSpeaker H: So yeah, but we don't know. We have to get some reference point to where we, what's a reasonable number?\nSpeaker H: Perhaps because it's too late or large.\nSpeaker D: Well, I don't think we're completely off the wall. I mean, I think that if we have, I mean, the ultimate fallback that we could do, if we find, I mean, we find that we're not really going to worry about the MLP.\nSpeaker D: Yeah, if the MLP ultimately, if they're all said and done, doesn't it really help? And we have it in. If the MLP does, we find help us enough in some conditions.\nSpeaker D: We might even have more than one MLP. We could simply say that that is done on the server.\nSpeaker D: And it's, we do the other manipulations that we're doing before that.\nSpeaker D: So I think that's, that's okay.\nSpeaker D: So I think the key thing was this plug-in doji.\nSpeaker D: What are they, what are they going to be working on? We know what they're going to be working on while we take their features.\nSpeaker H: They're starting to work on some kind of multiband. So that was pretty bad. Sunil, what was he doing?\nSpeaker H: Sunil?\nSpeaker H: Yeah. He was doing something new.\nSpeaker I: I don't think so.\nSpeaker I: I'm trying to tune.\nSpeaker F: What? Networks?\nSpeaker H: I think they were also mainly, well, working a little bit of new things like network and multiband, but mainly trying to tune their system as it is now.\nSpeaker H: Just trying to get the best from this architecture.\nSpeaker D: Okay, so I guess the way it would work is that you get, there be some point where you say, okay, this is their version one or whatever.\nSpeaker D: And we get these VAD labels and features and so forth for all these test sets from them.\nSpeaker D: And then that's what we work with.\nSpeaker D: We have a certain level, we try to improve it with this other path.\nSpeaker D: And then when it gets to be January, at some point, we say, okay, we have shown that we can improve this in this way.\nSpeaker D: So now, what's your newest version?\nSpeaker D: Maybe they'll have something that's better and then we combine it.\nSpeaker D: This is always hard.\nSpeaker D: I mean, I used to work with folks who were trying to improve a good HMEM system with the neural net system.\nSpeaker D: And it was a common problem that you, oh, and this is actually a true not just for neural nets, but just for the general people working with rescoring and best lists or lattice systems.\nSpeaker D: And then you get something from the other side at one point and you work really hard on making it better with rescoring.\nSpeaker D: But they're working really hard too.\nSpeaker D: So by the time you have improved their score, they have also improved their score.\nSpeaker D: And now there isn't any difference because the other.\nSpeaker D: So I guess at some point we'll have to, I don't know, I think we're integrated a little more tightly than happens.\nSpeaker D: A lot of those cases, I think, at the moment they say that they have a better thing.\nSpeaker D: What takes all the time here is that we're trying so many things, presumably, in a day we could turn around taking a new set of things from them and rescoring.\nSpeaker D: Well, OK, I think this is good. I think that the most wide open thing is the issues about the different trainings, training targets and devices and so forth.\nSpeaker H: So we can forget combining multiple features and MSG perhaps or focus more on the targets and on the training data.\nSpeaker D: Yeah, I think for right now, I really like MSG. I think that one of the things I like about it is it has such different temporal properties.\nSpeaker D: And I think that there is ultimately a really good potential for bringing in things with different temporal properties.\nSpeaker D: But we only have limited time and there's a lot of other things we have to look at. It seems like much more core questions are issues about the training set and the training targets and fitting in what we're doing with what they're doing.\nSpeaker D: But with limited time, yeah, I think we have to start cutting down. So I think so. And then, you know, once we have been going through this process and trying many different things, I would imagine certain things come up that you are curious about that you're not getting to.\nSpeaker D: So when that does settles from the evaluation, I think that would be time to go back and take whatever intrigued you most, you know, got your most interested and work with it for the next round.\nSpeaker D: As you can tell from these numbers, nothing that any of us are going to do is actually completely solve the problems.\nSpeaker D: That's still exciting to do. Very, very quiet. Well, I figured that out.\nSpeaker D: What were you involved in in this primarily?\nSpeaker B: Helping out preparing, well, they've been kind of running all the experiments and stuff. I've been doing some work on the preparing all the data for them to train and to test out.\nSpeaker B: Yeah, right now I'm focusing mainly on this final project I'm working on Jordan's class. What's that?\nSpeaker B: I'm trying to, so there was a paper in ICSLP about this multi-band belief in that structure. This guy did basically use two HMMs with a dependency arrow between the two HMMs.\nSpeaker B: I'm going to try coupling them instead of having the arrow that flows from one subband to another subband, trying having the arrows go both ways.\nSpeaker B: I'm just going to see if that better models AC-creening in any way or... No? Yeah. That sounds interesting.\nSpeaker D: Anything to do? No. Okay. Sound partner in the meeting. We got laughed out of them. Okay, everyone must contribute to our sound files here.\nSpeaker D: Okay, so speaking of which, if we don't have anything else, we're happy with where we are. No, no, no, no, we're going. I think so. Yeah. Yeah. Do you happy? Yeah. Okay. Everyone, should be happy.\nSpeaker D: You only have to be happy, you're almost done. Yeah. Okay.\nSpeaker A: Actually, I should mention about the Linux machine in Sweden. Yeah. So it looks like the neural net tools are installed there. And Dan Ellis, I believe, knows something about using that machine.\nSpeaker H: So if people are interested in getting jobs running on that, maybe I could help with that. Yeah, but I don't know if we really need a lot of machines.\nSpeaker D: But we could start computing another niche table. Well, yeah, I think we want a different table, please. Yeah, sure.\nSpeaker D: There's some different things that we're going to get at now. But yeah. So as far as you can tell, you're actually okay on CPU for training and so on.\nSpeaker H: Yeah, I think so. Well, more is always better. But I don't think we have to train a lot of networks. You know, we just select what works fine.\nSpeaker D: I'm okay. I'm okay. I'm okay. I'm disc.\nSpeaker H: Okay. Yeah. Well, sometimes we have some problems. But they're correct. Yeah. We're starting to script busy. Yeah. Yeah.\nSpeaker D: That's okay. Alright. So, we didn't give a channel out for you. You don't have to read the digits, but rest of it.\nSpeaker D: Is it not? Well, I think I won't touch anything because I'm afraid of making the driver crash.\nSpeaker D: It seems to do. Okay. So, I'll start off the connect my battery slow. Let's hope it works. Maybe you should go first.\nSpeaker H: I'm reading transcript. 2571 to 590. 1943. 263.\nSpeaker H: 24. 5. 7. 1. 8. 3067. 4. 4. 9. 3048. 4. 9. 9. 8. 4. 0. 1. 2. 0. 2. 8. 6. 9. 1. 4. 1. 5. 6. 9. 0. 7. 5. 4. 6. 6. 7. 8.\nSpeaker H: 9. 00. 1. 2. 7. 3. 0. 7. 3.\nSpeaker D: Transcript. 2. It's battery. Going down too. Oh, okay. I want you to go next.\nSpeaker I: Transcript number 2511-2530. 1905. 004. 7. 1307. 2637. 2554.\nSpeaker I: 3314. 354. 3314. 7. 0. 5. 6. 7. 0. 7. 9. 1. 0. 4. 0. 4. 7. 1. 7. 5. 2. 0. 4. 0.\nSpeaker I: 234. 0. 5. 5. 2. 6. 8. 6. 8. 9. 0. 9. 0. 7. 6. 4. 7. 8. 6. 1. 9.\nSpeaker D: Transcript. 2531-2550. 0. 0. 0. 0. 1035217. 311-0656. 4336-463.\nSpeaker D: 5811-244. 699-7801. 8. 00. 9. 0. 1. 6. 0. 1. 2. 3. 8. 1. 2. 6. 0. 4. 3. 9. 4. 9. 5. 6. 0. 4. 8. 3. 8. 2. 4. 2. 9. 3. 3. 3. 0. 5.\nSpeaker A: I'm reading transcript 2551-2570. 0. 0. 6. 6. 9. 5. 4. 7. 1. 2. 3. 0. 0. 5. 1. 6. 2. 5. 7. 7. 7. 7. 8. 9. 0. 0. 0. 5. 0. 8. 9. 0. 0. 8. 9. 0. 0. 0. 5. 0. 8. 9. 0. 0. 8. 9. 0. 0. 0. 8. 9. 9. 0. 0. 8. 9. 0. 0. 0. 8. 9. 0. 0. 0.\nSpeaker A: 6 2 1 3 0 5 3 5 7 2 3 6 0 4 3 5 6 9 9 6 7 7 0 6 8 0 6 5 0 0 3 8 5 4 0 2\nSpeaker B: I'm reading transcript 2491-2510 on Channel 2 8 9 0 7 9 0 2 1 4 0 6 2 6 3 1 3 8 2 205 8 4 4 5 6 0 8 8 1 2 0 6 9 6 2 0 4 4 3 0 7 3 0 7 1 2 0 5 4 1 5 1 5 6 6 9 7 5 9 1 3 8 7 2 9 7 2 6\nSpeaker D: I guess we can turn off our microphones.\nNone: I am always well gathered\n"}, {"meeting_id": "Bro011", "summary": "The meeting began with the team catching up the professor, who had missed the previous meeting. Then, the team moved onto talking about echo cancelling techniques. They were trying to get the voice of other people from the primary person's microphone. Then, the team discussed how they could improve the performance on the Aurora task by playing around with the model. The team ended the meeting with various comments on different topics.", "dialogue": "None: I am Guess so radio to\nSpeaker E: All right\nSpeaker D: It works really well people say this thing Everybody's on yeah So you guys had a meeting with Henik which I unfortunately had to miss And And I guess Chuck you weren't there either so I was there or you were there with Henik yeah, so everybody knows what happened except me Maybe somebody should tell me\nNone: All right Well first we discussed about some of the points That I was addressing in the mail I sent last week So yeah But well the don't sampling problem Yeah And about the feet of the filters\nSpeaker D: So what was the time sampling problem again I think\nSpeaker F: Is that the fact that there is no low pass filtering before the don't sample well there is because there is a LDA filtering but that's perhaps not the best\nSpeaker D: But it depends but it's a Consicure requesting this yeah so you could do a you could do a structure one maybe yeah so we discussed about this about\nSpeaker F: Was there any conclusion about that? Or try it yeah I guess\nSpeaker D: Yeah so again this is the this is the down sampling of the feature vector stream And yeah I guess the LDA filters they were doing do have It's so that the feature vectors are calculated every 10 milliseconds so The question is how far down they are 50 50 hertz Sorry at 25 hertz and so down sampling right too So does anybody know the frequency characteristic is\nSpeaker F: We don't have yet So yeah we should have a look for that perhaps yeah the modulation spectra So there is this there is the length of the filters So the idea of trying to find filters with shorter delays We started to work with this And the third point was the yeah the online normalization where Well the recursion for the min estimation is a filter with some kind of delay Yeah that's not taken into account right now Yeah and there again for this conclusion the fin acquiesce\nSpeaker D: We can try it but try try what?\nSpeaker F: So try to take into account the delay of the recursion for the min estimation Okay And this we've not worked on this yet And so while discussing about these LDA filters some issues appeared Like well the fact that if we look at the frequency response of these filters it's Well we don't know really what's the important part in the frequency response And there is the fact that in the very low frequency these filters don't really remove a lot Compared to the standard Rasta filter And that's probably the reason why yeah online normalization helps because Yeah to remove this maze Yeah but perhaps everything could be in the filter I mean the min normalization Yeah so Yeah so basically that's what we discussed about Good things to do also generally good stuff to do for the research And this was this LDA tuning perhaps and in fact proposed again to these traps\nSpeaker D: Okay I mean I guess the key thing for me is figuring out how to better coordinate between two sides Because I was talking with Henrik about it later And the sense sort of that neither group of people wanted to bother the other group too much And I don't think anybody is closed in and they're thinking or I'm willing to talk about things But I think you were sort of waiting for them to tell you that they had something for you And then they expected they would do certain things and they didn't want to bother you And we were sort of waiting for you and we ended up with this thing Where they were filling up all of the possible latency for themselves And they just hadn't thought of that Yeah I mean it's true that maybe no one really thought about that this latency thing would be such a strict issue\nSpeaker F: And yeah I don't know what happened really but I guess it's also the time constraints Because we discussed about this problem and they told us well we will do all that's possible to have enough space for network But then the wraps never ends But the rest was a problem of communication So we will try to talk more\nSpeaker D: So there's... All right well maybe we should just... I mean you're busy, other than that you folks are busy doing all the things that you're trying that we talked about before And machines are busy and you're busy Okay well let's... I mean I think that as we said before one of the things that we're imagining is that there will be in system There will be something to explicitly do something about noise in addition to the other things that we're talking about And that's probably the best thing to do And there was that one email that said that it sounded like things looked very promising up there in terms of...\nSpeaker D: I think they were using... approach or something in addition to... they're doing some noise removal thing\nSpeaker F: So yeah we will start to do this also So Carmen is just looking at the Ericsson code\nSpeaker A: I modified it, modified it I studied there is some code in the world To take only the first step the spectral substitution And we have some... the feature for Italian database And we will try with this feature with the filter to find the result But we have the result under this moment Yeah, sure We are working in this also and maybe try another type of spectrarset ratio\nSpeaker D: When you say you don't have a result yet you mean it's just that it's in focus or it finished and it didn't get a good result\nSpeaker A: No, no, we have... we have the experiment Or we have the feature, the feature but... this experiment... we have not made this experiment Maybe it would be good to resend all the resend... we don't know\nSpeaker D: Yeah So I suggest actually now we move on and tear what's happening in another areas like... what's happening with your investigations Oh, I don't know about echoes and so on\nSpeaker B: I haven't started running the test yet, I'm meeting with Adam today And he's going to show me the scripts he has for running recognition on the meeting recorder digits I also haven't got the code yet I haven't asked for the quiz code yet So that's how Adam Down on the pieces And I don't really understand what he's doing yet It sounded like the channel normalization part of his thesis was done in a bit of... I don't know what the word is A bit of a rough way, it sounded like he... he wasn't really fleshed out and maybe he did something that was interesting for the test situation But I'm not sure if it's what I'd want to use So I have to read it more, I don't really understand what he's doing yet\nSpeaker D: I haven't read it in a while so it's not going to be too much help unless I read it again So...\nSpeaker D: And then you're also going to be doing this echo cancelling between the question and the...\nSpeaker D: We're calling cheating experiment\nSpeaker B: I'm hoping... I'm hoping you ask me what to do\nSpeaker D: Oh, okay Delegate It's good to be Delegate\nSpeaker B: I think he's at least planning to do it for the close mic crosstalk And so maybe I can just take the other set of he has and use it\nSpeaker D: Great, great Yeah, actually he should...\nSpeaker D: Like maybe it's the analysis going to be doing a different cancellation One of the things that people working in the meeting task want to get at is We'd like to have cleaner close mic recordings So this is especially true for the lapel but even for the close mic cases We'd like to be able to have other sounds from other people on so forth removed So when someone isn't speaking you'd like to part whether or not speaking to actually So what they're talking about doing is using echo cancellation like techniques It's not the way I go but just taking the input from other mics and using an adaptive filtering approach to remove the effect of that other speech So what was it? There was some point where Eric or somebody was speaking And he had lots of silence in his channel and I was saying something to somebody else Which is in the background and it was not... it was recognizing my words which was the background speech On the close mic\nSpeaker C: Oh, we talked about yesterday Yeah, that was actually in my eyes I was wearing the lapel and you were sitting next to me Yeah, and I only said one thing But you were talking and it was picking up all your words\nSpeaker D: Yeah, so they were like clean channels And for that purpose, so let's pull it out So I think that's something that somebody was working with That's going to work out So Right And I don't know if we've talked lately about the plans You're developing and we talked about this morning If we talked about that last week or not But you just do work with Ryze and probably I guess I'm just pointing So\nNone: We're going to Next day It's on the way It's really hot What about the stuff that Miriam has been doing?\nSpeaker C: And Sean, yeah.\nSpeaker C: Yeah.\nSpeaker B: What's good about you and I think that's good.\nSpeaker C: So they're training up nets to try to recognize these acoustic features, I see.\nSpeaker D: But that's a certainly relevant study and you know what are the features that they're finding.\nSpeaker D: We have this problem with the overloading of the term features.\nSpeaker D: What are the variables we're calling this one?\nSpeaker D: What are the variables that they're finding useful?\nSpeaker C: And their targets are based on canonical mappings of phones to acoustic.\nSpeaker D: Right. And there's certainly one thing to do and we're going to try and do something more fine than that.\nSpeaker D: So I guess I was trying to remember some of the things we were saying.\nSpeaker D: Yeah. So some issues we were talking about was just getting a good handle on what good features are.\nSpeaker C: What did Larry Saul use for this sonarant detector?\nSpeaker C: How did he do that?\nSpeaker C: What was his detector?\nSpeaker E: Yeah. It was a tonnage.\nNone: It was variable.\nSpeaker C: It was a measure of the hand to the left.\nSpeaker C: A thing.\nSpeaker E: Actually, it was a measure of the correlation.\nSpeaker E: So how did he combine all these features?\nSpeaker C: What classifier did he use?\nSpeaker C: What did he use for this test?\nSpeaker D: What are the variables that you use?\nSpeaker D: You combine them using the software or something more complicated.\nSpeaker D: And then the other thing was where to get the targets from.\nSpeaker D: The initial thing is just the obvious that we're discussing is starting off with phone labels from somewhere and then doing the transformation.\nSpeaker D: But then the other thing is to do something better.\nSpeaker D: What did you tell us about this database?\nSpeaker D: I don't know.\nSpeaker C: I guess if you had people who had like, like, a computer, a computer, a computer.\nSpeaker C: Pierce Tongues.\nSpeaker D: You just mounted it to that and they wouldn't even notice.\nSpeaker D: Welled it.\nSpeaker C: That's right.\nSpeaker D: Okay.\nSpeaker C: There's a bunch of data around people who have done studies like that way, way back.\nSpeaker C: I can't remember where Wisconsin or someplace that used to have a big database.\nSpeaker C: Remember there was this guy at AT&T Randolph?\nSpeaker C: Researcher at AT&T, a while back that was studying, trying to do speech recognition from these kinds of features.\nSpeaker D: Mark Randolph.\nSpeaker C: Oh, is he?\nSpeaker F: I can't remember exactly what he was using now.\nSpeaker C: I just remember it had to do with, you know, positional parameters and trying to, you know, speech recognition based on them.\nSpeaker D: The only hesitation I had about it since I haven't seen the data is it sounds like it's continuous variables and a bunch of them.\nSpeaker D: And so I don't know how complicated it is to go from there, but you really want these binary labels just a few of them.\nSpeaker D: And maybe there's a trivial mapping.\nSpeaker D: I worry a little bit that this is a research project in itself.\nSpeaker D: Because if you did something instead that like having some manual annotation by, you know, agristic students, this would, there'd be a limited set of things that you could do as per our discussions with John before.\nSpeaker D: But the things that you could do like anxiety and voicing a couple other things, you probably could do reasonably well.\nSpeaker D: And then it would really be this binary variable.\nSpeaker D: Of course, then that's the other question is do you want binary variables?\nSpeaker D: So the other thing you could do is boot trying to get those binary variables and take the continuous variables from the data itself there.\nSpeaker C: But I'm not sure.\nSpeaker C: Could you cluster the, just do some kind of clustering?\nSpeaker C: Yeah, then I'm up into different categories.\nSpeaker D: So anyway, that's another whole direction that could be looked at.\nSpeaker D: I mean, in general, it's going to be for new data that you look at, it's going to be hidden variable because we're not going to get everybody sitting in these meetings to where the pellets and.\nSpeaker C: So you're talking about using that data to get instead of using canonical mappings of phones.\nSpeaker C: So you use that data to give you sort of what the true mappings are for each phone.\nSpeaker D: Yeah, where this fits into the rest in my mind, I guess, is that we're looking at different ways that we can combine different kinds of printed representations in order to get robustness and do difficult or even typical conditions.\nSpeaker D: And part of it, this robustness seems to come from multi-stream and multi-band sorts of things and Saul seems to have a reasonable way of looking at it, at least for one particular toy feature.\nSpeaker D: The question is, can we learn from that to change some of the other methods we have since any one of the things that's nice about what he had, I thought was that it, the decision about how strongly trained the different pieces is based on a reasonable criterion with it variables rather than just assuming that you should train every detector with equal.\nSpeaker D: With equal strength towards it being this phone or that phone.\nSpeaker D: So he's got these.\nSpeaker D: The ends between these different features.\nSpeaker D: It's a soft end, I guess, but in principle, you want to get a strong concurrence of all the different things that indicate something. And then he oars across the different soft oars across the different multi-band channels.\nSpeaker D: And the weight, the target for the training of the AND ended things is something that's kept as a hit variable and is learned with the app.\nSpeaker D: Whereas what we were doing is taking the phone target and then just back propagating from that, which means that it could be, for instance, that for a particular point in the data, you don't want to train a particular band, train the detector for a particular band, you want to ignore that band.\nSpeaker D: That's a band is a noisy measure. And we're still going to try to train it up in our scheme. We're going to try to train it up to do as well as it can at predicting.\nSpeaker D: Maybe that's not the right thing to do.\nSpeaker C: So he doesn't have to have truth marks?\nSpeaker D: Well, at the talent, he has to know where it's sonarant. But what he's not training up, what he doesn't depend on his truth is, I guess, one way of describing it would be, if a sound is sonarant, is it sonarant in this band?\nSpeaker D: It's hard to even answer that, what you really mean is that the whole sound is sonarant. So then it comes down to what extent should you make use of information from particular band towards making your decision?\nSpeaker D: I see. And we're making, in a sense, sort of this hard decision that you should use everything with equal strength. And because in the ideal case, we would be going for posterior probabilities, if we had enough data to really get posterior probabilities.\nSpeaker D: And if we also had enough data so that it was representative of the test data, then we would, in fact, be doing the right thing to train everything as hard as we can.\nSpeaker D: But this is something that's more built up along an idea of robustness from the beginning, so you don't necessarily want to train everything up towards the...\nSpeaker C: So where did he get his high level targets about what sonarant and what's not?\nSpeaker E: From canonical mappings, from a person, then it's unclear. Using timet? Using timet, right?\nSpeaker E: Yeah, and then he does some fine tuning.\nSpeaker D: I mean, we have a kind of iterative training because we do this embedded with Ruby. So there is something that's adjusted based on the data.\nNone: I think it's seemed like quite the same. Because then whatever that line is, is that wrong?\nNone: No, bands. Well, that's quite...\nNone: I'm trying to do something. That'll be a little more like it.\nSpeaker D: But it's still quite the same because then it's a target based on what you'd say the sound begins in a particular band where he's not labeling per se.\nSpeaker D: It might be closer, I guess, if we did a soft target embedded training, we'd have done a few times before we did a forward calculations to get the gammas and train our bells.\nSpeaker C: What's next? I can say a little bit about stuff I've been playing with.\nSpeaker C: So I wanted to do this experiment to see what happens if we try to improve the performance of the backend recognizer for the Aurora task and see how that affects things.\nSpeaker C: I think I sent around last week a plan I had for an experiment, this matrix, where I would take the original system.\nSpeaker C: So there's the original system trained on the Mel Kepstrow features and then optimize the HTK system and run that again. So look at the difference there.\nSpeaker C: And then do the same thing for the XE-OGI front end.\nSpeaker E: Which test was this?\nSpeaker C: If I look at it, I'm looking at the Italian right now. So as far as I've gotten, I've been able to go through from beginning to end the full HTK system for the Italian data and got the same results that Stefan had.\nSpeaker C: So I started looking at the point where I want to know what should I change in the HTK backend in order to try to improve it.\nSpeaker C: One of the first things I thought of was the fact that they use the same number of states for all of the models. And so I went online and I found a pronunciation dictionary for Italian digits and just looked at the number of phones in each one of the digits.\nSpeaker C: So the canonical way of setting up an HMM system is that you use three states per phone. And so then the total number of states for a word would just be the number of phones times three.\nSpeaker C: And so when I did that for the Italian digits, I got a number of states ranging on a low end from nine to the high end 18.\nSpeaker C: Now you have to really add two to that because in HTK there's an initial null and a final null. So when they use models that have 18 states, they're really 16 states. They've got this initial and final null states.\nSpeaker C: And so their guess of 18 states seems to be pretty well matched to the two longest words, the Italian digits, the four and five, which according to my sort of off the cuff calculation should have 18 states each. And so they had 16. So that's pretty close.\nSpeaker C: But for the most of the words are much shorter. So the majority of them want to have nine states. And so there's sort of twice as long. So my guess. And then if you, I printed out a confusion matrix for the well matched case.\nSpeaker C: And it turns out that the longest words are actually the ones that do the best. So my guess about what's happening is that if you assume a fixed the same amount of training data for each of these digits and a fixed length model for all of them.\nSpeaker C: But the actual words for some of them are half as long. You really have half as much training data for those models.\nSpeaker C: Because if you have a long word and you're training it to 18 states, you've got, you've got the same number of Gaussian's. You've got a train in each case. But for the shorter words, you know, the total number of frames is actually half as many.\nSpeaker C: So it could be that, you know, for the short words, there's because you have so many states, you just don't have enough data to train all those Gaussian's. So I'm going to try to create more word specific prototype HMMs to start training from.\nSpeaker D: Yeah, I mean, it's not at all uncommon. You do where send long words on short words and long words anyway, just because you're accumulating more evidence.\nSpeaker C: Yeah, so I'll, I'll, the next experiment I'm going to try is to just, you know, create models that seem to be more matched to my guess about how long they should be.\nSpeaker C: And as part of that, I wanted to see sort of how the, how these models were coming out, you know, when we train up the model for one, which wants to have nine states, you know, what is the, what are the transition probabilities look like in the self loops look like in those models.\nSpeaker C: And so I talked to Andreas and he explained to me how you can calculate the expected duration of an HMM just by looking at the transition matrix.\nSpeaker C: And so I wrote a little MATLAB script that calculates that. And so I'm going to sort of print those out for each of the words to see what's happening, you know, how these models are training up, you know, the long ones versus the short ones.\nSpeaker C: I did, quickly I did the silence model and, and that's coming out with about 1.2 seconds is its average duration. And the silence model is the one that's used at the beginning and the end of each of the string of digits.\nSpeaker C: Lots of silence.\nSpeaker C: Yeah, yeah. And so the SP model, which is what they put in between digits, I haven't calculated that for that one yet.\nSpeaker C: So they basically they're their model for a whole digit string is silence digit, SP digit, SP, blah blah and then silence at the end.\nSpeaker C: So are the SPs optional?\nSpeaker C: I have to look at that, but I'm not sure that they are. Now the one thing about the SP model is really it only has a single emitting state to it.\nSpeaker C: So if it's not optional, you know, it's it's not going to hurt a whole lot. And it's tied to the center state of the silence.\nSpeaker C: It's not a little, it doesn't require some training data, it just shares that state.\nSpeaker C: So I mean it's pretty good the way that they have it set up, but, so I want to put that a little bit more curious about looking at, you know, how these models have trained and looking at the expected durations models.\nSpeaker C: And I want to compare that into the well matched case, the unmatched case and see if you can get an idea just from looking at the durations of these models, you know, what's happening.\nSpeaker D: Yeah, I'm going to think, yeah, I'm going to think, yeah, I'm going to think, yeah, it's good to, so I'm not doing anything really tricky.\nSpeaker D: Not doing anything really finely. The premise is kind of if you have a good person look at this for two weeks and what you come up with.\nSpeaker C: And he nick when I told him about this, he had an interesting point and that was the final models that they ended up training up have, I think probably something on the order of six Gaussian per state.\nSpeaker C: So they're fairly, you know, hefty models and he was saying that, well, probably in a real application, you wouldn't have enough compute to handle models that are very big or complicated.\nSpeaker C: So in fact, what we may want are simpler models and compare how they perform to that.\nSpeaker C: But, you know, it depends on what the actual application is and it's really hard to know what your limits are in terms of how many Gaussian you can have.\nSpeaker D: Right. And at the moment, that's not the limitation. So, I mean, what I thought you were going to say, but what I was thinking was where did six come from?\nSpeaker D: Probably came from the same place, 18 came from.\nSpeaker D: Yeah. So that's another parameter, right?\nSpeaker D: Yeah. Maybe, you really want to do one thing.\nSpeaker C: If I start reducing the number of states for some of these shorter models, that's going to reduce the total number of Gaussian. So in a sense, it will be a simpler system.\nSpeaker D: Yeah. But I think right now, again, the idea is doing just very simple things, how much better can you make it?\nSpeaker D: And since there are only simple things, there's nothing that you're going to do that is going to blow up the amount of computation. So if you found that nine was better than six, that would be okay.\nSpeaker C: I really wasn't even going to play with that part of the system yet. I was just going to change the work with the models.\nSpeaker C: Yeah. Just look at the models and see what happened.\nSpeaker C: Yeah.\nSpeaker C: So.\nSpeaker D: Cool. Okay. So what's I guess your plan for you guys playing for the next next week is just continue on these same things you've been talking about.\nSpeaker F: Yeah.\nSpeaker F: I guess you can try to have some kind of new baseline for next week, perhaps with all these minor things.\nSpeaker F: And then you modify it, and then do other things play with the spectrospection and retry to a MFG.\nSpeaker D: Yeah. Yeah. We have a big list.\nSpeaker D: You have a big list of things to do.\nSpeaker D: So that's good. I think that after all of this confusion settles down. At some point a little later next year, there will be some sort of standard.\nSpeaker D: We'll get out there and hopefully a little have some effect from something that has been done by a group of people.\nSpeaker D: But even if it doesn't, there's going to be standards after that.\nSpeaker C: Does anybody know how to run MATLAB sort of in batch mode like you send it to commands to run it? Is it possible to do that?\nSpeaker E: I think Mike tried it.\nSpeaker E: Yeah.\nSpeaker E: And he said it was impossible. So he went to active.\nSpeaker E: I thought it was the Unix clone of MATLAB.\nSpeaker E: Which you can dodge.\nSpeaker C: Okay. Great. Thank you.\nSpeaker C: Yeah.\nSpeaker C: I was going crazy trying to do that.\nSpeaker F: Yeah. What is that?\nSpeaker E: It's a free software.\nSpeaker E: I think we have it here running somewhere.\nSpeaker F: Great.\nSpeaker F: And it does the same syntax.\nSpeaker F: I think it's a little behind it.\nSpeaker E: It's a little behind in that MATLAB went to these like, like, half-sales and you can implement object-oriented type things.\nSpeaker E: Active doesn't do that yet.\nSpeaker E: I think you've, like, active MATLAB or point something.\nSpeaker C: If it'll do a lot of the basic matrix and vector stuff.\nSpeaker C: That's perfect.\nSpeaker C: Great.\nSpeaker D: Okay. I guess we're done.\nSpeaker D: I don't find it.\n"}, {"meeting_id": "ES2014a", "summary": "Project Manager gave the team members a brief introduction of the new remote control project, then the team started an ice breaking game by drawing their favourite animals. The expected selling price of the remote control is 25 Euros and the production costs would be limited to twelve and a half Euro per unit. Marketing plan would depend on overhead costs. The team agreed that the remote needs to be robust, and they wanted it to be a device for all with accessible buttons. Project Manager suggested it to be only a television remote control instead of a multi-functional one.", "dialogue": "None: Can you help me?\nSpeaker C: No.\nNone: You okay?\nSpeaker A: You okay?\nSpeaker A: Okay?\nSpeaker A: Is that okay?\nNone: Come on.\nSpeaker A: Okay?\nSpeaker A: Okay.\nSpeaker A: All right, so start of the first meeting.\nSpeaker A: All right, so agenda of the first meeting.\nSpeaker A: Are we...\nSpeaker A: We have 25 minutes for this meeting.\nSpeaker A: We...\nSpeaker A: Are to get acquainted.\nSpeaker A: So, does everyone want to say who they are?\nSpeaker B: And that's sensible.\nSpeaker C: I'm Robin. I'm a marketing manager.\nSpeaker A: I'm Lisa. I'm a music designer.\nSpeaker A: I'm Nick. I'm the industrial designer.\nSpeaker A: And I'm Amostar. I'm on the project leader.\nSpeaker A: Project plan.\nSpeaker A: So, does anyone have any thoughts?\nSpeaker A: As to the tool training that is required.\nSpeaker C: I'm nice. I'll show you what you mean by tool training.\nSpeaker A: Neither am I.\nSpeaker A: Oh, I see. So, we shouldn't really be...\nSpeaker A: All right. Okay.\nSpeaker A: So, we have the project team, which is to basically to come up with a new remote control device.\nSpeaker A: We have...\nSpeaker A: The starting base was the original.\nSpeaker A: Which has been an existence now for a period of time.\nSpeaker A: And our idea is to make the new remote control device more user-friendly than the previous one.\nSpeaker A: And to be trendy or to be with it, and therefore to get a bigger market share and bigger audience.\nSpeaker A: So, the method of doing this is split up.\nSpeaker A: As you can see into the functional design, the conceptual design, and the detailed design.\nSpeaker A: So, in each of these phases, we'll basically be handing over to yourselves the designers of this...\nSpeaker A: This device.\nSpeaker A: And having meetings so that we can, during the course of the day, come up with a better implement than we had before.\nSpeaker A: And therefore have a successful conclusion to the D.\nSpeaker A: And you'll be doing various designs throughout the day to meet this end.\nSpeaker A: So, we've got two chain.\nSpeaker A: Try out whiteboard.\nSpeaker A: So, we'll...\nSpeaker A: Right. So, basically everyone's to supposedly draw their favourite animal over on the whiteboard over there.\nSpeaker A: I guess this is...\nSpeaker A: Make sure the whiteboard works.\nSpeaker A: I don't know who wishes to go first.\nSpeaker A: You wish to go...\nSpeaker A: Have a first bash at whatever.\nSpeaker C: Um...\nSpeaker A: Uh...\nSpeaker A: Good. I've got pocket study.\nSpeaker A: But now you move out from the microphone, and the camera...\nSpeaker A: What about you? That's right now, do you think?\nSpeaker A: I would guess so.\nSpeaker D: You've asked the microphone.\nSpeaker A: Technical problems. I mean, you designers are meant to come up with it.\nSpeaker A: So...\nSpeaker A: Oh, no.\nSpeaker C: Okay.\nSpeaker C: I think that I would have to see that my favourite animal is the cat.\nSpeaker C: What's my look at there?\nSpeaker C: Um...\nSpeaker C: And this would be because the very independent, the very intelligent, the compote the dogs maybe.\nSpeaker C: And it can be very, very affectionate. Some people don't think so, but are not very affectionate cats.\nSpeaker C: Um...\nSpeaker C: And they come back to themselves.\nNone: Right.\nSpeaker A: Next.\nSpeaker C: Shall I look like that? Oh, actually.\nSpeaker A: I don't see. There's any need to. There's plenty of space.\nSpeaker A: I mean...\nSpeaker B: Whatever. I'm a hominatry.\nSpeaker D: Exactly.\nSpeaker D: So I can get the casserole at this time.\nSpeaker A: We've had more time to prepare over this site, so we've all stuck our bits in pieces in the pockets.\nSpeaker D: I didn't think that.\nSpeaker D: Okay.\nSpeaker D: Three pencered.\nSpeaker D: I'll try the red pen.\nSpeaker D: Okay.\nSpeaker D: Um...\nSpeaker D: I'm gonna go for the bath.\nSpeaker D: Wait, something else is draw very well.\nSpeaker D: I'll have a bath.\nSpeaker A: You get marks for artistic impression.\nSpeaker D: Oh, I lost it there.\nSpeaker D: I have to look something like a...\nSpeaker A: Um... So you're just doing the face.\nSpeaker A: I'll go for a...\nSpeaker D: A small, small, bare button.\nSpeaker D: Um...\nSpeaker D: And I like my animal that looks nothing like a bath because, um...\nSpeaker D: Has no, maybe because there's so many cartoon characters who've made a bath.\nSpeaker B: Right.\nSpeaker B: Hello. Um...\nSpeaker B: I'm gonna go for the dog.\nSpeaker B: I'm gonna draw one badly as well.\nSpeaker B: Uh...\nSpeaker B: Looks like he's going to get ducks and does something.\nSpeaker C: That's very good.\nSpeaker C: Right.\nSpeaker B: There's my dog.\nSpeaker B: Um... I like dogs because they're very loyal and they're always happy.\nSpeaker B: So whenever you're feeling sort of a bit...\nSpeaker B: Down or tired, they're always coming up and they're always, um, quite excited.\nSpeaker B: So, um...\nSpeaker B: You can always have a lot of fun with the dog and they're also good for exercises for you so they get out and they never get tired.\nSpeaker B: And when they're tired, they're quite cute as well.\nSpeaker B: Okay.\nSpeaker B: That's why I like dogs.\nSpeaker A: Alright.\nSpeaker A: Well, I've not actually had too many pets over my time because, to be honest with you, uh...\nSpeaker A: Not too keen on them.\nSpeaker A: Anyway, not to worry.\nSpeaker A: So what my daughters have got at the moment is they've got a few fish.\nSpeaker A: And so hopefully, um, won't prove too difficult to draw.\nSpeaker A: Uh...\nNone: Mm-hmm.\nNone: Mm-hmm.\nSpeaker A: As you can see, my artistic work is useless as well.\nSpeaker A: Anyway, um...\nSpeaker A: One of the best things about fish is that they don't really take, uh, too much looking after because, uh, with most of the animals, if you're going away in holiday or whatever, you've got to spend money or get a friend or whatever to look after them for you.\nSpeaker A: Whereas if you've got fish, you just gotta put the food in a dripper feed, which feeds them over a couple of weeks at your way and, uh...\nSpeaker A: Change the water every couple of months and buying a few plants, so other than the fact they keep dying, uh, fish are not... are reasonable pets in that, uh, they're low maintenance.\nSpeaker A: Right.\nSpeaker A: Okay.\nSpeaker A: Or still, what's...\nSpeaker A: Right, okay.\nSpeaker A: So, work has been done on, uh...\nSpeaker A: This project whereby, um...\nSpeaker A: 25 euros is the expected, uh...\nSpeaker A: Sound price?\nSpeaker A: Information has come from our marketing manager here.\nSpeaker A: Mm-hmm.\nSpeaker A: So, we're looking to sell internationally, not just in Europe.\nSpeaker A: We're looking at, um...\nSpeaker A: Having a production costs limited to, uh, 12.5 euro per unit.\nSpeaker A: And therefore making a profit margin of, uh, well, not actually a profit margin, it's, uh...\nSpeaker A: Because obviously you're going to have overheads in your southern costs to, uh, take, uh, from, uh...\nSpeaker A: From that to give you your profit margin per unit.\nSpeaker A: And so depending on what the, uh, the overhead costs are, will determine, uh...\nSpeaker A: How many units were, uh, looking to sell, or projecting to sell at this point in time.\nSpeaker A: So, um...\nSpeaker A: Experience with remote control, first idea is...\nSpeaker A: You're remote. So, I guess we're looking at, um...\nSpeaker A: I'm going to be discussing at this point in time to help you, um...\nSpeaker A: Folks design our new model, as it were. So, uh...\nSpeaker A: Any, any thoughts?\nSpeaker D: Um...\nSpeaker D: I maybe found myself about controls a bit, but it's a little small.\nSpeaker D: But I had to pass. So we could make something with a...\nSpeaker D: This is passed, but...\nSpeaker D: As I was making a function.\nSpeaker A: Okay, so...\nSpeaker A: So basically we're looking for some, um...\nSpeaker A: We're looking for a device that is, um...\nSpeaker A: Robost.\nSpeaker A: And...\nSpeaker A: And therefore, uh...\nSpeaker A: I'm going to get damaged too easily.\nSpeaker A: Um... I'm looking for a device.\nSpeaker A: That is, uh...\nSpeaker A: What was the other thing you just said there?\nSpeaker D: Um... So, easy to use.\nSpeaker D: Easy to use.\nSpeaker D: That's what I was able to say.\nSpeaker D: Yeah.\nSpeaker A: Yes.\nSpeaker D: This is easy to use and see.\nSpeaker A: And see.\nSpeaker A: Okay.\nSpeaker A: Uh...\nSpeaker A: Could we just check?\nSpeaker C: Is this just a television?\nSpeaker C: Because a lot of them...\nSpeaker C: Systems are kind of TV video combined now. TV, TV, TV combined.\nSpeaker C: And one of the most annoying things is having like five and more to the house.\nSpeaker C: So if you've got a combined system, it could be a combined law.\nSpeaker C: Or is it just a television that's supposed to do?\nSpeaker A: Um...\nSpeaker A: They still get back to you on that, but it seems to me sensible because, as you rightly said, there's nothing more annoying than having three or four devices littered about the, uh...\nSpeaker A: About the room and, uh...\nSpeaker A: So, a device for...\nSpeaker A: For all remotes.\nSpeaker D: So, it's a little...\nSpeaker B: Okay. Yeah.\nSpeaker B: Um... One of the things we found in the market research is that people have to get confused by the number of buttons on them as well.\nSpeaker B: Because there's quite often lots and lots.\nSpeaker B: And, um...\nSpeaker B: Sometimes, uh...\nSpeaker B: They'd sort of remote controls to be their own purpose because you're sat in the chair and the remote is somewhere else in the room.\nSpeaker B: So, whereas in the past you don't have to get up to change the channel.\nSpeaker B: Now you have to get up to sort of pick up the remote.\nSpeaker B: So, I don't...\nSpeaker B: We need to sort of maybe think about how...\nSpeaker B: We could maybe, uh...\nSpeaker B: The remote control which moves around the room.\nSpeaker B: That's maybe something for the future when you talk to your television.\nSpeaker A: That is, in a sense, it's mutually exclusive.\nSpeaker A: You can't have both the one device and then have a few buttons on it too.\nSpeaker A: Because you want simplicity as well.\nSpeaker A: You want any idiot to be able to use it.\nSpeaker A: Whilst at the same time you want, as you rightly say, one remote for all.\nSpeaker A: And so, these are probably mutually exclusive options.\nSpeaker A: That, uh...\nSpeaker A: You could argue that experience of using devices and similar devices as people get more and more used to using remotes.\nSpeaker A: Therefore, they're more often with handling them.\nSpeaker A: Therefore, you can make them more complicated as time goes on.\nSpeaker D: They would have better instructions with the remote.\nSpeaker D: Or we just do the design of the remote control itself or the instructions that we can with it.\nSpeaker A: There's instructions.\nSpeaker B: Yeah.\nSpeaker B: I mean, we've done some research about sort of, you know, what the cutting edge sort of handheld devices are.\nSpeaker B: And a lot of them sort of use, you know, they're like mini laptops.\nSpeaker B: So it's possible that we could devise a system where you're basically sort of holding a miniature computer which is controlling all your sort of your television, your stereo.\nSpeaker B: And where, you know, if you buy a new thing, then it's sort of, you can link it to that as well.\nSpeaker A: Okay.\nSpeaker A: Well, we've got five minutes before the end of the meeting.\nSpeaker A: So, we have to start winding up.\nSpeaker A: Is there next meeting in 30 minutes?\nSpeaker A: Okay.\nSpeaker A: So, um, right.\nSpeaker A: So we've got ID, the command of mine.\nSpeaker B: You just click return.\nSpeaker B: Okay.\nSpeaker B: Get rid of the message.\nSpeaker D: Or not.\nSpeaker B: If you hit just hit return and it should get rid of the message.\nSpeaker A: Oh, there we go.\nSpeaker A: Yeah.\nSpeaker A: That's what I was looking for.\nSpeaker A: Right.\nSpeaker A: So we've got functional.\nSpeaker C: What happened to the.\nSpeaker C: Yeah.\nSpeaker A: Sorry about that.\nSpeaker A: Okay.\nSpeaker A: So, um, the working design for ID, the, the technical functions design, marketing, the user requirement specification specific instructions sent to you about your person, by your personal coach.\nSpeaker A: So we all clear what objectives we're looking to meet in the next 30 minutes.\nSpeaker A: And I guess I'll try and write up some minutes of this meeting to, uh, to give it to you for the next meeting.\nSpeaker C: I'm not necessarily clear on what we're designing.\nSpeaker C: Is it a multi functional one or do we decide that ourselves?\nSpeaker C: So the more we work on it.\nSpeaker D: I think you just said the start of the television about television.\nSpeaker A: That's true.\nSpeaker A: During the course of our day, we might make decisions based on information or meetings that would change.\nSpeaker A: Okay.\nSpeaker A: Where we're going.\nSpeaker A: But at this point in time, I think you're right.\nSpeaker A: So make it just a TV.\nSpeaker A: Okay.\nSpeaker A: So.\nSpeaker A: Depart.\nSpeaker A: We will stay here and, uh, and break off.\nSpeaker A: So we'll see you in half now.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Right.\nNone: Okay.\n"}, {"meeting_id": "ES2014c", "summary": "The project manager gave a brief review of the last meeting. After the Industrial Designer's presentation about the components of the remote and team's discussion about it, the team agreed that the kinetic power supply, simple push buttons and a simple circuit board and a finder feature are better choices for the remote. Then, the team discussed the design of the remote. They made the decision that standard rubber case and rubber buttons and a single curve are better choices for the remote, and the shape of buttons should be simple.", "dialogue": "Speaker A: I'm not sure if you're signed off a page.\nSpeaker A: I'll just kind of press the box and see what happens.\nSpeaker A: I don't know.\nSpeaker A: Right.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: Right.\nSpeaker D: I'm going to go to the concert.\nSpeaker D: Okay.\nSpeaker D: Right. Well, from the last meeting, I was trying to send you the minutes, but it didn't work out too well.\nSpeaker D: So maybe in sort of quick summary of the last meeting, I can quickly give you what we had.\nSpeaker D: Right.\nSpeaker D: So, right. So we had the fact that we're going to have the logo, the company logo in its color scheme, corporate it onto the device, the remote device. We had made our decisions about the device itself.\nSpeaker D: There's going to be simple to make it a complete project in time. We're going to have effectively two pages, the front page, which had the features that the customer's most wanted, and then the backup features on the second page so that it could meet the technical requirements and the customers would have to look at them too often only as and when required.\nSpeaker D: So, so basically what decisions have we made?\nSpeaker D: Have there been any changes?\nSpeaker C: I think we all have a presentation again.\nSpeaker C: Three presentations.\nSpeaker C: I'll just go first.\nNone: Okay.\nSpeaker C: So, basically, for the components design, the next step is basically the way the remote is going to work is still the same ideas before.\nSpeaker C: So, I still have the user interface, which is all the buttons we're going to incorporate.\nSpeaker C: Then there is a chip and still the sender.\nSpeaker C: So, yes, there's just a list of that. I think we're in the power supply as well.\nSpeaker C: I'll go on to my findings in each of these areas.\nSpeaker C: First in the power supply, we have the option of just the standard battery.\nSpeaker C: There's a dynamo. If you think of the old tortures which you wind up, like no idea.\nSpeaker C: There's a kinetic option, which then if you've seen those new watches which you power up by waving around, which requires a small amount of movement, which would mean the batteries wouldn't have to be replaced.\nSpeaker C: That's one option, but I think that was going to cost a little more.\nSpeaker C: And then the solar cells, there's a final option.\nSpeaker C: For the buttons, we have an integrated push button, which is just to say all these are supplied by real reaction.\nSpeaker C: So, I guess for the ease of quickness needs, we should take them from at least one of these options.\nSpeaker C: Okay.\nSpeaker C: So, if the button is an integrated push button, which I guess is just the same as the standard ones.\nSpeaker C: It says it's a similar to the button on the mouse for a normal, like one computer.\nSpeaker C: There's a scroll wheel, which is, you know, the new mouse is just going to be the center section, which you can scroll up and down, which may be for the volume.\nSpeaker C: Do that.\nSpeaker C: One issue for the buttons is depending on which material we use, if we use rubber buttons, then it requires a rubber case to have to go into consideration.\nSpeaker C: Moving on to the...\nSpeaker C: What would be the cost, do we know?\nSpeaker C: That's one.\nSpeaker C: I think there wasn't too much difference in the cost, but the relation to the actual buttons, but it does affect the printed circuit board, which is the next action.\nSpeaker C: Basically for the circuit board, which is the middle, let's just see it down there, the chips, and the workings of the actual remote.\nSpeaker C: The firm supplies a simple, irregular, and advanced sensor board, and there's different prices according to each.\nSpeaker C: So, if we got the scroll wheel for one of the buttons, that would require a slightly more advanced circuit board than if we just had a standard push button.\nSpeaker C: One final thing for the camera was some information on the speech recognition.\nSpeaker C: There's a small unit available through the company, which obviously would be an extra cost, but it wouldn't affect the size of the remote too much.\nSpeaker C: I guess that would require a more advanced circuit board, so there is an extra price in that sense.\nSpeaker C: There isn't so much cost in that sense.\nSpeaker C: Going to my personal preferences, I thought possibly for power we could use Kinetto, which is the idea of the watches. You move the remote around to power that.\nSpeaker C: This would avoid batteries running out of home to replace batteries.\nSpeaker C: For the buttons, I thought we'd probably get away with just having the standard push buttons, rather than the scroll wheel.\nSpeaker C: For the circuit board, again, it depends on which features we want in the actual remote.\nSpeaker C: If we wanted the scroll wheel and wanted the voice recognition, then we'd have to get a more costly circuit board.\nSpeaker C: That's it.\nSpeaker D: John, with the printed circuit board you were going for the...\nSpeaker C: It kind of depends. If we can have a speech recognition, we'd have to probably get an advance when I'm guessing.\nSpeaker C: But I don't know, as long as some people have a speech.\nSpeaker C: But are we going?\nSpeaker D: That's what decision follows.\nSpeaker D: So we're able to make that decision now, in a sense, that this is the point at which we're discussing that issue, so would it not be best to...\nSpeaker D: Rather than... I mean, one way is to do each of the presentations and then make decisions going back to the various presentations as they were.\nSpeaker D: The other way would be to do the presentation and then make decisions at that point in time.\nSpeaker A: And that's probably the better one to discuss it, it's great way.\nSpeaker D: Because at that point then you've got the details up there, so if we wanted to know, for instance, that the scroll wheel required the regular or what required advance, then if we were able to see that down and we could make the decision at that point in time and then that would be the end of that issue.\nSpeaker D: Does that make sense?\nSpeaker A: I have a lot of information there, it might not be very clear.\nSpeaker A: Unless you want to forget back in the years.\nSpeaker C: Good to you.\nSpeaker C: Yes.\nSpeaker C: As I say, it only specified that we need a more advance or for the scroll wheel than the voice recognition came as a set of piece of information.\nSpeaker D: Now the scroll wheel required the regular, so that...\nSpeaker C: Yeah, if you let down, it's just this bit of bond which I've highlighted, but the scroll requires a minimally regulated chip, which is in a higher price range.\nSpeaker D: Okay.\nSpeaker D: The display requires an advance chip, the display required an advance chip, which in turn is more or less expensive.\nSpeaker C: And the lobby.\nSpeaker D: Okay.\nSpeaker C: And the push button just requires a simple chip, so that would keep the price down.\nSpeaker A: Yeah, and if we're going for a sticky, sexy, I think the scroll wheel is maybe a bit kind of bulky.\nSpeaker A: I've got some pictures, so I've seen pictures where they're kind of sticking off the side of it.\nSpeaker A: Right.\nSpeaker C: Okay.\nSpeaker C: Simple push button and that would...\nSpeaker D: So, so we're going for...\nSpeaker D: Okay.\nSpeaker D: So, the answer word is...\nSpeaker D: So we're going for the simple one.\nSpeaker B: Yeah, simple push button.\nSpeaker C: We did everyone get this from the speech recognition.\nSpeaker C: There was basically what we said before, the idea that you were calling a set message and then it picks up our message and replies to.\nSpeaker C: So it is basically the concept we discussed before.\nSpeaker C: But then we don't know for sure whether it would require more complicated circuit or I guess it would have been like the definition of information.\nSpeaker C: Maybe we should go on with this.\nSpeaker D: So we go for the simple push button.\nSpeaker D: So effectively we're going for the simple printed circuit board.\nSpeaker D: Are we going for the regular?\nSpeaker C: If it's just a push button and it just needs a simple circuit.\nSpeaker D: But is there any other...\nSpeaker D: I mean, okay, that's true for the... for that element.\nSpeaker D: But we have to take all elements into consideration.\nSpeaker D: And so if there is one element that requires more expensive one or say the regular one or the more advanced, then that would have to be the same for all of them.\nSpeaker B: I suppose I need really to find out what circuit board's that requires.\nSpeaker B: Maybe we probably need a decision.\nSpeaker A: But I'm going to put that.\nSpeaker A: It doesn't seem to send out a signal to the tele.\nSpeaker A: It's like a parrot just would reply.\nSpeaker A: It would reply in your message.\nSpeaker C: So maybe that would be something separate.\nSpeaker C: So I don't think it would affect our circuit board.\nSpeaker C: No.\nSpeaker C: Okay, so we'd have a simple circuit board and that would be extra.\nSpeaker C: That would be the bit.\nSpeaker A: Yeah.\nSpeaker A: And I don't think it could really perform any of the remote functions with this.\nSpeaker A: Because the example that we've given there is good morning coffee machine, good morning dual.\nSpeaker A: It's useful to see like, where are you in the morning?\nSpeaker A: Here I am.\nSpeaker A: Sure.\nSpeaker A: I think that's maybe as far as that one could go.\nSpeaker C: Okay.\nSpeaker C: Yeah, that makes sense.\nSpeaker C: So we stick the simple circuit board and then think of the speech recognition as next.\nSpeaker A: Yeah, just as a fun way to find it.\nSpeaker D: Simple circuit board.\nSpeaker D: Simple push button.\nSpeaker D: Okay.\nSpeaker D: That seems like a good thing.\nSpeaker D: You were wanting to go for the kinetic person.\nSpeaker C: Yeah, I thought so just for ease of moving to replace the batteries.\nSpeaker D: And how does it get charged up?\nSpeaker C: It's a thing that works on the basis to have some kind of ball bearings inside.\nSpeaker C: There's someone watches which you kind of shape to power that somehow the mechanism inside powers that to your movement.\nSpeaker D: Okay, so the speech recognition was, are we going for a speech recognition?\nSpeaker D: No, because that required the advanced.\nSpeaker A: I think it would be helpful to find it.\nSpeaker C: But yeah, we decided to affect the circuit board just to the call.\nSpeaker D: I had speech recognition requires advanced.\nSpeaker C: I thought that's why I thought that maybe maybe doesn't know.\nSpeaker C: Okay.\nSpeaker D: Speech recognition you reckon then is.\nSpeaker D: Is this a simple.\nSpeaker D: And so we would want it in as an extra because it doesn't appear to cost too much.\nSpeaker D: And that be.\nSpeaker C: Okay.\nSpeaker A: So, that wouldn't really cost anymore would it.\nSpeaker A: Actually, I don't think it really says anything about the cost, but it says it's already in the coffee machines.\nSpeaker A: So.\nSpeaker A: It's already.\nSpeaker C: I assume it would cost X a bit.\nSpeaker D: Maybe we'll find out how much that does cost.\nSpeaker D: And then after change.\nNone: Change on everything at the last minute.\nNone: Okay.\nNone: Okay.\nSpeaker A: So, very brief presentation.\nSpeaker A: I haven't actually got the examples of this robot in there.\nSpeaker A: But there's some curve cases that you can see.\nSpeaker A: And the same just happened with the buttons.\nSpeaker A: Sorry.\nSpeaker A: It just seems to be skipping on with it.\nSpeaker A: Yeah, I've found that.\nSpeaker D: And I need to try and get it back.\nSpeaker C: Have you right click and then go into it previously.\nSpeaker A: Oh, sorry.\nSpeaker A: There wasn't much more to say about that.\nSpeaker A: Just round.\nSpeaker A: And some of the remotes that I looked at, one of the models that did actually have voice recognition where you could.\nSpeaker A: Where it was connected to the remote control functions.\nSpeaker A: And it was quite a swish model where it can control four devices to be the satellite video, editing, audio.\nSpeaker A: So that's a bit of competition there.\nSpeaker A: So.\nSpeaker A: I mean, maybe it's better not to try and compete with that sort of thing.\nSpeaker A: And just to market it as a completely different, like different view pointers, the kind of finally lost control rather than trying to compete with the platform.\nSpeaker A: The scroll buttons as I already mentioned them.\nSpeaker A: And those examples of those were the don't look as sleek as other models.\nSpeaker A: And there's no real advantage.\nSpeaker A: And because it impacts on the materials in the price, it's not great.\nSpeaker D: So you're saying the scroll buttons.\nSpeaker A: Yeah.\nSpeaker A: There was a specialist type of report that we could think about.\nSpeaker A: There was children's remote where they just had a very limited range of buttons in the world, bright and colorful.\nSpeaker A: And you could program them so that they could only look at certain channels.\nSpeaker A: But I don't know if that's really in our field.\nSpeaker C: I guess we're going to the biggest market and maybe not.\nSpeaker C: Was it specified that we went to the biggest?\nSpeaker D: Or to go for the international market rather than a local market, but that wouldn't necessarily preclude.\nSpeaker D: There's one thing that you can often do with products is you can make small modifications.\nSpeaker D: You have your basic model, which you would sell at, whatever.\nSpeaker D: And then you could have additional features in, you know, like a model one model two and model three.\nSpeaker D: And therefore you can subdivide your market up, but that's really more in your fields.\nSpeaker C: Maybe the children most would like an ex-step, maybe.\nSpeaker C: Or you could add on for an extra package.\nSpeaker D: But on this basic one, I'm writing that we're going for the basic model to be discussed here and that you would have for future reference, the possibility of adding in extra features at extra cost to take care of specialist market segments.\nSpeaker D: Okay.\nSpeaker D: So what are we deciding to do here?\nSpeaker A: I think because it's already a very good voice recognition technology out there.\nSpeaker A: Because ours might not cover the same functions that the movie fans do.\nSpeaker A: It might be a good idea to market it as a finder function.\nSpeaker D: Yeah, finder function rather than as a speech function to find your remote.\nSpeaker C: Yeah.\nSpeaker C: So maybe older people might not like the speech recognition.\nSpeaker A: Oh yeah.\nSpeaker C: So if it's the languages, they might not be compatible.\nSpeaker C: It would make it quite complicated.\nSpeaker C: Whereas this keeps it very simple.\nSpeaker A: Yeah, because I think you program this one yourself, like to see whatever you want to your question.\nSpeaker B: Yeah, that was a cheap device.\nSpeaker B: I don't know how much we'll be able to put into it.\nSpeaker D: So you have a finder feature rather than a voice recognition feature in your talk.\nSpeaker D: That's something else comes up.\nSpeaker D: And you're talking about scroll buttons.\nSpeaker A: Yeah, I think we've decided that it's going to increase the cost and give real real attention to the benefit.\nSpeaker A: And it's going to decrease from the sleepless order.\nSpeaker D: All right, so we're just going to have the rubber button.\nSpeaker A: And just to be aware that the archivist specialist functions and specialist remote will probably don't want to focus on those.\nSpeaker D: Okay, so not to be focused on.\nSpeaker A: Yeah, there was a mention just as kind of a warning about button design.\nSpeaker A: Just to avoid ambiguity.\nSpeaker A: So I gave the example of say your volume buttons, but up and down.\nSpeaker A: They might both have a V on for volume.\nSpeaker A: I think I did this.\nSpeaker D: Good. In flip it round at 90 degrees, 180 degrees and have it up and down.\nSpeaker D: And upside down V.\nSpeaker D: So that would show that volume was going up.\nSpeaker D: Whereas the one underneath would see the volume going down.\nSpeaker A: I think the thing was that if you decide to do this to have triangular buttons, somebody might look at this one and see all this triangular button is pointing up.\nSpeaker A: And that's the first thing that the C.\nSpeaker A: No, I think.\nSpeaker A: Oh, no, they might see.\nSpeaker A: Yeah, they might see this pointing down.\nSpeaker A: I think right, that's going to turn the volume down.\nSpeaker A: Whereas the actual button is pointing up.\nSpeaker A: So the function is to turn the button up.\nSpeaker A: So be careful what you put on the buttons and be careful of the shape that you make them because it might be kind of two contradicted shapes.\nSpeaker C: So maybe we have volume written on the side, then up and down on the button.\nSpeaker D: You can have volume up and volume.\nSpeaker D: Volume up down.\nSpeaker D: And because the idea was to have limited, it was to have sizable amount of information on it.\nSpeaker D: Yeah.\nSpeaker D: Limited number of buttons.\nSpeaker D: Because it was 16 buttons, wasn't it?\nSpeaker C: Yeah, we were going down to that.\nSpeaker A: And I think that's all I had to see before that.\nSpeaker A: So what was the decision on the design of the volume button?\nSpeaker C: I'm going to go through the design of all the buttons at the moment.\nSpeaker B: I've got some things to say about possible design.\nSpeaker B: Okay.\nSpeaker B: So I'm going to turn the volume down the direction.\nSpeaker B: Right.\nSpeaker B: So I'm going to turn the volume down the direction.\nSpeaker B: I'm going to turn the volume down the direction.\nSpeaker B: about remote controls and what was good about them, what was bad, what they used.\nSpeaker B: And we've also been looking at sort of fashion, what people are wanting out of consumer goods at the moment. So we've had people in Paris and Milan watching via fashion trends.\nSpeaker D: You know, go to yourself.\nSpeaker B: So just to summarize, the most important thing is you just came out of the remote control market investigation.\nSpeaker B: The most important thing was that the things sort of look and felt fancy rather than just functional.\nSpeaker B: And second, there should be some technological innovation.\nSpeaker B: And then third, and less important than the other two, that there should be an ease of use as well.\nSpeaker B: And apparently the fashion trends are that people want sort of clothes and shoes and things with fruit and vegetables.\nSpeaker B: But the feel of the material should be spongy, which is country to last year apparently.\nSpeaker B: And presumably not spongy last year.\nSpeaker B: So we need to emphasize the fancy design on our remote control above all else.\nSpeaker B: And then also try and add in technological innovation, which could be sort of find a thing with a hand clap.\nSpeaker B: And then we need to make it easy to use as a third priority.\nSpeaker B: So perhaps a few buttons and functions as we've discussed.\nSpeaker B: And then maybe find a way to incorporate these trends so that we sort of capture people's imaginations.\nSpeaker B: So maybe we could make the buttons shaped like fruit and veg, or the buttons could be spongy somehow.\nSpeaker B: Maybe we could make them out of rubber rather than sort of hard plastic.\nSpeaker B: And then sort of even wacky than that, we could maybe have a fruit or vegetable shaped remote, say in the shape of a banana or something like that.\nSpeaker B: Right, so maybe a banana or a courgette or something.\nSpeaker B: So how far we actually want to go along and sort of follow the trends, do you think the trends are particularly important for this type of gadget or do they not matter that much?\nSpeaker A: I think if you start making the buttons fruit shift, it might make it more complicated to use.\nSpeaker D: Well, you were just talking about you've got to be careful about how you shape your buttons because you're going to miss direct people.\nSpeaker D: And I would have thought the functionality because the people get cheesed off by having to read instructions, etc.\nSpeaker A: Maybe just one button, say the standby button is quite kind of separate from all the other functions.\nSpeaker A: Maybe that could be a little apple.\nSpeaker A: And I wouldn't get in the way of like kind of one with nine and it wouldn't cost you numbers.\nSpeaker D: Yeah, standby button.\nSpeaker D: No, that incorporates the trend, both at the same time not confusing people.\nSpeaker D: If you're looking for functionality.\nSpeaker C: I guess maybe through a vegetable, maybe pop in the loam, but that's going to be next year.\nSpeaker D: But okay, but you can incorporate that.\nSpeaker D: If you change all the buttons, then you've got the problem that this year's fruit and veg next year's, I was going to say animals are elephants or whatever.\nSpeaker D: That means you're constantly changing your production schedule and you've got to make different molds and everything else.\nSpeaker D: So that's not a good idea. I would suggest.\nSpeaker B: What the time scale with selling the product over it.\nSpeaker C: I mean, the scene is realistic. The remote control market isn't the kind of thing which takes those kinds of fashion.\nSpeaker C: So it keeps us something which is maybe more universal.\nSpeaker A: Well, as long as it's quite a subtle design.\nSpeaker C: Even if it's a design, you can apply it.\nSpeaker C: The spongy feels.\nSpeaker D: Well, yeah.\nSpeaker C: Maybe saw the rubber design.\nSpeaker D: Was that in the sort of fashion sense that this or was the spongy feel was that sort of fashion it was, wasn't it?\nSpeaker D: Yeah.\nSpeaker A: So, the...\nSpeaker A: It's like you're going to have rubber cases as well as buttons.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker A: And like you can make them curved or double curved.\nSpeaker A: That would be the kind of thing.\nSpeaker C: Oh, yeah.\nSpeaker C: One of the things that if you had rubber buttons and you had to have rubber case.\nSpeaker C: Oh, no, no, no.\nSpeaker C: So it's if you use a rubber double curved case and you must use rubber buttons.\nSpeaker C: That's the way I'm doing.\nSpeaker C: You have the rubber case and you have to have the rubber buttons to go with it.\nSpeaker C: Sure.\nSpeaker C: Thanks.\nSpeaker C: Thanks.\nSpeaker D: Remember buttons require rubber case.\nSpeaker A: I know.\nSpeaker D: The main problem is how frequently do the fashions change.\nSpeaker D: Because in essence in the production you want things to stay.\nSpeaker D: You want to basically mint them out.\nSpeaker D: Because if you've got fashion changes and that you're incorporating, then it means that your stock is, you know, last year's stock and therefore you're selling it or having to sell it at a discounted rate, which you wouldn't want to do.\nSpeaker D: Whereas if you kept the product the same, but you could have a difference from year to year.\nSpeaker D: It seems to me that you could incorporate a fashion statement if you like, rather than changing the whole, getting caboodle, you're just changing one aspect like this down by button or something like that.\nSpeaker D: And especially because then you could make it something that.\nSpeaker C: So, you might be limited if I were to slasper the logo on the edge.\nSpeaker D: Well, you might be limited in space.\nSpeaker D: Yes.\nSpeaker D: Well, you two are obviously going to find out fairly quickly when you move over to your kit modeling stage as to how much, how much, how much, how pliable is plasticine.\nSpeaker C: Moving the cases like changing the fashions like the Nokia phones where you could take the casing off the outside.\nSpeaker C: Like, whether that would be too much to incorporate into that.\nSpeaker C: Whether that would just make it more complicated.\nSpeaker D: So, you're talking about changing the casing.\nSpeaker C: Yeah, the actual sort of look from the outside.\nSpeaker C: So, the buttons would stay the same.\nSpeaker C: And the general function of the remote would stay the same, but then you could change the way it would.\nSpeaker A: Yeah, and then you could have, or but you still would have to have the logo on every nucleus.\nSpeaker A: Yeah, but you could have like pinky since for the old ones.\nSpeaker D: Yeah, you could do a color change.\nSpeaker D: So, therefore you would, yeah, yeah.\nSpeaker D: I mean, that's effectively what they do with the mobile phones, which is to have some in blue, some in red, some in red, rather than all in black or, you know, which four do you want as long as it's black.\nSpeaker D: So, so...\nSpeaker C: So, it's most dirty.\nNone: We are supposed to use the, um, Comprom, color, or how many?\nSpeaker D: Yes, oh, that's true.\nSpeaker D: I mean, I don't know.\nSpeaker A: I mean, what do you both put in my classroom?\nSpeaker D: Well, not necessarily because you could have your company, or might be finishing up.\nSpeaker D: You could have your company badge.\nSpeaker D: I mean, a lot of computers, for instance, like, like on the one you've got there.\nSpeaker D: It actually has a sort of stick on badge.\nSpeaker D: So, what you would all, all you would really need, whether, you know, whether the casing be any color, could be any color.\nSpeaker D: And that badge would then have to stick out on top of it, so that, in a sense, with a, with a logo like that, because it's on a white background, the only color that it might not stick out somewhere well on would be a white casing.\nSpeaker D: Because you, you know, you're sort of, you're badging it.\nSpeaker D: In fact, a lot of companies get somebody else to make them and literally just badge them themselves with their own badge over the top.\nSpeaker D: And in fact, the way they've got that there, even if you had that on a white, which is the predominant color of the, the windows badge, you'd still be able to see it clearly from, you know, a white casing product.\nSpeaker D: So,\nSpeaker A: we'd have a big in the market to have this kind of like secondary market selling the cases.\nSpeaker B: People say that it's the look, the fancy look of the in-bounds. Yeah, I'm not convinced on whether having change book covers would be something that people buy into. I think with the mobiles, it's the, you know, it's a communication device. People see you with it all about.\nSpeaker C: Yeah, I suppose we're a bit of a cute little motor.\nSpeaker B: It's in the house, isn't it? Yes, it was.\nSpeaker D: Okay, so if we just went for one color, so don't change case, change case color and we're sort of saying no to that.\nSpeaker B: Do we decide on the other case? The sponge you feel I did, I think that might go as a trend. Well, it was different last year, the trend was different last year, apparently, it was not, not the sponge you feel, but I don't know whether the trends have changed. I don't know whether it's from those things that, like, sort of having all group shaped keys that that probably would go out of fashion very quickly. Yeah, just the fact that it was a rubber case is probably less, less or less in my own sense.\nSpeaker B: You're going to end up hating it a year.\nSpeaker D: If you're going for fashion trends like that, you'd have to have interchangeable cases so that you could, or because otherwise someone's going to have to buy a complete new remote rather than just a case.\nSpeaker C: This seems to make sense if we just make a stick with the standard rubber case and then have the standard rubber buttons as well. Okay.\nSpeaker C: Well, haven't we talked about the curvature of the case? This flat, the single curve, and this double curve? I'm not exactly sure what it looks like.\nSpeaker C: Maybe kids give it a funny more aesthetic deal with a double curve when requires to perform miracles for the class.\nSpeaker D: Well, it's a tough one.\nSpeaker B: When you say, when you say double curve, what exactly does that mean?\nSpeaker A: I'm not exactly sure. I'll show you the motor. See how the one, I'm not familiar with that.\nSpeaker D: One thing to consider is that in some ways you want, by having a fairly standard case, it means they can all fit together on top of each other.\nSpeaker D: Therefore, for storage purposes and shops and the like, it makes it easier if you can store them on top of each other.\nSpeaker D: Whereas if you do fancy things, whether you then got to put it in a packaging box that does that.\nSpeaker D: And the cost of packaging could be quite important vis-a-vis the total cost of the product.\nSpeaker C: So, I think quickly and just work out what we decided on the way after.\nSpeaker D: So, but...\nSpeaker A: It's not very clear up there, but it's not a similar kind of bulges, except one and the end.\nSpeaker A: When it's a curve there. I don't know if a double curve is maybe a comes up slightly.\nSpeaker A: I'm sorry, I don't worry, Kerve.\nSpeaker D: So, do you want to go for curves, more curves?\nSpeaker D: We're meant to be finishing this meeting in about a minute or so.\nSpeaker A: We'll go for single curves, just to call it minus.\nSpeaker D: Okay, curved or double curve, so it's a single curve.\nSpeaker C: So, did we decide on the kinetic power supply?\nSpeaker C: Yeah, I think we're going to go for simple stuff before just to keep the cost down.\nSpeaker C: I think we can't buy not having anything to complicate.\nSpeaker A: Yeah, and the voice recognition, we're going to use that car, we're just to find it.\nSpeaker C: Yeah, yeah, sure.\nSpeaker C: We could always decide against this if something comes up.\nSpeaker C: That's just something to move out.\nSpeaker B: Okay.\nSpeaker B: And then we're going for sort of one button, which is like a group.\nSpeaker A: Yeah, that's nice.\nSpeaker A: You wouldn't do too much harm in a couple of years.\nSpeaker D: So, we've got spongy feel buttons as well, have we?\nSpeaker D: Or was that rubber buttons?\nSpeaker D: Yeah, so it's rubber buttons, so it's not really spongy feel buttons, it's just rubber buttons.\nSpeaker D: Yeah, it's a rubber case.\nSpeaker D: And the standby button is going to be different.\nSpeaker A: Yeah, okay.\nSpeaker A: So, what shape are we making the standby button?\nSpeaker D: Apple.\nSpeaker D: Apple.\nSpeaker A: Oh, sorry.\nSpeaker A: We'll give it a little bit of a...\nSpeaker B: I think it's quite a big one as well.\nSpeaker B: Oh, it could be red.\nSpeaker B: Can we red apple?\nSpeaker B: Yeah.\nSpeaker B: I don't know.\nSpeaker B: A red apple?\nSpeaker A: Yeah, because...\nSpeaker A: I want to incorporate a bit of colour if we can.\nSpeaker A: Once we find out.\nSpeaker A: Yeah, and then we're going to...\nSpeaker C: You're going to work on keeping the buttons like, quite simple.\nSpeaker C: Just like the...\nSpeaker C: Just working out what we're going to do for the next time.\nSpeaker A: Yeah.\nSpeaker A: Yeah, that seems pretty straightforward.\nSpeaker A: Because most of them will just be kind of...\nSpeaker A: mainly circular, or very pleased.\nSpeaker D: Okay.\nSpeaker D: So, what was that last thing in there?\nSpeaker A: Just to keep the shape of the button is simple.\nSpeaker D: Right, well, you have much option on that.\nSpeaker D: I thought you were going for a single curve and...\nSpeaker A: Just the shape of the buttons?\nSpeaker D: All right.\nSpeaker C: Yeah, just keeping the sort of the label and the labeling of them.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: And sort of the size.\nSpeaker D: All right, so the shape of buttons is simple.\nNone: Okay.\nSpeaker D: So, that's it, I guess.\nSpeaker D: You should now go away and get these things sorted out.\nSpeaker D: I guess you two are on plasticine duty or whatever.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker C: Okay.\nSpeaker D: Okay.\n"}, {"meeting_id": "TS3006c", "summary": "The meeting began with the members' personal presentations on the conceptual remote control. The Marketing would like the remote control to be with an appealing appearance and light material to attract more young consumers. The User Interface proposed to add a speak recognition system onto the remote control so that it would be able to function according to the user's instruction. After the presentations, they spent some time on the button layout and the shape of the remote control. Though the discussion was mostly smooth, they hardly reached an agreement towards the materials. At the end of the meeting, the group talked about the company features on the product, including a slogan and a logo.", "dialogue": "None: Hello.\nNone: You braced that in my?\nNone: Yes, I made it.\nNone: English from now on.\nSpeaker E: Drawing or?\nSpeaker D: Yeah, just testing.\nSpeaker D: Yeah.\nSpeaker E: I'm not feeling that much.\nSpeaker E: To all the stress.\nNone: Just kidding.\nSpeaker E: English.\nSpeaker E: So annoying.\nSpeaker E: I'm not feeling that much.\nSpeaker C: I'm not feeling that much.\nNone: I'm not feeling that much.\nSpeaker C: I'm not feeling that much.\nSpeaker C: So annoying.\nSpeaker D: Break is over.\nSpeaker E: Ooh, it works.\nSpeaker E: Spicy.\nSpeaker E: Spicy.\nSpeaker E: Or all the other presentations.\nSpeaker C: I just put it in the shared folder.\nSpeaker C: Which is conceptual.\nSpeaker C: Yeah, I think so.\nSpeaker C: Yeah, conceptual design.\nSpeaker E: Or whatever it does.\nSpeaker E: I see only my own presentation.\nSpeaker C: No, no, no. Can you go back?\nSpeaker C: Yeah.\nSpeaker C: Because it has to be PowerPoint.\nSpeaker C: Yeah, components design.\nSpeaker E: That's it.\nSpeaker E: I just put it in there.\nSpeaker A: He's coming.\nSpeaker C: I did get a bit more done than the last time.\nSpeaker C: Because I knew that I didn't have time.\nSpeaker C: So I just copy and paste everything into it.\nSpeaker E: I can kind of paste it into the other folder.\nSpeaker D: You can look at the final report.\nSpeaker D: Because I have to record everything.\nSpeaker D: I'm not sure if you are deciding on the search.\nSpeaker D: So move to me.\nSpeaker D: I'm trying to write it down between everything else.\nSpeaker C: Sometimes I have these pop-ups or these sounds.\nSpeaker C: And there's nothing there.\nSpeaker C: Yeah, me too.\nSpeaker C: Also, I don't know how to use PowerPoint.\nSpeaker C: So it takes me forever to get something done.\nSpeaker B: You've got the same problems.\nSpeaker D: Yeah.\nSpeaker D: Yeah, here we go again.\nSpeaker D: Welcome.\nSpeaker D: You have again three presentations.\nSpeaker D: And then we have to decide on what concepts the mobile phone has to, for the remote control has to support.\nSpeaker D: So let's go.\nSpeaker E: Yes.\nSpeaker E: Who wants to start?\nSpeaker D: Me first again or?\nSpeaker D: Yeah, sure.\nSpeaker E: Doesn't matter.\nSpeaker E: Yeah.\nSpeaker E: All right.\nSpeaker E: If you open it already or?\nNone: No.\nNone: Yes.\nSpeaker E: Yes.\nSpeaker E: So welcome to the marketing presentation once again.\nSpeaker E: This time about trend watching.\nSpeaker E: Well, there has been investigation again in the remote control market.\nSpeaker E: It shows a number of developments.\nSpeaker E: I will address them in a moment.\nSpeaker E: Fashion watchers have detected the trends for young public because that's our public.\nSpeaker E: Well, fruit and vegetables will be the most important team for clothing, shoes and furniture.\nSpeaker E: And the field material is expected to be spongy.\nSpeaker E: So the developments, I will address them.\nSpeaker E: Oh, yeah, well, this is the fruit and vegetables.\nSpeaker E: Lukes are fresh, bright colors to give you an ID.\nSpeaker E: Well, the developments, development one.\nSpeaker E: Well, the most important aspect for remote control happens to be a fancy look and feel instead of the current functional look and feel.\nSpeaker E: Well, fancy stands for an original look and feel of the case and the interface.\nSpeaker E: And the second most important aspect is that remote control should be technological, innovative.\nSpeaker E: Well, it stands for the use of technical features that do not exist in current remote controls.\nSpeaker E: I think we pretty much covered that with our screen and speed recognition.\nSpeaker E: So I don't expect that to be a problem.\nSpeaker E: The third developments is that the remote control should be easy to use.\nSpeaker E: Well, the first aspect was twice as important as the second aspect, which was twice as important as the third aspect.\nSpeaker E: So that kind of gives you these ratios.\nSpeaker E: So fancy look and feel is the most important point of attention.\nSpeaker E: So the fruits and vegetables in combination with the sponge material.\nSpeaker E: Well, technological innovation, we covered that pretty much, I guess.\nSpeaker E: And easy to use.\nSpeaker E: I don't think that would be a problem.\nSpeaker E: So my point of attention is especially this part that this will be correct.\nSpeaker E: So that was the marketing presentation.\nSpeaker E: I had only one document left.\nSpeaker B: So I go first.\nSpeaker E: So kind of this.\nSpeaker E: Yeah, sure.\nSpeaker E: So a small example kind of this.\nSpeaker E: Nothing about the buttons, but just sponge kind of thing and some fruit and colors, I don't know.\nSpeaker E: Just made a quick design.\nSpeaker E: Cool.\nSpeaker D: All right.\nSpeaker D: You're just using the instructions.\nSpeaker C: Yeah, components.\nSpeaker B: Yeah, layout.\nNone: Yeah.\nSpeaker C: It's okay.\nSpeaker D: You probably opened this.\nSpeaker C: Yeah, true.\nSpeaker C: F5.\nSpeaker C: F5.\nSpeaker C: All right.\nSpeaker C: So I'm dealing with the components design.\nSpeaker C: Let's see.\nSpeaker C: I used some design examples we had from similar products.\nSpeaker C: I used possibilities from our manufacturing department about current components, which will have to be implemented in the design.\nSpeaker C: That's why I wanted to go first.\nSpeaker C: Well, they gave me an idea about what people want.\nSpeaker C: We're mainly focusing on this group, but I want to make the distinction clear.\nSpeaker C: I could not drag the pictures into the slide.\nSpeaker C: So I don't have examples of how it looks like, but it comes down to what you think we should do with the spongy and the fruity-looking type.\nSpeaker C: If the young dynamic people want soft primary colors, which looks like fruits, you know, you can, and shapes that are curved and not solid, straight-lined anymore.\nSpeaker C: So this basically goes on to what you were mentioning earlier.\nSpeaker C: There's a lot of factors involved in choosing the components.\nSpeaker C: There's a lot of options that we have to discuss.\nSpeaker C: For example, the energy source.\nSpeaker C: We have four types, the basic battery.\nSpeaker C: We have a hand dynamo, which we Dutch referred to it as the Kneipkot.\nSpeaker C: The kinetic provision of energy, which means if you move the thing, you shake it, which will be fun for toddlers, right?\nSpeaker C: If they want to use the...\nSpeaker C: And of course, solar cells, but I don't know how we would use that into the design of the actual product.\nSpeaker C: So...\nSpeaker C: Also, in countries where there isn't much light, like in Scandinavia, they wouldn't be able to use it half of the year.\nSpeaker C: That's not cool, either.\nSpeaker C: So, for the case, there's the traditional uncarved flat-heart case, single-curved, which means that it has curves in one dimension or the double-curves.\nSpeaker C: I wasn't able to finish my personal preferences sheet, but, well, in all, we will have to go for the double-curved, because it's daring and different from what we have now.\nSpeaker C: The case materials, well, we have all kinds of hard materials, like the hard plastic, the wood and the titanium.\nSpeaker C: I would definitely go for rubber, because it fits most in what people want to see nowadays.\nSpeaker C: Phew, this is a lot of text. I wasn't able to organize this yet.\nSpeaker C: We have several interface designs.\nSpeaker C: We can use scroll buttons for the menus, but we already kind of decided to go for the push buttons for the arrow buttons.\nSpeaker C: So, that's not really interesting.\nSpeaker C: Electronics.\nSpeaker C: Yeah, maybe we want to decide on what electronics to use.\nSpeaker C: The advanced chip, I think, is easiest to implement for the production, because they can print it better.\nSpeaker C: I think this is about it. I was working on some personal preferences.\nSpeaker C: I first chose for the battery, because I'm traditional, and that's the most obvious, easiest choice to go to.\nSpeaker C: But I really think that we should maybe think about the kinetic energy, where you have to move the thing to be able to use it.\nSpeaker B: So, I think that we should be able to use the traditional feature, or combine both with one.\nSpeaker D: I guess we have maybe one source of energy.\nSpeaker C: I can imagine that the kinetic type energy source would be more expensive to make, but it is more long-lasting that people don't have to ever buy batteries again.\nSpeaker E: That is more fun.\nSpeaker C: And it's also more fun. I always chuck my remote control around.\nSpeaker E: I think it's a good thing with it, and especially when the material is rubber, it can be done.\nSpeaker C: You can't harm it, so it's a perfect combination.\nSpeaker C: You don't have to be scared about bouncing it off the floor and breaking it, or whatever.\nSpeaker C: So, that's the end of it.\nSpeaker E: So, double-curve is like this.\nSpeaker C: It means curved in two dimensions. So, single-curved, let's say, would be a square box, but then with curves on one dimension.\nSpeaker C: And double-curved would mean that it would have curves in every direction, like 3D.\nSpeaker B: One very important thing I was thinking about, speech option.\nSpeaker B: We were going to use that.\nSpeaker B: That's one thing which I'm not sure of how to implement it.\nSpeaker D: The visual representation is not there with speech.\nSpeaker B: It has to be combined with a manual for functions.\nSpeaker D: I think you can just match the speech commands with the features that are already present.\nSpeaker D: We don't think you have to design anything else.\nSpeaker B: Do the speech just for the basic options, for the simple buttons, for everything, also for fun options.\nSpeaker B: This is all a very basic, trend design, everybody says it.\nSpeaker B: In the last meeting, we were putting the simple and advanced options separated.\nSpeaker B: It's obvious.\nSpeaker B: Pressing the menu option will disable all other options on your remote control.\nSpeaker B: Only the LCD panel will light up and then you can only change the buttons that you need to control.\nSpeaker B: Design has to be very attractive with that.\nSpeaker B: You have to delete this.\nSpeaker B: Display on the upper side with the menu button and maybe some sort of cancel button or save button.\nSpeaker B: That would be the back.\nSpeaker D: You did read the minutes I wrote?\nSpeaker B: A little bit, but not more.\nSpeaker D: I hate doing work for nothing.\nSpeaker B: This is the basic design for the buttons.\nSpeaker B: I wanted to categorize everything with speech display sounds, everything you noted in your minutes.\nSpeaker B: Every push button has its own lead light.\nSpeaker B: You can make it more trendy for younger people.\nSpeaker B: If there are older people, they want it more luxurious.\nSpeaker B: That's also an option.\nSpeaker D: That was it.\nSpeaker D: I'm again.\nSpeaker D: What we have to do is what kind of components do we use?\nSpeaker D: Energy source, chip type, case type and user interface.\nSpeaker D: I didn't see a clear distinction between this.\nSpeaker D: I think what we have is okay.\nSpeaker D: We already decided that kinetic would be the choice for energy.\nSpeaker D: The case would be doubly curved.\nSpeaker D: The only thing we have left.\nSpeaker C: We need the chip on print to be able to support the screen and audio function.\nSpeaker E: Is that the advanced chip?\nSpeaker C: Otherwise, we would have a simple chip just for pressing buttons.\nSpeaker C: We need more.\nSpeaker C: I'm not sure what this is going to cost to be able to.\nSpeaker D: I didn't get any info on this.\nSpeaker C: We need to sell it for 25 euro a piece.\nSpeaker E: It's going to be difficult.\nSpeaker E: The cost of making it should be 12.5.\nSpeaker B: I didn't get any information about that.\nSpeaker C: We don't produce it in China.\nSpeaker D: Child labor, we love it.\nSpeaker B: It's quite difficult because we haven't got all the options.\nSpeaker D: Do you have a picture of doubly curved case?\nSpeaker D: Could you put it in the group folder?\nSpeaker C: Let me see what it says.\nSpeaker E: If you go to the homepage or something.\nSpeaker C: I'm going there now.\nSpeaker E: Get your own information.\nSpeaker E: I got my fresh and fruity picture.\nSpeaker E: You didn't enjoy it yourself.\nSpeaker E: Maybe it's the menu.\nSpeaker E: Maybe it's easy with an arrow so that indicates that there's a menu under that menu.\nSpeaker C: Maybe it's easier if you guys come over here.\nSpeaker C: This is the standard traditional type where the form serves the function.\nSpeaker C: It's like really basic.\nSpeaker C: It's more appealing to old people and we don't want that.\nSpeaker C: This is what we're looking for.\nSpeaker C: That means curved in both dimensions.\nSpeaker C: Not only like this, but it has to be...\nSpeaker C: It has to be kind of instead of the PlayStation module.\nSpeaker C: It has to be like the GameCube.\nSpeaker C: Your thumbs would be laying in the instrument.\nSpeaker C: It has to be nice to hold.\nSpeaker B: It has to be luxurious for rich people.\nSpeaker B: This looks a little bit like only for children.\nSpeaker C: Yeah, but that's the problem.\nSpeaker C: The dilemma actually.\nSpeaker C: We want to appeal to the young public with flashy colors and with a lot of shape.\nSpeaker C: The rubber, it will look cheap always.\nSpeaker B: The colors you can make the colors with.\nSpeaker B: It lets beneath the buttons, pressure button, and you can disable the core and let for people that don't like it.\nSpeaker C: There's mobile phones in which you can change the color also of the lights.\nSpeaker C: Maybe we should consider this function to customize it.\nSpeaker C: Kids can make it look more flashy with different kinds of colors and people want something different or more design.\nSpeaker C: They can go for one color.\nSpeaker C: For example, this photo camera.\nSpeaker E: Cool.\nSpeaker D: Underwater.\nSpeaker C: Personally, I think it's really ugly.\nSpeaker C: Just give me the thing that's inside there.\nSpeaker C: Maybe I have to hold for this stuff.\nSpeaker D: Yeah.\nSpeaker E: Yeah.\nSpeaker D: Okay.\nSpeaker D: So, I think we're all...\nSpeaker D: I have bright colors.\nSpeaker C: This is with the curves that I mean.\nSpeaker D: Yeah.\nSpeaker D: That's a singly curve.\nSpeaker E: Yeah.\nSpeaker E: It should be nice.\nSpeaker D: But we could make a compromise between that, but I don't know if it's worth the effort.\nSpeaker C: A compromise between what?\nSpeaker D: Instead of doubly carved, we take a single curved.\nSpeaker D: So, to be only a more to the...\nSpeaker C: This will be a single curved.\nSpeaker C: Yeah.\nSpeaker E: This is only in this dimension.\nSpeaker E: Yeah.\nSpeaker E: Like this.\nSpeaker E: So, curve here.\nSpeaker E: Not the eye.\nSpeaker E: Exactly.\nSpeaker E: Yeah.\nSpeaker E: So, we keep it singly, single curved.\nSpeaker D: That would be an option, I don't know what you think.\nSpeaker C: I think that, I mean, our aim is to make something different, right?\nSpeaker C: To make something new.\nSpeaker C: I would go for the double curved.\nSpeaker C: Okay.\nSpeaker C: And I'm thinking, you know, a drawing palette where you have the shape for your thumb.\nSpeaker C: So, it kind of holds nicely, something like that.\nSpeaker B: Yeah.\nSpeaker B: Well, this is really your decision.\nSpeaker B: We can make more options for buttons if you have it in your hand.\nSpeaker B: You are not only...\nSpeaker B: Yeah, you don't need one dimension, but you can use other dimensions for pressing the buttons.\nSpeaker C: Yeah.\nSpeaker C: So, you can make a trigger button or something like that.\nSpeaker B: Something to shoot at your television.\nSpeaker C: Yeah.\nSpeaker C: Or that is the confirmation button or something that you scroll with your thumb with the arrows and then confirm.\nSpeaker C: That would be a nice way to use it.\nSpeaker C: But, I mean, I'm thinking big already and we need something that you can able to use in one hand, I think.\nSpeaker C: Oh, yeah.\nSpeaker E: So, from top view looks kind of like this.\nSpeaker E: But from the side, you can have it also in a kind of shape.\nSpeaker D: Maybe because the screen is on top, you can have buttons to interface.\nSpeaker D: If you turn it a little.\nSpeaker E: Maybe you can have this kind of shape.\nSpeaker E: A little upwards so that the screen is more...\nSpeaker E: Oh, yeah.\nSpeaker E:...through it yourself so you can easily see your screen.\nSpeaker E: Oh, well, you have it in your hand.\nSpeaker E: So, it's kind of an angle from your eyes to the screen.\nSpeaker E: So, then you have double...\nSpeaker E: Double carves in some way.\nSpeaker E: So, this...\nSpeaker E: This is so the screen is positioned over here.\nSpeaker E: Oh, something like that.\nSpeaker E: And the buttons are more... Oh, it's very thick now.\nSpeaker C: Yeah, I understand what you mean.\nSpeaker C: That's...\nSpeaker C: How about we do a pop-up screen like the laptop?\nSpeaker C: So that only the simple functions will be visible at first?\nSpeaker C: Yeah, that you can press it and then it comes up or...\nSpeaker D: Yeah, something like that.\nSpeaker D: So, you have the side view?\nSpeaker E: But then the side view can be straight if you have a pop-up screen.\nSpeaker E: I don't know if that's too expensive.\nSpeaker D: Yeah, it's too much.\nSpeaker D: Yeah, it's too much.\nNone: You want to be able to...\nSpeaker D:...make this...\nSpeaker C: No, like...\nSpeaker C: I would draw it like this.\nSpeaker C: Let's say this is the side view that you have a screen that will come up here and can go down that way.\nSpeaker C: You know what I mean.\nSpeaker C: So that it will come up like that.\nSpeaker D: Okay, so the buttons are on top here and you flip it over that way.\nSpeaker C: Yeah, or preferably even keep the simple buttons here.\nSpeaker C: Yeah.\nSpeaker C: And then under the screen even you...\nSpeaker C: Oh, the advanced buttons.\nSpeaker E: Yeah, yeah, yeah.\nSpeaker B: Right.\nSpeaker B: Yeah, that's good to be.\nSpeaker B: More advanced options wherever for the menu.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: For the LCD menu.\nSpeaker B: Okay.\nSpeaker B: Yeah.\nSpeaker B: So...\nSpeaker D:...to hide them all.\nSpeaker D: Yeah.\nSpeaker D: Not all, because you need most of them.\nSpeaker D: There are buttons.\nSpeaker D: But you can hide the okay in the back.\nSpeaker D: Yeah, yeah.\nSpeaker D: And then you can do it a little bit and also because when you flip it over...\nSpeaker D: Yeah.\nSpeaker D:...maybe if you drop the menu button because maybe if you flip it open it will...\nSpeaker D:...how to activate the mode.\nSpeaker C: I'll do it a little bit.\nSpeaker E: Activate the...\nSpeaker E: Yeah.\nSpeaker E: So...\nSpeaker E: Okay, but you have to...\nSpeaker E:...you need some button to flip this.\nSpeaker C: Right.\nSpeaker C: You choose...\nSpeaker C:...open...\nSpeaker C:...make it mechanical.\nSpeaker C: But you can make it...\nSpeaker E:...yeah, you can make a trigger here.\nSpeaker E: Yeah.\nSpeaker E:...you do a simple...\nSpeaker E: Yeah.\nSpeaker B:...with the...\nSpeaker B:...with the...\nSpeaker B:...with the...\nSpeaker B:...one...\nSpeaker B:...yeah, if you drop it one time...\nSpeaker B: True.\nSpeaker E: It can go open.\nSpeaker C: Yeah, the idea of it was...\nSpeaker C:...is that because you close it, you cover the LCD screen and it won't be vulnerable to scratches or whatever.\nSpeaker E: And it will be covered in some kind of thin rubber layer or something like that.\nSpeaker E: Exactly.\nSpeaker C: Exactly.\nSpeaker C: We just have to make sure that the closing mechanism won't break.\nSpeaker C: It's very solid, yeah.\nSpeaker C: Okay.\nSpeaker D: So that may work.\nSpeaker C: That actually will offer some extra protection...\nSpeaker C: Okay.\nSpeaker D:...we still have the...\nSpeaker D:...the thing of the...\nSpeaker D:...the shape.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I was thinking, if you have your hand, if this is your...\nSpeaker A:...harder.\nSpeaker E: Yeah, the lower part doesn't work, I guess.\nSpeaker E: So maybe you should try it over there.\nSpeaker C: Okay.\nSpeaker C: If this is your thumb and this is your hand like that...\nSpeaker C:...with your wrist, that it would be kind of shaped like this, you know?\nSpeaker C: So it's easier to hold in your hand to...\nSpeaker C: Yeah.\nSpeaker B:...with your left hand, that's your problem.\nSpeaker C: Yeah.\nSpeaker C: Yeah, of course.\nSpeaker C: Yeah, then you would have to...\nSpeaker C: Yeah.\nSpeaker C:...to make it like this, like you drew here.\nSpeaker C: Yeah.\nSpeaker C: And maybe then make this thicker also than the center.\nSpeaker C: Give it...\nSpeaker C: I would give it a female shape.\nSpeaker C: Yeah.\nSpeaker D: Yeah.\nSpeaker D: The female shape, yeah.\nSpeaker D: Obviously.\nSpeaker C: It could make some more appealing to guys.\nSpeaker B: Yeah.\nSpeaker B: The mouse, which you can change.\nSpeaker D: Oh, yeah.\nSpeaker D: So if you...\nSpeaker E: Yeah.\nSpeaker E: Yeah, but that's optional for later, I guess.\nSpeaker C: Yeah, but we have hardware inside, which is...\nSpeaker C:...so it has to have some sort of basic shape.\nSpeaker C: Yeah, we better...\nSpeaker C:...and also the screen, you can unmold it.\nSpeaker C: Yeah, it's okay.\nSpeaker E: We should better choose one shape.\nSpeaker E: Yeah.\nSpeaker E: But that's kind of the idea, so it lays good in the hand and then...\nSpeaker E:...it can be inside with your thumb.\nSpeaker E: You can place the screen here, you can use...\nSpeaker E:...co-on.\nSpeaker D: Yeah, you can use the button.\nSpeaker D: So the key wraps are primary colors.\nSpeaker C: But then I would...\nSpeaker C: Spongy?\nSpeaker C: Spongy.\nSpeaker C: I would do the arrows here, kind of thing.\nSpeaker C: Spongy can be written by myself.\nSpeaker C: And the simple buttons here, so...\nSpeaker E: Yeah.\nSpeaker E: And the control things in the middle, the...\nSpeaker E:...the arrows?\nSpeaker C: Yeah, that's what I mean.\nSpeaker C: The arrows over here.\nSpeaker C: Yeah.\nSpeaker C: And there are the numbers.\nSpeaker C: Simple.\nSpeaker E: Yeah.\nSpeaker D: Yeah, I think that's...\nSpeaker D: That's a nice design.\nSpeaker D: Pretty design, yeah.\nSpeaker E: That's cool.\nSpeaker E: Yeah, that looks pretty fancy.\nSpeaker D: Yeah.\nSpeaker C: Or do we want to make a shape look like a banana or something?\nSpeaker C: I don't know.\nSpeaker C: Yeah, it is.\nSpeaker E: Banana is a weird shape and other fruits also.\nSpeaker E: Yeah.\nSpeaker E: It's better to have some sort of basic print and then a fruit print in some...\nSpeaker E:...primary colors.\nSpeaker E: I don't know what colors you'd be like.\nSpeaker E: Like some soft green or something?\nSpeaker B: Or...\nSpeaker B:...and then...\nSpeaker E:...dark blue or...\nSpeaker E: Oh, yeah, yeah, dark blue.\nSpeaker E: And then...\nSpeaker E:...you should use...\nSpeaker E:...a very bright yellow banana or an orange, a green apple, stuff like that.\nSpeaker E: Very bright tones, I guess.\nSpeaker E: Yeah, we need very primary colors.\nSpeaker B: Like bright red, bright yellow.\nSpeaker B: Yeah, but the primary colors we can create with the lats on the...\nSpeaker B:...on the...\nSpeaker B:...\nNone:...\nSpeaker B: Yeah, okay, yeah.\nSpeaker B: If you make it just...\nSpeaker B:...give it a blue color, then it's just a neutral color.\nSpeaker B: Also for the more...\nSpeaker B:...yeah, for the older people.\nSpeaker B: That doesn't really work to draw.\nSpeaker E: It's...\nSpeaker E: I guess.\nSpeaker E: Oh, what's this?\nSpeaker D: This text.\nSpeaker D: Right.\nSpeaker E: No, you have text.\nSpeaker D: You can also make the line this.\nSpeaker E: Yeah, two hours further.\nSpeaker E: Yeah, yeah.\nSpeaker E: Well, thickness.\nSpeaker B: So that's blue?\nSpeaker E: Oh, why not go for the 20?\nSpeaker E: Yeah, that's what I call painting.\nSpeaker E: So that's a dark blue basic color, I guess.\nSpeaker E: Yeah, it's pretty nice.\nSpeaker E: And then...\nSpeaker E: Yeah, with some yellow banana...\nSpeaker C:...and how about some flashing standby lights?\nSpeaker C: Like, yeah, if only something...\nSpeaker C:...well, I don't like to call brands phones.\nSpeaker C: You know, that...\nSpeaker C:...cause I remember from one of our first meetings that people lose their remote controls often.\nSpeaker C: Yeah.\nSpeaker C: So not only in the colors of the lats, that we want something to keep it visible at all times.\nSpeaker E: How do you mean?\nSpeaker D: Sometimes flashing colors, so you can't lose it.\nSpeaker C: Yeah, exactly.\nSpeaker D: I think it's a bit too much.\nSpeaker E: Yeah, an orange.\nSpeaker E: Oh, all right.\nSpeaker E: Well, this is more like purple, I guess.\nSpeaker E: Yeah.\nSpeaker E: It should be more real dark blue.\nSpeaker E: So the contrast with the fruit of Jaxxis is pretty high.\nSpeaker E: Yeah.\nSpeaker E: So, yeah, it would be a nice device, I guess.\nSpeaker B: Which color should I give the display?\nSpeaker E: I mean, the color of the background of the display.\nSpeaker E: Well, I don't guess it has to be a...\nSpeaker A: 60...\nSpeaker E:...65,000 colors, so.\nSpeaker E: Yeah, too expensive.\nSpeaker E: So just a blue-blue backlight or something like that.\nSpeaker E: Green is too old-fashioned.\nSpeaker E: Not blue.\nSpeaker C: As long as you use high contrast between the background and foreground color.\nSpeaker C: Yeah, maybe I would say white.\nSpeaker B: White backlight.\nSpeaker E: White backlight and dark.\nSpeaker B: Dark letters.\nSpeaker E: Yeah.\nSpeaker E: Whatever, which is visible.\nSpeaker E: I know you should test it under light conditions.\nSpeaker E: I mean, it's hard to tell.\nSpeaker C: And also for people who are a bit color blind.\nSpeaker D: Yeah.\nSpeaker D: That's mostly red and green, I believe.\nSpeaker E: Which color should the buttons be?\nSpeaker B: That's adjustable.\nSpeaker E: Whoa.\nSpeaker E: White adjustable.\nSpeaker E: All buttons? Yeah.\nSpeaker B: No, that's how we are going to make it more trendy.\nSpeaker B: Or just a little expensive.\nSpeaker E: Maybe, I mean, they have to have some color, right?\nSpeaker E: And if the background is very dark blue, maybe green.\nSpeaker C: But don't we want to make the background the bright colors?\nSpeaker C: So the total, the thing is very bright.\nSpeaker C: Like the pictures I showed you guys.\nSpeaker C: Those things were all like bright red, bright red, fleshy.\nSpeaker E: So more like this color.\nSpeaker C: Yeah, something like that.\nSpeaker E: And then it stands out more.\nSpeaker E: Then yellow and orange and red objects on it or something.\nSpeaker E: But then again, which color should the buttons be?\nSpeaker E: The pressed buttons.\nSpeaker E: Should it be white or black or?\nSpeaker B: Red, maybe.\nSpeaker B: It looks quite cheap.\nSpeaker B: That's color, I think.\nSpeaker B: It's not green.\nSpeaker E: Yeah, but it's pretty fresh on the other side.\nSpeaker C: It's actually pretty trendy color at the moment.\nSpeaker C: It's trendy.\nSpeaker E: Yeah.\nSpeaker E: My couch is in that color.\nSpeaker D: Ooh, this is it?\nSpeaker D: Couch.\nSpeaker E: Yeah.\nSpeaker E: Well, it works pretty well.\nSpeaker E: And then time was up.\nSpeaker D: Thank you.\nSpeaker C: Do you get a pop-up?\nSpeaker D: Yeah, within five minutes.\nSpeaker D: That you have five minutes left or?\nSpeaker D: Yeah, left and then left to kick you off.\nSpeaker E: So something like this.\nSpeaker E: That should be pretty nice color.\nSpeaker E: But maybe the buttons in red is maybe a bad contrast for kind of blind people.\nSpeaker B: No, that's a bad option.\nSpeaker B: But the buttons had their own lads or not.\nSpeaker C: Yeah, the red and green are actually the easiest to discriminate, even for color blind.\nSpeaker C: Yeah.\nSpeaker C: They will see one of each as gray.\nSpeaker C: Yeah.\nSpeaker C: But if you use green on blue, those kind of colors will look the same.\nSpeaker C: Okay.\nSpeaker E: So red buttons are okay.\nSpeaker C: I think so.\nSpeaker B: Okay, that's the default setting.\nSpeaker B: The red buttons.\nSpeaker D: Yeah, I think you need to keep in mind that the lads are just extra light.\nSpeaker E: How do you mean?\nSpeaker E: Yeah.\nSpeaker D: They don't determine the color.\nSpeaker D: Okay.\nSpeaker B: I think about the red and blue lads.\nSpeaker B: That's too busy.\nSpeaker D: You have a background.\nSpeaker D: Each number is transparent.\nSpeaker D: Yeah.\nSpeaker D: Partly.\nSpeaker D: But you have to print on the number or the sign.\nSpeaker D: So it comes.\nSpeaker E: You better keep the left and the button itself in the same color, I guess.\nSpeaker E: So just an extra bit of light.\nSpeaker C: What we should do, I think, is make kind of a see-through plastic button with one color left behind it.\nSpeaker C: Yeah.\nSpeaker C: So that the whole button will shine as the color.\nSpeaker C: And if you think about easy to use buttons, we have to make it the shape so that it's easy to hold for both hands, but also that you can reach the buttons with your thumb if you hold the machine.\nSpeaker D: Okay.\nSpeaker D: Don't mean to discourage you.\nSpeaker D: Basically, you two are going to work on Smartboard.\nSpeaker D: Next 30 minutes to design.\nSpeaker D: All right.\nSpeaker D: And you will do the evaluation.\nSpeaker E: Of the product, which you don't have yet.\nSpeaker D: Yeah.\nSpeaker D: About.\nSpeaker E: So how should I do that?\nSpeaker D: Yeah, I don't know.\nSpeaker D: Okay.\nSpeaker D: You can't do me.\nSpeaker E: Or you send it to me.\nSpeaker E: Or just because you are going to design it on this board.\nSpeaker D: Right?\nSpeaker D: Yeah, but maybe you can do the evaluation.\nSpeaker E: You don't have to evaluate yet, but you can make a procedure for them.\nSpeaker E: Yeah.\nSpeaker E: I probably get instructions on how to do that.\nSpeaker D: So I make another presentation, I guess.\nSpeaker D: I know what's going to happen in life.\nSpeaker D: Yeah.\nSpeaker D: So you have a basic ID.\nSpeaker C: And you two are going to do this.\nSpeaker E: Yeah.\nSpeaker E: All right.\nSpeaker D: So that's.\nSpeaker D: Just make a new page and be creative.\nSpeaker E: Yeah.\nSpeaker B: All right.\nSpeaker B: We have to do it at this moment after.\nSpeaker D: Yeah, you have 30 minutes.\nSpeaker D: Okay.\nSpeaker D: Then we have to see something.\nSpeaker D: Yeah.\nSpeaker D: All right.\nSpeaker B: Show to the management.\nSpeaker B: This is basically what we are thinking about.\nSpeaker C: Yeah.\nSpeaker C: Shall we make a new.\nSpeaker B: Yeah.\nSpeaker B: Just delete all next.\nSpeaker C: Yeah.\nSpeaker C: Just make a new one.\nSpeaker C: All right.\nSpeaker D: Save this board.\nSpeaker C: Yeah.\nSpeaker C: I'll just keep it there.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nSpeaker D: Just press save.\nSpeaker D: It will be fine.\nSpeaker E: On the left.\nSpeaker E: Sorry.\nSpeaker D: We also need a good flip art.\nSpeaker C: Yeah.\nSpeaker C: So.\nSpeaker C: Okay.\nSpeaker C: Shall we make some outline sketches of the basic shape first?\nSpeaker C: Yeah.\nSpeaker B: Because then off that we can make the user own.\nSpeaker B: Yeah.\nSpeaker C: Because I have to focus on the basic look and feel design.\nSpeaker C: So how it's going to look.\nSpeaker C: And you have to think how we're going to put the screen in it.\nSpeaker C: And that's kind of thing.\nSpeaker C: So if I'm drawing and you think, okay, I'm not going to be able to put the screen in there.\nSpeaker C: Yeah.\nSpeaker C: I have to.\nSpeaker C: Correct.\nSpeaker C: Do you mind if I draw in black then for normal sketches?\nSpeaker E: I don't even use your interactions.\nSpeaker C: I don't even use your interactions.\nSpeaker C: Fine.\nSpeaker C: So we kind of want the polish.\nSpeaker C: Yeah.\nSpeaker E: So the plastic buttons also help because of their only...\nSpeaker E: I'm not so good at it.\nSpeaker E: During interaction.\nSpeaker E: So.\nSpeaker C: Excuse me?\nSpeaker E: Well, it's a good thing that the buttons aren't...\nSpeaker E: Well, that they are plastic because then you can light up the light on red when they are usable.\nSpeaker C: Yeah.\nSpeaker E: Maybe you should draw it very large.\nSpeaker E: Yeah.\nSpeaker B: Just part into filtering.\nSpeaker B: All right.\nSpeaker C: How do we erase?\nSpeaker C: Now our insert text.\nSpeaker D: I know maybe just start typing.\nSpeaker A: That's a bit big.\nSpeaker D: You also do the other sides, not only the front side.\nSpeaker D: Yeah, exactly.\nSpeaker D: The side view.\nSpeaker B: And let's make first all the views. The front view, side view.\nSpeaker C: I thought for the side view that the base exception would be rather...\nSpeaker C: Do you like a little bit?\nSpeaker C: A bit thicker than the middle.\nSpeaker C: Let's see.\nSpeaker C: Because your fingers have to fit underneath.\nSpeaker C: The middle has to be very small.\nSpeaker E: But then you sign that the meeting is over yet.\nSpeaker E: Exactly.\nSpeaker C: But the upper side has to be a bit more like that I think.\nSpeaker C: Yeah.\nSpeaker C: The display we will put in here, the basic functions and here where it's most reachable.\nSpeaker C: The arrow functions.\nSpeaker B: Exactly.\nSpeaker C: This is hard.\nSpeaker C: But don't have to draw it exactly, do we?\nSpeaker C: No, it's okay.\nSpeaker C: Wait, let me try it one more time.\nSpeaker C: It's easier if I draw it in once.\nSpeaker C: Okay, of course, it will become way more ugly.\nSpeaker B: You can make it larger.\nSpeaker B: Maybe it's easier to draw one.\nSpeaker C: So take forever.\nSpeaker C: It's fun to work with this band.\nSpeaker C: So larger.\nSpeaker C: That's the basic idea.\nSpeaker C: So five minutes left before the meeting ends.\nSpeaker C: Okay.\nSpeaker C: Other views?\nSpeaker E: All right.\nSpeaker B: Only if you are going to put buttons on the sides of the unit.\nSpeaker B: That's the question.\nSpeaker C: Let's fill in the buttons later.\nSpeaker C: So this is going to be from the side.\nSpeaker E: Because the screen goes up like that, right?\nSpeaker E: So then it's like this.\nSpeaker E: That's not convenient because then you have to screen like this and you look like this.\nSpeaker E: It's better to have it somewhat like this.\nSpeaker E: Or does it flip all the way?\nSpeaker B: Maybe it's easier to integrate the LCD screen just into this bubble.\nSpeaker B: Okay.\nSpeaker B: Because it doesn't have to flip then.\nSpeaker B: Because it's more space for making an LCD.\nSpeaker E: It's better to have this like this, I guess, and then flip it like this.\nSpeaker B: Why do we need the flipping?\nSpeaker C: You can adjust the angle to which it flips.\nSpeaker C: So it can also from this angle it can flip all the way up to there.\nSpeaker E: It can flip it up to there if you want.\nSpeaker E: So yeah.\nSpeaker E: But we still keep the flipping mechanism.\nSpeaker C: Think so?\nSpeaker C: Yeah, we keep the flip.\nSpeaker C: Because I think it will feel weird if you would make this smaller and this bigger or something that I don't know.\nSpeaker B: The shape is okay.\nSpeaker B: I don't see the, yeah, why we should use the flipping mechanism.\nSpeaker C: Yeah, I thought it would be cool.\nSpeaker E: Okay, yeah, but maybe we should then.\nSpeaker B: Because we have enough space.\nSpeaker B: Here we go to the basic function.\nSpeaker B: Yeah, there are the middle arrow.\nSpeaker E: Shouldn't we integrate it?\nSpeaker E: Yeah, and then we should.\nSpeaker E: Like, oh.\nSpeaker E: Dude.\nSpeaker E: So this is the shape.\nSpeaker E: It doesn't.\nSpeaker C: Yeah, it doesn't seem so well.\nSpeaker E: Okay, but the screen is a bit lower because if it falls on the other side, it doesn't fall on the screen.\nSpeaker E: So there's a layer of rubber on the sides.\nSpeaker E: Yeah, definitely.\nSpeaker E: So no flipping, but just.\nSpeaker B: No flipping?\nSpeaker B: No flipping?\nSpeaker B: You wanted the flipping.\nSpeaker C: Yeah, I guess, but I mean, most of those count, right?\nSpeaker B: If you drop it, it just breaks.\nSpeaker B: And it has to be very strong because of the.\nSpeaker E: Throw in the kinetic.\nSpeaker E: So it better make it like this.\nSpeaker C: Yeah, if you're going for the kind of.\nSpeaker B: Yeah, for people not 16 years.\nSpeaker E: They're already responsible enough to have a mobile phone.\nSpeaker E: It's also to do with their remote control.\nSpeaker E: Yes.\nSpeaker C: Yeah.\nSpeaker C: Well, I'm just thinking totally different designs also.\nSpeaker C: Remember that the weird pocket phone thing here, which looked like kind of a game boy.\nSpeaker C: Maybe we should try something like that.\nSpeaker D: I thought up a name for our product.\nSpeaker D: Yeah.\nSpeaker D: It's called the real remote.\nSpeaker D: Oh, all right.\nSpeaker D: Copyright signed off the real.\nSpeaker D: Yeah.\nSpeaker E: All right.\nSpeaker C: Yeah, I like it.\nSpeaker E: Good.\nSpeaker E: We should work in our own room, right?\nSpeaker C: Oh, yeah.\nSpeaker C: We have to include that in our designs as well.\nSpeaker C: Yeah.\nSpeaker E: How's that?\nSpeaker E: Okay.\nSpeaker D: See you two and a half an hour.\nSpeaker D: Okay.\nNone: Good luck.\nSpeaker B: Yep.\nSpeaker B: Okay.\nSpeaker B: That's a side.\nSpeaker B: It's okay.\nSpeaker B: But do we want options on the side with a little bit of buttons?\nSpeaker C: Yeah.\nSpeaker B: Or I think we do.\nSpeaker B: We could make an volume button scroll.\nSpeaker B: Yeah.\nSpeaker C: Or also scrolling for the.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker C: I don't know.\nSpeaker C: We kind of wanted to stick with the.\nSpeaker C: Yeah.\nSpeaker C: I'm just thinking if we want to make something different, right?\nSpeaker C: So with the scroll is more futuristic than the standard arrow buttons.\nSpeaker C: Yeah.\nSpeaker C: I think.\nSpeaker C: Yeah.\nSpeaker B: You are going to design it.\nNone: Yeah.\n"}, {"meeting_id": "IS1002c", "summary": "This was a conceptual design meeting, at which some decisions on product components, functional design, and pricing were made. The group spent a lot of time discussing the components of the device, especially the scroll wheel and power source. The main topics in the discussion about functional design were speech recognition and the function of rolling through the user's favourite channels. The former was abandoned while the latter was adopted by the group for its novelty and feasibility. After discussing the functions, the group thought they were approaching the high-end market, so they could push up the price. In spite of that, the project manager supposed the profit expectation might not be so important as opening up a new market and promoting their brand. In the end, they agreed to price the product at thirty-five to fifty Euros.", "dialogue": "Speaker B: So do we need to retrain Mike on how to put his mic on?\nSpeaker B: Can he get it all by himself this time?\nSpeaker D: I don't know, I'm feeling like a bumble.\nSpeaker D: Probably not.\nSpeaker B: I believe I can fly.\nSpeaker C: We got some exciting stuff for you guys.\nSpeaker C: Or not.\nSpeaker D: Or not.\nSpeaker B: Just when I needed something exciting.\nSpeaker B: Remember I'm an old man.\nSpeaker A: Okay, ready to go.\nSpeaker A: Alright, ready.\nSpeaker A: Okay, so we've got a conceptual design meeting.\nSpeaker A: Hopefully we've all got exciting ideas now.\nSpeaker A: We do.\nSpeaker A: Exciting ideas.\nSpeaker A: Okay, so here's our agenda.\nSpeaker A: Our agenda.\nSpeaker A: I've got it open.\nSpeaker A: I'm going to talk for a bit about what we're going to do.\nSpeaker A: I'm going to take some notes.\nSpeaker A: I'm going to do a presentation.\nSpeaker A: And then hopefully we're going to make some decisions now.\nSpeaker A: Yep.\nSpeaker A: Well, when I say hopefully we have to.\nSpeaker A: So I'm going to let you guys talk before we make decisions.\nSpeaker A: Does anyone really want to go first?\nSpeaker C: I guess I'll go first.\nSpeaker A: You too?\nSpeaker C: Yeah.\nSpeaker C: What component I think.\nSpeaker D: Yep, that's it.\nSpeaker D: Presented by name.\nSpeaker C: My name is Jose.\nSpeaker C: My name is name.\nSpeaker D: My name is Indigamontaya.\nSpeaker D: Indigamontaya.\nSpeaker D: Sorry, I did this in a bit of a ride.\nSpeaker C: Nani!\nSpeaker C: So here's a look inside your really old looking remote control.\nSpeaker C: You've got a printed circuit board here.\nSpeaker C: And you've got all these buttons which kind of press down little rubber nubbies into these little holes.\nSpeaker C: We've all broken a remote control.\nSpeaker C: It's all broken a remote control.\nSpeaker C: So you've also got your chip here, your batteries here, and some sort of electronics.\nSpeaker B: I just love you tech guys.\nSpeaker D: Yeah, there's a thingy in the game.\nSpeaker C: Well, do jabber.\nSpeaker C: So here's a transistor and this amplifies your signal.\nSpeaker C: You've got the LED here on the end of the printed circuit board.\nSpeaker C: You've got a couple of diodes here for, I don't know, who and what not.\nSpeaker C: Exactly.\nSpeaker C: So we've got in this drawing here, this example here, this is an 18 pin chip.\nSpeaker C: Yeah, I don't know.\nSpeaker C: It's two AA batteries.\nSpeaker C: This is pretty standard remote.\nSpeaker C: So here are options for our power sources.\nSpeaker C: You can use a basic battery, which we've already discussed.\nSpeaker C: Our tech department also said we have the option of doing some kind of hand dynamo, or maybe you crank it or something like that.\nSpeaker C: I don't know if that's really...\nSpeaker C: I don't know, we've got some crazy guys down there in that department.\nSpeaker B: I got a flashlight.\nSpeaker B: You shake it?\nSpeaker B: Yeah, but it's interesting because you shake it like this.\nSpeaker B: Something like this.\nSpeaker C: Yeah, so that's the next bullet.\nSpeaker B: That's on the camera.\nSpeaker C: The kinetic provision of energy.\nSpeaker C: It's like that flashlight where you have to shake it.\nSpeaker C: We've got solar cells, which I don't think is a very good idea because you could not use your remote at night.\nSpeaker C: Dude, this doesn't make a lot of sense.\nSpeaker C: And finally, we've got our power cradle idea.\nSpeaker A: Okay, so we basically have battery versus cradle here.\nSpeaker C: Battery versus cradle, I think is...\nSpeaker D: I like the kinetic.\nSpeaker D: I think you would.\nSpeaker A: It's actually a novel thing because you could sell it as a novelty.\nSpeaker A: Just to be actually serious for a minute here.\nSpeaker C: Well, it is more eco-friendly than the cradle because you're still using power off the grid with the cradle.\nSpeaker C: So, our case design.\nSpeaker C: We have choices in materials and choices in the general shapes that we can do.\nSpeaker C: Our material choices are plastic latex type or plastic, a rubber latex type thing, wood or titanium.\nSpeaker C: If we go with titanium, we're going to be limited in the amount of shapes we can do because it's tough to shape the titanium.\nSpeaker A: And yeah, wood gave us a lot of the marketing.\nSpeaker C: I think wooded...\nSpeaker C: I can't see anybody wanting to use a wooden remote.\nSpeaker C: It's just anti-technology really.\nSpeaker B: To me, in a marketing sense, it's not relative.\nSpeaker B: We can accentuate whatever product you put in there.\nSpeaker B: We can find a way to accentuate.\nSpeaker C: And what we may be able to do, and I think this might be the best option, is to combine a couple of these.\nSpeaker C: My recommendation personally would be to do some kind of a plastic inner shell with a rubber outer shell.\nSpeaker C: To make it like a thick plastic inner shell and kind of a...\nSpeaker C: To have that rubber outer shell to make it more durable.\nSpeaker C: And also maybe it feels a little better than the plastic.\nSpeaker A: You get your wrap on the rubber.\nSpeaker B: And if you make it from that super rubber, when you drop it on the floor, it can bounce right back up in your hand.\nSpeaker B: You don't even need to lean down to get it.\nSpeaker C: The advantage of working with plastic and rubber is we'll have a lot more options in terms of shape because you can extrude plastic and basically any shape you want.\nSpeaker C: And then we can cover the breakable bits with rubber.\nSpeaker C: But basically these are curved and double curved.\nSpeaker C: I believe the tech department in their message to me that they were referring to the number of curves in the bottom.\nSpeaker C: I have no idea exactly what they're talking about.\nSpeaker C: But that's what they told me.\nSpeaker C: Uncurved, flat, curved or double curved.\nSpeaker C: I would guess this pen would be kind of like a double curved.\nSpeaker C: Where it's curved on multiple axes.\nSpeaker C: I think curved means just curved in one axis and double curved is curved in two axes.\nSpeaker A: I think it might mean something like that sort of shape because of double cover.\nSpeaker A: Oh, okay.\nSpeaker A: That's where they're shaped.\nSpeaker C: That makes sense.\nSpeaker C: Okay.\nSpeaker C: With Interface we have the following options.\nSpeaker C: We can use push buttons.\nSpeaker C: We can use a scroll wheel with an integrated push button and LCD display on multiple scrolling wheels.\nSpeaker C: So these are all options that the user interface guy has his disposal to put together a user interface.\nSpeaker C: Okay. For electronics we have these very technical descriptions here.\nSpeaker C: A simple chip which is the least expensive but I have no numbers to give you.\nSpeaker C: A regular chip which is like the medium porridge.\nSpeaker C: The medium.\nSpeaker A: Did we have actually any concept of what the difference is between a simple chip and an advantage chip?\nSpeaker C: Yes. The difference is with a simple chip, a simple chip will operate.\nSpeaker C: So when is it going to scroll up?\nSpeaker C: A simple chip is required to operate push buttons.\nSpeaker C: An advanced chip is required to operate the LCD display.\nSpeaker C: It didn't say specifically but I have a hunch that a regular chip is going to be the scroll wheel and the multiple scroll wheels.\nSpeaker C: Okay.\nSpeaker C: So, but yeah.\nSpeaker A: Yeah, that makes sense.\nSpeaker A: Let's see in my presentation.\nSpeaker A: Presentation from I guess design which go best.\nSpeaker A: Next.\nSpeaker A: Technical functions are interface concept.\nSpeaker D: I think.\nSpeaker D: Oh, interface concept.\nSpeaker A: Yeah, that's right.\nSpeaker A: Yes, but it has your name on it.\nSpeaker D: Well, that doesn't board well for it as well.\nSpeaker D: So, somehow the thing is too big.\nSpeaker D: But, okay.\nSpeaker D: Our manufacturing division wanted a speech recognition.\nSpeaker D: They say they could put it to work but we don't think so.\nSpeaker D: No.\nSpeaker D: But you'll be, you know, be affected by the other schemes.\nSpeaker C: TV is working.\nSpeaker C: Yeah.\nSpeaker C: It's just, I mean, if somebody says up in the middle of the television show, it's going to change the channel.\nSpeaker C: Yeah.\nSpeaker D: And then fighting for the remote would not be fun anymore.\nSpeaker D: And I think that's one of the things we want.\nSpeaker A: But what if he actually had to press a button to make it recognize?\nSpeaker A: So if you press that and run up.\nSpeaker C: That kind of would just press the up button.\nSpeaker A: Yeah.\nSpeaker A: That would kind of.\nSpeaker A: But if it's just one thing with a button, you can just go.\nSpeaker C: That's right.\nSpeaker C: Even still there's going to be interference.\nSpeaker C: There still will be interference from the TV.\nSpeaker C: It might not be completely confusing but I think you'll still, I don't think it's practical at all.\nSpeaker C: I think it's a bad idea, Frank.\nSpeaker D: Okay.\nSpeaker D: I'm thinking that away.\nSpeaker D: You guys may understand.\nSpeaker D: The robot buttons are more reliable.\nSpeaker D: It's the ones that would allow us to market our product as being less prone to damage and more resistance to things like spillage of liquids over it or mistreat, misuses as it happens to remote controls.\nSpeaker D: As for the point that we're making about losing it, well, we want a small remote control on side because we want it to be cool and designed.\nSpeaker D: But apparently the market shows that bigger, bigger remote gets less lost than I would\nSpeaker C: believe.\nSpeaker D: But yeah, I think we need to compromise between those two and somehow we'll do that somehow. But what I would propose is something more or less in the direction of what is to you the right of that slide but with a less complicated design.\nSpeaker D: The numbers, the volume control and channel control and teletext access, the volume and channel control can just become the four button array as in if it was a round dial.\nSpeaker D: If you just think about it as the one to the right but with the numbers and the four buttoned plus maybe a center one with teletext.\nSpeaker A: So we're suggesting a scroll wheel thing for the volume?\nSpeaker D: No, it's not a scroll wheel.\nSpeaker D: It's just four buttons that are on the cross.\nSpeaker D: Okay, basically you can control all of the important tasks from that.\nSpeaker C: Instead of place, stop rewind and fast forward there.\nSpeaker C: That's up, down, louder and quieter.\nSpeaker D: Yeah.\nSpeaker D: So I think we go for something mid-size, something that looks good and not too prone to get lost.\nSpeaker D: That be it.\nNone: Okay.\nSpeaker A: So on to...\nSpeaker A: Functional requirements are trained watching.\nSpeaker B: I don't know.\nSpeaker A: Train watching is a later date there.\nSpeaker B: It looks like it's been so long.\nSpeaker B: I don't know what to say.\nSpeaker B: When I see the product, I don't want to buy it.\nSpeaker B: I see so many of them out there.\nSpeaker B: There's nothing about that product that makes me want to choose that product over other products that are out there.\nSpeaker B: Are you talking about the picture?\nSpeaker C: That's not a design.\nSpeaker C: That's just something a graphic he used to show you the layout of what the layout of the buttons might be like.\nSpeaker B: Okay, because right now I don't have too much to say about how to market this product because we don't have a product to market yet.\nSpeaker B: And from talking to Mike is we can market a more expensive product now.\nSpeaker B: That's what I understand.\nSpeaker A: I haven't said yes.\nSpeaker B: Hello.\nSpeaker B: I'm a little bit stuck right now.\nSpeaker B: What is it that I'm going to market?\nSpeaker B: Without special or increased marketable features, I don't believe the product has a consumer demand.\nSpeaker B: I like the idea of the scroll.\nSpeaker B: There are so many people making these products at this price right now.\nSpeaker B: What are we going to do to make this one special and unique?\nSpeaker C: What's special and unique about a scroll?\nSpeaker B: I don't kill.\nSpeaker B: I just see it as different.\nSpeaker B: I don't say it's special.\nSpeaker B: I say that it's different.\nSpeaker B: What I'm looking for is marketing is give me something different.\nSpeaker B: It gives me a lower price, a higher price, some new technology.\nSpeaker B: Don't give me the same thing that everybody else is putting out there on a shelf.\nSpeaker B: It's the same price.\nSpeaker B: I need something to market about this thing.\nSpeaker A: If we do make the decision to go with the cradle, we have that as well.\nSpeaker B: When we have something like the cradle or something, give that as a marketing standard, I need something to market to make this product unique.\nSpeaker C: I think the two big points that we have so far are having the cradle and also having the actual design of the case itself, having this rubber shell with a plastic interior, having it look really nice and also be really durable.\nSpeaker B: Just remember, when I made up this report, I didn't have the information that we're discussing here.\nSpeaker B: When we have a cradle, when we have some kind of design, from my perspective, I don't have a product to market right now.\nSpeaker B: My personal preference is that we make some adjustment in the cost, either lower or develop and integrate new technology.\nSpeaker B: That is the next step.\nSpeaker B: There's technology and then there's technology, which we're moving into the next phase.\nSpeaker B: We're going to have some new technology to enhance the marketability.\nSpeaker B: Again, I'm not sold on the product because we don't have a product in my opinion yet.\nSpeaker B: Let's get a product.\nSpeaker B: I need a product to market.\nSpeaker B: Whatever products you guys put together, we'll find a way to market it.\nSpeaker B: That I'm not concerned with.\nSpeaker B: Now, if you give me a cut out of what everybody else has, then I need to find a lower cost.\nSpeaker A: Our big questions here really are cradle or not cradle.\nSpeaker A: Do we go basic or do we go for features?\nSpeaker A: Does anybody really want to do anything with the scroll wheel or should we ditch that?\nSpeaker C: My question is what would the scroll wheel do?\nSpeaker C: Function wise, what does that do?\nSpeaker A: That was in your presentation.\nSpeaker A: What would you imagine doing?\nSpeaker C: It's just another way to do the exact same thing that the buttons do.\nSpeaker B: What I see with the scroll wheel is everybody has buttons.\nSpeaker B: From a marketing standpoint, I have another door to walk through when we have something unique.\nSpeaker B: If it creates something more complex or more expensive, then I don't know if that makes it a marketing necessity.\nSpeaker B: Again, from a marketing perspective, I want as much new and different about this thing as possible because it's a very competitive market.\nSpeaker A: I understand that the scroll wheels are quite expensive to make.\nSpeaker C: They are going to be more expensive.\nSpeaker C: On the other hand, it is an alternative.\nSpeaker C: If you think that's going to sell some...\nSpeaker C: If we're going to make up the extra cost by extra sales, I think maybe it's worth it to do it.\nSpeaker C: If I was going to do this, I would just use the scroll wheel for channels up and channels down.\nSpeaker C: Do you think volume would be?\nSpeaker D: I would first involve it.\nSpeaker C: No, we can do multiple scroll wheels.\nSpeaker A: I like the idea of basically focusing the product on the idea of the stock to target environmentally friendly ideas.\nSpeaker A: It makes it different to market.\nSpeaker A: I think so.\nSpeaker A: The whole product is going to need to be more durable.\nSpeaker A: You're not going to be ditching it as often.\nSpeaker A: It's going to need to be more expensive because of the cradle.\nSpeaker A: If we can market it in terms of that.\nSpeaker B: I think we've got some exclusivity in it.\nSpeaker B: We've got something that nobody else has right now.\nSpeaker B: That means we can make some adjustments in the retail or wholesale price if we need to.\nSpeaker B: It also can create its own demand from its uniqueness.\nSpeaker A: Our big decision then is how do we do...\nSpeaker A: We have to decide on the details and how to decide on...\nSpeaker A: Well, not exactly the details, but do we have what type of casing?\nSpeaker A: I personally like what you were saying about the plastic with the rubber.\nSpeaker C: It's something almost like these pens where you have plastic bits.\nSpeaker C: It's not really rubber, but just a fused plastic with rubber on the outside.\nSpeaker A: Just a very thin sheet for a grip.\nSpeaker C: Maybe a bit more than is on here.\nSpeaker C: Just as an inspiration, these pens I think are kind of...\nSpeaker A: So then, do we want to do anything more basic with the more advanced with the user interface in terms of...\nSpeaker A: Do we want to go for buttons or do we want...\nSpeaker A: If we're ditching the scroll fields, do we want to go LCD?\nSpeaker A: Do we want to give... do we want to have anything else on it?\nSpeaker D: It's only a TV.\nSpeaker C: From my perspective, I think LCD is a mistake because this is a universal remote and all you're doing...\nSpeaker C: You're not going to get any information back from the television, so the only information you can display in this remote is what channel you just sent it at last.\nSpeaker C: There's just not a lot of information there.\nSpeaker C: It receives no information.\nSpeaker C: So I don't see any reason for having an LCD because it's an increased cost.\nSpeaker C: Unless you can think of something interesting to do with it.\nSpeaker B: Right now, there's nothing coming out of the TV to receive.\nSpeaker B: So unless technology changes and information starts coming out of a cable box or something, I don't see the necessity of an LCD either.\nSpeaker B: I don't see it.\nSpeaker A: Okay, so the question is now, I guess, we need to decide.\nSpeaker A: You guys basically have to now go and figure out the details of this thing.\nSpeaker A: So what we need to know in terms of marketing and project management are...\nSpeaker A: Are there any other questions that we need to answer now before you guys can go and build this?\nSpeaker A: What overall things have we not decided on?\nSpeaker C: Well, I think for me it's still not exactly clear exactly what the user interface is going to be.\nSpeaker C: There's a scroll wheel in or out.\nSpeaker C: What do you guys like in the user interface?\nSpeaker A: Maybe in terms of marketing, is that going to make it off?\nSpeaker A: Again.\nSpeaker B: That's going to justify the cost.\nSpeaker B: Well, I think it's the more uniqueness you can bring to the product, the easier it is, I believe for me to market.\nSpeaker B: Again, the push buttons I see are everywhere.\nSpeaker B: So we can go with the same thing, but we're going to be competing in a broader market than if we go with something unique.\nSpeaker B: The other thing I thought about was, do we go to something like this?\nSpeaker B: Make a remote that doesn't look like a remote.\nSpeaker B: It's just an idea, and I don't know...\nSpeaker C: I don't know about the flip phone idea, because I think as far as durability, it's not a big...\nSpeaker B: I mean, what I see, one of the things you brought up in an earlier presentation is, and you've got children, there's stuff that gets inside the circuitry, they get dirty, they get messy with drinks and stuff, this.\nSpeaker B: And what I keep throwing out there, I just keep throwing out ideas to try to make this thing look or act, or in some way, identify itself as unique or different.\nSpeaker B: But you guys are the ones that have to ultimately create the product within cost.\nSpeaker A: Yeah, and I guess the question that you're being asked right now is whether, is the dock enough of a unique feature to be able to go out and sell that as a very different product?\nSpeaker A: Or do we really need the scroll wheel as well?\nSpeaker A: Oh, okay.\nSpeaker A: I said quite a fast...\nSpeaker B: Yeah.\nSpeaker B: I think you've got to get into cost effectiveness.\nSpeaker B: I think if you can have the base station with a locator, I think those are two very strong features.\nSpeaker B: If that's something that can be integrated without a much extra cost.\nSpeaker C: The locator is going to require a radio transmitter, which will...\nSpeaker A: Is that going to need a better chip as well?\nSpeaker C: We're probably going to have to go with it like a medium chip, I would imagine.\nSpeaker C: But we will need a receiver in antenna.\nSpeaker B: Integrated, yeah.\nSpeaker B: So, it's just... I think that's...\nSpeaker C: It should be a really simple signal, though.\nSpeaker B: I know personally that would be a very attractive feature is to have a button I can push to find my remote control.\nSpeaker A: I don't know if it's a little bit of a signal, but I guess a little tiny speaker is going to be quite cheap.\nSpeaker A: It's not going to need quality, is it?\nSpeaker C: Oh, yeah.\nSpeaker C: It'll be really cheap.\nSpeaker B: I remember you could have a smell of ramma, you know, you push a button and it puts out a stink.\nSpeaker B: Maybe not.\nSpeaker A: Makes your 11 room more fresh as well.\nSpeaker A: Okay, we're doing well for time here. We've got about another 10 minutes.\nSpeaker A: I think that...\nSpeaker C: So, the scroll wheel.\nSpeaker C: Interrupt.\nSpeaker A: My personal preference is out.\nSpeaker A: I don't think the cost is justified for a little bit more uniqueness when we already have that.\nSpeaker A: And I don't think it's obvious and natural how it would be used.\nSpeaker D: Yeah, I think we have, like, it's not very usable and it will bring down the robustness of the whole thing.\nSpeaker D: Yeah, it breaks down easier.\nSpeaker C: For me, I think the scroll wheel actually might not be so bad.\nSpeaker C: I don't know exactly what the increased cost is going to be, but I think he does have a point.\nSpeaker C: It might push somebody over the edge when they're looking at our remote versus something else when they see this one has a scroll wheel to go up and down on the channels.\nSpeaker C: I think it might be kind of neat to be able to do it like that.\nSpeaker A: But then that surely depends a little on the TV because some TVs are quite slow at changing channels.\nSpeaker A: Sure.\nSpeaker A: So if you've got a scroll here and you have to scroll it really, really slowly just so that you're actually keeping in pace with the TV's ability to change channels.\nSpeaker A: You have to go through and you wait for it.\nSpeaker A: You scroll a bunch of times and then you wait for it.\nSpeaker C: Well, I think what it would be is like this where it's maybe a digital wheel, where it's quantized into a circle.\nSpeaker A: Ooh, yes, that's where you go.\nSpeaker A: My intuition about the scroll wheel would be.\nSpeaker A: Ah, see, see if something about now.\nNone: Okay.\nSpeaker C: It's just a look and feel thing.\nSpeaker C: It has the same exact functionality as two buttons.\nSpeaker B: I think there are so many people today that are surfing, or television surfers, and I see the scroll as a great mechanism for surfing instead of going button to button.\nSpeaker D: Yeah, you just said they're going.\nSpeaker B: Yeah, I really think that's a really cool thing for surfing.\nSpeaker C: Now, just so you know though, you did bring up a point which is very valid, is a lot of TVs won't respond exactly the same.\nSpeaker C: Some of them are going to be kind of slow switching.\nSpeaker C: So you may like queue up like 15 channel changes and it'll be like flip, flip, flip.\nSpeaker A: Well, there's not necessarily you could basically make it so that it'll, I mean, it's just going to be sending a signal to the TV.\nSpeaker A: So if you send about five flip channels, if you did that, it's going to flip once.\nSpeaker C: Yeah, it just might be frustrating where you can't make it go as fast as you want.\nSpeaker C: But I think once people get used to it, I do like the idea of the scroll wheel though.\nSpeaker A: And if we're marketing it as a high niche product, then we're going to be selling it to people who are buying good TVs as well.\nSpeaker A: Yeah, presumably.\nSpeaker A: Yeah, I think so.\nSpeaker B: I think so.\nSpeaker B: The only thing on the interface side of it is that I see the dilemma.\nSpeaker B: But if we have the option of scrolling, check it at any particular speed or the option of jumping direct.\nSpeaker B: Okay?\nSpeaker B: So I can go presuming I have on my television something that tells me what channel I'm on.\nSpeaker B: I can scroll direct from channel five to channel 32.\nNone: Shunk.\nSpeaker B: I know what because it's on a television.\nSpeaker B: The television tells me what channel it's on when I change it.\nSpeaker B: So I don't know that really that's going to be a problem.\nSpeaker B: Because a television can go automatically from channel five to channel 32 with a push of a button.\nSpeaker B: Okay?\nSpeaker B: So can we create that kind of interface within a scroll system?\nSpeaker B: You understand what I'm saying?\nSpeaker C: I think I know what you might be getting it.\nSpeaker C: Oh, I see.\nSpeaker A: So maybe if we had an LCD of the top that just did a number on it, right?\nSpeaker A: Or we could read it from a television.\nSpeaker C: What about this?\nSpeaker C: What about if you can program in your favorite channels into this scroll wheel and you can just like roll through your favorite channels?\nSpeaker A: That's great.\nSpeaker A: You need that display on the...\nSpeaker A: Why?\nSpeaker A: It'll tell you when you flip the channel.\nSpeaker B: The television can tell you.\nSpeaker B: Yeah.\nSpeaker A: Can.\nSpeaker A: Oh, yeah, yeah.\nSpeaker A: No, I see what you're talking about.\nSpeaker A: Yeah, that's kind of cool.\nSpeaker C: How would you program it though?\nSpeaker C: Oh, it's just one extra button.\nSpeaker C: You say program start and then type in.\nSpeaker C: Because you still have the typing, you know, you'll still have the keypad where you can type in manually.\nSpeaker C: So the program start 0 1, enter 0 5, enter 38, enter program end.\nSpeaker A: Okay.\nSpeaker A: And yeah, and that just basically flips between it and it will go.\nSpeaker A: It sends out 0 5 and then 36 and then 0 1.\nSpeaker A: Yeah, that's kind of cool.\nSpeaker A: I like that.\nSpeaker B: And again, we have another great marketing tool.\nSpeaker B: We have about three.\nSpeaker A: We have three or four things here.\nSpeaker A: That's not going to be too expensive because that's going to be...\nSpeaker A: You're going to be able to nab that off of computer mouse manufacturers really.\nSpeaker A: You can basically come up with a partnership to be able to produce that quite cheaply.\nSpeaker B: I don't know.\nSpeaker B: And maybe we maybe even have this as in-house technology.\nSpeaker B: This may be something that's available through our own services.\nSpeaker A: Maybe.\nSpeaker A: But that's not going to be such a costly feature.\nSpeaker A: The problem we're going to have is making it robust.\nSpeaker C: Well, we also have to determine in some manner how to switch between modes, between going through your favorites list and just hitting up 1, up 2.\nSpeaker B: We go directional.\nSpeaker B: So if there's a button...\nSpeaker B: We go this way for one, we go this way for the other.\nSpeaker C: No, because you want to be able to go up and down through the channels.\nSpeaker B: Yeah, people are going to have a favorite sort of whether they do that or whatever they do.\nSpeaker B: Well, then you just have a different...\nSpeaker B: You have a mode switch.\nSpeaker C: I think we'll need a mode switch.\nSpeaker C: But then if we have a mode switch, we're going to need some kind of indicator to switch.\nSpeaker A: It's a white behind buttons.\nSpeaker A: And LED.\nSpeaker C: And LED.\nSpeaker C: Okay.\nSpeaker A: Does that work?\nSpeaker A: Yeah.\nSpeaker A: Is that...\nSpeaker A: Okay, we have five minutes.\nSpeaker A: So, right, details that we've talked about here are that we want a scroll wheel.\nSpeaker A: We want a mode indicator.\nSpeaker A: We want backlit buttons.\nSpeaker A: And if we're making backlit buttons, period, do we want that just for the mode indicator or maybe to indicate what button you're pressing at the time so that you know if it's actually pressed or not?\nSpeaker A: I've seen some remotes do that.\nSpeaker C: Okay.\nSpeaker C: Just so you know, I think...\nSpeaker C: I mean, it was my understanding that before we were going to stay in the mid-market range, it seems we're kind of approaching a higher end range.\nSpeaker C: I think we are.\nSpeaker B: I want to make sure everybody's okay with that.\nSpeaker B: Well, you had acknowledged that we have more money for this.\nSpeaker A: Yeah, well, we don't have...\nSpeaker A: It's not that we have more money, we can push up the price.\nSpeaker A: That's what I mean.\nSpeaker B: We can increase the cost.\nSpeaker B: Okay.\nSpeaker B: So, I don't know.\nSpeaker B: I don't know how we're having to...\nSpeaker C: Just want to make sure everybody's on board with it.\nSpeaker A: Yeah.\nSpeaker A: So, we really need to be sure as to what we can push the cost up to, though we haven't got to a stage where we're ready to pin down the price of components.\nSpeaker A: But, I'd say, what sort of price are we looking to be able to sell, something like that, and what sort of price can we make it for?\nSpeaker B: Well, I think the design and technology has to come back and say, okay, to create this product, we see it's going to cost us this much for this volume, because we do have a volume target of 50 million profit.\nSpeaker A: Yeah.\nSpeaker A: No, I think that's where we really should be more flexible than anything else, because as we said in the last meeting, our management's really looking for us to push our brand.\nSpeaker A: Right.\nSpeaker A: We're entering a new market here, so I think the profit expectation for this one product is maybe not as important as being able to...\nSpeaker B: Okay.\nSpeaker B: I'm okay with that.\nSpeaker B: I guess what I just want...\nSpeaker B: To me, the next step is for these guys to come in with a design proposal with the cost estimate attached.\nSpeaker B: And then we have to take this to the next level.\nSpeaker A: That's a ball-parkest of what we'd be able to sell, something like that.\nSpeaker A: Well, yeah.\nSpeaker C: Let's try to think now.\nSpeaker C: How much would you pay for... with all these features, how much would you pay if you went to the store and you were in the market to replace your TV remote?\nSpeaker C: How much would you pay for that?\nSpeaker A: Well, you got to think who our target market is, because I'm not our target market.\nSpeaker A: I'm a student.\nSpeaker A: But if, on the other hand, I would think...\nSpeaker A: With my project manager salary.\nSpeaker A: You could probably afford this.\nSpeaker A: Yeah, I could probably afford this.\nSpeaker A: But maybe if I had a cool new TV and this was looking really slick and it had the dock and it had the scroll wheel, which I think is a really cool idea, that would sell me on it a little.\nSpeaker A: Then maybe...\nSpeaker A: I don't think I'd go over 100 euros, certainly.\nSpeaker A: Oh, no.\nSpeaker A: That would be way too much.\nSpeaker A: But I would be happy paying over 40 for it, I guess.\nSpeaker C: I would say 35 to 40.\nSpeaker B: Yeah.\nSpeaker B: I was going to say 35 to 50.\nSpeaker B: Okay.\nSpeaker A: 35 to 50 euros is a sales bracket.\nSpeaker A: Okay.\nSpeaker B: One of the things we're marketing about this product is this is the last one you will ever need to buy for your television.\nSpeaker B: It's one of the marketing features in this.\nSpeaker A: That's why the scroll needs to be really robust.\nSpeaker D: Okay.\nSpeaker D: So we'll come up with something between 35 and 50 that is rug the rubber, robust with scroll wheel with the new facilities of the scroll wheel like favorite stuff, favorite channels.\nSpeaker C: With the cradle, radio transmitters.\nSpeaker C: And the locator.\nSpeaker C: And backlit buttons and it's going to look sexy.\nSpeaker C: Or not.\nSpeaker C: Or not.\nSpeaker C: It might look like clay.\nSpeaker D: Okay.\nSpeaker D: So you can market.\nSpeaker D: Yeah.\nSpeaker B: That's easy.\nSpeaker B: That's not a... that's a... no.\nSpeaker B: Because we have about six... six, seven features in that alone.\nNone: Yep.\nSpeaker B: Under the title of uniquenesses.\nSpeaker B: Hmm.\nSpeaker A: Okay.\nSpeaker A: The next meeting starts in 30 minutes, although it does it, it starts at 3.21 the next meeting.\nSpeaker A: So we've got more than 30 minutes.\nSpeaker A: We've got like 50.\nSpeaker B: What?\nSpeaker B: It's my 3.21.\nSpeaker B: It's the next meeting.\nSpeaker A: The next meeting is 3.21, yeah.\nSpeaker B: That's almost 50 minutes.\nSpeaker A: Yeah.\nSpeaker A: That is.\nSpeaker A: You guys have changed the times from the presentations.\nSpeaker B: You guys can create a...\nSpeaker B: Probably...\nSpeaker C: We'll let you know when we're done.\nSpeaker A: If we can go earlier.\nSpeaker A: Thanks, Chef.\nSpeaker A: Okay.\nSpeaker A: So you guys will be getting your modelling done now.\nSpeaker A: And...\nNone: Okay.\nSpeaker A: Are there any other questions with regards to what this thing's going to do?\nSpeaker A: Look like how it's going to work that need to be addressed before we really look at this in a lot of detail.\nSpeaker C: I don't think so.\nSpeaker C: If we need questions, we'll just email you, I guess.\nNone: Okay.\nSpeaker D: Yeah.\nSpeaker D: I think it pretty much everything's covered.\nSpeaker A: Okay.\nSpeaker A: This one was quite easy.\nSpeaker A: Could have been worse.\nSpeaker A: It's always the optimist.\nSpeaker A: Yes, I am.\nSpeaker A: Okay.\nSpeaker B: Thanks, guys.\nNone: Thank you.\nSpeaker B: I don't have any emails.\nSpeaker B: This means I can go home.\nSpeaker A: It's going to be easy, yeah.\nNone: Oh, we gotta leave.\nSpeaker B: Maybe I just keep working.\nNone: Lift the pen.\nNone: Maybe I just keep working.\nNone: I'm on the mother node.\nNone: I didn't care about it.\n"}, {"meeting_id": "TS3010c", "summary": "Firstly, User Interface suggested trendy and one-color cases, good icons on the big buttons since they would be recognizable and excite the young people. Secondly, Industrial Designer introduced that young people favored soft materials and primary colors, but trendy design on the button and the use of LCD display would increase cost since the budget had always been a big problem. Moreover, the team would aim at the younger market based on the decision of the former meeting and agreed that to make the device more profitable they should sell the product for longer than one year. The team also discussed the button function of the remote control but they had not come to an agreement. Lastly, Project Manager decided on a normal battery and no kinetic to control the budget.", "dialogue": "None: I'm assuming.\nNone: Okay, let's use my, state onwards the input.\nNone: The statue is black, just bling.\nSpeaker B: Okay.\nSpeaker B: Well, let's start.\nSpeaker B: What are you doing?\nNone: Up.\nNone: Ah, pinball.\nSpeaker A: Okay.\nNone: We're not doing.\nSpeaker B: I have my screen back too.\nSpeaker B: Very good.\nSpeaker B: Okay.\nSpeaker B: We have presentations.\nSpeaker B: So, first, it's your turn.\nSpeaker B: Mine.\nSpeaker B: Yeah, great.\nSpeaker B: It's amazing.\nSpeaker D: Very interesting.\nSpeaker B: Industrial designer interface concept.\nSpeaker E: Yes, well, let's talk about the interface concept.\nSpeaker E: First, I'll discuss the buttons.\nSpeaker E: We just show you some samples, discuss some colors and design, maybe, already, and my personal preferences.\nSpeaker E: Well, we chose the power button to switch the television on and off, the mute button, to switch the volume on and off.\nSpeaker E: The channel buttons, one to nine, and zero to nine, and the button to choose higher channels than nine.\nSpeaker E: The volume and channel quadrant left and right, up and down arrows to the volume and channel.\nSpeaker E: And the menu button to manipulate the LCD display.\nSpeaker E: I found some interesting samples.\nSpeaker E: Well, what's pretty standard is that they're all pretty high.\nSpeaker E: Large, yeah, large, and pretty thin and long.\nSpeaker E: Power buttons are mostly at the top, left or right.\nSpeaker E: We see the same arrows, like there.\nSpeaker E: Yeah, well, arrow buttons can be blue.\nSpeaker E: And what's interesting is the icons on the buttons.\nSpeaker E: Some buttons have icons, like the play and stop, but we don't use that.\nSpeaker E: We have to choose the right icons or letters.\nSpeaker E: This is the V for volume, but they're both a V.\nSpeaker E: So it's not very clear what's the function of that.\nSpeaker E: Yeah, so that's...\nSpeaker B: Can you go back one page?\nSpeaker B: For the menu, what do we use for that?\nSpeaker B: We don't have buttons for the menu.\nSpeaker B: Or we have to use the volume and channel quadrant.\nSpeaker E: Yeah, I thought that was our idea.\nSpeaker E: Okay, so...\nSpeaker B: You have to put it on the...\nSpeaker E: Like this, or this.\nSpeaker E: That the menu button is okay.\nSpeaker B: Yeah, but it's to be clear that you can use the arrows.\nSpeaker E: Yeah, okay.\nSpeaker E: So the icons on the arrows as well.\nSpeaker E: Yeah, it's a second one.\nSpeaker E: Yeah.\nSpeaker E: Well, that's something to think about.\nSpeaker A: Okay, maybe I'll have something in my presentation.\nSpeaker E: Okay, I can see it.\nSpeaker E: Well, I don't know if we have to discuss this already or in the next meeting, but as we have to design the case and the whole remote control in our corporate company colors and logo, I would recommend the yellow case around the edges, the logo at the bottom.\nSpeaker E: And, well, maybe it's a set of buttons as its own color, so it's good recognizable.\nSpeaker E: So, I think...\nSpeaker E: Not too much colors.\nSpeaker E: No, not too much, but not flower power.\nSpeaker E: No, no, no, but it has to be trendy and...\nSpeaker E: Okay.\nSpeaker E: Yeah, so good icons on the buttons and big buttons is my personal opinion.\nNone: Okay.\nSpeaker B: That was that.\nSpeaker B: Thank you.\nSpeaker B: So, you're next.\nSpeaker B: I'm next, okay.\nSpeaker A: Yes.\nSpeaker A: Here we go.\nSpeaker A: First, we will say something about what younger people want, our group, we want to sell our remote control soon.\nSpeaker A: And then I'll discuss what my opinion is about the costs, about what battery is in it, what kind of buttons also.\nSpeaker A: First, the younger people, they want like soft materials and primary colors, like totally yellow, totally red.\nSpeaker A: So, it's visible.\nSpeaker A: The shapes are curved and round, like you also said.\nSpeaker A: Maybe it's nice to get remote control, not like all the other ones, straight and flat and long, but to give him a shape of your hand so it's easier to use or something like that.\nSpeaker A: But that's just an idea.\nSpeaker A: And then I'll have to discuss about the costs of all the things for the remote control.\nSpeaker A: The battery, there are a few options.\nSpeaker A: I think the best option is to use the basic battery so everybody can buy it at the supermarket.\nSpeaker A: Or use a kinetic battery, like within a watch, when you shake it a few times, it's loaded.\nSpeaker A: The form of the remote control, I think it's also nice to have it curved and maybe like it's hand shaped.\nSpeaker A: So, you take it here in your hand and here are the buttons.\nSpeaker A: Material, you use plastic, hard plastic, because it won't have to burst in one time.\nSpeaker A: And also rubber, because the younger people like that, what we see in the research.\nSpeaker A: The push buttons, we have one new thing discovered.\nSpeaker A: It's a scroll push thing like a mouse.\nSpeaker A: Maybe it's easy to use for the channels.\nSpeaker A: When you want to go move up, you just scroll up and click on the button if you want to see the next channel.\nSpeaker A: And also for the mouse, for the volume, it's also easy to use.\nSpeaker A: Just scroll a bit up, scroll a bit down.\nSpeaker A: And that's also easy, just when you have a thing like this and you get it here.\nSpeaker A: You can do it with your thumb.\nSpeaker A: If you're left hand, you can push the buttons, if you push channel one, you can see channel one.\nSpeaker A: The electrics with scroll push button, we must use regular chips.\nSpeaker A: There are also simple chips, they are cheaper.\nSpeaker A: But then we have just a basic remote control.\nSpeaker A: And I think there are a lot of those things and people won't buy it anymore.\nSpeaker A: They have seen enough of it.\nSpeaker A: And you have also advanced chips, but that's with the LCD screen.\nSpeaker A: And the costs will increase a lot more.\nSpeaker A: And I think our budget is too low to use an LCD and the chip is more expensive.\nSpeaker A: And maybe it's also thoughtful if we use different kind of shapes for the remote control that we then use the primary colors.\nSpeaker A: Like you get a yellow remote control, red one, blue one, etc.\nSpeaker A: You have any more questions about this?\nSpeaker A: I think the main thing is when you look at the costs and not the basic remote control, everybody already has.\nSpeaker D: This is with an LCD?\nSpeaker D: No.\nSpeaker A: Not with an LCD.\nSpeaker B: No.\nSpeaker B: There's this easy when you use the scroll buttons.\nSpeaker B: You can scroll, you see what number, then you push.\nSpeaker A: But then what I say, the costs will get a lot higher.\nSpeaker B: But then it's not easy to use scroll, will.\nSpeaker A: Then you'll see it on the television.\nSpeaker B: Yes, but then you go one down one up.\nSpeaker A: Yeah, but when you see a menu on the television, it's like you see one to twenty, you go scroll up.\nSpeaker B: Like we said before, it has to be used on every television.\nSpeaker B: So you may not be the television misduded.\nSpeaker A: I think the younger people will have newer televisions which can provide our remote control.\nSpeaker B: Younger people have them on their room.\nSpeaker B: And mostly they are smaller.\nSpeaker A: Yes, so, but that will be a problem.\nSpeaker B: I think most of the times that are not as fun as televisions.\nSpeaker A: But then we'll get to the regular remote controls and I think what I said, everybody has them already.\nSpeaker A: And they go to a supermarket and buy them for two euros and get the most cheapest thing.\nSpeaker A: And I think we must look further to develop something new.\nSpeaker D: Can you give an indication in the cost difference between the chip with LCD or without?\nSpeaker A: I got it on the screen and it was higher, but I don't know how much higher.\nSpeaker D: I think if we have an LCD, it will also sell a lot better.\nSpeaker A: That's true.\nSpeaker D: And that might bring back the cost.\nSpeaker A: But then we'll, I think, we must discuss what will be better if we have a better shape of the remote control or better options on it.\nSpeaker A: With a scroll menu, a scroll thing and an LCD and then a flat remote control or a more hand-shaped remote control.\nSpeaker A: With a scroll without LCD.\nSpeaker B: Maybe you can look how much it is.\nSpeaker B: I can look on my LCD.\nSpeaker B: Very important.\nSpeaker F: There we go.\nSpeaker A: Now that the push button requires a simple chip, a scroll wheel requires minimally a regular chip, which is a higher price range.\nSpeaker A: The display requires a advance chip, which in turn is more expensive than a regular chip.\nSpeaker B: More expensive, but how much?\nSpeaker A: That's from my manufacturing.\nSpeaker A: Deficient.\nSpeaker B: Okay.\nSpeaker B: Well, thank you.\nSpeaker D: So, so my presentation is about trend watching.\nSpeaker D: I did some trend watching. It's very important to keep up with latest trends, because if you don't, you won't sell.\nSpeaker D: So, well, I will do that.\nSpeaker D: Well, we made an investigation of the market by trend watches.\nSpeaker D: They watch it in cities like Paris and Milan, of course well known for their trend.\nSpeaker D: And well, what did you find?\nSpeaker D: We have two groups, young and 20, and the old and rich, and young and 20.\nSpeaker D: They starting to like food and vegetables as a team for clothes, shoes, and also products.\nSpeaker D: And material, the chute feel as a pointy feeling.\nSpeaker D: And you get a feeling for what this is, and with an image of it.\nSpeaker D: Then the old and rich, they like dark colors and simple, nice blue shapes.\nSpeaker D: And they also like familiar material, especially wood.\nSpeaker D: Another picture.\nSpeaker D: You get a feeling for this.\nSpeaker D: Well, then already come to my personal preferences.\nSpeaker D: We aim at the younger market, so we should also be a look at their trends.\nSpeaker D: However, with trends, it's always, it's now, in my last one year, and next year, it can be totally different.\nSpeaker D: And I think we want to sell our products for longer than one year.\nSpeaker D: So, we must not just only look at the product trend is now, as it might be totally different next year.\nSpeaker D: So, that's one thing to keep in mind.\nSpeaker F: Changing covers.\nSpeaker D: Any questions?\nSpeaker D: No.\nSpeaker B: No.\nNone: It's clear.\nSpeaker B: So, no.\nSpeaker B: Now, we have to decide.\nSpeaker B: Well, we have to decide on the concepts.\nSpeaker B: So, we have to look at the next components, and use the interface concept.\nSpeaker B: Now, we have to make some concepts. Maybe one of you can paint it on the board.\nSpeaker B: First, use the interface.\nSpeaker A: How we make it?\nSpeaker B: Yes, a concept.\nSpeaker A: Sure, we first discuss about what...\nSpeaker B: Okay.\nSpeaker B: Well...\nSpeaker B: Something like this, or shapes, or what do we need?\nSpeaker C: Can make some concepts.\nSpeaker B: Yes, okay.\nSpeaker D: We have to have some idea in shape of your end.\nSpeaker A: More like something.\nSpeaker A: Yes.\nSpeaker A: I can't draw with.\nSpeaker A: I have to.\nSpeaker A: I'm not a designer.\nSpeaker A: I have 3D like...\nSpeaker A: This is the remote control.\nSpeaker A: And then you'll have something like this under it.\nSpeaker A: So, it's easier to get it like this.\nSpeaker B: It's like a gun.\nSpeaker B: So, that's to be soft.\nSpeaker A: And it has to be soft, yeah.\nSpeaker A: So, you can squeeze in buttons.\nSpeaker A: Sorry?\nSpeaker A: Buttons.\nSpeaker A: Buttons on top of it.\nSpeaker A: And here, the scrolling.\nSpeaker A: You can do it with your thumb.\nSpeaker E: But that's the only scroll button on it.\nSpeaker A: Now, we use one scroll button, and the other one is here, one till...\nSpeaker A: Zero till nine.\nSpeaker B: Okay, but one for the sound and one for the channel.\nSpeaker A: Or two buttons.\nSpeaker A: Okay, if you go to...\nSpeaker A: If we use one, then we'll have just switch on it.\nSpeaker A: You'll just switch it.\nSpeaker A: And now it's the sound switch back.\nSpeaker D: It's more difficult.\nSpeaker D: If you have a menu, how do we use other options?\nSpeaker A: With the menu button.\nSpeaker A: And then you also can scroll in it.\nSpeaker A: But like all the other ones with this thing and here and arrow, here and arrow, here and arrow, here and arrow.\nSpeaker A: Because from 100 remote controls, 99 half it.\nSpeaker B: But if we don't have LCD, we don't have a menu.\nSpeaker A: Then we have it on the TV menu.\nSpeaker B: But again, maybe...\nSpeaker B: How do we know that TV can handle it?\nSpeaker B: You don't know.\nSpeaker B: So there's no menu.\nSpeaker A: It's like some sort of a teletext option, but we don't have teletext.\nSpeaker B: Well, you can't use it.\nSpeaker A: And if we put an LCD thing on it, then the costs will paint much higher.\nSpeaker B: Okay, we make two concepts.\nSpeaker B: One with LCD, one without LCD.\nNone: Okay.\nSpeaker A: But you all like this kind of thing.\nSpeaker B: Good concept.\nSpeaker A: With the scroll button.\nSpeaker A: And this one has to be soft.\nSpeaker A: And this has to be harder because when it falls, it must not burst.\nSpeaker A: Or some kind of rubber around it.\nSpeaker E: One, two.\nSpeaker E: Number two.\nSpeaker E: You can make the power button as a trigger.\nSpeaker E: That's nice.\nSpeaker A: Trigger.\nSpeaker F: No.\nSpeaker B: It's one way or two.\nSpeaker B: I put it on and off.\nSpeaker B: Not good to use.\nSpeaker E: Right.\nSpeaker E: Yeah, but...\nSpeaker B: How's that?\nSpeaker B: Fucking out.\nSpeaker B: Yeah.\nSpeaker B: No, it's not good.\nSpeaker B: Yeah.\nSpeaker B: No, second concept.\nSpeaker B: One with LCD, one without LCD.\nSpeaker B: Then a...\nSpeaker B: Paint it.\nSpeaker B: Paint it.\nSpeaker A: With the scroll thing on...\nSpeaker B: One with two scroll buttons and one without...\nSpeaker B: Yeah.\nSpeaker B: One without...\nSpeaker B: With a menu and one without a menu.\nSpeaker B: And one with...\nSpeaker B: With a menu as an LCD.\nSpeaker B: Draw it.\nSpeaker B: Unbelievable.\nSpeaker B: Do I have to do everything?\nSpeaker B: Blank.\nSpeaker B: I have...\nNone: Not so difficult.\nNone: But if you push the menu button...\nNone: Or this menu...\nNone: Yeah, they have the LCD screen.\nSpeaker B: Yes, but you don't know which of the...\nSpeaker E: It's called buttons you have to choose.\nSpeaker B: You have to...\nSpeaker B: For the menu.\nSpeaker B: Yeah.\nSpeaker E: Mm-hmm.\nSpeaker B: One...\nSpeaker B: That way and one...\nSpeaker B: That way.\nSpeaker B: So...\nSpeaker B: Then it depends on the coast.\nSpeaker B: On and off.\nSpeaker A: But is it easy to use when you're on your left side?\nSpeaker B: When it's not too big.\nSpeaker B: Just like a phone.\nSpeaker D: Maybe it's better if the scroll wheels are...\nSpeaker D: Separate, more separated.\nSpeaker D: Like you have the menu built in between the...\nSpeaker A: Yeah, on the left...\nSpeaker A: The scroll button and on the right the scroll button.\nSpeaker A: But what it would be easy to use then, if it's like...\nSpeaker A: You have a big...\nSpeaker B: It's better.\nSpeaker B: When you are on the menu you have to go...\nSpeaker A: I also think this concept is not what the young people were looking for.\nSpeaker A: They were like round curves.\nSpeaker B: Yeah, okay. That's the outside.\nSpeaker B: But now the first buttons.\nSpeaker D: You have now two buttons missing the mute button.\nSpeaker D: Sorry.\nSpeaker D: You have to use missing the mute button and...\nSpeaker D: You have to...\nSpeaker B: Mute and the other.\nSpeaker B: Not so difficult.\nSpeaker A: Personally I think two scroll buttons aren't easy to handle.\nSpeaker B: But how do you want to solve it?\nSpeaker A: With the switch button.\nSpeaker B: Yeah, but in the menu that's not easy.\nSpeaker B: No, like down.\nSpeaker B: You switch, you go into the right, you switch, you go down.\nSpeaker A: All you mean like that?\nSpeaker A: Then you can also have like...\nSpeaker B: Joystick.\nSpeaker A: Yeah, and joystick thing.\nSpeaker B: But this is just a break.\nSpeaker B: Joystick.\nSpeaker B: Or a small one just like in a laptop.\nSpeaker A: Like in a laptop.\nSpeaker A: Some sort of thing.\nSpeaker A: A little bit bigger.\nSpeaker B: I mean it's better.\nSpeaker B: But how expensive is it?\nSpeaker B: I don't know.\nSpeaker B: Why do I pay for it?\nSpeaker B: Well...\nSpeaker B: Better, I'd ease.\nSpeaker A: Or no scroll.\nSpeaker A: I think it's just a shape.\nSpeaker A: And...\nSpeaker B: Young people's I think scroll buttons good.\nSpeaker B: So, I think we have to keep them.\nSpeaker A: Or a remote control more like joystick.\nSpeaker B: Yeah, that's not expensive then that joystick is better.\nSpeaker B: A small one.\nSpeaker A: Like this like in a Nintendo.\nSpeaker B: No, just like in a playstation thing.\nSpeaker B: A laptop.\nSpeaker B: Small round.\nSpeaker B: Then it's not so big.\nSpeaker A: No, no, no.\nSpeaker A: I mean the shape of the remote control.\nSpeaker A: It's like a playstation.\nSpeaker B: Then you can...\nSpeaker B: You have to use it with one hand.\nSpeaker F: Yeah.\nSpeaker B: So, maybe if it's possible, it's not too expensive.\nSpeaker B: I think joystick is better.\nSpeaker B: A small one.\nSpeaker B: So, please look at it.\nSpeaker B: Okay.\nSpeaker C: And on the LCD, how much it's got?\nSpeaker C: It's close to extra.\nSpeaker A: They're not in details.\nSpeaker A: It's more expensive or less expensive.\nSpeaker B: I think you get it.\nSpeaker B: So, after this meeting you have half an hour to fix it.\nSpeaker A: And after going with it.\nSpeaker A: Yes.\nSpeaker A: I got my personal cash.\nSpeaker A: I don't know the costs.\nSpeaker B: You're a problem.\nSpeaker B: I'm not mine.\nSpeaker A: And I'll make something up.\nSpeaker B: Okay.\nSpeaker B: So, do we have other concepts?\nSpeaker B: Then for the component, we use a normal battery.\nSpeaker B: Then, yeah.\nSpeaker A: Or the kinetic.\nSpeaker A: No, no battery.\nSpeaker B: I think it's more expensive.\nSpeaker B: So, we use a normal battery.\nNone: Yeah.\nSpeaker B: Chip depends on the LCD.\nSpeaker A: It depends on the scroll.\nSpeaker A: If we use a scroll, then we have the regular chip.\nSpeaker A: If we don't use a scroll, then we can use the simple chip.\nSpeaker B: And if you use the LCD, we have the most expensive.\nSpeaker A: Okay.\nSpeaker B: So, it depends on the screen.\nSpeaker A: If we...\nSpeaker A: Now, okay, scroll, we also have this.\nSpeaker A: So, it will be the advanced chip or the regular chip.\nSpeaker B: Okay.\nSpeaker B: So, the shapes of the design.\nSpeaker B: It depends on the LCD.\nSpeaker B: But it has to be small.\nSpeaker A: I think...\nSpeaker A: Or shall we just put it on the pistol thing and then just put also LCD on it?\nSpeaker B: If you have pistol, and LCD is not easy.\nSpeaker A: You should use your thumb.\nSpeaker A: Yeah, but if you use your thumb.\nSpeaker A: Yeah.\nSpeaker B: I use my thumb.\nSpeaker B: But then you have it.\nSpeaker B: If you have pistol, you have to sell.\nSpeaker B: And the screen is...\nSpeaker B: If you have a joystick on...\nSpeaker B: Keep it this way to look at the screen.\nSpeaker A: If you have like a...\nSpeaker A: A PlayStation game controller.\nSpeaker A: And you move up, forward, down, left.\nSpeaker A: Then you have just a little bit curved.\nSpeaker A: It's not just a...\nSpeaker A: No, no, no.\nSpeaker A: That's how we're using.\nSpeaker A: That's why they make joysticks like that.\nSpeaker A: I think...\nSpeaker B: Yeah, but then you look forward and then you can...\nSpeaker F: Yeah.\nSpeaker B: Now, if you have to look at it...\nSpeaker A: Okay.\nSpeaker A: You know how to design it.\nSpeaker D: We have something standing here with the LCD.\nSpeaker B: Yeah.\nSpeaker B: Then it goes like this.\nSpeaker A: No, why not?\nSpeaker B: No.\nSpeaker B: It's for the younger people.\nSpeaker A: Yes, of course.\nSpeaker B: It's something new.\nSpeaker B: That's good.\nSpeaker B: It may not break.\nSpeaker A: Then we put rubber around it.\nSpeaker B: Okay.\nSpeaker B: Is this possible?\nSpeaker A: Hard plastic, the shape, and around it, around it's rubber, and the hand shape is also rubber.\nSpeaker B: Okay.\nSpeaker E: Or you can turn it inside.\nSpeaker D: But the each of...\nSpeaker D: The each of you wasn't the most important aspect of it.\nSpeaker E: No, that's true.\nSpeaker D: For which it's about to sell it.\nSpeaker B: This is something new.\nSpeaker B: Okay.\nSpeaker B: Then this is the design and the buttons are on the next page.\nSpeaker B: So it depends on the cost.\nSpeaker B: So...\nSpeaker B: Close it, okay.\nSpeaker B: We have one minute.\nSpeaker B: One.\nSpeaker B: I think.\nSpeaker B: No, more.\nSpeaker B: More.\nSpeaker B: Seven.\nNone: Still 10.\nSpeaker B: Next meeting.\nSpeaker B: Thirty minutes.\nSpeaker B: So, very up.\nSpeaker B: Ah, that's us together.\nSpeaker B: You two stay here.\nSpeaker B: Pained it.\nSpeaker B: Okay.\nSpeaker B: Now you have to.\nNone: So...\nSpeaker B: I think it's clear.\nSpeaker B: Check your mail.\nSpeaker B: So...\nSpeaker B: Yes, it has to be ready.\nSpeaker B: Next meeting.\nSpeaker B: So, cooking.\nSpeaker B: Okay.\nSpeaker B: Next meeting is called the detailed design.\nSpeaker B: So...\nSpeaker B: Everything has to be ready.\nSpeaker B: Okay.\nSpeaker B: Thanks for your attention.\nSpeaker B: Okay.\nSpeaker D: Thirty-five.\nSpeaker D: Three over the next meeting.\nSpeaker D: All right.\nNone: Five minutes to finish.\nNone: Couple now.\nNone: Oh, there's two batteries in the laptop.\nNone: Two?\nNone: Nice.\nNone: Five seconds.\nSpeaker A: Both in front, I think.\nSpeaker A: Yeah.\nSpeaker F: Hmm?\nSpeaker F: Yeah.\nNone: Okay.\nNone: I'll leave.\nNone: Okay.\nNone: Okay.\nNone: Good luck.\nNone: I think you're right.\nNone: The little key.\nNone: Note.\nSpeaker B: Done.\nNone: Oh, oh, oh, oh, oh.\nNone: Okay.\nNone: I don't know, I don't know.\nNone: Safe.\nNone: No.\nNone: In this room, big English...\nNone: That's naughty.\nNone: We speak some sort of de-elect.\nNone: Okay, it's got Dutch.\nSpeaker B: It has to be ready, in thirty minutes.\nSpeaker B: See you for play.\nNone: All right.\nSpeaker F: Oh.\nSpeaker B: Anyway, you bet please.\n"}, {"meeting_id": "ES2002c", "summary": "This meeting was about the conceptual design of the remote control. Firstly, Marketing presented on the market trend and suggested a remote control with a fancy look and feel, technology innovation, and spongy material. Next, User Interface presented on conceptual design and recommended the control itself should use an inconsistent colour scheme. Then, Industrial Designer presented on the component design and that the cost, the complexity, and the size would have an impact on their remote design and reminded there would be a restriction on grouping plastic and rubber cases together. Lastly, the group discussed remote concepts and decided to have a small size remote with the speaker, the chip on print for the power, and a rubber case.", "dialogue": "Speaker C: to do now is decide how to fulfill what your stuff is.\nSpeaker C: OK.\nSpeaker B: Sure.\nSpeaker B: OK.\nSpeaker B: It kind of does make sense, isn't it?\nSpeaker B: Because when we get into the end of meeting, we're kind of talking about action and design, as opposed to background.\nSpeaker B: Everything I have is kind of background.\nSpeaker F: OK, we're all ready to do.\nSpeaker F: We'll have it in this meeting then.\nSpeaker F: If we just recap on the minutes from the last meeting, we decided on our target group being 15 to 35.\nSpeaker F: We decided that it was going to be normally chargeable battery power, that we're going to group our audio, visual, and other functions into those categories.\nSpeaker F: And I told you guys about the three new requirements about ignoring teletext, ignoring everything, accept the TV, and trying to incorporate the corporate color and slogan.\nSpeaker F: Start with the last meeting.\nSpeaker F: Is there anything?\nSpeaker F: Have I forgotten anything?\nSpeaker F: No.\nSpeaker F: So everything.\nSpeaker F: OK.\nSpeaker F: So if we have these three presentations, and then if you have anything to kind of that you know you're going to want to discuss, maybe just make a note of it, and then we'll have all the discussion at the end.\nSpeaker F: I want you to have your idea this time.\nSpeaker F: And so if we start off with Angie and then Craig, and then David, is that all right?\nSpeaker F: And then after that we'll have to make some decisions.\nSpeaker F: I've got stuff.\nSpeaker F: Yeah.\nSpeaker F: Right.\nSpeaker E: I don't know.\nSpeaker F: Get that.\nSpeaker F: Take this.\nSpeaker F: Screening quick, take it away.\nSpeaker F: How did we leave it with speech recognition?\nSpeaker F: No, we should say we're going to try maybe incorporate, but we hadn't made a definite decision on that.\nSpeaker F: We should also point out that the final objective of this meeting is to reach a decision on the concepts of the product.\nSpeaker F: OK.\nSpeaker F: So that's kind of the end result, hopefully.\nSpeaker E: OK.\nSpeaker B: All right.\nSpeaker B: So does it function F8?\nSpeaker F: OK.\nSpeaker F: I think it's working.\nNone: Oh, no, we're good.\nSpeaker B: OK, great.\nSpeaker B: So we just start this.\nSpeaker B: OK, great.\nSpeaker B: So let's move on.\nSpeaker B: Oh, where did it all go?\nNone: It's not good.\nNone: OK.\nNone: OK.\nSpeaker B: So let's move on.\nSpeaker B: Oh, where did it all go?\nSpeaker B: It's not good.\nSpeaker B: OK.\nSpeaker B: Let me just see where I find it.\nSpeaker B: Is it one like it?\nSpeaker B: I think I just opened up the template.\nNone: So we know that.\nNone: OK.\nNone: All right.\nNone: So let's have a look here.\nNone: OK.\nNone: So this was the method that I've taken.\nSpeaker B: Basically what I want to do here.\nSpeaker B: OK.\nSpeaker B: So we're going to do that.\nSpeaker B: OK.\nSpeaker B: So we're going to do that.\nSpeaker B: OK.\nSpeaker B: So we're going to do that.\nSpeaker B: OK, so this is the method that I've taken.\nSpeaker B: Basically what I want to do here, before we get into it too far, I want to show you all the background information I have that I think we need to acknowledge if we want this to be successful.\nSpeaker B: OK.\nSpeaker B: And then sort of go through some of the way that I've dealt with that information and then sort of bring us all together into it to see how this fits in with the overall vision.\nSpeaker B: So I've tried to take a whole lot of market research and summarize it for us.\nSpeaker B: And then identify trends that are sort of in sync and are important to our project plan that we have so far.\nSpeaker B: And then initiate a kind of discussion on design options.\nSpeaker B: So that sort of helps us to narrow in on aspects that will inform other elements of the project.\nSpeaker B: Does that make sense?\nSpeaker B: Yeah.\nSpeaker B: That sort of strategy.\nSpeaker B: I thought that that will impact on the rest of what we do.\nSpeaker B: So that's why I suggest we get in.\nSpeaker B: OK, so out of different figures and ratings of people in general, consumers in general, the number one thing that was found was that the television remote control of a fancy-looking feel, OK, and not specified, not a functional look or feel.\nSpeaker B: Fancy.\nSpeaker B: However, this is where we kind of have to be very, I think, creative about it.\nSpeaker B: Number two was that it be innovative.\nSpeaker B: OK, so that tells me that we have to find a way to be innovative without adding just unnecessary sort of functional bits to it.\nSpeaker B: And third priority for ease of use.\nSpeaker B: So again, that kind of gives us a general picture of how it has to be quite user-friendly while still having technology.\nSpeaker B: So I'll just say right away as a bit of a foreshadowing into how we proceed with this in terms of marketing is that I think what we should think about is how the, about how the innovation contributes to the look and feel and not so much to the functionality of it.\nSpeaker B: For example, like when you pick it up and push it, like it all lights up or something, you know what I mean?\nSpeaker B: Or it's got something else to it that just seems innovative because obviously the thing that the message here is ease of use.\nSpeaker B: So how do you make innovation make something more easy to use?\nSpeaker B: Well, that's, I guess, where we're going to go with this.\nSpeaker B: OK, then there's the other aspect of the back of the market research I have here is on fashion style, which, as we've agreed is a priority.\nSpeaker B: Top European fashion trend that I read about says there's this emerging theme of fruit and vegetable.\nSpeaker B: OK, especially in clothes and furniture.\nSpeaker B: And when I first saw that I thought, well, do we want to actually try and think about this trend and how we add something to it?\nSpeaker B: Or we get right into it or we completely steer away from it?\nSpeaker B: OK, OK.\nSpeaker B: My feeling is that we do want to observe this trend, but we want to think also about the fact that it sort of has to fit in with something which is not specifically electronics.\nSpeaker B: Because I think what we're in, what we're in is partly sort of home decor, partly something like a computer.\nSpeaker B: So I think we might want to be careful about how quickly we create like a remote control in the shape of an apple or something. I think that would be pushing it.\nSpeaker B: And then in terms of material trends are for things to be soft and spongy and sort of, I might say ergonomic or friendly to handle, which is, which also indicated that last year this was, this was not the case.\nSpeaker B: So probably a lot of the competition on the market will be still in last year's mode. So if we try and really capitalize on that, I think that'll be in our favor.\nSpeaker B: So this is the summary of everything.\nSpeaker B: Style is number one thing in the market of who we're selling to.\nSpeaker B: Innovative design technology is also a must in that it seemed to be seem to be cutting edge.\nSpeaker B: But ease of use has to be ensured throughout. That was like the number three thing.\nSpeaker B: And then then there are vibrant natural colors.\nSpeaker B: That's the way I interpreted it anyway. Softness and material shape and function.\nSpeaker B: And so I wrote and written at the bottom to give us sort of a context of discussion.\nSpeaker B: Mac iPods. Something which is, I'd have to say very high tech. Ten gigabytes, whatever.\nSpeaker B: But when you hold it in your hand, there's like no buttons.\nSpeaker B: You know what a Mac has changed?\nSpeaker B: I'm thinking however Mac iPod is sort of last year's because it's very hard and sort of glassy and glossy.\nSpeaker B: So I'm thinking if we imagine that we're taking some of the features of a Mac iPod and we're then making it's more of like a more of like a comfortable type of, or more of like a maybe more vibrant friendly thing to have.\nSpeaker B: And then so this is with all that information, what I'm suggesting in this slide here is that we take these ideas and as we get into more the more.\nSpeaker B: Like sort of production side of things that we think about shape materials and themes or series that go throughout sort of like.\nSpeaker B: I don't know like we think of some kind of a theme that unifies it all that we agree on sort of like a marketing identity.\nSpeaker B: Does that make sense? Yeah.\nSpeaker B: So, so like I threw out a few ideas there just to kind of get us thinking along those lines like lemon line, I don't know, green colors, whatever just ideas I'm thinking that some of these ideas will seem quite coherent if we use them in terms of their, what people associate them with in terms of texture shape colors, things like that.\nSpeaker B: Okay.\nSpeaker B: Like the ones, the ones which I'm most fond of in terms of giving like a theme to it would be like lemon or something like that, you know, something which is like you see a lot in other areas like I see lots of websites and things that seem to associate with like lemon line.\nSpeaker B: So, anyway, just just an idea. I'm thinking maybe we could incorporate some of these features into a fairly into something which is, which seems to have something to it which is almost gimmicky because like, like something to do with like lighting within it.\nSpeaker B: Like, you know, just this in the simple sense when you pick up a phone and touch a button, it lights up.\nSpeaker B: Usually the buttons light up. How can we build on that? Maybe like it could light up in different colors or something or people could buy the by the control and then it comes with different like covers or something.\nSpeaker B: Okay.\nSpeaker B: Anyway, that's all I have, but hopefully we can revisit those ideas when we get into.\nSpeaker B: Okay.\nSpeaker F: Great.\nSpeaker F: Thank you for that.\nSpeaker F: Craig, do you want to plug yours in?\nSpeaker F: Okay.\nSpeaker E: Okay.\nSpeaker C: It's probably not sending.\nSpeaker F: Oh, something coming out.\nSpeaker A: So, little bit of concept.\nSpeaker A: To research it, I have a company, the homepage again, it's providing me with more examples of previous existing control controls.\nSpeaker A: There's a wee bit of discussion about the existing ones there.\nSpeaker A: So, I'm taking the suggestions from them and try to incorporate them into this.\nSpeaker A: And the side of the control and the buttons, the shape of the control and wherever the button should be located on the control.\nSpeaker A: I found from the research is that most of the current controls are just basically bricks with loads of buttons all over them.\nSpeaker A: They're not very attractive to look at and they're not very comfortable to hold.\nSpeaker A: They're just holding big bricks and they're very even lost.\nSpeaker A: They tend to be very dark colors so if they're shadowed places, they're in the catches, you can't really see them.\nSpeaker A: The controls themselves tend to use a very consistent color scheme.\nSpeaker A: For instance, the standby button isn't almost red, it really should be.\nSpeaker A: It's something that the user then identify with.\nSpeaker A: They say red switch off, that's how it should be.\nSpeaker A: I'm not sure if there's any other examples of that.\nSpeaker A: So from that, I've got a couple of preferences for the end control.\nSpeaker A: Again, with the red color button for standby and the other examples of that.\nSpeaker A: The button should be large, they should be tiny little things like heat and some mobile phones, they should be easy to prep, very comfortable.\nSpeaker A: One of the examples given on the home beach was there's an up and down volume button, both of them have a B on them.\nSpeaker A: So the up volume button looks like it should be a down volume button, not as confusing.\nSpeaker A: It should avoid things like that.\nSpeaker A: If the corporate color scheme allows it, we should have a very bright color so that it can be identified anywhere.\nSpeaker A: Obviously, it's trying to avoid being tacky there, but it could be timed very easily with your alignment.\nSpeaker B: Do we have a corporate color scheme?\nSpeaker F: I think it's yellow because the website is yellow and there's the band at the bottom.\nSpeaker F: The playdough's yellow.\nSpeaker F: Definitely food for the not there, but keep going and we'll discuss it.\nSpeaker A: Any extra features we add, the only basic ones should be hidden.\nSpeaker A: They shouldn't be on the, should be visible without something to be opened or special or extra effort.\nSpeaker A: If we do decide to go for a voice activation, the show will be about the alternative, possibly hidden in there, the open top section.\nSpeaker A: In case something was wrong with it, I can use voice to be a bit of cold.\nSpeaker A: We should definitely avoid the big square block book.\nSpeaker A: That's just wrong.\nSpeaker A: We've got an email from the research department.\nSpeaker A: They've said the voice control can now talk back if you ask a question.\nSpeaker A: It could be good to have them confirm any action you take.\nSpeaker A: I don't know where the spoken cue go.\nSpeaker A: I think it's fashion and dealt funnings.\nSpeaker A: We don't know how to explore the color scheme.\nSpeaker A: You can see you want the corporate colors, but you don't see if you can use any other colors at all.\nSpeaker E: Okay.\nSpeaker E: Great.\nSpeaker B: Lots of good information there.\nSpeaker F: That was very good.\nSpeaker B: It's a shame the cable wasn't just in the middle.\nSpeaker B: I know it would be handy, wouldn't it?\nSpeaker F: Oops.\nSpeaker F: Do you want to sit in the line of sight of this?\nSpeaker C: Okay.\nSpeaker E: Let me just get this.\nSpeaker C: There it is.\nSpeaker C: Okay.\nSpeaker C: I guess the same thing again.\nSpeaker C: I started with something very basic.\nSpeaker C: So just so you guys have some idea of what's involved in my process.\nSpeaker C: Then you can just work through it and either modify it or start from scratch, depending on what your needs are.\nSpeaker C: The components are exactly the same.\nSpeaker C: I think like what you guys said, the most input that's needed is basically in the user interface.\nSpeaker C: The rest of the components, they do have an impact in terms of cost and complexity.\nSpeaker C: Like you said, time to market was a problem.\nSpeaker C: And how many components are physically in their cost?\nSpeaker C: And the power is basically a factor of that.\nSpeaker C: And the lower components of the power, the logic, the transmitter, and the infrared, they affect you in terms of the size of your device.\nSpeaker C: And that would have some impact on how you hold rather than the actual use using the remote control because like we said, we've defined like we only want the basic things that to be visible in the rest of them.\nSpeaker C: We try to hide.\nSpeaker C: So, you know, it's just a matter of working out space.\nSpeaker C: So I guess three things.\nSpeaker C: Cost, complexity, and the size.\nSpeaker C: These are the three things that will have an impact on you.\nSpeaker C: So just go through the components.\nSpeaker C: These are the options that are available to you.\nSpeaker C: I'm not very sure about the voice thing because I got another email and it was quite sketchy on what the voice options are.\nSpeaker C: It said it could talk to you, but it never said anything about being able to listen.\nSpeaker C: It said something about a sensor, but it never clarified that.\nSpeaker C: So maybe if you could see the other email that they sent you because they got back to me in a different requirement.\nSpeaker C: Or different offerings of what components are available.\nNone: Okay.\nSpeaker C: So your basic components are buttons.\nSpeaker C: Okay.\nSpeaker C: And you have a wheel available like a mouse scroll wheel.\nSpeaker C: Okay.\nSpeaker C: There's an LCD display.\nSpeaker C: I think these are quite standard things.\nSpeaker C: They're standard, I think.\nSpeaker C: No.\nSpeaker C: Well, in a sense, these are all the options available for you.\nSpeaker C: I explained to you the complexity and the cost thing again a bit later.\nSpeaker C: Okay.\nSpeaker C: Then there's how the case actually looks.\nSpeaker C: It can actually be flat or it can be curved.\nSpeaker C: And then the different types of materials that you can use.\nSpeaker C: I don't think you can use them in a combination, but I could check back for you, but I don't think you can actually use them in a combination.\nSpeaker F: You can have like plastic rubber.\nSpeaker C: I think plastic and rubber would be fine, but plastic rubber and wood.\nSpeaker C: I wasn't.\nSpeaker C: I'm not very sure about the titanium.\nSpeaker C: They had some restrictions on using the rubber and the titanium.\nSpeaker C: The rubber was a restriction on the kind of power source you could use, but the titanium had a different kind of things on the shape of the.\nSpeaker C: Okay.\nSpeaker C: So I think that there is some restriction on.\nSpeaker C: I think you could probably group plastic and rubber together.\nSpeaker C: Wood and titanium, but you know, it might be easier from a cost perspective and a complexity just to use one.\nSpeaker C: You know, as opposed to two.\nSpeaker C: And the other components are logic chips.\nSpeaker C: Again, I'll go back to the component chips.\nSpeaker C: How complex or how easy the logic is.\nSpeaker C: It depends on how many functions you have on the unit.\nSpeaker C: And that impacts cost.\nSpeaker C: I don't think the logic chip has an issue about size because they should be about the same size.\nSpeaker C: Power consumption should be about the same.\nSpeaker C: I think the main impact is complexity.\nSpeaker C: And the other thing is the power options.\nSpeaker C: The first one is the standard battery.\nSpeaker C: Okay.\nSpeaker C: The second one, I think, is more of a gimmick than actually a usable thing.\nSpeaker C: It's a wind up.\nSpeaker C: You know, a crank.\nSpeaker C: Just by moving.\nSpeaker C: Yeah.\nSpeaker C: But that might be something I think that's more of a look and field decision because I don't think you can have one power source if you're using the alternative power sources.\nSpeaker C: I think whatever this you still need a battery because I don't think anybody wants to keep doing one thing.\nSpeaker C: Okay.\nSpeaker C: The other ones are a solar power cell, which may not be a great idea in Europe or any country that has seasons because half the year you've been dead.\nSpeaker C: So like what I said, you probably need like a battery and something else.\nSpeaker C: And the kinetic one, I guess for me is the most interesting one because it's movement and people like to fit over there.\nSpeaker C: And it's a nice sales gimmick, I think, from a marketing gimmick.\nSpeaker C: It's a technology thing.\nSpeaker C: It's a shaker.\nSpeaker C: It doesn't shake it, knock it or something.\nSpeaker C: You know, you know, you have those balls, you know, those stress balls where you bounce the ball and it lights up and it goes, you know, that might be a gimmick combined with rubber.\nSpeaker C: You know, just to, if you get frustrated with remote control, you can throw it kind of, you know, just the, you know, so, okay, my, from my role, I don't think there are personal preferences, but role preferences, I think something comfortable to hold.\nSpeaker C: Small and slim, I guess that's more in the sense of small and slim in terms of comfortable, not so small, you can't, you know, like a phone or too small phone. And the other thing is from a production point of view, the less components we use and the simpler the components means you reduce your cost and you increase your profit.\nSpeaker C: And also the time to market and the complexity of developing, designing and debugging it.\nSpeaker C: So, okay, let me just go back and talk about some of the restrictions.\nSpeaker C: The user interface restrictions basically means that if you use more complicated features like the buttons are standard.\nSpeaker C: Okay, the LCD panel and the scroll wheel, you need more complicated logic.\nSpeaker C: The case, okay, with a rubber case, you can't have the solar panels.\nSpeaker C: Okay, with the titanium case, let me just check that.\nSpeaker C: Titanium case can't be curved, it has to be square.\nSpeaker C: Okay, there's no restriction on the plastic.\nSpeaker B: It can't be curved.\nSpeaker C: It can't be curved on the wood.\nSpeaker C: So that's, again, I don't think you can use them in a combination.\nSpeaker C: Okay, especially the titanium, I suspect they're very fixed to a particular need.\nSpeaker C: So mixing them may not be a good idea.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker B: Question on, can I ask question?\nSpeaker F: Yeah, well, yeah, just quite keen to get the discussion going with the time we've left.\nSpeaker B: Okay.\nSpeaker B: But yeah, can we power a light in this?\nSpeaker B: Can we get a strong and a battery to power a light?\nSpeaker C: I think we could, because the LCD panel requires power and the LCD is a form of a light.\nSpeaker C: Okay.\nSpeaker B: So maybe one of the things we can just try and include is a really good battery.\nSpeaker F: Why, what kind of light do you want?\nSpeaker B: Well, I mean, I'm thinking it might be that for this to be a high tech thing, it's going to have to have something high tech about it.\nSpeaker B: Yeah.\nSpeaker B: It's going to take battery power and to make that, to make that a realistic goal.\nSpeaker B: I think one of the issues that will come up later is a battery power.\nSpeaker C: Are you thinking of a light in the sense of light light or a light in the sense of a low kind of Frankenstein?\nSpeaker B: Well, I'm thinking along the lines of you, you're in the dark watching a DVD and you, you find the thing in the dark and you go like this.\nSpeaker B: And that's what everybody does.\nSpeaker B: Always the volume button in the dark.\nSpeaker B: Okay.\nSpeaker B: And you just touch it or you just pick it up and it lights up or something like a phone.\nSpeaker B: Okay.\nSpeaker B: Like a backlight.\nSpeaker B: Okay.\nSpeaker B: Whereas with phones, people charge them once a week.\nSpeaker B: Yeah.\nSpeaker B: We're going to need to put in a really good battery so people don't have to charge their remote control every few days.\nSpeaker C: That's why I think the option of the, the kinetic thing, which basically means as long as you shake it like a watch, like an on the watch.\nSpeaker C: But if people are going to want to shake their movie control, it's probably sensitive enough when you fill it.\nSpeaker C: Right.\nSpeaker C: So you could trigger that to a light, like I said, the bouncing ball thing or you could trigger that to use that to power the light as opposed to, so when they pick it up right.\nSpeaker C: And then that sort of triggers.\nSpeaker C: Right.\nSpeaker C: Okay.\nSpeaker F: Okay.\nSpeaker F: Well, let's just go right back to the marketing ideas for a start.\nSpeaker F: And it's keeping an idea on the time we've got about 15 minutes to play with at most.\nSpeaker F: Okay.\nSpeaker F: So, um, yeah.\nSpeaker F: So just to bear in mind that the ultimate goal of this meeting is to reach an absolute decision on the, the concepts of the product.\nSpeaker F: So back to your idea about, um, incorporating the idea of like fruit and veg and the corporate color and things like that.\nSpeaker F: Um, I mean, what does anybody think about?\nSpeaker F: Does anybody have any ideas about how we can fit all that in together?\nSpeaker F: I mean, that's kind of the user interface type of thing.\nSpeaker F: What are your thoughts on that?\nSpeaker A: I think I'm a corporate.\nSpeaker A: Um, maybe not our, our fruit ship, the, uh, very sort of curvy type ship, then you can have this intra-textual and color as stuff.\nSpeaker F: So maybe do it, are we thinking something that like could sit in your hand comfortably or sit you tooled on to comfortably or?\nSpeaker A: Um, possibly sitting in your hand comfortably.\nSpeaker A: So something quite curvy.\nSpeaker F: Okay.\nSpeaker F: Um, right.\nSpeaker F: Okay.\nSpeaker F: Color wise.\nSpeaker F: I mean, you, it was a year.\nSpeaker F: I can't remember who made the point about how if you have a nice bright color, you're not losing it.\nSpeaker F: Is that?\nSpeaker F: What's that mean?\nSpeaker F: About how if you have a bright color, you'll not lose it so much.\nSpeaker F: Yeah.\nSpeaker F: Um, and when the corporate color is yellow, I mean, maybe we can think about, about the color of the whole product being yellow, I don't know.\nSpeaker F: Um, and then obviously the, uh, the materials, but as anybody got like an overall picture in their mind about what, what might work.\nSpeaker B: Well, I mean, I'm thinking that what we need to do is have something that kind of unifies a lot of the different concepts.\nSpeaker B: And if we think that what we are, our number one marketing motive is, um, the look and feel.\nSpeaker B: So for the look and feel to seem coherent and not just sort of bits and bits and pieces of, of concept and technology or whatever, or fashion, then we should have it kind of come back to one thing that we kind of all sort of can visualize.\nSpeaker B: Um, maybe what we could do is think about a concept which touches back to the color, you said company color, yellow.\nSpeaker B: I mean, if we think of something like, I was saying also lime and lemon, you know, what can we come up with something where we, we try and associate it with, with like the series.\nSpeaker B: We just come up with something like that.\nSpeaker B: We kind of use it as a theme to inspire the shapes and things.\nSpeaker B: Right.\nSpeaker C: Is there a particular shape that you're interested in like, this marketing, how many research on, does it need to be long, does it need to be a square thing?\nSpeaker C: And I think circular and shape or, yeah, because that will help narrow down the choice of,\nSpeaker F: choice of material. I was kind of thinking about as well, you know how you get these shock resistant mobile phones and they're plastic but they also have like rubber around the outside that kind of feels, it feels kind of warmer to the touch, feels a bit more comfortable.\nSpeaker F: Maybe it could incorporate plastic on rubber and to it and then we could have carb shapes because we're titanium yet it's going to have to be boxy and rectangular and I think we might be moving away from that, you know.\nSpeaker F: So, do we want it in like the shape of a lemon or?\nSpeaker F: No, no.\nSpeaker B: No, no.\nSpeaker B: No, I don't think we do.\nSpeaker B: It's more, more just that we, we think about like, what it is we're trying to achieve and then we have one, one sort of theme that we stick with.\nSpeaker F: Right, well, so thoughts about the actual shape is a thing.\nSpeaker A: I'm quite like a sort of snowman type shape.\nSpeaker F: A snowman shape.\nSpeaker A: And so, sort of larger bits it's new hands.\nSpeaker A: And then you've got the yellow, the top, or just any other thing you need.\nSpeaker F: That's quite a distinctive shape that we go with.\nSpeaker D: Yeah.\nSpeaker F: Yeah.\nSpeaker F: So, yeah, should we go with that?\nSpeaker B: Can we, yeah, like, to any, like, do you have a...\nSpeaker B: Do you want to draw it on the board?\nSpeaker B: Yeah, just, we can visualize it.\nSpeaker A: Something like that.\nSpeaker A: You've got two groups there.\nSpeaker A: Maybe you could fold up and you got a third group inside.\nSpeaker A: Okay.\nSpeaker A: That'd be good.\nSpeaker A: You have one which was about there.\nSpeaker F: Okay.\nSpeaker F: So, call that the snowman shape, trademark.\nSpeaker F: That's cool.\nSpeaker F: And, I mean, color wise, which is everything.\nSpeaker F: I think it is quite important to get yellow in there somewhere.\nSpeaker F: I mean, do you want the whole thing yellow?\nSpeaker F: Do you want something?\nSpeaker A: I think it's quite nice if you just add this around here.\nSpeaker A: They start background yellow and then have a nice bold color for the buttons.\nSpeaker F: Okay.\nSpeaker F: Cool.\nSpeaker F: And also, I mean, how are we going to incorporate this lugan in?\nSpeaker F: The fact that it talks to you, I mean, it might be quite cool if when you first start using it, it says, what is it putting fashion into electronics or something?\nSpeaker F: I don't know.\nSpeaker F: Or when you, like, turn it off or something.\nSpeaker F: If it can speak, if you could actually say this lugan, it might be a bit more powerful than just having it written on it somewhere.\nSpeaker F: And any thoughts on that at all?\nSpeaker A: I think that probably scare me.\nSpeaker A: Turn on your...\nSpeaker F: I know.\nSpeaker F: Unless, I mean, if you, also, like, were, if we wanted to incorporate an LCD display, where would you put that?\nSpeaker F: Will you put that on the inside?\nSpeaker C: Do we need an LCD display?\nSpeaker C: What's the functionality of that?\nSpeaker F: It's bound to increase the cost of it a lot.\nSpeaker C: Yeah, but the question is, what are we using it? What would we achieve from it?\nSpeaker C: Putting in lights is cheap, but putting in LCD panels just to make it glow is a bit...\nSpeaker B: When you mentioned LCD, I think that would be about...\nSpeaker B: And the thing I could see it helping with the...\nSpeaker B: If it was somehow connected with listings.\nSpeaker B: So as you scroll through, because we said we might have a jog dial, as you scroll through your stations, you can...\nSpeaker B: It actually tells you what it is.\nSpeaker C: I think that will be a problem because we don't have an input device to get the listings into it.\nSpeaker C: So it's a bit nuts to get the user there.\nSpeaker C: One day, Tuesday, Wednesday.\nSpeaker C: So no need for an LCD display?\nSpeaker C: I'm not saying there's no need for an LCD display, but it's...\nSpeaker C: What would it tell the user because the LCD tends to be an output as opposed to an input?\nSpeaker C: So there's a remote control need to talk back to the user.\nSpeaker C: We have the option of the speaker as well.\nSpeaker C: The same thing goes for the speaker. Is there a need for a remote controller?\nSpeaker F: Why don't I... There is really...\nSpeaker F: Talk back.\nSpeaker F: I would say no need for a talk back. Does anybody disagree with that?\nSpeaker C: We could put a game on it.\nSpeaker C: Easy.\nSpeaker C: When the TV dies, you can play with the remote control.\nSpeaker F: Right, so you're going to have these three different sets of functionalities.\nSpeaker F: I mean, do you want to group them into head of the snowman, and the body of the snowman inside of the snowman? Is that what you're thinking?\nSpeaker A: Well, I think the advanced ones, the ones you don't usually use can be head inside.\nSpeaker A: Okay.\nSpeaker A: I think they're the three groups we had, where we're pretty basic ones.\nSpeaker A: They'd have to go on there, they're fun somewhere.\nSpeaker F: Okay. Right.\nSpeaker F: What else do you need to talk about?\nSpeaker C: Well, where would you physically position the buttons?\nSpeaker C: I think that has some impact on many things.\nSpeaker C: Maybe you want to draw into that.\nSpeaker F: So I'm just going to pop this in here because I have a slide about decision making.\nSpeaker F: Yep.\nSpeaker F: I'll try to figure out the word.\nSpeaker F: Oh, go on, we've got five words.\nSpeaker F: Okay.\nSpeaker F: Back again.\nSpeaker F: Energy.\nSpeaker F: Do you think that's suggesting we're hard-pouring the thing?\nSpeaker F: I really like the idea of this kinetic thing, where you'd have the backup of the battery, but have kinetic power.\nSpeaker B: I've had kinetic things before, and the one issue we need to keep in mind with them is that you're committing the user to moving it.\nSpeaker B: No, like I said, we have that.\nSpeaker C: Watch is hybrid kind of thing, so it's not going to charge the battery.\nSpeaker C: Right.\nSpeaker C: Okay.\nSpeaker C: It's just...\nSpeaker B: It's just worth pointing out because I've known people to have kinetic watches that they wear all the time, and it's just like magic because it's always powered and there's no battery.\nSpeaker B: I've also known people to have things like a jewelry watch they wear from time to time, and they eventually just say it's just too much of a nuisance because I don't wear it all the time.\nSpeaker B: Remote control is similar.\nSpeaker B: Your way on vacation, I don't know.\nSpeaker B: Whatever.\nSpeaker B: Something...\nSpeaker B: And it just starts to get worn down.\nSpeaker F: Let's face that, if you're away and you're not using it, then you're not using any power either.\nSpeaker F: So you'd have the battery as to kind of keep it ticking over idea.\nSpeaker F: I'm really sorry we're going to have to wrap up quite quickly.\nSpeaker F: I don't know as much time as I thought.\nSpeaker F: So I think that's what energy is referring to here.\nSpeaker F: Chip on print.\nSpeaker F: Is that an industrial design thing?\nSpeaker F: Yes.\nSpeaker F: Okay.\nSpeaker F: As for the case, kind of discuss that.\nSpeaker B: And this size here, I'd suggest this be small.\nSpeaker B: Yeah.\nSpeaker F: And we're going to have rubber buttons that feel kind of...\nNone: Okay.\nSpeaker B: Just a lot of...\nSpeaker B: I mean, one of the things running through my mind right now, I realize we're being efficient to wrap up the meeting and have lots of decisions made.\nSpeaker B: But we are leaning quite a bit to the side of being low tech rubber buttons, plastic frame.\nSpeaker B: It's almost like we're reproducing the same old remote control that's out there.\nSpeaker B: Should we think about how we are actually getting this high tech user-friendly theme?\nSpeaker B: Like what is it that we're using to achieve those goals?\nSpeaker F: I think that backlighting the buttons and stuff like that.\nSpeaker C: Or even a clear case.\nSpeaker C: A glowing yellow type case where the yellow showable, but in the dark, it's sort of...\nSpeaker C: It's a light.\nSpeaker C: It's slightly transparent.\nSpeaker F: So it's yellow, like tinted yellow, but you can maybe see threads out there.\nSpeaker C: Or there might be a light running through it like a mouse.\nSpeaker C: You know, you have colors nice and they don't need that much power.\nSpeaker C: So the battery in that sense, where you have one or two strategically placed lights.\nSpeaker C: Sure.\nSpeaker C: That sort of...\nSpeaker C: Yeah, they emanate a light through it.\nSpeaker C: Yeah, but because the case is transparent, so it gives it a little bit of a glow.\nSpeaker B: That's a make it freaky.\nSpeaker B: And then the other thing that we've committed ourselves achieving is simplicity.\nSpeaker B: And so I'm thinking maybe should we try and think about having something like some kind of an innovative concept about how the volume and the channels are controlled.\nSpeaker B: Because that's the main thing people just want to do.\nSpeaker B: Could we use like a jog dial, like a nice just sort of round somewhere on it where you just roll it?\nSpeaker C: When you're rolling it, how do you want to roll it? Do you want them to roll it like that? Do you want them to roll it like that?\nSpeaker C: Because in a mouse, your hands in a position to roll it.\nSpeaker C: Whereas the other thing about having a jog dial this way, it tends to get moved accidentally.\nSpeaker C: Why don't we do it like a mouse?\nSpeaker F: You could do that, couldn't you? If you're holding it in your hand.\nSpeaker C: That's a very unnatural motion to... Yeah.\nSpeaker C: Okay.\nSpeaker C: Can you imagine you have to scroll a lot?\nSpeaker C: We might work for volume and maybe some of the brightness controls and stuff like that.\nSpeaker C: But not for channels, right? If you have a tele-west box, you don't have to buy all the channels. You can go out to 50 channels.\nSpeaker C: You can imagine trying to... Okay.\nSpeaker C: And I don't think having that too quick, too slow, it's confusing to the...\nSpeaker A: I don't know.\nSpeaker A: For skipping watch, you have to skip the channel button.\nSpeaker C: And users tend to want to use that and once they lose out and then use the experience, it becomes the most accessible thing.\nSpeaker B: But that's not a bad thing, is it?\nSpeaker B: Because if you think about it, the alternative is to push the button.\nSpeaker B: Yeah.\nSpeaker B: Jog dials are much easier than that.\nSpeaker B: Okay.\nSpeaker F: Right.\nSpeaker F: Well, we do need to make a decision on whether we want to incorporate a job dial in nice and quickly.\nSpeaker F: I'm all for them, actually. I think they're quite unique, very quick to use.\nSpeaker F: So, does anybody oppose the idea of incorporating one into the design at all? No.\nSpeaker F: And the other thing was, can we think of any way of getting the slogan into this thing?\nSpeaker B: Yeah. Like, I mean, if we keep coming back to this board here, I wouldn't be surprised if we could take this idea.\nSpeaker B: Imagine that, I don't know, that it's within the shape of the hand, it's quite small.\nSpeaker F: Ooh, okay. We really got to wrap up.\nSpeaker B: Yeah.\nSpeaker B: It's small, and that we've got the slogan somewhere on the casing, the side.\nSpeaker B: Okay. What if we can do that, great?\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: I mean, isn't that what we just said we just have to decide now?\nSpeaker F: Yeah.\nSpeaker F: Let's try and get the slogan on there.\nSpeaker C: And then I can just spend the shape of the...\nSpeaker B: And then like a jog dial, somewhere that fits in with the shape of it.\nSpeaker B: Like, I don't know, like here.\nSpeaker C: That might have one problem in terms of...\nSpeaker B: You never get bumped.\nSpeaker C: In terms of whether you're left-handed or you're right-handed, you might be locking yourself in.\nSpeaker B: Or maybe just fitting in like down the middle here.\nSpeaker C: Okay.\nSpeaker C: Can I just jump in and suggest something to the audience?\nSpeaker F: Yeah, I'm really going to have to hurry on here because we're actually over time.\nSpeaker F: It's kind of...\nSpeaker F: Is there anything anybody's unsure about?\nSpeaker F: Just for enclosing, just the next meeting is going to be in 30 minutes, and so you can see in the screen here what each of you are going to hopefully be doing.\nSpeaker F: And I know that the designers are going to be working and playing out on that.\nSpeaker F: So that'll be good.\nSpeaker F: And I'll get the minutes as soon as possible.\nSpeaker F: Anything at all you think we haven't discussed that we need to?\nSpeaker F: Everybody kind of happy about what they're going to be doing?\nSpeaker C: I think one thing would be the jog dial because that's going to have quite a big impact on the thing.\nSpeaker B: I think the jog does.\nSpeaker B: You know, just after you do that, what if it was flat and you just...\nSpeaker B: Yeah, that's what I was thinking.\nSpeaker C: That would be great.\nSpeaker C: A slide.\nSpeaker C: Because then you don't have to put the hand.\nSpeaker C: I think incorporating the logo is quite straightforward.\nSpeaker C: There's lots of space for it.\nSpeaker C: That's kind of a design thing that you guys can discuss.\nSpeaker C: But it's also a marketing and a function.\nSpeaker B: Okay.\nSpeaker B: And materials we sort of said we do plastic and rubber, didn't we?\nSpeaker B: Yeah.\nSpeaker B: And I think maybe we should try and stay away from just the big protruding rubber buttons because that'll just be so standard.\nSpeaker F: Do we have something flush with the case?\nSpeaker F: Something a bit more flush.\nSpeaker B: Okay.\nSpeaker B: Maybe have rubber incorporated into the case as well.\nSpeaker B: Okay.\nSpeaker B: And also plastic I've seen can get really textured.\nSpeaker B: You can get plastics that actually feel soft in your hands.\nSpeaker B: Okay.\nSpeaker B: It's kidding.\nSpeaker B: You can feel kind of like...\nSpeaker B: No.\nSpeaker B: You get balloons now and then you'd think that they were rubber but they're not.\nSpeaker B: They're actually just plastic that's textured.\nSpeaker B: Kind of a little bit on that.\nSpeaker F: I'm going to have to call this to your clothes because we're away every time.\nSpeaker F: So that's really good.\nSpeaker F: Like we've had so much to talk about that pretty much run out of time to do so.\nSpeaker F: So what do you do?\nSpeaker F: Okay.\nSpeaker F: I'm going to have to do some design stuff.\nSpeaker F: Play it all down.\nSpeaker F: Yeah.\nNone: Quite jealous actually.\nSpeaker C: You got to choose first.\nSpeaker C: No, we're kidding.\nSpeaker C: Can I just swipe your power cable?\nSpeaker C: I don't think it matters.\nSpeaker C: I'm going to pull everybody out first and then put in whoever needs to be left.\nSpeaker C: It's you.\nSpeaker E: Yeah.\nSpeaker C: This is real.\nSpeaker C: It has all the...\nSpeaker C: Oops.\nSpeaker C: I'm going to take the microphones because it's too lazy to take them off again.\nSpeaker D: Cool.\n"}, {"meeting_id": "Bro007", "summary": "The meeting participants discuss results from experiments and challenges that the model was facing. There was a significant server side delay, so they could not accommodate silence detection. Members noted that reducing model dimensions had a detrimental effect on model performance. The professor wanted to know the size of words that the word error rate was calculated on and explained that there would be no new training. From this point, they can start talking about future directions and work on fixing the system.", "dialogue": "Speaker D: Okay, so, today, looking at a number of things we're trying and, fortunately, for listeners to this, some of it's visual, but got tables in front of us. What is combo mean? So, combo\nSpeaker E: is a system where we have these features that go through the network and then these same three more features, but low pass filter with the low pass filter use in the MSG features.\nSpeaker E: So these low pass filter goes through another MLB and then the linear outputs of these two MLBs are combined just by adding the values and then there is this guilty and the output\nSpeaker D: is used as features. So, let me try to restate this and see if I have it right. There's the features, there's the OGI features and then those features go through a contextual, let's take this bottom arrow, one point by the bottom arrow. Those features go through a contextualized KLT. Then these features also get low pass filtered.\nSpeaker E: Yeah, so, yeah, I could grab through this and blackboard. Yeah, that's good. So, we have these features from OGI that goes through three paths, three, okay. The first is a KLT using several frames of the features. The second path is MLB, also using nine frames, several frame of features. The third path is this low pass filter, MLB. Adding the output is just like in the second proposal for the first evaluation. And then MLB and those\nSpeaker D: two together. Okay, so that's this bottom one. And then the one at the top and I presume these things that are in yellow are in yellow because overall they're the best. Yeah, that's the reason. Oh, that's focused on them. So, what's the block diagram for the one above it?\nSpeaker E: For the first yellow line, you mean? Yeah. So, it's basically the same except that we don't have this low pass filtering. So, we have only two streams. Well, there's no low pass processing. Use this additional feature stream. Do you, they mentioned, when I was\nSpeaker D: on the phone, what's the note? They mentioned some waiting scheme that was used to evaluate\nSpeaker E: all these numbers. Actually, the waiting seems to... Well, it's 40% 40, I think it's 60% for all the speech that comes with all these languages. The well-match is 40, medium 35, even. I miss much 25. And we don't have the TI digits. No. Okay. But yeah, generally what we observe with TI digits is that the result are very close, whatever the system. Okay.\nSpeaker D: And so, have you put all these numbers together into a single number representing that?\nSpeaker D: Not yet. Okay. So, that should be pretty easy to do. Yeah. And that would be good to make a compare the two and say what was better. Yeah. And how does this compare to the numbers?\nSpeaker E: Oh, so, OGI2 is just the top. Yeah. Top row. So, yeah. So, actually, OGI2 is the baseline with the OGI features. But this is not exactly the result that they have because they've still made some changes in the features. Okay. But actually, our results are better than their results. I don't know by how much because they did not send us their new results.\nSpeaker D: Okay. Okay. So, the one place where it looks like we're messing things up a bit is in\nSpeaker E: the highly mismatched Italian. Yeah. Yeah. There is something for any of you here because there are 36 and then sometimes we are around 42. Now, one of the ideas that you had mentioned\nSpeaker D: last time was having a second silence detection. Yeah. So, there are some results here.\nSpeaker E: So, the third and the fifth line of the table. So, felt is what that is. Yeah. So, it seems for the well-matched mismatched condition, it brings something. But, actually, apparently there is no room left for any silence detector at the server side because of the delay.\nSpeaker D: Oh, we can't do it. Okay. Yeah. Too bad. Good idea. But, anyway. Okay. Except, I don't know\nSpeaker E: because I think they are still working well. Two days ago, they were still working on this, trying to reduce the delay of the silence detector. So, but yeah, if we had time perhaps, we could try to find some kind of compromise between the delay that's on the end set and on the server side, but try to reduce the delay on the end set. But, well, for a moment, they have this large delay on the feature computation. Okay. So, all right. So, for now,\nSpeaker D: at least that's not there. You have some results with low pass filter, capstone. It doesn't have a huge effect, but it looks like it maybe can help in a couple places. Yeah. A little bit. And, yeah. And, let's see. Well, just that we have in there. I guess, at this point, this is, I guess I should probably look at these others a little bit. You yellow these out, but, oh, I see, yeah, that one you can't use because of the delay. That looks pretty good. Let's see. That one.\nSpeaker D: Well, even just the second row doesn't look that bad, right? That's just, that looks\nSpeaker E: like an interesting one, too. Actually, the second line is pretty much like the first line in yellow, except that we don't have this guilty on the first, on the left part of the diagram, just add the features. Yeah. Yeah. So, when we do this weighted measure,\nSpeaker D: we should compare the two because it might even do not better. And, it's a little slightly simpler. So, there's, so I would put that one also as a maybe. And, yeah. And, it's actually, it does, does significantly better on the highly mismatched Italian. So, and love a worse on the, it's on the MM case, but, well, yeah, it's worse on a few things.\nSpeaker D: So, see how that, see how that comes out on their measure. And, are we running this for TI digits or? Yeah. Now, is TI, is that part of the result that they get for the development, the results that they're supposed to get at the end of the month, the TI digits are there also? It's here. Okay. Okay. Let's see what else there is here. Oh, I see the one that I was looking down here at the row below the lower yellow one. That's, that's with the reduced KLT size. Yeah, reduced functionality. What happens there is it's around the same.\nSpeaker D: And so, you could reduce the dimension. What you're saying? Yeah, it's, it's, it's significant, but, but it's significantly worse. It's, it's, it's, it's, it's, it's, it's mostly worse. Yeah. I did this a little, I mean, not, not by a huge amount. I don't know.\nSpeaker D: What, what are the sizes of any of these sets? I'm sure you told me before, but I forgot.\nSpeaker D: So, you know, how many words are in one of these test sets? Yeah. Well, it depends, well,\nSpeaker E: the one much is generally larger than the other sets. And I think it's a row. Two thousand, two thousand, two thousand. What? I don't know. Sentences. Sentences. Some sets have 500\nSpeaker D: sentences. Yeah. So, so the sets, so the test sets are between 500 and 2,000 sentences, let's say, each sentence on the average has four or five digits or is it most of them\nSpeaker C: longer? Yeah. For Italian, even 70, but sometimes the sentences have only one digit and three times, like the number of credit cards, something like that. Right. So, between one and 16, see\nSpeaker D: the, I mean, the reason I'm asking is is we have all these small differences and I don't know how seriously to take them right. So, yeah. If you had, just, you know, to give an example, if you had, if you had a thousand words, then a tenth of a percent would just be one word, right. So, so, it means anything. So, yeah, be kind of, I can't, I can know what the sizes of these tests are. The size of the, yeah. Since these, well, also just to know the numbers, right. So, these, these are word error rates. So, this is on how many words.\nSpeaker C: Yeah. We have the result of the output of the H-tika. Yeah. The number of sentences, not\nSpeaker D: the numbers. Yeah, sure. Yeah. So, anyway, if you could just mail out what those numbers are and then that would be great. What else is there here? See, the second, the second from the bottom, it says S-I-L. This is some different type of silence or thing. What\nSpeaker C: was that? The output silence of the MLP. It's only one small experiment to know what happened. To fly also, to include also the silence of the MLP. We have the 56 phone and the silence to pick up the silence and to include also. This silence plus the K-L-T output.\nSpeaker C: Oh, so you're only using the silence? Yeah, because when we apply the K-L-T output.\nSpeaker E: No, I think there is this silence in addition to the K-L-T output. In addition to the K-L-T output.\nSpeaker E: Because we, we just keep, we don't keep all the dimensions after the K-L-T. And we're not sure if we try to add the silence also in addition to the 28 dimensions. Okay. What, what's OGI 45? It's OGI 2. It's the features from the first line. Yes. Right, but I mean, what is the last row mean? So, it's basically this, but without the K-L-T on the left. I\nSpeaker D: thought that was the one. I thought that was the second row. What's the difference between\nSpeaker E: the second line? You don't have these combos stuff, so you just... Oh, so this is like the\nSpeaker C: second line, but with the combo stuff. And with the output of the combo. Yeah. Okay, so,\nSpeaker D: all right, so it looks to me, I guess, the same given that we have to take the field ones out of the running because of this delay problem. It looks to me like the ones you said, I agree, there are the ones to look at, but it just would add the second row. Yeah. One. And then, if we can... Oh, yeah, also, when they're using this weighting scheme of 40, 35, 25, is that on the percentages or on the raw errors, I guess it's probably on the percentages, right?\nSpeaker E: I guess, yeah. Yeah, I kind of guess. All right, it's that query. Okay, maybe they'll argue about it.\nSpeaker D: Okay, so if we know how many words are in each, and then they've, they've promised to get something tomorrow, which will be there as far as they've gotten, right? And we'll operate with that.\nSpeaker D: How long did it... I guess if we're not doing all these things, if we're only doing...\nSpeaker D: I guess since this is development data, it's legitimate to do more than one, right?\nSpeaker D: I mean, or now, if in final test data, you don't want to do several and take the best. That's that's not proper, but if this is development data, we could still look at a couple. Yeah.\nSpeaker E: Yeah, we can, yeah, sure. But we have to decide, I mean, we have to fix the system.\nSpeaker E: Yes, on this data. Choose the best. Right. The question is when do we fix the system?\nSpeaker D: Do we fix the system tomorrow, or do we fix the system on Tuesday?\nSpeaker D: I think we fix on Tuesday, yeah. Okay, except that we do have to write it up.\nSpeaker E: Yeah, well, basically it's this way. Perhaps some kind of printing on some of the pets.\nSpeaker D: Right, so maybe what we do is we get the data from them, we start the training and so forth, but we start the write-up right away, because as you say, there's only minor differences between these.\nSpeaker E: I think we could start to, yeah. Yeah, there's something. Yeah.\nSpeaker D: And I would, yeah, I would kind of like to see it. Maybe I can add it a bit.\nSpeaker D: The money in the situation is my forte, which is English.\nSpeaker D: So, yeah, have you seen, do they have a format for how they want the system descriptions or anything? Not really. Okay.\nSpeaker E: There is the format of the table, which is quite impressive.\nSpeaker D: I see. Yes, for those who are listening to this and not looking at it, it's not really that impressive, it's just tiny.\nSpeaker D: It's all these little categories.\nSpeaker D: So they said B, said C, multi-conditioned clean.\nSpeaker D: No mitigation. Wow.\nSpeaker D: You know what, no, no mitigation means.\nSpeaker E: It should be...\nSpeaker E: Oh, that's probably the general error.\nSpeaker D: This is probably a channel error stuff.\nSpeaker D: Oh, this is... Right, it says right above your channel error resilience.\nSpeaker D: Yeah.\nSpeaker D: So recognition performance is just the top part, actually.\nSpeaker D: And they have the aspect between scene databases and non-scene, so basically, due development and evaluation.\nSpeaker D: And so, right, it's presumed there's all sorts of tuning that's gone on on the...\nSpeaker D: Let's see what they call scene databases.\nSpeaker D: And there won't be tuning for the unseen...\nSpeaker D: Multi-condition. Multi-condition.\nSpeaker D: So they have... Looks like they have...\nSpeaker D: So they're splitting up between the TI digits and everything else, I say.\nSpeaker D: So they everything else is the speech debt car. That's the...\nSpeaker E: Multi-conditioned. Yeah, so it's not divided between languages, you mean, or...\nSpeaker D: Well, it is. It is. But there's also these tables over here for the TI digits and these tables over here for the car, which is, which is, I guess, all the multi-lingual stuff.\nSpeaker D: And then there's...\nSpeaker D: They also split up between multi-conditioned and clean only.\nSpeaker E: Okay. Forty-indigets. Yeah. Actually, yeah. Forty-indigets, they want to train clean and unnoisy.\nSpeaker E: Yeah.\nSpeaker D: So we're doing that also, I guess.\nSpeaker E: Yeah. But...\nSpeaker E: We actually... Do we have the features? Yeah, for the clean-conditioned, but we did not test it yet.\nSpeaker E: The clean training stuff.\nSpeaker E: Okay.\nSpeaker D: Well, anyway, sounds like there'll be a lot to do just to work with our partners to fill out the tables for the next few days. I guess they have to send it out. See, the 31st is Wednesday.\nSpeaker D: And I think it has to be there by some hour European time on Wednesday. So I think basically...\nSpeaker C: It was a long time when they maybe, because... Excuse me?\nSpeaker C: That the difference in the time may be... It's so long different of the time. Maybe the 12th of the night of the Thursday one is not valid in Europe. We don't know exactly.\nSpeaker D: Yeah. So I mean, I think we have to actually get it on Tuesday, because I think...\nSpeaker E: Except if it's the 31st of midnight, we don't know.\nSpeaker E: Can't still do some work on Wednesday.\nSpeaker D: Yeah, well... Is it been... I thought it was actually something like 5pm on...\nSpeaker D: It was like, I thought it was 5pm or something. I didn't think it was midnight.\nSpeaker D: I thought they said they wanted it at 5pm.\nSpeaker D: Well, so 5pm their time is...\nSpeaker C: No, like 3pm. 3pm.\nSpeaker C: 3pm. 3pm.\nSpeaker D: Alright, that's six in the morning here.\nSpeaker C: No. 3... 3... 8... 3pm?\nSpeaker E: No, we were wondering about the... the... the... the... the... the... the... we have to...\nSpeaker C: Oh yeah, yeah, yeah.\nSpeaker E: I don't know if it's 3pm.\nSpeaker C: 3pm here is in Europe, with 9.\nSpeaker C: Yeah, it's 3pm here.\nSpeaker D: Yeah, it's 3pm here. But I didn't think it was midnight that it was...\nNone: Oh.\nSpeaker D: I thought it was due it's some hour during the day, like 5pm or something.\nSpeaker D: In which case...\nSpeaker D: So I... well, we should look, but my assumption is that we basically have to be done Tuesday.\nSpeaker D: Yeah. So, then next Thursday we can sort of have a little aftermath.\nSpeaker D: But then we'll actually have the new day, which is the German and the Danish.\nSpeaker D: But that really will be much less work, because the system will be fixed.\nSpeaker D: Yeah.\nSpeaker D: So all we'll do is take whatever they have and...\nSpeaker D: and run it through the process.\nSpeaker D: We won't be changing the training on anything.\nSpeaker D: So there'll be no new training, there'll just be new HDK runs.\nSpeaker D: So that means in some sense we can kind of relax from this after Tuesday.\nSpeaker D: And maybe next meeting we can start talking a little bit about where we want to go from here in terms of the research.\nSpeaker D: You know, what things...\nSpeaker D: Did you think of when you were doing this process that just didn't really have time to adequately work on?\nSpeaker E: Yeah.\nSpeaker B: Oh, Stefan always had these great ideas.\nSpeaker B: Oh, but I would have time.\nSpeaker E: I'm not sure the other great ideas.\nSpeaker E: That's right.\nSpeaker C: But there are ideas.\nSpeaker C: It was a great idea.\nSpeaker C: To apply the sick guilty was a great idea.\nSpeaker D: Yeah, that was good.\nSpeaker D: And also it's still true that I think it's true that we at least got fairly consistent improve results by running the neural net transformation in parallel with the features rather than in sequence which was your suggestion and that seems to have been born out.\nSpeaker D: The fact that none of these are enormous is not too surprising.\nSpeaker D: Most improvements aren't enormous.\nSpeaker D: Some of them are.\nSpeaker D: But when you have something really, really wrong and you fix it, you can get really enormous improvements.\nSpeaker D: It's our best improvements over the years that we've gotten from finding bugs.\nSpeaker D: Okay, well, I think I see where we are and everybody knows what they're doing.\nSpeaker D: Is there anything else we should talk about?\nSpeaker E: I think it's okay.\nSpeaker E: So basically we will, I think we will try to focus on these three architectures.\nSpeaker E: And perhaps I was thinking also a fourth one with just a single guilty because we did not really test that removing these guiltys in particular single guilty at the end.\nSpeaker D: Yeah, I mean that would be pretty low maintenance to try it if you can fit it in.\nSpeaker D: Oh, I have, yeah, I do have one other piece of information which should tell people outside of this group too.\nSpeaker D: I don't know if we're going to need it.\nSpeaker D: But Jeff at the University of Washington has gotten a hold of some kind of server farm of 10 multi-processor IBM machines or 6,000s.\nSpeaker D: And so I think each one is four processors or something.\nSpeaker D: I don't know, 800 megahertz or something. There's four processors in a box and there's 10 boxes and there's some kind of type. So he's got a lot of processing power.\nSpeaker D: And we'd have to schedule it but if we have some big jobs and we want to run them, he's offering it.\nSpeaker D: So it's a, when he was here, he used not only every machine here but every machine on campus as far as I could tell. So it's got his payback. But again, I don't know if we'll end up with if we're going to be CPU limited and anything that we're doing in this group but if we are, that's an offer.\nSpeaker D: Okay, well, guys doing great stuff. So that's really neat.\nSpeaker D: I don't think we need to.\nSpeaker D: Oh, well, the other thing I guess that I will say is that the digits that we're going to record momentarily are starting to get into a pretty good size collection.\nSpeaker D: And in addition to the speech data stuff, we will have those to work with really pretty soon now.\nSpeaker D: So that's another source of data which is, understand what better control and that we can make measurements of the room that, you know, if we feel there's other measurements we don't have, we'd like to have. We can make them.\nSpeaker D: Dave and I were just talking about that a little while ago.\nSpeaker D: So that's another possibility for this kind of work.\nSpeaker D: Okay, if nobody else has anything else, maybe we should go around. Do our digits, do our digits, doodie. Okay, I'll start.\nSpeaker D: 3231-3250. Let me say that again, transcript number 3231-3250.\nSpeaker D: 780-03819-0598-1794-2345-74737-84416-9988-083-08556-1-2003.\nSpeaker D: 4246-5234-403-670276.\nSpeaker A: I'm reading transcript 3251-3270.\nSpeaker A: 898559-006-16150-243-278-4350-4756-479-567-606-70000-904.\nSpeaker A: Oh, sorry, 914-07-06-19-23-407-23-744-624-7479-605-8451.\nSpeaker E: Transcript 3291-3310-07-0898-187-2-301-55636-798-8866-868636-9-006-3867-186-866.\nSpeaker E: 329-00332-99-2636-35348-76-495-6-78090-03060.\nSpeaker B: Transcript 3171-3190-5980-659817175394-89292836-9-011825-3604515158.\nSpeaker B: 365-360-45156-700928-77-05430506589-126434.\nSpeaker C: Transcript 3191-321-061113-728992897-96100109-0100142336-47250-5455.\nSpeaker C: 8992897 961000 0901014233647250 0.5446787078041907 0.43461753329347 0.4367257\nNone: 0.2347 quite very clean spring contours\n"}, {"meeting_id": "Bro013", "summary": "The meeting focused on the results from the existing model, speeding up the backend for feature direction, and a few tangential topics being developed for future exploration. The team was unsure why their model was not performing as well as they expected it to, and they suspected that their smaller dataset might be at fault. They also discussed how they could improve the performance of the backend model by improving feature detection. The meeting ended with some students sharing the tangential fields they were exploring and a brief discussion on the different pronunciations of zero.", "dialogue": "None: We're going.\nNone: Okay.\nNone: Close your door.\nSpeaker F: Thanks.\nSpeaker F: Door in the way up.\nSpeaker F: Yeah, I'm going to get this other door.\nSpeaker F: Okay.\nSpeaker F: So, what are we talking about today?\nSpeaker D: Well, first of all, perhaps these meeting record or digits?\nSpeaker F: Oh, yeah.\nSpeaker F: That was kind of interesting.\nSpeaker F: Both the SRI system and for one thing, that sure shows the difference between having a lot of training data or not.\nSpeaker F: The best kind of number we have on English, on near microphone only is 3 or 4%.\nSpeaker F: And it's significantly better than that using fairly simple front ends on the SRI system.\nSpeaker F: So, I think that that's using pretty huge amount of data.\nSpeaker F: Mostly not digits, of course, but then again, well, yeah, in fact, mostly not digits.\nSpeaker F: So, in this case, what is this using digits in the digital image?\nSpeaker F: Did anybody mention about whether the SRI system is doing the digits?\nSpeaker F: The word as a word model or as a sub-phone.\nSpeaker D: I guess it's a lot of fault models.\nSpeaker D: Yeah, I think so because it's very huge system.\nSpeaker D: But so, there is one difference.\nSpeaker D: Well, the SRI system, the result for the SRI system that are presented here are with adaptation.\nSpeaker D: So, there is a complete system including online, supervised adaptation.\nSpeaker D: If you don't use adaptation, the error rate is around 50% more.\nSpeaker F: Okay, is that much?\nSpeaker D: Yeah, it's quite significant.\nSpeaker F: Still.\nSpeaker F: But what I think I'd be interested to do given that is that we should take, I guess, as somebody is going to do this right, is to take some of these tandem things, feed it into the SRI system.\nSpeaker D: Yeah.\nSpeaker D: Yeah, because...\nSpeaker D: But I guess the main point is the data because I'm not sure or back down this is very simple, but still know what the attempts to improve it, that's a thing that what I mean.\nSpeaker D: Which I tried to do.\nSpeaker F: Yeah, but he's doing with the same data, right?\nSpeaker F: So there's two things to be infected.\nSpeaker F: One is that there's something simple that's wrong with the back end, playing number states.\nSpeaker F: I don't know if he got to the point of playing with the number of Gaussian's.\nSpeaker F: But, you know, but yeah, so far he hadn't gotten any improvement, but that's all with the same data, which is pretty small.\nSpeaker D: So you could retrain some of these tandem?\nSpeaker F: Well, you could do that, but I'm saying even with that part not retrained.\nSpeaker F: Just using having the HMMs, much better HMMs, yeah.\nSpeaker F: But just train those HMMs using different features, features coming from RRAS now.\nSpeaker D: But what would be interesting to see also is perhaps it's not really the amount of data but the recording conditions, I don't know, because it's probably not the problem of noise, because our features are supposed to be robust today.\nSpeaker D: It's not the problem of channel because there is normalization with respect to the channel.\nSpeaker F: So, I'm sorry, what is the problem that you're trying to explain?\nSpeaker D: The fact that the result with the tandem or a system model.\nSpeaker D: So much worse?\nSpeaker D: No, so much worse.\nSpeaker F: Oh, but I'm almost certain that it has to do with the amount of training data.\nSpeaker F: It's orders of magnitude on.\nSpeaker D: Yeah, but retrained on the own digits and it's a digit task.\nSpeaker F: But having a huge, if you look at what commercial places do, they use a huge amount of data.\nSpeaker F: This is a modest amount of data.\nSpeaker F: So, ordinarily you would say, well given that you have enough occurrences of the digits, you can just train the digits rather than with it.\nSpeaker F: But the thing is if you have a huge, you know, the word models, but if you have a huge amount of data, then you're going to have many occurrences of similar to the alpha one.\nSpeaker F: And that's just a huge amount of training point.\nSpeaker F: So, I think it has to be that because as you say, this is near microphone.\nSpeaker F: It's really pretty clean data.\nSpeaker F: Now, some of it could be the fact that, let's see, in these multi-trained things, did we include noisy data in the training?\nSpeaker F: I mean, that could be hurting us actually for the quick.\nSpeaker D: Yeah, actually we see that the clean train for proposals are better.\nSpeaker F: It is better.\nSpeaker F: Yeah, because this is clean data.\nSpeaker F: So, that's not too surprising.\nSpeaker D: But, I guess what I meant is that, well, let's say if we add enough data to train on the meeting record digits, I guess we could have better examples of this.\nSpeaker D: What I meant is that perhaps you can learn something from this, what's wrong?\nSpeaker D: What is different between the IDGids and these digits?\nSpeaker F: What kind of numbers are we getting on the IDGids?\nSpeaker D: It's 0.8 persons.\nSpeaker F: Oh, I see.\nSpeaker F: So, on the actual TIDGIDGID database, we're getting 28 percent.\nSpeaker F: And here we're getting three or three, let's say three for this.\nSpeaker F: Yeah, sure.\nSpeaker F: But I mean 0.8 percent is something like double or triple, what people have gotten who have worked very hard at doing that.\nSpeaker F: And also, as you point out, there's adaptation in these numbers also.\nSpeaker F: So, if you put the adaptation off, then for the English near, you get something like 2 percent.\nSpeaker F: And here you had something like 3.4.\nSpeaker F: And I could easily see that difference coming from this huge amount of data that was trained on.\nSpeaker F: So, I don't think there's anything magical here.\nSpeaker F: It's a simple HGK system with a modest amount of data.\nSpeaker F: And this is a modern system.\nSpeaker F: It has a lot of nice points to it.\nSpeaker F: So, the HGK is an older HGK, you know, so.\nSpeaker F: Yeah, it's not that surprising.\nSpeaker F: For me, it just meant a practical point that if we want to publish results on digits that people pay attention to, we probably should, because we've had the problem before, we can show some nice improvement on something that seems like too large a number.\nSpeaker F: And people don't necessarily think it's so serious.\nSpeaker F: Yeah, so the 3.4 percent for this is, looks, why is it, it's an interesting question of us.\nSpeaker F: Why is it 3.4 percent for the digits reported in this environment, as opposed to the 0.8 percent for the original TI digits database?\nSpeaker F: Yeah, given the same, yes, so ignoring the SRA system for a moment, just looking at the NM system, if we're getting 0.8 percent, which, yes, it's high.\nSpeaker F: It's awfully high, but it's, you know, it's high.\nSpeaker F: Why is it 4 times it's high?\nSpeaker F: More.\nSpeaker F: I guess this much would be harder.\nSpeaker F: Yeah, I guess.\nSpeaker F: Right, I mean, even though it's close mic, there still really is background noise.\nSpeaker F: And I suspect when the TI digits were recorded if somebody fumbled or said something wrong or something, they probably made them take it over.\nSpeaker F: I mean, there's no attempt to have it be realistic in any sense at all.\nSpeaker D: And it's quite different.\nSpeaker D: Yeah, it's very, very clean.\nSpeaker D: And it's like, it's got a Uricorn, whereas it's a mid-range, it's a mid-range, sometimes you have breath noise.\nSpeaker F: Right, yeah, so I think they were, it's got a little extra.\nSpeaker F: Yeah, I think it's, it's, so, yeah, I think it's, it's an indication it's hard.\nSpeaker F: Yeah, again, that's true either way.\nSpeaker F: I mean, so take a look at the, yeah, so I result in a much better, but still you're getting something like 1.3% for things that are the same data as in TI digits, the same text.\nSpeaker F: And I'm sure the same system would get 0.3, 0.4% or something on the actual TI digits.\nSpeaker F: So this, I think, on both systems, these digits are showing up as harder.\nSpeaker F: Which I find so interesting, because I think this is closer to, I mean, still red, but I still think it's much closer to what, what people actually face.\nSpeaker F: And they're dealing with people saying digits over the telephone.\nSpeaker F: I don't think, I mean, I'm sure they wouldn't erase the numbers, but I don't think that the, the, the company is, the telephone speech get anything like 0.4% on their digits.\nSpeaker F: I'm sure they get, I mean, for one thing people do, from now prove don't have middle America access.\nSpeaker F: So you ask it has, as many people.\nSpeaker F: It's not many different ways.\nSpeaker F: Okay, that was that topic. What else we got?\nSpeaker F: Did we end up giving up on, on, in a year speech submissions?\nSpeaker F: Man, Otilo and Dan Elles are submitting something.\nSpeaker D: Yeah, I guess the only thing with these, the meeting recorder.\nSpeaker D: So I think, yeah, basically give it.\nSpeaker F: Now, for the, for the, we do have stuff for a run, right?\nSpeaker F: Yeah, yeah, yeah, for sure.\nSpeaker D: Yeah. Well, that's fine.\nSpeaker F: So, so we have a couple, couple of little things on meeting recorder.\nSpeaker F: You don't, we don't have to fly it with papers.\nSpeaker F: We don't have to prove anything to anybody.\nSpeaker F: That's fine.\nSpeaker F: Anything else?\nSpeaker D: Yeah, well, so perhaps the place that we've been working on.\nSpeaker D: Yeah, we have put the good VAD in the system.\nNone: It really makes a huge difference.\nSpeaker D: So yeah, I think this is perhaps one of the reason why our system was that, not the best because with the new VAD, there is a similar to the, that still come with ourselves.\nSpeaker D: Perhaps even better sometimes.\nSpeaker D: So there is this point.\nSpeaker D: The problem is that it's very big and still have to think how to, where to put it.\nSpeaker D: Because it, what is VAD, yield some delay.\nSpeaker D: And we, if we put it on the server side, it doesn't work because of the server side features.\nSpeaker D: So you already have the LDA applied from the terminal side to your cumulative delay.\nSpeaker D: So the VAD should be before the LDA, which means perhaps on the terminal side and then smaller.\nSpeaker F: So where does good VAD come from?\nSpeaker D: It's from OGI.\nSpeaker D: So it's the network with the huge amount of, the huge units and nine input frames compared to the VAD that was in the proposal, which has a very small amount of hidden units and fewer inputs.\nSpeaker F: This is the one they had originally.\nSpeaker F: Yeah, but they had to be rid of it because of the space.\nSpeaker D: But the assumption is that we'll be able to make a VAD that's smaller than that.\nSpeaker F: Well, so that's the problem.\nSpeaker F: But the other thing is to use a different VAD entirely.\nSpeaker F: I don't know what the thinking was amongst the LCPOQ, but if everybody agreed to use this VAD.\nSpeaker D: Or they just one apparently they don't want to fix it really because they think there is some interaction with you.\nSpeaker D: Feature extraction and video frame dropping.\nSpeaker D: But they still want to just to give some requirement from this VAD because it will not be part of the standard.\nSpeaker D: So it must be at least somewhat fixed but not completely.\nSpeaker D: So there just will be some requirements that are still not yet.\nSpeaker F: But I was thinking that sure there may be some interaction, but I don't think we need to be stuck on using our OGI's VAD with somebody else's smaller.\nSpeaker F: Yeah.\nSpeaker F: That's good.\nSpeaker D: So there is this thing.\nSpeaker D: Yeah.\nSpeaker D: I designed a new filter because when I designed the filters, we showed a delay from the LCPOQ filters.\nSpeaker D: There was one filter with 60ms delay and the other 10ms.\nSpeaker D: Right. You'd like to suggest that both could have 65.\nSpeaker D: I think it's 65.\nSpeaker D: Yeah.\nSpeaker D: But should have 65 because I didn't gain anything.\nSpeaker D: So I did that and it's running.\nSpeaker D: Let's see.\nSpeaker D: Let's see if I have one.\nSpeaker D: But the filter is of course closer to the reference filter.\nSpeaker E: Yeah.\nSpeaker F: So that means logically, the principle should be better, so probably it will be worse.\nSpeaker F: Or the basic perverse nature of reality.\nSpeaker F: Yeah.\nSpeaker F: Okay.\nSpeaker F: Yeah.\nSpeaker D: Yeah.\nSpeaker D: And then we've started to work with this.\nSpeaker D: Vice to the vice.\nSpeaker D: And we will perhaps try to have a new system with MSG stream.\nSpeaker D: So something that's similar to the proposal too, but with MSG stream.\nSpeaker E: Okay.\nSpeaker B: Now, we're going to play with Matt Lat and to found some parameter robust for voice and voice decision, but only to play.\nSpeaker B: And we found that maybe when it's a classical parameter, the variance between the FFT of the signal and the small spectrum of time after the Melfilter bank.\nSpeaker B: And while it's smaller, it's good for clean speech.\nSpeaker B: It's quite good for noise speech.\nSpeaker B: But we must have a big statistic with timid.\nSpeaker B: And it's not right yet to use, well, I don't know.\nSpeaker D: Yeah.\nSpeaker D: So basically we want to look at something like excitation signal.\nSpeaker D: Right.\nSpeaker D: Which is a variance of 15.\nSpeaker B: Yeah.\nSpeaker B: I have here for one thing, for one frame.\nSpeaker B: The mix of the two noise and noise.\nSpeaker B: And the signal is this, clean and this noise.\nSpeaker B: They are the two, the mix, the big.\nSpeaker B: The signal is for clean.\nSpeaker F: Well, there's no, these axes are labeled.\nSpeaker B: So I don't know what this axis is.\nSpeaker B: This axis is frame.\nSpeaker F: And what's this?\nSpeaker B: This is energy, logarithm energy of the spectrum.\nSpeaker B: No, this is the variance, the difference between the spectrum of the signal and FFT of each frame of the signal and the small spectrum of time after the Melfilter.\nSpeaker B: For the two.\nSpeaker B: And the here they are to signal this is for clean and this is for noise.\nSpeaker F: Oh, there's two things on the same graph.\nSpeaker B: Yeah, I don't know.\nSpeaker B: I think that they have another graph.\nSpeaker E: Which is clean?\nSpeaker E: Which is noise?\nSpeaker D: I think the lower one is noise.\nSpeaker B: The lower is noise and the height is clean.\nSpeaker F: Okay, so it's harder to distinguish.\nSpeaker F: It's hard.\nSpeaker F: But it's worth the answer, of course.\nSpeaker B: Oh, I must have clean.\nSpeaker B: I don't have two different.\nSpeaker F: Yeah, presumably when there is a...\nSpeaker D: So this should be voiced.\nSpeaker B: Yeah, this height is voiced portion.\nSpeaker B: And this is the noise portion.\nSpeaker B: And this is more or less like this.\nSpeaker B: But I must have to have two pictures.\nSpeaker B: This is for example for one frame, the spectrum of the signal.\nSpeaker B: And this is the smooth version of the spectrum after ML filter bank.\nSpeaker B: Yeah.\nSpeaker B: And this is...\nSpeaker B: This is not the different.\nSpeaker B: This is trying to obtain with LPC model, the spectrum, but using MATLAB without going factor and...\nSpeaker B: Not pre-efficient.\nSpeaker B: Not pre-efficient, nothing.\nSpeaker B: And I think that this is good.\nSpeaker B: This is quite similar.\nSpeaker B: This is another frame, how I obtain the envelope, the same envelope, with the ML filter bank.\nSpeaker F: Right.\nSpeaker F: So now I wonder, I mean, do you want to...\nSpeaker F: I know you want to get it something orthogonal from what you get with the smooth spectrum.\nSpeaker F: But if you were really trying to get a voicetant voice, do you want to totally ignore that?\nSpeaker F: I mean, do you...\nSpeaker F: I mean, clearly a very big, very big cues for a voicetant voice come from a spectrum slope and so on.\nNone: Right.\nSpeaker D: Yeah, well, this would be...\nSpeaker D: This would be perhaps an addition of parameters.\nSpeaker D: Yeah, I see.\nSpeaker B: Yeah, because when the noise is clear, in this section is clear, if a high value is indicative that this voice frame and no one...\nSpeaker F: Yeah, we probably won't.\nSpeaker F: Certainly, if you want to do a good voicetant voice, it's actually a need of few features.\nSpeaker F: Each feature is myself.\nSpeaker F: I don't know, but people look at it slow.\nSpeaker F: First-dollar correlation coefficient, by by power, or...\nSpeaker F: There's...\nSpeaker F: I guess we probably don't have enough computation to do a simple pitched detector or something.\nSpeaker F: A pitched detector, you could have an estimate of what the...\nSpeaker F: Or maybe you could just do it going through the FFTs, or you could get out some probable harmonic structure.\nSpeaker F: Right?\nSpeaker B: You have read, you have a paper.\nSpeaker B: The paper just gives me just the...\nSpeaker B: I don't know, but...\nSpeaker B: They are some problem.\nSpeaker B: Yeah, that's a matter of...\nSpeaker B: It's another problem.\nSpeaker D: Yeah, there is...\nSpeaker D: This fact, actually, if you...\nSpeaker D: Look at this spectrum.\nSpeaker D: Yeah.\nSpeaker D: What's this again?\nSpeaker D: Is it the main field?\nSpeaker B: Yeah, like this.\nSpeaker B: Uptime like this.\nSpeaker D: So, the envelope here is the output of the MFFFF.\nSpeaker D: And what we clearly see is that in some cases, and clearly appears here, and the harmonics are resolved by the...\nSpeaker D: Well, there is still a pair after MFFF.\nSpeaker D: And it happens for a pitched noise, because the width of the low frequency MFFF is sometimes even smaller than pitch.\nSpeaker D: Yeah, it's a problem.\nSpeaker D: Right?\nSpeaker D: 150 hertz.\nSpeaker D: And so what happens is that this...\nSpeaker D: Add additional variability to this envelope.\nSpeaker D: Yeah.\nSpeaker D: So we were thinking to modify the mass spectrum to have something that's smaller on low frequencies.\nSpeaker F: That's a separate thing, yeah.\nSpeaker E: Yeah, it's a separate thing, yeah.\nSpeaker E: Yeah, maybe so.\nSpeaker F: Yeah, so, yeah, but that was time I was just starting with the FFT.\nSpeaker F: You could do a very rough thing to estimate pitch.\nSpeaker F: And given that, you could come up with some kind of estimate of how much of the...\nSpeaker E: The energy was explained by...\nSpeaker E: by those harmonics.\nSpeaker F: It's very...\nSpeaker F: The...\nSpeaker F: The metal does give a smooth thing, but as you say, it's not that smooth here.\nSpeaker F: So if you just subtract it off, you're the S of the harmonics, then something like this would end up with quite a bit lower energy.\nSpeaker F: First, 50 hertz or so.\nSpeaker F: And if it was noisy, the proportion of it would go down to B.\nSpeaker E: And if it was on lowest, there's nothing.\nSpeaker E: So you had to be able to pick out voiced segments.\nSpeaker F: At least it should be another cue.\nSpeaker E: Okay.\nSpeaker F: That's what's going on.\nSpeaker C: So, I went to talk with Mike Jordan this week, and shared with him the ideas about extending the Larry Saul work.\nSpeaker C: And I asked him some questions about factorial HMM.\nSpeaker C: So like later down the line, when we come up with these feature detectors, how do we model the time series that happens?\nSpeaker C: And we talked a little bit about factorial HMMs and how when you're doing inference, or when you're doing recognition, there's like simple, but turbid stuff that you can do for these HMMs.\nSpeaker C: And the great advantages that a lot of times the factorial HMMs don't over-learn the problem there.\nSpeaker C: They have a limited number of parameters and they focus directly on the subproblems at hand.\nSpeaker C: So you can imagine five or so parallel features transitioning independently.\nSpeaker C: And then at the end, you couple these factorial HMMs with undirected links based on some more data.\nSpeaker C: So he seemed like really interested in this and said this is something very dual wall and learned a lot.\nSpeaker C: And I've just been continuing reading about certain things, thinking maybe using modulation spectrum stuff to as features also in the sub-air, because it seems like the modulation spectrum tells you a lot about the intelligibility of certain words and stuff.\nSpeaker C: So yeah, just about it.\nSpeaker A: Okay, and so I've been looking at Avandano's work.\nSpeaker A: I'll try to write up in my next status report, and I'll write down what he's doing, but it's an approach to deal with reverberation, or the aspect of his work that I'm interested in.\nSpeaker A: The idea is that normally, and analysis frames are too short to encompass reverberation effects in full.\nSpeaker A: You miss most of the reverberation tail in a 10 millisecond window.\nSpeaker A: And so you'd like it to be that the reverberation response is simply convolved in, but it's not really with these 10 millisecond frames.\nSpeaker A: But if you take, say, a two millisecond window, I'm sorry, a two second window, then in a room like this, most of the reverberation response is included in the window.\nSpeaker A: And then things are more linear.\nSpeaker A: It is more like the reverberation response is simply convolved.\nSpeaker A: And you can use channel normalization techniques.\nSpeaker A: Like in his thesis, he's assuming that the reverberation response is fixed, he just does mean subtraction, which is like removing the DC component of the modulation spectrum.\nSpeaker A: And that's supposed to deal pretty well with the reverberation.\nSpeaker A: And the neat thing is you can't take these two second frames and feed them to a speed recognizer.\nSpeaker A: So he does this method training, the spectral resolution for time resolution.\nSpeaker A: And synthesizes a new representation, which is with, say, 10 second frames, but a lower frequency resolution.\nSpeaker A: So I don't really know the theory.\nSpeaker A: I guess these are called time frequency representations.\nSpeaker A: And he's making the time finer grain in the frequency resolution.\nSpeaker A: Let's find grain.\nSpeaker A: So I guess my first stab actually in continuing his work is to re-implement this thing, which changes the time and frequency resolutions, because he doesn't have code for me, so that it will take some reading about the theory.\nSpeaker A: I don't really know the theory.\nSpeaker A: Oh, and another first step is, so the way I want to extend his work is make it able to deal with a time varying reverberation response.\nSpeaker A: And we don't really know how fast the reverberation response is varying in the meeting recorder data.\nSpeaker A: So we have this blockly squares, echo counselor implementation.\nSpeaker A: And I want to try finding the response between a near mic and the table mic for someone using the echo counselor and looking at the echo counselor tabs.\nSpeaker A: And then see how fast that varies from block to block that should give an idea of how fast the reverberation response is changing.\nSpeaker E: Okay.\nSpeaker F: I think we're sort of done.\nSpeaker F: Yes, we did it.\nSpeaker F: We did it and go home.\nSpeaker F: Okay.\nSpeaker D: I'm reading transcript L-40, 3, 4, 9, 1, 0, 5, 1, 8, 3, 6, 2, 0, 5, 7, 0, 4, 1, 9, 6, 0, 1, 4, 4, 9, 5, 7, 3, 7, 6, 1, 4, 2, 0, 5, 2, 7, 8, 9, 9, 3, 6, 6, 3, 8, 9, 7, 1, 7, 8, 3, 1, 9, 3, 0, 1, 9, 5, 7, 5, 5, 1, 8, 0, 8, 2, 9, 8, 4, 6, 1, 9, 4, 8, 1, 2,\nSpeaker A: So I think you read some of the zeros as always and some of zeros. Is there a particular way we're supposed to read them?\nSpeaker D: There are only zeros here.\nSpeaker F: No, oh, it's 0 and 0, two ways that we say that digit.\nSpeaker D: So it's perhaps in the sheets, it should be another sign photo.\nSpeaker D: If we want to, the guy to say photo.\nSpeaker F: No, I mean, I think people will do what they say.\nSpeaker F: It's okay.\nSpeaker F: I mean, digit recognition is done before.\nSpeaker F: You have two pronunciations for that value.\nSpeaker D: It's perhaps more difficult for the people to prepare the database.\nSpeaker D: If, because they already put me in the order.\nSpeaker D: No, they just write pronounce 0 or 0.\nSpeaker F: They write down OH or they write down ZRO.\nSpeaker D: Yeah, but if each other, she'd be prepared with a different sign photo.\nSpeaker F: But people wouldn't know what that, well, I mean, there is no convention for it.\nSpeaker F: Yeah.\nSpeaker F: I mean, you'd have to tell them, okay, when we write this, say it, they just want people to read the digits as you would in early work.\nSpeaker F: And people say it different ways.\nSpeaker A: Okay, is this a change from the last batch of, of, um, forms?\nSpeaker A: Because in the last batch, it was spelled out, which one you should read.\nSpeaker F: Yes, that's right.\nSpeaker F: It was spelled out and they decided they wanted to get at more of the way people would really say.\nSpeaker F: Okay.\nSpeaker F: That's also why they're, they're bunched together in these different groups.\nSpeaker F: Okay.\nSpeaker F: So it's, yeah, so it's, it's, it's, it's, it's, they're being spying.\nSpeaker F: Transcript L-39.\nSpeaker F: 2326-1014-2475.\nSpeaker F: 938-726-2627.\nSpeaker F: 6734-2224.\nSpeaker F: 2964040882.\nSpeaker F: 879-94082.\nSpeaker F: 780-395123.\nSpeaker F: 559-8142.\nSpeaker F: 0209-2926.\nSpeaker F: Actually, let me just, since it's spread up, I was just, it was hard not to be self-conscious about that one.\nSpeaker F: The reasons we just discussed it.\nSpeaker F: But I realized that, that, um, when I'm talking on the phone, certainly, and, and seeing these numbers, I almost always say zero.\nSpeaker F: And because, because, uh, it's too solvable, so it's, it's more likely to understand what I said.\nSpeaker F: That, that's the habit I'm in.\nSpeaker F: But some people say, oh, okay.\nSpeaker C: Yeah, I normally say, oh, it's easier to say.\nSpeaker F: Yeah, it's true.\nSpeaker F: Yeah, so, so, uh, no, don't think about it.\nSpeaker F: Oh, no.\nSpeaker C: Okay, I'm reading transcript L-38.\nSpeaker C: 545-032-858.\nSpeaker C: 338-904109.\nSpeaker C: 850-711-140.\nSpeaker C: 2161-82-5678.\nSpeaker C: 576-82004.\nSpeaker C: 710-0587-756.\nSpeaker C: 561-371913.\nSpeaker C: 436-009-9220.\nSpeaker A: I'm reading transcript L-37.\nSpeaker A: 0519-0327-1669.\nSpeaker A: 627-026-4510.\nSpeaker A: 542-9501.\nSpeaker A: 711-271-8123.\nSpeaker A: 4084-57-622-9.\nSpeaker A: 823-6726-764.\nSpeaker A: 927-3123.\nSpeaker A: 936-861-9177.\nSpeaker B: Transcript L-36.\nSpeaker B: 044-1629-0.\nSpeaker B: 477-3845654688.\nSpeaker B: 312-525459.\nSpeaker B: 970-698-5851.\nSpeaker B: 0832-973-145.\nSpeaker B: 6493-844223.\nSpeaker B: 4553-0245-422.\nSpeaker B: 682-189-819.\nSpeaker F: I have no access in postcode.\nNone: There is some technical problem here.\n"}, {"meeting_id": "Bro012", "summary": "The meeting began with the professor discussing the technical features of their recordings. The interaction between the hardware and the model had to be mediated through different techniques. Some members on the team were working on low-pass down-sampling as a replacement for LDA filters. Low-pass seemed to perform better, but the team was uncertain about its robustness. The professor thought that the team should submit something to Eurospeech and that they should further study reverberation. The professor thought that studying spectrograms would be a good place to start. The team then discussed VAD and line mean normalization as techniques for getting rid of noise. The meeting ended with a brief discussion on feature detection.", "dialogue": "Speaker H: Hello. Hey, we're on.\nSpeaker H: Okay, so I had some interesting mail from Dan Ellis. Actually, I think he redirected everybody also. So the PD-A mics have a big bunch of energy at five hertz. Where this came up was that I was showing off these waveforms that we have on the web and I just sort of had noticed this, but the major component in the wave in the second waveform, in the pair of waveforms, is actually the air conditioner.\nSpeaker H: So I have to be more careful about using that as a good illustration. In fact, it's not of effects of rumor or vibration. It isn't a bad illustration of the effects of room noise on some mics.\nSpeaker H: And then we had this other discussion about whether this affects the dynamic range, because I know, although you start off with 32 bits, you end up with 16 bits and, you know, are we getting hurt there, but Dan's pretty confident that we're not that the quantization error is still not a significant factor there.\nSpeaker H: So there was a question of whether we should change things here, whether we should change a capacitor on the put box for that or whether we should. Yeah, he suggested a smaller capacitor, right?\nSpeaker H: Right. He then I had some other thing discussed with him and the feeling was once we start munching with that, and many other problems could happen. And additionally, we already have a lot of data that's collected with that.\nSpeaker H: Yeah. The simple thing to do is he has a, I forget if this wasn't that mayor in the following mail, but he has a simple filter, a digital filter that he suggested. We just run over the data, where we deal with it.\nSpeaker H: The other thing, I don't know the answer to, but when people are using Fickalk here, whether they're using it with the high pass filter option or not. And I don't know if anybody knows.\nSpeaker H: You can go check. Yeah. So when we're doing all these things using our software, there is, if it's based on the RASTA PLP program, which does both PLP and RASTA PLP, then there is an option there, which then comes up through to Fickalk, which allows you to do a high pass filter.\nSpeaker H: And in general, we like to do that because of things like this. And it's pretty, it's not a very severe filter, doesn't affect speech frequencies, even pretty low speech frequencies at all.\nSpeaker E: What's the cutoff frequency that you use?\nSpeaker H: Oh, I don't know why I wrote this a while ago. Is it like 20? Something like that.\nSpeaker H: I mean, I think there's some effect above 20, but it's mild.\nSpeaker H: So, I mean, it's probably, there's probably some effect up to 100 hertz or something, but it's pretty mild.\nSpeaker H: I don't know, in the strut implementation of this stuff, is there a high pass filter or pre-impicist or something?\nSpeaker F: I think we use a pre-impic basis.\nSpeaker H: So, we want to go and check that for anything that we're going to use the PDA mic for.\nSpeaker H: He says that there's a pretty good roll-off in the PCM mics.\nSpeaker H: So, we don't need to worry about them, I'm aware of the other, but if we do make use of the cheap mics, we want to be sure to do that that filtering before we process them.\nSpeaker H: And again, if it's depending on the option that our software is being runway, it's quite possible that's already being taken care of.\nSpeaker H: But I also have to pick a different picture to show the effects of reverberation.\nSpeaker E: Did somebody notice it during your talk?\nSpeaker H: No. Well, they may have, but they were...\nSpeaker H: They were nice. But, I mean, the thing is, since I was talking about reverberation and showing this thing that was noise, it wasn't a good match, but it certainly was still an indication of the fact that you get noise with distant mics.\nSpeaker H: It's just not a great example because not only isn't a reverberation, but it's a noise that we definitely know what to do.\nSpeaker H: It doesn't take deep, new bold new methods to get rid of the fiber-hearts noise.\nSpeaker H: But so, it was a bad example in that way, but it's still a real thing that we did get out of the microphone in a distance, so it wasn't wrong in the scene appropriate.\nSpeaker H: But someone noticed it later, right out to me, and they went, oh, man, why didn't I notice that?\nSpeaker H: So I think we'll change our picture around the web when we go that way.\nSpeaker H: One of the things I was trying to think about, what's the best way to show the difference?\nSpeaker H: I had a couple thoughts. One was that spectrogram that we show is okay, but the thing is, the eyes, and the brain behind them, so good at picking out patterns from noise, that in first glance, you look at them, it doesn't seem like it's that bad, because there's many features that are still preserved.\nSpeaker H: So one thing to do might be to just take a piece of the spectrogram where you can see that something looks different, and blow it up, and have that be the part, just to show as well.\nSpeaker H: Some things are going to be hurt.\nSpeaker H: Another I was thinking of was taking some spectral slices, like we look at with the recognizer, look at the spectrum or capstrom that you get out of there, and the reverberation does change that, so maybe that would be more obvious.\nSpeaker D: Spectral slices?\nSpeaker H: What do you mean?\nSpeaker H: I mean, all the recognizers look at frames, so that one instant time.\nSpeaker H: So it's one point in time, or over 20 milliseconds, or something, you have a spectrum or a capstrom, that's why I meant slice.\nSpeaker E: You could just throw up the MFCC feature vectors, one from one, one from the other, and then you can look and see how different the numbers are.\nSpeaker E: Right, what else are I saying either?\nSpeaker E: See how different these sequences are numbers.\nSpeaker H: Or I could just add them up and get a different total. It's not the square.\nNone: What else is going on?\nSpeaker F: Yeah, at first I had to remark, why, I am wondering why the PDA is so far. We were always meeting at the beginning of the table.\nSpeaker H: I guess because we had one of the moves. We could move us.\nSpeaker F: Yeah, so since the last meeting, we've tried to put together the clean, the OPS, the new filter that's replacing the Filters, and also the delay issues, so that we can consider the delay issue for the online organization.\nSpeaker F: So we've put together all this, and we have results that are not very impressive, but there is no real improvement.\nSpeaker H: But it's not worse, and it's better latency, right?\nSpeaker F: Actually, it's better. It seems better when we look at the mismatch case, but I think we are like cheated here by this problem that in some cases when you modify a slight, slightly modified initial condition, you end up completely somewhere error, somewhere else in the space.\nSpeaker F: Yeah.\nSpeaker F: Well, the other system, for instance, for Italian, is at 78% recognition rate on the mismatch.\nSpeaker F: This new system has 89%, but I don't think it indicates something really.\nSpeaker F: I don't think it means that the system is more robust, it's simply affected.\nSpeaker H: Well, the test would be if you then tried it on one of the other tests.\nSpeaker H: So this was Italian, right?\nSpeaker H: So then if you take your changes, then...\nSpeaker F: From the 78% recognition rate system, I could change the transition probabilities for the first HMM between the NWC and the MWC.\nSpeaker F: By using 0.5 instead of 0.6, 0.4, I think the HMMC is great.\nSpeaker E: Yeah, I looked at the results when Stefan did that, and it really happens.\nSpeaker E: I mean, the only difference is you change the self-loop transition probability by a tenth of a percent, and it causes 10% difference in the word of a percent.\nSpeaker E: Yeah, from point...\nSpeaker E: I'm sorry, you change a point one.\nSpeaker E: And then not a tenth of a percent, one tenth.\nSpeaker E: So from 0.6 to 0.5, and you get 10% better.\nSpeaker E: And I think it's what you basically hypothesized in the last meeting about just being very...\nSpeaker E: And I think you've mentioned this in your email too. It's just very... you know, you get stuck in some local minimum and this thing throws you out of it, I guess.\nSpeaker H: Well, what are... according to the rules, what are we supposed to do about the transition probability? Are they supposed to be 0.5 or 0.6?\nSpeaker E: I think you're not allowed to... yeah, it's supposed to be 0.6.\nSpeaker H: 0.6. It's supposed to be 0.6.\nSpeaker E: Yeah, but changing into 0.5, I think, is... which gives you much better results, but that's not allowed.\nSpeaker F: Yeah, but even if you miss 0.5, I'm not sure if we'll always give you the better results.\nSpeaker E: Yeah.\nSpeaker E: That's sad.\nSpeaker E: Right, we only tested it on the medium mismatch, right?\nSpeaker E: You said on the other cases you didn't notice.\nSpeaker F: I think the reason is... yeah, I...\nSpeaker F: So I see my name, I think.\nSpeaker F: So it's the fact that the mismatch is trained only on the far microphone.\nSpeaker F: Well, for the mismatch case, everything is... using the far microphone training testing.\nSpeaker F: Whereas for the I-N-Mismatch training is done on the close microphone, so it's clean speech-based.\nSpeaker F: So you don't have this program of looking at the microphone.\nSpeaker F: And for the well-match, it's a mix of close microphone and distant microphone.\nSpeaker E: I did notice something...\nSpeaker E: So it's not just the whole difficult training.\nSpeaker E: Somebody, I think, was Morgan suggested at the last meeting that I actually count to see how many parameters and how many frames.\nSpeaker E: And there are almost 1.8 million frames of training data and less than 40,000 parameters in the baseline system.\nSpeaker E: So it's very, very few parameters compared to how much training data.\nSpeaker H: Yeah, so that says that we could have lots more parameters.\nSpeaker E: Yeah, I did one quick experiment just to make sure I had everything worked out.\nSpeaker E: And I just... for most of the... for all of the digit models, they end up at three mixtures per state.\nSpeaker E: And so I just did a quick experiment, or I changed it so that it went to four.\nSpeaker E: And it didn't have any significant effect at the medium mismatch and high mismatch cases.\nSpeaker E: And it was just barely significant for the well-match better.\nSpeaker E: So I'm going to run that again, but with many more mixtures per state.\nSpeaker H: Yeah, because at 40, you could have... yeah, easily four times as many parameters.\nSpeaker E: And I think also, just seeing what we saw in terms of the expected duration of the silence model when we did this tweaking of the self-loop, the silence model expected duration was really different.\nSpeaker E: And so in the case where it had a better score, the silence model expected duration was much longer.\nSpeaker E: So it was like, it was a better match.\nSpeaker E: I think, you know, if we make a better silence model, I think that will help a lot too for a lot of these cases.\nSpeaker E: So, but one thing I wanted to check out before I increased the number of mixtures per state was in their default training script, they do an initial set of three re-estimations.\nSpeaker E: And then they build the silence model.\nSpeaker E: And then they do seven iterations.\nSpeaker E: Then they add mixtures. And then they do another seven. Then they add mixtures.\nSpeaker E: Then they do a final set of seven. And they quit.\nSpeaker E: Seven seems like a lot to me. And it also makes the experiments go take a really long time.\nSpeaker E: I mean, to do one turn around of the well-matched case takes like a day.\nSpeaker E: And so, you know, in trying to run these experiments, I noticed, you know, it's difficult to find machines, you know, compute to run on.\nSpeaker E: And so one of the things I did was I compiled HTK for the Linux machines.\nSpeaker E: Because we have this one from IBM that's got like five processors in it.\nSpeaker E: Right.\nSpeaker E: And so now you can run stuff on that. And that really helps a lot.\nSpeaker E: Because now we've got, you know, extra machines that we can use for compute.\nSpeaker E: And if I'm doing running an experiment right now where I'm changing the number of iterations from seven to three, to see how it affects the baseline system.\nSpeaker E: And so if we can get away with just doing three, we can do many more experiments more quickly.\nSpeaker E: And if it's not a huge difference from running with seven iterations, you know, we should be able to get a lot more experiments done.\nSpeaker E: And so I'll let you know what happens with that.\nSpeaker E: But if we can, you know, run all of these back ends with many fewer iterations.\nSpeaker E: And on Linux boxes, we should be able to get a lot more experimenting done.\nSpeaker E: So I wanted to experiment with cutting down the number of iterations before I increased the number of Gaussian.\nSpeaker H: So how is it going on the?\nSpeaker H: So you did some things.\nSpeaker H: They didn't improve things in a way that convinced you you'd substantially improved anything.\nSpeaker H: But they're not making things worse and we have reduced latency, right?\nSpeaker F: Yeah, but actually, actually it seems to do a little bit worse for the one-match case.\nSpeaker F: And we just noticed that actually the way the final score is computed is quite funny.\nSpeaker F: It's not a mean of word error rate.\nSpeaker F: It's not a weighted mean of error rate.\nSpeaker F: It's a weighted mean of improvements.\nSpeaker F: So which means that actually the weight on the well-match is...\nSpeaker F: Well, what happened is that if you have a small improvement or a small bit on the well-match case, it will have a huge influence on the improvement compared to the reference because the reference system is quite good for well-match case also.\nSpeaker E: So it weights the improvement on the well-match case really heavily compared to the improvement on the other cases?\nSpeaker F: It's a weighting of the improvement, not of the error rate.\nSpeaker E: Yeah, and it's hard to improve on the best case because it's already so good, right?\nSpeaker F: Yeah, but what I mean is that you can have a huge improvement on the HM case, like 5% absolute.\nSpeaker F: And this will not affect the final score, or most... this will not affect the final score because the improvement relative to the baseline is small.\nSpeaker H: So they do improvement in terms of accuracy rather than word error rate?\nSpeaker F: Improvements.\nSpeaker F: It's compared to the word error rate.\nSpeaker H: Okay, so if you have 10% error and you get 5% absolute improvement, then that's 50%.\nSpeaker H: Okay, so what you're saying then is that if it's something that has a small word error rate, then even a relatively small improvement on it, and absolute terms will show up is quite large in this. Is that what you're saying? Yes. Okay, but yeah, that's the notion of relative improvement.\nSpeaker H: Yeah.\nSpeaker H: Word error rate.\nSpeaker F: Sure, but when we think about the weighting, which is 0.5, 0.3, 0.2, it's an absolute... on relative figures.\nSpeaker F: Yeah.\nSpeaker F: So we look at this error rate?\nSpeaker H: That's why I've been saying we should be looking at word error rate and not at accuracy.\nSpeaker H: I mean, we probably should have standardized the 9 all the way through.\nSpeaker E: Well, I mean, it's not that different, right? I mean, just subtract the accuracy.\nSpeaker H: Yeah, but when you look at the numbers, your sense of the relative is quite different.\nSpeaker H: If you had 90% correct and 5% 5 over 90, it doesn't look like it's a big difference, but 5 over 10 is big.\nSpeaker H: So just when you're looking at a lot of numbers, getting a sense of what was important.\nSpeaker E: That makes sense.\nSpeaker F: Well, anyway, so yeah, so it hurts a little bit on the well match.\nSpeaker H: What's a little bit like?\nSpeaker F: Like, it's difficult to say because again, I'm not sure I have to...\nSpeaker E: Hey Morgan, do you remember that signif program that we used to use for testing signif?\nSpeaker E: Is that still valid? I've been using that.\nSpeaker H: Yeah, it was actually updated.\nSpeaker H: It was updated some years ago and cleaned it up, made some things better.\nSpeaker E: Okay, I should find that new one. I just used my old one from 92 or whatever.\nSpeaker H: Yeah, I'm sure it's not that different, but...\nSpeaker H: He was a little more rigorous as I recall.\nSpeaker F: So it's around like 0.5...\nSpeaker F: 0.6% absolute on the Daniel.\nSpeaker H: Worst?\nSpeaker H: Worst, yeah.\nSpeaker H: Out of what?\nSpeaker F: We start from 94.64, and we go to 94.04.\nSpeaker H: So that's 6...\nSpeaker E: 93.64, right?\nSpeaker E: Is the baseline.\nSpeaker F: Oh, the baseline.\nSpeaker F: Yeah.\nSpeaker F: But I'm not talking about the baseline.\nSpeaker E: Oh, oh, I'm sorry.\nSpeaker F: My baseline is the submitted system.\nSpeaker E: Oh, okay.\nSpeaker F: For finish, we start 93.84, and we go to 93.74.\nSpeaker F: And for Spanish, we were at 95.05, and we go to 93.61.\nSpeaker H: Okay, so we are getting hurt somewhat.\nSpeaker H: And is that what, you know what piece?\nSpeaker H: You've done several changes here.\nSpeaker H: Do you know what piece?\nSpeaker F: I guess it's the filter.\nSpeaker F: Because, well, we don't have complete result, but the filter...\nSpeaker F: So the filter with a shorter delay, earns on the Italian and the light.\nSpeaker F: And the other things like...\nSpeaker F: Don't sampling something from simple words.\nSpeaker F: I'm...\nSpeaker F: You're like normalization either.\nSpeaker E: So...\nSpeaker E: I'm really confused about something.\nSpeaker E: If we saw that making a small change, like, you know, a tenth to the self-loop had a huge effect, can we really make any conclusions about...\nSpeaker E: Yeah, that's differences in the stuff.\nSpeaker E: I mean, especially when they're this small.\nSpeaker F: I think we can be completely fooled by this thing.\nNone: But...\nSpeaker F: Well, yeah...\nSpeaker F: There is first this thing, and then...\nSpeaker F: I computed the confidence level on the different test sets.\nSpeaker F: And for the well-matched, they are around 0.6 percent.\nSpeaker F: For the mismatched, they are around...\nSpeaker F: I see 1.5 percent.\nSpeaker F: For the well-m, HM, they are also around 1.5.\nSpeaker H: But, okay, so these degradations you were talking about, were on the well-matched case.\nSpeaker H: Yeah.\nSpeaker H: Does the new filter make things better or worse for the other cases?\nSpeaker F: About the same, it doesn't hurt.\nSpeaker H: Doesn't hurt, but it doesn't get a little better or something.\nSpeaker H: No.\nSpeaker H: Okay, so...\nSpeaker H: I guess the argument one might make is that, yeah, if you look at one of these cases, and you jiggle something and it changes, then, you know, I'm quite sure what to make of it.\nSpeaker H: You look across a bunch of these, and there's some pattern.\nSpeaker H: I mean, so, here's all the...\nSpeaker H: If in all these different cases, it never gets better, and there's significant number of cases where it gets worse, then you're probably...\nSpeaker H: For any things.\nSpeaker H: That would say.\nSpeaker H: So, I mean, at the very least, that would be a reasonable prediction of what would happen with a different test set that you're not jiggling things with.\nSpeaker H: So, I guess the question is, if you can do better than this.\nSpeaker H: If we can approximate the old numbers while still keeping the latency down.\nSpeaker H: So...\nSpeaker H: So, what I was asking though is, what's the level of communication with the OGI gang now about this?\nSpeaker F: Yeah. When we are exchanging data, so, as we see, we have significant results.\nSpeaker F: For the moment, they are working on integrating the spectrosuppraction from Eric's.\nSpeaker F: Yeah. And so, yeah. We are working northside on other things, also trying to set spectrosuppraction, but...\nSpeaker F:...the spectrosuppraction.\nSpeaker F: Yeah, so I think it seems okay.\nSpeaker H: It's depending further discussion about this idea of having some sort of source code control.\nSpeaker F: Yeah. Well, for the moment, everybody is quite...\nSpeaker F: There is this Eurospeg deadline.\nSpeaker F: I see.\nSpeaker F: Yeah. But, yeah. As soon as we have something that's significant, that's better than what was submitted, we will fix this.\nSpeaker E: Okay, so, what is your answer to this?\nSpeaker H: Do you want me to take this on?\nSpeaker H: Yeah. Sounds like a good idea, but I think that that keeps saying people are scrambling for your Eurospeg deadline.\nSpeaker E: But that will be done in a week, so maybe after.\nSpeaker E: Wow, already a week, man.\nSpeaker H: You're right. It's amazing.\nSpeaker H: I think anything for your speech or...\nSpeaker F: We are trying to do something with the meeting recorded digits.\nSpeaker F: And the good thing is that there is this first deadline.\nSpeaker F: Yeah.\nSpeaker F: Well, some people from OGI are working on paper for this, but there is also the special session about Aurora, which is an extended deadline.\nSpeaker H: Oh, for your speech?\nSpeaker F: Yeah. So far.\nSpeaker H: Oh, special dispensation. That's great.\nSpeaker E: Where is your Eurospeg this year?\nSpeaker H: Alborg. Alborg.\nSpeaker H: So the deadline, once a deadline.\nSpeaker F: Once a deadline?\nSpeaker H: That's great.\nSpeaker H: So we should definitely get something after that.\nSpeaker H: But on meeting digits, maybe.\nSpeaker F: Yeah.\nSpeaker H: So I think that you could certainly start looking at the issue, but I think it's probably on...\nSpeaker H: It's from what Stefan is saying. It's unlikely to get sort of active participation\nSpeaker E: from the two sides until after they've... Well, I could at least...\nSpeaker E: Well, I'm going to be out next week, but I could try to look into like this CVS over the web. That seems to be a very popular way of people distributing changes and over, you know, multiple sites and things.\nSpeaker E: So maybe if I can figure out how to do that easily and then pass the information on to everybody that's easy to do is possible and people won't interfere with their regular work, then maybe that would be good.\nSpeaker E: And I think we could use it for other things around here too.\nSpeaker G: Good. That's cool.\nSpeaker D: And if you're interested in using CVS, I've set it up here.\nSpeaker D: Oh, great. Okay.\nSpeaker E: I used it a long time ago, but it's been a while, so maybe I can ask you some questions.\nSpeaker D: So I'll be away tomorrow and Monday, but I'll be back on Tuesday or Wednesday.\nNone: Okay.\nSpeaker H: You do the other thing, actually, is business about this waveform.\nSpeaker H: Maybe you and I can talk a little bit at some point about coming up with a better demonstration of the effects of reverberation for our web page.\nSpeaker H: So the...\nSpeaker H: Actually, the...\nSpeaker H: It made a good audio demonstration because when you play that clip, the really obvious difference is that you can hear two voices in the second one.\nSpeaker E: You can just...\nSpeaker G: Like, talk into a pet.\nSpeaker G: Yeah.\nSpeaker E: It's not good reverb.\nSpeaker H: No, I mean, it sounds pretty reverber, but I mean, when you play back in a room with a big room, nobody can hear that difference, really.\nSpeaker H: They hear that it's lower amplitude and they hear this second voice.\nSpeaker H: But that...\nSpeaker H: Actually, that makes for a perfectly good demo because that's real obvious thing.\nSpeaker H: Not a good two voices.\nSpeaker H: Well, that's okay, but for the visual, just, you know, like to have the spectrogram again because your visual abilities as a human being are so good, you can pick out...\nSpeaker H: You look at the good one, you look at the screwed up one, and you can see the features in it.\nSpeaker E: I noticed that in the pictures I thought, my initial thought was, this is not too bad.\nSpeaker E: Right.\nSpeaker H: If you look at it closely, you see, well, here's a place where this one has a big format.\nSpeaker H: The major formats here are moving quite a bit, and then you look in the other one, and they look practically flat.\nSpeaker H: So, I mean, that's why I was thinking in a section like that, you could take a look at just that part of the spectrogram, and you could say, oh, yeah, this really distorted it quite a bit.\nSpeaker E: The main thing that struck me in looking at those two spectrograms was the difference in the high frequencies.\nSpeaker E: It looked like, for the one that was far the way, you know, everything was tenuated.\nSpeaker E: Right.\nSpeaker E: I mean, that was the main visual thing that I noticed.\nSpeaker H: Right.\nSpeaker H: But it's, so, yeah, so there are clearly spectral effects.\nSpeaker H: Since you're getting all this indirect energy, then a lot of it does have reduced high frequencies.\nSpeaker H: But the other thing is the temporal courses, the things really are changed, and we want to show that in some obvious way.\nSpeaker H: The reason I put the waveforms in there was because they do look quite different.\nSpeaker H: And so I thought, oh, this is good, but I just, after you put in there, I didn't really look at them anymore.\nSpeaker H: So I just, they're different.\nSpeaker H: So, what's something that has a more interesting explanation for why they're different?\nSpeaker D: So maybe we can just substitute one of these waveforms, and then do some kind of zoom in on this spectrogram on an interesting area.\nSpeaker H: Something like that.\nSpeaker H: Yeah. The other thing that we had in there that I didn't like was that the most obvious characteristic of the difference when you listen to it is that there's a second voice.\nSpeaker H: And the cuts that we have there actually don't correspond to the full waveform.\nSpeaker H: It's just the first, I think there was something where he was having some trouble getting so much in, or I forget the reason behind it, but it's the first six seconds or something of it, and it's in the seventh or eighth second or something where the second voice comes in.\nSpeaker H: So we would like to actually see the voice coming in too, I think.\nSpeaker H: Since that's the most obvious thing when you listen to it.\nSpeaker G: So.\nSpeaker F: Yeah.\nSpeaker F: Yeah.\nSpeaker F: So.\nSpeaker F: Figures here. Well, we started to work on spectrograms of traction.\nSpeaker F: And the preliminary results were very bad.\nSpeaker F: So the thing that we did is just to add spectrograms of traction before this, the whole process which contains a monomine vision.\nSpeaker F: And it hurts a lot.\nSpeaker F: And so we started to look at things like this, which is, well, it's.\nSpeaker F: So you have the C0 bar meters for one Italian utterance, and I plotted this for two channels.\nSpeaker F: Channel zero is the closed-wave microphone, which is like just a microphone.\nSpeaker F: And it's perfectly synchronized.\nSpeaker F: And the sentence contains only one word, which is doing.\nSpeaker E: This is, this is a lot of C0, the energy.\nSpeaker F: There is a lot of C0 when we don't use spectrograms of traction.\nSpeaker F: And when there is no online normalization.\nSpeaker F: So there is just some filtering with the LDA.\nSpeaker F: So that's an example.\nSpeaker E: C0 is the closed-talking, the closed channel.\nSpeaker F: And channel one is the.\nSpeaker F: So C0 is very key.\nSpeaker F: Yeah.\nSpeaker F: Then when we apply normalization, it looks like the second figure.\nSpeaker F: Which is good, well, the noise part is around zero.\nSpeaker F: The third figure is what happens when we apply mineralization and variance.\nSpeaker F: So what we can clearly see is that on the speech portion, the two channels become very close.\nSpeaker F: But also what happens on the noisy portion is that the variance of the analysis.\nSpeaker E: This is still being a plot of C0.\nSpeaker E: Can I ask, what does variance normalization do?\nSpeaker E: What is the effect of that?\nSpeaker H: Normalization is the variance.\nSpeaker F: What does that mean?\nSpeaker F: No, I understand that.\nSpeaker E: No, I understand what it is.\nSpeaker E: What is it?\nSpeaker E: What's the rationale?\nSpeaker E: Yeah.\nSpeaker H: Why do it?\nSpeaker H: Well, I mean, because everything, if you have a system based on Gaussian, everything is based on means and variances.\nSpeaker H: So if there's an overall reason, you know, it's like if you're doing image processing.\nSpeaker H: And some of the pictures you were looking at, there was a lot of light.\nSpeaker H: And some there was low light.\nSpeaker H: You'd want to adjust for that in order to compare things.\nSpeaker H: And the variance is just sort of like the next moment.\nSpeaker H: So what if one set of pictures was taken so that throughout the course, it was one through daylight and night.\nSpeaker H: Ten times, another time.\nSpeaker H: I mean, it's, you know, how much, how much very, or, no, I guess a better example would be, how much of the light was coming in from outside rather than artificial light.\nSpeaker H: So if it was a lot, if more was coming in from outside, then it would be the bigger effect of the change.\nSpeaker H: So every mean, every, all of the parameters that you have, especially the variances, are going to be affected by the overall variance.\nSpeaker H: And so it's a bit, okay.\nSpeaker H: If you remove that source, then, you know, you can...\nSpeaker E: So the major effect is that you're going to get is by normalizing the means.\nSpeaker E: But it may first order, but first order, thank you.\nSpeaker H: But then the second order is the variances.\nSpeaker H: Because again, if you, if you're trying to distinguish between E and B, if it just so happens that the E's were more, you know, recorded when the energy was large.\nSpeaker H: It was, was, was larger or something, or the variation in it was larger.\nSpeaker H: Uh-huh.\nSpeaker H: Then the B's, then this will give you some, some bias.\nSpeaker H: So it's removing these sources of variability in the data that have nothing to do with the linguistic component.\nSpeaker H: Gotcha.\nSpeaker H: Okay. Sorry, dinner.\nSpeaker H: But the, let me, I just ask you something.\nSpeaker H: Is, if, if you have a good voice activity, a checker, isn't, isn't it going to pull that out?\nSpeaker F: Yes. Sure.\nSpeaker F: Sure. If you have a good, yeah. Well, what it shows is that, yeah, perhaps a good voice activity detector is, is good before magnumization.\nSpeaker F: And that's what we already observed.\nSpeaker F: But, yeah, voice activity detection is not...\nSpeaker E: But after you do this, after you do the variance normalization, I mean, I don't know, it seems like this would be a lot easier than this signal.\nSpeaker F: So, to work with... Well, I prefer to look at the second bigger than the third one, because you clearly see where a speech is.\nSpeaker F: Yeah. Yeah.\nSpeaker F: But the problem is that on the speech portion, channel zero and channel one are more different than when you use variance on the evaluation, where channel zero and channel one become closer.\nSpeaker E: But for the purposes of finding this speech, you're more interested in the difference between the speech and the non-speech, right?\nSpeaker F: So, I think, yeah, for... I think that perhaps it shows that the parameters that the voice activity detector should use, after use should be different than the parameters that have to be used for speech recognition.\nSpeaker H: Yeah. So, basically, you want to reduce this effect. So, you can do that by doing the voice activity detection. You also could do it by special spectral subtraction before the variance normalization, right?\nSpeaker F: Yeah, but it's not clear, man.\nSpeaker F: So... Well, it's just to... Yeah.\nSpeaker F: The number that are here are recognition experiments on the Italian, HM and M, with these two kinds of parameters.\nSpeaker F: But it's better with variance normalization.\nSpeaker H: Yeah. Yeah, so, that's good, better even out of luck, sogley.\nSpeaker H: But does this have the voice activity detection in it? Yeah.\nSpeaker F: Okay. So, the fact is that the voice activity detector doesn't work on the channel one, so... Yeah, channel one.\nSpeaker E: Yeah.\nSpeaker E: What stage is the voice activity detector applied? Is it applied here or after the variance normalization?\nSpeaker F: It's applied before variance normalization, so it's a good thing. Because I guess we're setting the detection on this. Yeah. Is it applied all the way back here?\nSpeaker F: It's applied... Yeah, something like this.\nSpeaker E: Maybe that's why it doesn't work for channel one.\nSpeaker F: It could perhaps do just mean normalization before... Nice guy.\nSpeaker H: Sort of a couple of other questions, which is if most of what the OGI folk are working with is trying to integrate this other spectrum subtraction, where are we working about it?\nSpeaker F: Speckless subtraction. Yeah. It's just... Well, it's another... They are trying to use the...\nSpeaker F: Ericsson, and we're trying to use something, nothing else. Yeah, and also to understand what happened, because...\nSpeaker F: Okay.\nSpeaker F: Well, when we do spectrum subtraction, actually, I think this is the two last figures.\nSpeaker F: It seems that after spectrum subtraction, speech is more emerging than...\nSpeaker G: Mm-hmm. Speech is more what?\nSpeaker F: Well, the difference between the energy of the speech and the energy of the spectrum subtracted no expression is larger.\nSpeaker F: Well, if you compare the first figure with this one, I'm trying to just get it's not the same, but if you look at the numbers, you clearly see that the difference between the zero of the speech and the zero of the nice portion is larger.\nSpeaker F: But when it advances that after spectrum subtraction, you also increase the variance of the zero.\nSpeaker F: So if you apply variance or my addition of this, it completely is...\nSpeaker F: Ericsson, everything. Yeah. So yeah.\nSpeaker F: And what they did at RGI is just...\nSpeaker F: They don't use online on my addition for the moment of spectrum subtraction, I think.\nSpeaker F: Yeah. I think as soon as they were trying to do an online addition, it would be a problem.\nSpeaker F: So yeah, we're working on the same thing, but I think...\nSpeaker F: With different systems.\nSpeaker H: Right. I mean, the...\nSpeaker H: In election, it's interesting to work on things one way or the other, but I'm just wondering if...\nSpeaker H: The list of things that there are to do, if there are things that we won't do because we got two groups doing the same thing.\nSpeaker H: That's...\nSpeaker H: Just asking.\nSpeaker E: There also could be... I mean, I can maybe see a reason for both working on it too, if...\nSpeaker E: You know, if you work on something else and you're waiting for them to give you spectral subtraction, I mean, it's hard to know whether the effects that you get from the other experiments you do will carry over once you've then bring in their spectral subtraction module.\nSpeaker E: So it's almost like everything's held up waiting for this one thing.\nSpeaker E: I don't know if that's true or not, but I could see how. Maybe that's what you were thinking.\nSpeaker H: I mean, we still evidently have a latency reduction plan, which isn't quite what you'd like it to be.\nSpeaker H: That seems like one prominent thing.\nSpeaker H: And then what there are issues of having a second stream or something, that was...\nSpeaker H: There was this business that we could use up at 4,800 bits.\nSpeaker F: Yeah, but I think we want to work on this day, also want to work on this.\nSpeaker F: Yeah, we will try MSG.\nSpeaker F: But...\nSpeaker F: I think they want to work on this second stream or something.\nSpeaker F: Some kind of...\nSpeaker F: Not get that one.\nSpeaker F: And they call it crap.\nSpeaker G: You know, it's crap.\nSpeaker H: Okay.\nSpeaker H: Do you remember when the next meeting is supposed to be?\nSpeaker H: In June.\nSpeaker H: Yeah.\nSpeaker H: Yeah, the other thing is that you saw that mail about the VAD.\nSpeaker H: VAD is performing quite differently.\nSpeaker H: So there was this experiment of what if you just take the bass line, a feature, just Melcapster, and you incorporate the different VADs.\nSpeaker H: And it looks like the French VAD is actually better.\nSpeaker H: It improves the bass line.\nSpeaker E: Yeah.\nSpeaker F: I don't know which VAD they use.\nSpeaker F: If they use the small VAD, I think it's easy to do better because it doesn't work at all.\nSpeaker F: So...\nSpeaker F: Which one is pretty better than the bass as well?\nSpeaker A: Yeah.\nSpeaker F: We should ask which VAD.\nSpeaker A: I think that he said with the good VAD of Prongoji with the Arcade VAD.\nSpeaker A: And the experiment was sometimes better, sometimes worse.\nSpeaker F: Yeah, but I think you're talking about the other way that you use VAD on the reference features.\nSpeaker F: Yes.\nSpeaker H: And on that one, the French one was better.\nSpeaker H: I mean, it was enough better that it would account for a fair amount of the difference between a performance, actually.\nSpeaker H: So if they have a better one, we should use it.\nSpeaker H: You know, it's...\nSpeaker H: You can't work on everything.\nSpeaker F: Yeah, so we should find out if it's really better, if it's...\nSpeaker F: Yeah.\nSpeaker F: Compared to the small, on a big band.\nSpeaker F: Yeah.\nSpeaker F: And perhaps we can easily prove if we put, like, the normalization before...\nSpeaker G: Yeah.\nSpeaker H: He can't go back in town the week after next, back in the country.\nSpeaker H: So start organizing more visits and connections and so forth.\nSpeaker H: Working towards students.\nSpeaker A: Also, Stefan was thinking that maybe it was useful to sing about voice and voice to work here in voice and voice detection.\nSpeaker A: And we are looking at the singing.\nSpeaker F: Actually, when we look at all the proposals, everybody is still using some kind of spectrum hand look.\nNone: Right?\nSpeaker H: No use of pitch, basically.\nSpeaker F: Yeah, well, the pitch, but to look at the...\nSpeaker F: Fine.\nSpeaker F: Well, that's just everyone to find the pitch and the sound.\nSpeaker F: It's another feeling that when we look at the...\nSpeaker F: Yeah, well, there is no way you can tell if it's for it and then for it.\nSpeaker F: If there is some...\nSpeaker F: It's easy in clean speech because voice sound, I'm all over it, I'll see.\nSpeaker F: So there will be more.\nSpeaker F: Yeah.\nSpeaker F: There is the first problem, just the larger.\nSpeaker F: And then voice sound or more eye frequencies because it's vacation.\nSpeaker F: Right.\nSpeaker F: But, yeah.\nSpeaker F: When you have noise, you have a low frequency, you know, you could be even for voice speech.\nSpeaker H: Yeah, you can make these mistakes, but...\nSpeaker E: Isn't there some other...\nSpeaker E: I think it would be good.\nSpeaker E: I'm not going to say, isn't there... aren't there lots of ideas for doing voice activity or speech non-speech rather by looking at, you know, I guess, harmonics or looking across time?\nSpeaker H: I think you sound about the voice non-voiced though.\nSpeaker E: Yeah.\nSpeaker E: Or that, you know, even with voice non-voiced and voiced, I thought that you were...\nSpeaker H: I'm really was talking about...\nSpeaker H: Well, yeah, we should have finished with it.\nSpeaker F: Okay.\nSpeaker F: It's good.\nSpeaker F: Yeah, so yeah, I think if we try to develop a second stream...\nSpeaker F: Well, there would be one stream that is the envelope and the second, it would be interesting to have something that's more related to the fact structure of the spectrum.\nSpeaker F: Yeah, so I don't know. We were thinking about like using ideas from Larry Sohn.\nSpeaker F: Have a good voice detector or...\nSpeaker F: Have a good voice speech detector.\nSpeaker F: Let's work in...\nSpeaker F: Larry Sohn could be an idea.\nSpeaker F: We were thinking about just kind of taking this background and computing the variance.\nSpeaker F: I have a resolution spectrum.\nSpeaker H: Okay, so today's something about that.\nSpeaker H: We had a guy here some years ago who did some work on making use of voicing information to help in reducing the noise.\nSpeaker H: So what he was doing is basically you do estimate the pitch.\nSpeaker H: Or you estimate fine harmonic structure, which are either way, it's more of the same.\nSpeaker H: The thing is that you then can get rid of things that are not...\nSpeaker H: If there is strong harmonic structure, you can throw away stuff that's not a harmonic.\nSpeaker H: And that is another way of getting rid of part of the noise.\nSpeaker H: So that's something that is sort of finer, brings in a little more information than just spectrum of subtraction.\nSpeaker H: And he did that sort of in combination with Rostovs, kind of like Rostovs, taking care of convolutional stuff.\nSpeaker H: And that's some decent results doing that.\nSpeaker H: But yeah, there's all these cues.\nSpeaker H: Actually, back when Chuck was here, we did some voice-to-voice classification using a bunch of these.\nSpeaker H: And works okay, obviously it's not perfect.\nSpeaker H: But the thing is that you can't, given the constraints of this task, we can't, in a very nice way, feed forward to the recognizer, the information, the probabilistic information that you might get about, whether it's voice-to-own voice, we can't affect the distributions or anything.\nSpeaker H: But what we, I guess we could.\nSpeaker E: Didn't the head dude send around that message?\nSpeaker E: I think you sent us all a copy of the message where he was saying that I'm not sure exactly what he was saying, but something having to do with the voice activity detector and that people shouldn't put their own in or something was going to be...\nSpeaker H: Okay, so that's voice activity detector as opposed to voicing detector.\nSpeaker H: So we're talking about something a little different.\nSpeaker H: I'm sorry.\nSpeaker H: I'm sorry.\nSpeaker H: I guess what you could do, maybe this would be.\nSpeaker H: So if you have, if you view this second stream, before you do KLTs and so forth, if you do view it as probabilities.\nSpeaker H: And if it's an independent, so if it's not so much envelope based, but fine structure based, looking at harm and acidity or something like that.\nSpeaker H: If you get a probability from that information and then multiply it by, you know, multiply it by all the voiced outputs and all the unvoiced outputs, you know.\nSpeaker H: Then use that as the take the log of that or pre-nonlinearity.\nSpeaker H: And do the KLT on that.\nSpeaker H: Then that would, I guess, be a reasonable use of independent information.\nSpeaker H: So maybe that's what you meant.\nSpeaker H: That would be...\nSpeaker F: Yeah, well, that would be...\nSpeaker F: Yeah.\nSpeaker F: So you'll mean that some kind of probability from the voice thing and that...\nSpeaker H: Right, so you have a second neural net, it could be pretty small.\nSpeaker H: Yeah, if you have a tandem system, you have some kind of, it could be pretty small, net.\nSpeaker H: We did some of this stuff.\nSpeaker H: I did some use go and you use...\nSpeaker H: The thing is to use information primarily that's different, as you say, it's more fine structure based than envelope based.\nSpeaker H: So then you can pretty much guarantee it's stuff that you're not looking at very well with the other one.\nSpeaker H: And then you only use it for this one distinction.\nSpeaker H: And so now you've got a probability of the two cases and you've got the probability of the finer categories on the other side, you multiply them more appropriate.\nSpeaker H: And if they really are from independent information sources, then they should have different kinds of errors and roughly independent errors.\nSpeaker H: That's a good choice for...\nSpeaker H: Yeah, that's a good idea.\nSpeaker F: Because yeah, well, spectrosuffraction is good and we could use the finer structure to let the finer scheme of the null.\nSpeaker F: But still, the very...\nSpeaker F: This issue with spectrosuffraction that seems to increase the variance of...\nSpeaker F: Yeah.\nSpeaker F: Well, this is a couple of...\nSpeaker F: Right.\nSpeaker F: I know in a few of you do some kind of a leg-on-eye addition at the end of the...\nSpeaker F: Well, spectrosuffraction has a leg-on-eye addition to...\nSpeaker H: What if you do the spectrosuffraction, do some spectrosuffraction first, then do some online normalization, then do some more spectrosuffraction?\nSpeaker H: I mean, maybe you can do it in layers or something, so it doesn't hurt too much or something.\nSpeaker H: But anyway, I think I was sort of arguing against myself there by giving that example.\nSpeaker H: I mean, because I was already sort of suggesting that we should be careful about not spending too much time on exactly what they're doing.\nSpeaker H: In fact, if you go into a harmonics-related thing, it's definitely going to be different than what they're doing and should have some interesting properties and noise.\nSpeaker H: I know that when people have done sort of the obvious thing of taking your feature vector and adding in some variables which are pitch-related or that it hasn't...\nSpeaker H: My impression is it hasn't particularly helped. It has not.\nSpeaker H: Yeah, but I think that's a question for this...\nSpeaker H: Extending the feature vector versus having different streams.\nSpeaker H: Was it nice and nice and conditioned for the example?\nSpeaker H: And it may not have been noisy conditions.\nSpeaker H: I don't remember the example, but it was on some DARPA data some years ago, so it probably wasn't, actually.\nSpeaker F: But we are thinking we're discussing very...\nSpeaker F: Perhaps...\nSpeaker F: I was thinking about some kind of cheating experiment. It's voicing bit.\nSpeaker H: Why don't you just do it with a rure?\nSpeaker H: Just in each frame...\nSpeaker F: We don't have a free-telling brush. We don't have a labeling...\nSpeaker F: We just have a labeling word model.\nSpeaker C: You have frame, frame, level.\nSpeaker H: But you can align so that it's not perfect, but if you know what was said...\nSpeaker E: The problem is that their models are all word-level models. There's no phone models that you get alignable.\nSpeaker E: You see, you can find out where the word boundaries are, but that's about it.\nSpeaker H: Yeah. I see.\nSpeaker C: But we could use the noisy version of Timit, which is similar to the noises found in the TI digits.\nSpeaker C: The portion of our aura.\nSpeaker F: I guess we can see nothing to be out.\nSpeaker F: If this voicing bit doesn't help...\nSpeaker F: I think we don't have to...\nSpeaker H: We want more about this.\nSpeaker H: Right.\nSpeaker H: In experiments, we did a long time ago, and it was probably a resource management or something.\nSpeaker H: I think you were getting something like still 8% or 9% error on the voicing as I recall.\nSpeaker H: So what that said is that sort of love to its own devices, like without a strong language model and so forth, that you would make significant number of errors just with your probabilistic machinery.\nSpeaker E: It also... I think there was one problem with that in that we used canonical mapping. So our truth may not have really been true to the acoustics.\nSpeaker H: Yeah. Well, back 20 years ago when I did this voiced-on-voiced stuff, we were getting more like 97% or 98% correct in voicing, but that was speaker-dependent.\nSpeaker H: Actually, we were doing training on a particular announcer and getting a very good handle on the features.\nSpeaker H: We did this complex feature selection thing where we looked at all the different possible features one could have for voicing and exhaustively searched all-size subsets.\nSpeaker H: For that particular speaker, you'd find the five or six features which really did well.\nSpeaker H: Doing all of that, we get down to 23% error, but that, again, the speaker-dependent with lots of feature selection and very complex sort of thing.\nSpeaker H: So I would believe that it was quite likely that looking at envelope-only that would be significantly worse than that.\nSpeaker F: And the speech-cordes?\nSpeaker F: Yeah, I do. What do they even have to detect first? The modern ones don't do a simple switch. They work on the code book, excitation.\nSpeaker H: Yeah, they do analysis by synthesis. They try every possible excitation they have in their code book and find the one that matches best.\nSpeaker F: Yeah.\nSpeaker E: Okay. Can I just mention one other interesting thing? One of the ideas that we had come up with last week for things to try to improve the system.\nSpeaker E: I guess I wrote this in after the meeting, but the thought I had was looking at the language model that's used in the HTK recognizer, which is basically just a big loop.\nSpeaker E: So it goes digit, and then that can either go to silence or go to another digit, which that model will allow for the production of infinitely long sequences of digits.\nSpeaker E: So I thought, well, I'm going to just look at the what actual digit strings do occur in the training data. And the interesting thing was it turns out that there are no sequences of two long or three long digit strings in any of their training data. So it's either one, four, five, six, up to 11, and then it skips, and then there's some at 16.\nSpeaker H: But what about the testing data?\nSpeaker H: I don't know. I didn't look at the test data yet. So if there's some testing data that has, has two or three.\nSpeaker E: Yeah, but I just thought that was a little odd that there were no two or three long.\nSpeaker E: So I just, for the heck of it, I made a little grammar, which had its separate path for each length digit string you could get.\nSpeaker E: So there was a one long path, and there was a four long and a five long. And I tried that and it got way worse. There were lots of deletions. So it was, you know, I didn't have any weights on these paths, or I didn't have anything like that.\nSpeaker E: And I played with tweaking the word transition penalties a bunch, but I couldn't go anywhere. But I thought, well, if I only allow, I guess I should have looked at to see how often there was a mistake where a two long or three long path was actually put out as a hypothesis.\nSpeaker E: So to do that right, you'd probably want to have a lot for them all, but then have weightings and things. So I just thought that was an interesting thing about the data.\nSpeaker H: Okay, so we're going to read some more just strings, I guess.\nSpeaker E: Yeah.\nSpeaker H: I'm going to go ahead and read it.\nSpeaker H: 5, 5, 4, 0, 1, 8, 8, 7, 2, 2, 4, 6, 3, 8, 5, 3, 9, 5, 6, 1, 6, 1, 4, 0, 2, 9, 4.\nSpeaker H: 737-339.\nSpeaker H: 3126-114850.\nSpeaker H: 685-3-741-3923.\nSpeaker H: 917-839-7546.\nSpeaker F: Transcript L-19.\nSpeaker F: 874-30-928.\nSpeaker F: 859-43471.\nSpeaker F: 020-975-009.\nSpeaker F: 677-601-5254.\nSpeaker F: 423-575-426.\nSpeaker F: 903-4-6231.\nSpeaker F: 380-08007-8351.\nSpeaker F: 471-929-180.\nSpeaker E: Transcript L-16.\nSpeaker E: 5608-425566.\nSpeaker E: 735-475-477.\nSpeaker E: 037-715-505.\nSpeaker E: 3255-8169-34.\nSpeaker E: 393-057-019.\nSpeaker E: 588-625-7698.\nSpeaker E: 850-33434.\nSpeaker E: 230-46550.\nSpeaker D: Transcript L-17.\nSpeaker D: 946-470139.\nSpeaker D: 671-268-209.\nSpeaker D: 7500-462280.\nSpeaker D: 527-1-7133-5202.\nSpeaker D: 616-1120-8959.\nSpeaker D: 031-522-71.\nSpeaker D: 388-420-8457.\nSpeaker D: 680-4835-00.\nSpeaker A: Transcript L-18.\nSpeaker A: 585-771-443.\nSpeaker A: 9241-91-301-1.\nSpeaker A: 784-528-3698.\nSpeaker A: 739-724-523-96.\nSpeaker A: 572-921-08.\nSpeaker A: 698-2.\nSpeaker A: 288-541.\nSpeaker A: 794-9.\nSpeaker A: 792-420-563.\nSpeaker A: 567-978-558-1.\nSpeaker B: Transcript L-21.\nSpeaker B: 010-453366.\nSpeaker B: 125-545-434.\nSpeaker B: 410-6960-7230.\nSpeaker B: 793-1-4150.\nSpeaker B: 417-083-532-4.\nSpeaker B: 165-687-594.\nSpeaker B: 377-5030-568.\nSpeaker B: 747-999-119.\n"}, {"meeting_id": "ES2002b", "summary": "This meeting was about the functional design of the remote control. Firstly, User Interface presented on the technical function design. For the previous remote, User Interface found it was clunky but the functions were basic, or the remote was too complicated with lots of buttons. Based on this, User Interface demonstrated two kinds of buttons access to advanced functions. Next, Marketing presented on market trends and proposed three inspirations including the fancy look and technology innovation based on the customer needs. Then, Industrial Designer presented on the working design and introduced the working design logic of the battery and power source. Lastly, the group discussed the remote functions. They decided to have fifteen to thirty-five-year-old target groups as well as to have video controls map to advanced functionality so that they could skip the basic channel button.", "dialogue": "Speaker A: Oh yes, I forgot my test.\nNone: Okay.\nSpeaker A: Is that all right now?\nSpeaker A: Okay.\nNone: Let's go.\nSpeaker D: Someone turn these on.\nSpeaker D: Sorry?\nSpeaker D: Someone turn these on.\nNone: Great.\nSpeaker A: Okay.\nSpeaker A: I'll set to start the meeting.\nSpeaker A: Okay.\nSpeaker A: We've got half an hour for this one.\nSpeaker A: I'll discuss the functional design.\nSpeaker A: Okay.\nSpeaker A: Thanks.\nSpeaker A: All ready to go? Okay.\nSpeaker A: Okay.\nSpeaker A: So hopefully, if I've been working away, and I've put the minutes of the last meeting in the project folder.\nSpeaker A: So I guess just to recap on what we did last time.\nSpeaker A: I got to nature a little bit and got familiar with all the equipment and started to discuss a bit about the project, and you've caused how much money we had to spend.\nSpeaker A: Just want to tell you that we have three new requirements, which is the first one, is that the companies decided that teletext is outdated because of how popular the internet is.\nSpeaker A: Nobody uses teletext very much anymore.\nSpeaker A: So we don't really need to consider that in the functionality of the remote control.\nSpeaker A: They've also suggested that we only use the remote control to control the television, not the VCR or DVD or anything else.\nSpeaker A: I think the worry is that if the project becomes too complex, it'll affect how long it takes us to get it into production, the time to market.\nSpeaker A: Okay.\nSpeaker A: So we're just going to keep it simple and it'll just control the TV.\nSpeaker A: The first thing was that the company wants the corporate color and slogan to be implemented in the new design.\nSpeaker A: I'm not entirely sure what the corporate color is.\nSpeaker A: It might be yellow because there seems to be a lot of yellow everywhere.\nSpeaker D: And the slogan, like the actual written slogan, or just to embody the idea of the slogan?\nSpeaker A: Well, that's the thing. I'm not sure.\nSpeaker A: Because on the company website, what does it say?\nSpeaker A: What's the thing in the fashion and electronics?\nSpeaker A: Yeah. I mean, is that something they want to actually written on?\nSpeaker A: Because it's quite long.\nSpeaker A: Or, yeah, just the idea that I'm not sure.\nSpeaker A: So that's something we can discuss as well.\nSpeaker A: So those are the three things.\nSpeaker A: Just not to worry about teletext, only control the TV and incorporate the color and slogan of the company.\nSpeaker A: So, is everybody okay with any of that?\nSpeaker A: I don't really need to recap at all.\nSpeaker A: But time for presentations, and who would like to go first?\nSpeaker C: I'm not going first.\nSpeaker E: Okay, cool.\nSpeaker C: All right, can I steal this from the back of your laptop?\nSpeaker A: Yeah, of course.\nSpeaker C: Okay.\nSpeaker C: Just like opaque, you can shoot \ufffd If I were to look online I had a look at the home page which has given us some inspiration from previous products.\nSpeaker C: I had a look at the previous products to see what they offer and I would like to ask you guys for your ideas about the design at the end of the meeting and unfortunately when I looked at all the talk I was thinking.\nSpeaker C: Okay, having a look at the existing products I found that it tends to come into two extremes.\nSpeaker C: There's either a very complicated one that's got lots of buttons, lots of colours, very confusing.\nSpeaker C: You don't know what you're doing.\nSpeaker C: In that case the label is going to be very bad.\nSpeaker C: There's an example I should eat the end.\nSpeaker C: I'll show you now.\nSpeaker C: I'll see you see here.\nSpeaker C: There's a button there and there.\nSpeaker C: This one's perg.\nSpeaker C: Sorry, that one's perg and that one's perg and it doesn't really tell you what it does.\nSpeaker C: I'm not sure if you had a look at the other controller now in that example.\nSpeaker C: It's a very simple one.\nSpeaker C: It's got only the basic functions but the same size as the hard to use one.\nSpeaker C: It's a very big and not very much used for actual buttons.\nSpeaker C: It's a very hard to access the advanced functions.\nSpeaker C: There's nothing for instance for a slow motion button.\nSpeaker C: My own preference is I prefer the clunky one.\nSpeaker C: It's very easy to use but unfortunately it does like the advanced functions which I quite like having on the controls.\nSpeaker C: I believe the advanced functions should maybe be hidden in the drawer.\nSpeaker C: I don't think I have any problem with it.\nSpeaker C: I've got to put your preferences unless you're a problem we'll go.\nSpeaker A: I would come to chat for five minutes or so I think.\nSpeaker A: I'd most discover that anyway.\nSpeaker D: A lot of what I've read and prepared for this meeting fits in really closely with what Craig has just gone over.\nSpeaker D: I could give you some of my personal preferences but I could also add some of this which is just about market research.\nSpeaker A: We'll stick to kind of your area.\nSpeaker C: I think it's supposed to be the same size but it's got much fewer buttons.\nSpeaker C: I think it's a valid point.\nSpeaker A: The one on the left looks quite complicated and that PROG thing is incredibly confusing.\nSpeaker A: I see why you might prefer the simpler design but you don't want to lose out on what it does so maybe you know you get a lot of remote controls where you kind of open up the thing and that's a good idea.\nSpeaker A: I think it's a good idea.\nSpeaker C: Do we have any functions we'd want on it?\nSpeaker C: So far I've got iron off so it's the channel up and down and the volume up and down.\nSpeaker C: They're just the very basic use by TV.\nSpeaker A: And then actual numbers for channels as well.\nSpeaker C: You see that's how the quad one or a question one which was that?\nSpeaker C: The numbers are the up down.\nSpeaker A: I would say that's required.\nSpeaker A: There's no way anybody's going to buy a remote control these days so she can't actually individually select channels.\nSpeaker A: I mean would anybody disagree with that?\nSpeaker A: What else?\nSpeaker A: So don't need to worry about teletext, don't need to worry about VCR.\nSpeaker A: Any kind of display controls at all do you think we need to worry about?\nSpeaker D: Is it brightness and contrast?\nSpeaker D: We're doing right now as we're categorizing we're saying we want this to be a product that offers all the more tricky features but we want them to be in another area.\nSpeaker D: Is that right?\nSpeaker D: Is that what we're kind of sorting with them?\nSpeaker D: Or are we actually eliminating things we just don't want product to have?\nSpeaker A: I think are you maybe kind of thinking what we absolutely have to have and what would\nSpeaker C: be nice? To start with, we need to find out exactly what we have to have.\nSpeaker C: Not that we can add things if they're possible.\nSpeaker A: Okay, right.\nSpeaker A: Well, do you want to maybe just at this point decide on what we absolutely must have as a function of this.\nSpeaker A: So so far just to recap we've got volume and channel control.\nSpeaker C: And volume and channel and skip to certain channels with the numbers.\nSpeaker A: Right, okay.\nSpeaker D: Well, one way I would look at this would be that we approach the different controls in terms of like control types so that for the user it's very clear what they want to do, where they go.\nSpeaker D: And I also think maybe a little bit about what we just want to be easily accessible.\nSpeaker D: For example, if we had audio controls those could be something people set up very rarely.\nSpeaker D: Maybe they're in a little area, but covered up.\nSpeaker D: Things like channel and volume are used all the time.\nSpeaker D: So we just have them right out on top.\nSpeaker D: Very just very self explanatory.\nSpeaker D: So maybe we need to think about having three or more groupings of controls.\nSpeaker D: You know, like one which are just the habitual ones that should be right within your natural grip and others that are also available and then others that are concealed.\nSpeaker A: Okay, and we're just to wrap up quickly on this little section.\nSpeaker A: Do you think maybe that's the only kind of essential requirements and there may be just things that would be nice if it could do would be things like audio setup and display setup and things like that.\nSpeaker A: Maybe like a mute button.\nSpeaker A: That's really saying any of the ins and outs of that at all.\nSpeaker A: No.\nSpeaker A: Okay.\nSpeaker A: Right.\nSpeaker A: Okay, if we can move on to next presentation then please.\nSpeaker A: Sure.\nSpeaker D: Can this reach, can this plug come across?\nSpeaker A: Probably not actually.\nSpeaker D: So why don't I just pick up and.\nSpeaker D: Yeah.\nSpeaker G:UK.\nSpeaker D: Can you go behind me?\nSpeaker D: I'm all in the audience.\nSpeaker D: It's very complicated.\nSpeaker A: I'm all in the audience.\nSpeaker A: It's very complicated.\nSpeaker A: It's very nice if everything was wireless in it.\nNone: Okay.\nSpeaker A: Right.\nSpeaker D: So I can say already, I don't know whether this is for good or for bad, but there will be a lot of kind of redundancy in the issues and the...\nSpeaker D: Oh, like overlap between what you said.\nSpeaker D: Oh, probably not necessarily bad thing, but what I've already started doing is I created a slide in my presentation here so that we kind of think, well, what's the cumulative effect of what we've taken from your ideas and mine, because certainly I have a hard time separating things completely.\nSpeaker D: It's hard to know what your land is in.\nSpeaker D: Obviously what you just told me impacts a lot on what, like market research that I've been...\nSpeaker D: So how do I get this up?\nSpeaker A: I put yeah, I put function in F8, yeah.\nSpeaker G: Okay.\nSpeaker G: All right.\nSpeaker G: So...\nNone: Actually...\nSpeaker F: I'm sharing the blue button.\nSpeaker B: Thanks for the control on the left.\nSpeaker B: Oh, and F8.\nSpeaker B: Okay.\nSpeaker G: Appreciate it together.\nNone: Mm-hmm.\nNone: Okay.\nNone: I think that's it.\nNone: Yeah.\nSpeaker D: Okay.\nSpeaker E: Okay.\nSpeaker C: Again, we need the one, the button might be this.\nSpeaker A: Okay.\nSpeaker E: Oh, here we go.\nSpeaker E: Okay.\nNone: Okay.\nSpeaker D: Okay.\nSpeaker D: Just before I bring this up, what I'll just say is what I've done is try to collect some information so that I can then relay this to you guys so that it now becomes a collective thing and then kind of lead us in the direction of deciding, okay, what are our options?\nSpeaker D: What should we decide?\nSpeaker D: And do you know what I mean?\nSpeaker E: Okay.\nSpeaker F: Okay.\nSpeaker F: Okay.\nSpeaker E: Right.\nSpeaker E: Can you...\nNone: Right.\nSpeaker A: Okay.\nNone: All right.\nNone: Okay.\nSpeaker A: All right.\nSpeaker A: Okay.\nSpeaker D: All right.\nSpeaker D: Okay.\nSpeaker D: So, um, that makes sense.\nSpeaker D: So what I basically got is I just looked into some information and sort of tried to think about how we could review it and how we could and what kind of decisions we could take away from it and then maybe by the end of just looking at some of these things we can think about what are our priorities because certainly there's lots of different information to go through.\nSpeaker D: So, um, I'm thinking here about primarily about customer needs that we start with the customer and what they want and what are issues with existing products.\nSpeaker D: I just think about trends and also about try and connect that as you see with the company vision which is about fashion and electronics.\nSpeaker D: And then as I say, we'd like to prioritize our design features from this and...\nNone: Okay.\nSpeaker D: So, this is what I've found here.\nSpeaker D: A lot of this is new to me.\nSpeaker D: So, we'll just read two together.\nSpeaker D: Users dislike the look and feel of current remote control.\nSpeaker D: So, if I find them ugly, most people find them ugly.\nSpeaker D: The vast majority would spend more money for it to look fancy as well.\nSpeaker D: We'll see later, the vast majority of people would spend more money for slightly more intuitive controls such as voice recognition.\nSpeaker D: Okay.\nSpeaker D: I'm going to...\nSpeaker D: We'll look at that in a second.\nSpeaker D: Most people use only a very slim portion of all the controls.\nSpeaker D: So, I guess what we're looking at here is people want this technology.\nSpeaker D: They tend to use the most simple controls and overall, they find remote controls to be something they don't...\nSpeaker D: Doesn't really appeal to them.\nSpeaker D: So, I think what we're doing is we're trying to take like...\nSpeaker D: For me, this is sort of like three different, different, um, inspirations.\nSpeaker D: One is that we want something that's high tech but we want it to seem easy.\nSpeaker D: And in spite of the primitive side of it and the very high tech side, we want it to just be an appealing piece of equipment in people's hands.\nSpeaker D: Frustrations, they get lost a lot as it came up in our last meeting.\nSpeaker D: Takes time to learn how to use them.\nSpeaker D: This is why I mentioned when Craig was showing us some ideas that we actually try and group control.\nSpeaker D: So, it doesn't just look like a big panel.\nSpeaker D: Kind of like when you look at, you know, a new computer keyboard or something that is quite explanatory if you want audio, if you want visual, and you have those.\nSpeaker D: And I will admit, I don't know what RSI stands for.\nSpeaker A: Headed it's train injury.\nSpeaker B: Is installing a remote control for many people?\nSpeaker D: No, that did not come up at all.\nSpeaker D: So here's another sort of a review here of the main things.\nSpeaker D: I also found most people would, adults at least, would pay more for voice recognition.\nSpeaker D: Apparently, we do have access to all the cutting edge technology and remote control.\nSpeaker D: So I don't know if that's possible.\nSpeaker D: We might consider getting into it.\nSpeaker D: And again, here as we sort of move, sort of things, start thinking about how we want to sell and market this, I think a recurring theme here is the company wants us to make something that's fashionable and sleek and trendy.\nSpeaker D: People additionally aren't liking the appearance of their product.\nSpeaker D: So we want to think about, as we take all the sort of the techie features, how we can put that into a unit which is, which people like, you know, they like the aesthetics and the ergonomics.\nSpeaker A: So once in the end, it looks good and it's easy to use.\nSpeaker A: Big priorities.\nSpeaker D: Yeah, so just looking at what Craig's ideas there, sort of tell me that maybe what we want to do is try and separate the different things that we want to include in this.\nSpeaker D: If we do say, well, we want there to be all the technology, but we try and make that almost be like optional technology.\nSpeaker D: No, it's like, I find a lot of TVs these days, something really nice about them is, if you want to just turn them on and off, you can.\nSpeaker D: But they have little panels where you click.\nSpeaker D: And there's just like tons of features.\nSpeaker A: So you want to grip all the different kind of types of functions together.\nSpeaker A: You know, that's the thing that I do.\nSpeaker D: That's sort of, but I hope here is that I'm putting out this information so we can then say, okay, well, how do we collectively move on with it?\nSpeaker D: I haven't brought out one specific marketing idea, although my sense is that what we should try and think about is what are the current trends in materials and shapes and styles and then use that, but not let that confine us technologically.\nSpeaker D: Okay.\nSpeaker A: Right.\nSpeaker D: All right.\nSpeaker D: Any comments on all of that?\nSpeaker A: Well, one of the things that we have to decide on by the end of the meeting is who we're going to be, who's our target audience or target market.\nSpeaker A: So if we want something that looks good and it's easy to use, but has, is fairly powerful product or whatever, who do we really want to aim that at?\nSpeaker A: Okay.\nSpeaker D: I mean, where's the money?\nSpeaker A: Yeah, who would have the money to spend?\nSpeaker A: Well, if like 25 euro is our selling price, and you can imagine, well, I'm not really sure how much that will retail at, but you want, it's somebody who's not going to just use the remote that comes with their tally.\nSpeaker A: I suppose they're going to actually go out and buy one.\nSpeaker A: So who do you think we're aiming this at?\nSpeaker B: I think it'll be the mid range to high end market in terms of people because 25 euros for a remote, how much is that?\nSpeaker B: It's about 16, 17 pounds, I think.\nSpeaker B: Is that a lot of money to buy an extra remote or a replacement remote?\nSpeaker B: Okay.\nSpeaker B: So maybe not the high end range, but maybe middle, middle, up-ish.\nSpeaker B: Okay.\nSpeaker B: So how much, I don't know, I guess you would be like 10, 10, 10, 10, 10, 10, 10, 10, 10, like a simple replacement.\nSpeaker B: Yeah.\nSpeaker B: I mean, if you lost your remote and the first thing you just want to go out and get-\nSpeaker D: Yeah. This kind of touches on your comments here, David.\nSpeaker D: These are the age groups which we have information on, and these are, this is a table of what people would pay more for a certain feature.\nSpeaker D: Okay.\nSpeaker D: Just give us a rough idea of where the will to spend money on TV equipment is.\nSpeaker D: Okay.\nSpeaker D: Mostly focused around the 25 age group.\nSpeaker A: Yeah.\nSpeaker A: So do you think we're aiming at a fairly young market then?\nSpeaker D: Yeah.\nSpeaker D: Sort of young professional.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Do you think then voice recognition is something we should really seriously consider?\nSpeaker A: What do you think, Craig?\nSpeaker C: Well, you don't see it was the adults that were going through the voice recognition.\nSpeaker F: Oh, yeah.\nSpeaker F: That's the- Yeah.\nSpeaker D: It does fit with the market that we're sort of identifying in terms of-\nSpeaker A: I think we are going to have to narrow it down to say, let's target these people and give them what they want.\nSpeaker A: Yeah.\nSpeaker A: Because you know, there needs to be some kind of selling point to it.\nSpeaker A: So anybody, anything to add, just kind of young professionals, like if we are going to include speech recognition, it's kind of between 15 and 35.\nSpeaker A: It seems to be like a really high response to that.\nSpeaker A: So we could say that was our target.\nSpeaker B: Yeah, so 5 to 35 is fair to add that in and that's more than half a group of people who are willing to at least try to use it.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So 15 to 35, look fairly young.\nSpeaker A: Yeah.\nSpeaker A: You know, they have a bit of expandable income to spend on this sort of thing.\nSpeaker B: I think perhaps that age group is significant as well, because those are people who use the computer who are familiar with the computers in the everyday world.\nSpeaker B: Yeah.\nSpeaker B: I think people who are ABBOP, I wouldn't say 35, but people are on a 48-ish and above now would not be so dependent and reliant on a computer or about phone or something like that.\nSpeaker B: So the people who are gadgety, right?\nSpeaker B: Yeah.\nSpeaker B: Growing up used to, you know, in schools and universities, when they go on to their working lives, they go forward.\nSpeaker A: Yeah.\nSpeaker A: So they're not shy away from something quite high-tech.\nSpeaker A: That's a good point.\nSpeaker A: Okay.\nSpeaker A: So shall we make the decision to include speech recognition?\nSpeaker A: If we can.\nSpeaker B: I think one thing we should try not to avoid is not to say, you have to use speech recognition right now.\nSpeaker B: Okay.\nSpeaker B: Why is that?\nSpeaker B: Based on what you, everybody's saying that you want something simple.\nSpeaker B: You want basic stuff and you want something that's easy.\nSpeaker B: You speech recognition might not be the simplest thing.\nSpeaker D: Right.\nSpeaker D: Okay.\nSpeaker D: Could it be an on-off thing?\nSpeaker D: Like if you want it.\nSpeaker B: We can act a bit and de-actress.\nSpeaker B: Yeah.\nSpeaker B: We're trying to lock ourselves into a particular kind of technology.\nSpeaker B: Yeah.\nSpeaker B: So we're focusing on exactly what are the features that we're going to say.\nSpeaker B: Okay.\nSpeaker B: Sure.\nSpeaker B: Say speech recognition is good for this.\nSpeaker B: Speech recognition is not good for this.\nSpeaker B: Sure.\nSpeaker B: Okay.\nSpeaker B: I suggest that we think about speech recognition.\nSpeaker B: Sure.\nSpeaker B: It's something that can be used to fulfill a function.\nSpeaker B: Uh-huh.\nSpeaker B: And then, you know, today we don't look at the technology but we'll get the function first.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Well, do you want to give us your presentation and then we can, I might have been a good idea to all deliver presentations and then discuss.\nSpeaker A: But this is a hybrid.\nSpeaker A: Good to get ideas out while there's a question in mind.\nSpeaker A: Yeah.\nSpeaker A: Exactly.\nSpeaker A: Yeah.\nSpeaker A: It's something that's just occurred to me as well as if we make it, um, speech recognition, if we incorporate speech recognition, that's appealing to people, um, maybe with, uh, physical disability as well.\nSpeaker A: Yeah.\nSpeaker D: And not losing.\nSpeaker D: And also it helps in terms of people not losing this, you know, they, they're saying, oh, it's, I lose it in the couch.\nSpeaker D: Yeah.\nSpeaker D: It's like we're kind of, we're, we're sort of getting into here is mating different, uh, design features together.\nSpeaker D: Okay.\nSpeaker C: One problem is speech recognition is, um, I've actually seen one of them used and, uh, the technology was, and that one wasn't, but I'm using it.\nSpeaker C: So you end up yelling control for hours, for example, up.\nSpeaker D: Oh, really, you've seen one before.\nSpeaker A: Do you think maybe we need like further advances in that kind of area until it's worthwhile incorporating it?\nSpeaker A: I think it'd probably be quite expensive to, to, to.\nSpeaker E: Of course not.\nSpeaker E: I think so.\nNone: Nice.\nSpeaker B: Well, this is just the working design.\nSpeaker B: Oh, this is just what, how I would go about it.\nSpeaker B: Um, I guess I tried to define like what we're doing now.\nSpeaker B: Try to define what we're trying to get done.\nSpeaker B: Um, I think in a practical way, we kind of know what it is.\nSpeaker B: We've used it with familiar with it, but we're, we can't, we have a narrow down, um, exactly what the things we're trying to fulfill.\nSpeaker B: But, um, I think in the back of our minds, we know what the basics are.\nSpeaker B: It's changed how it all does, it came falling, but in like specifics, which one of the basics are you trying to target?\nSpeaker B: Um, are there certain parts of the basics that are more important or less important than the basics?\nSpeaker B: Um, and I just, the idea is just to get everybody to, um, I usually have a, have a design as a, as a basic.\nSpeaker B: So, you know, things that, to start everything going, but I guess everybody does have some ideas.\nSpeaker B: I don't think, um, anything to that.\nSpeaker B: Um, okay, this finding things is a little bit confusing, so I'll go into the diagram.\nSpeaker B: It just explains how the process goes through from the, from the basic technology point of view, the basic steps that you need, um, in the diagram.\nSpeaker B: This slide probably works better.\nSpeaker B: Um, okay, you need some power source, get a battery or something to keep it going.\nSpeaker B: Um, and that power source is important because it ties you down to, um, how long the device will last?\nSpeaker B: Uh-huh.\nSpeaker B: I mean, it ties you down perhaps a bit later on in terms of technologies.\nSpeaker B: Um, how far you can transmit the signal or the complexity of the functions that you want.\nSpeaker B: Like for example, voice recognition, right?\nSpeaker B: That might be constrained because that you might need to power microphone, you might need to power other thing.\nSpeaker B: So that's one, perhaps, constraint, yeah.\nSpeaker B: Um, okay, the basic thing is, as a user in a program, you can, you know, you can, you can shut it apart.\nSpeaker B: Inipped that path using a device.\nSpeaker B: If one button has an opinion, when the object iservesin, um, looks blank to it because it applies slide, so the network actually is done.\nSpeaker B: If it is used forWhoosheduct, you've got, two questions, aAndView\u00e0, andickers.\nSpeaker B: a series of logic has to decide what the user is telling the device and the device has to you know, based on push button A, so I will do something with button A, so maybe button is a problem. And then it needs to be able to send it to the device itself, which is a secret here. And I think that's what it's about. It's fairly general. And I guess the purpose of this is also not to restrict you in the way you're thinking, like voice recognition, that if it's something which is important, then we just add more power, rather than having a thing that we don't have enough power, so it's not going to be a constraint in that sense.\nSpeaker B: I mean, these are functionally, you know, what they're going to do.\nSpeaker B: Okay. So I guess the rest of it, I think we should, maybe you want to go back to what the questions are. I think you're more relevant to discussion.\nSpeaker A: Do you want to finish up your presentation? Are you all done?\nSpeaker B: Oh, it's just putting the rest of it into words, but it's essentially a thing.\nSpeaker B: You have a translator and you can just write it to the computer.\nSpeaker D: Okay. And like, on the, I mean, since we're on the topic of the technology, are there any, like, what are our options? What's, what is in, is this the only way that we go about it?\nSpeaker D: Are there other?\nSpeaker B: These, these aren't technology options in that sense.\nSpeaker B: This is just the basic principle of principles and basic components that I needed.\nSpeaker B: Right.\nSpeaker B: For example, if you needed to add a voice recognition, right, then you use it in the base, we'll just put broken down more components, right?\nSpeaker B: You have a microphone, but we are something that.\nSpeaker A: So it's just how we're kind of modularizing the whole thing.\nSpeaker B: Okay.\nSpeaker B: So each component represents one function, but I think the basic functions are the logic, the transmitter, and the receiver, okay, and the power are things that you won't have to care about.\nSpeaker B: Yeah.\nSpeaker B: And those are things that based on what your user interface requires and will add more functionality to it.\nSpeaker B: Okay.\nSpeaker B: There might be one other consideration which would be the transmission between the remote control and the TV, for example.\nSpeaker B: Okay.\nSpeaker B: Are we going to restrict ourselves to using the traditional technologies as an indicator?\nSpeaker B: Okay.\nSpeaker B: Because that's something unique to actually be physically pointing to.\nSpeaker B: Okay.\nSpeaker A: Right.\nSpeaker A: Well, a worry that was expressed in the new requirements was that if we made this too complex, it would affect how long it took us to get this to market.\nSpeaker A: So I suspect it might be a good idea to restrict our kind of creative influence on this on the user interface and not worry so much about how we transmit it.\nSpeaker A: Because it's tried and tested and for ads.\nSpeaker B: So there might be a little problem with the transmission in particular analysis we're talking about voice recognition.\nSpeaker B: If somebody's going to talk to the device, you really want them to hold it to them.\nSpeaker B: You may not require that, but it's something very natural, I guess, to hold it to signal to the user and push a button to start talking about it.\nSpeaker B: Then you need to send the signal out, so if you're using infrared, the line of sight, I'll say the TV is at that chair and I'm standing in front of here and the transmitter here blocks it.\nSpeaker B: Okay.\nSpeaker B: So in that sense, there's not really a restriction, but it's something you may have to think about later on in the process.\nSpeaker D: Okay.\nSpeaker D: And just a clarification before we finish this, is our controller, have the option of being on a standard frequency as all of the other equipment?\nSpeaker D: So that the one controller can control several pieces of equipment?\nSpeaker B: There's not much specific information, but I think that one indication of infrared means that they're just targeting traditional devices.\nSpeaker B: Because infrared is something that you've been having.\nSpeaker A: Well, we've in the new requirements spec, they said just to focus on a TV, so that's what we should do for now, I think.\nSpeaker A: Something I was wondering about was the power.\nSpeaker A: Is it worth considering having a charging unit as opposed to just regular batteries?\nSpeaker A: I mean, is that something you really want to go into, do you think?\nSpeaker A: Or should we just consider running on regular batteries?\nSpeaker B: From a component point of view, there's added complexity, and you add cost to it.\nSpeaker B: And then there's probably the fact that you need another physical component, you need a docking cradle, for example.\nSpeaker B: Yeah.\nSpeaker B: Put it into charge.\nSpeaker B: Okay.\nSpeaker B: Or you need to get the user to plug it in.\nSpeaker B: Okay.\nSpeaker B: So most users are used already, use the value of buying batteries and putting it into the controller.\nSpeaker B: Yeah.\nSpeaker B: Unless the controller is going to consume a lot of batteries, like, it's going to run through batteries a month, then I don't think we charge about something that you're not doing.\nSpeaker A: Okay, so just stick to regular.\nSpeaker A: Okay.\nSpeaker A: Right.\nSpeaker A: So, basically, I'm just going to just recap.\nSpeaker A: When I said at the start, it was that the whole point of this meeting was to absolutely finalise who were going to aim this at and what exactly the product is going to do.\nSpeaker A: So, just to recap on, are we all happy about the idea of aiming the product at the 15 to 35 bracket?\nSpeaker A: Yeah.\nSpeaker A: And also the actual functions of what it's going to do, do you want to recap on that, Craig?\nSpeaker C: I think we used to say that it was going to be the most basic stuff possible.\nSpeaker C: On, off, up and down, down, up and down, volume, and skip to the channel.\nSpeaker A: Okay.\nSpeaker D: Is it going to include any of the more advanced features or are we going to eliminate those?\nSpeaker C: I think we include new, but apart from that, I think we just go for the simplest.\nSpeaker C: Okay.\nSpeaker D: Is it, is it, is it, is it not an option still that we include some things just as a sort of under, like, sort of under a door?\nSpeaker A: Yeah, it's as optional functions.\nSpeaker D: What I might be a bit worried about is if someone was had previously developed habits of expecting to control surround sound or this and that with their controller and then, and then they.\nSpeaker D: You know, they get ours.\nSpeaker D: It's, it doesn't have that.\nSpeaker D: I don't know if that'd be a problem.\nSpeaker B: And I think that you're saying about categorizing the people.\nSpeaker B: Maybe I could suggest we break them down in this reasonable kind of use.\nSpeaker B: Yeah.\nSpeaker B: One would be audio control.\nSpeaker B: One would be video control.\nSpeaker B: Okay.\nSpeaker B: Yeah, one would be a device.\nSpeaker B: This may not map very well to advanced functionality especially, but I think that from a manufacturer point of view, I think from a point of view of a person using it, you know, a TV is something to see and something to hear.\nSpeaker B: And it's something they do other things to like turn it on and turn it off.\nSpeaker B: Yeah.\nSpeaker B: Right.\nSpeaker B: What we could have is like two buckets.\nSpeaker B: We could throw things into like if you want this feature, that's where you can hear.\nSpeaker B: Yeah.\nSpeaker B: And then from there, it's basically what's non-basic.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: And it would actually help with the component.\nSpeaker B: Okay.\nSpeaker A: Okay.\nSpeaker A: Great.\nSpeaker A: Okay.\nSpeaker A: Well, I've got a kind of got five minutes to wrap up now.\nSpeaker A: Next thing we're doing is having lunch.\nSpeaker A: And then we're going to have 30 minutes of working on the next stage.\nSpeaker A: So I'll be pulling the minutes of this, this meeting into the project documents folder.\nSpeaker A: So I guess just to confirm that we know what we're doing in the next, well, in the 30 minutes after lunch anyway.\nSpeaker A: For our industrial designer, you're going to be thinking about the components concept.\nSpeaker A: User interface designer, you're going to be thinking about our user interface and marketing.\nSpeaker A: You're going to be thinking about trend watching.\nSpeaker A: And you'll all get specific instructions as well.\nSpeaker A: So I know just to ask now, if you've got anything else you thought about while we've been talking.\nSpeaker A: And do you want to start with David, then else to say at all.\nSpeaker A: No.\nSpeaker A: Okay.\nSpeaker D: Yeah.\nSpeaker D: I just wanted to ask them before we wrap up, should we agree for sake of sort of clarity and when we resume that we'll use this idea, David's proposal, we think of these three sort of buckets and.\nSpeaker D: Yeah, I think that's definitely a good idea.\nSpeaker D: Okay.\nSpeaker D: We're talking about this.\nSpeaker C: Should we do that?\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: So I know the three buckets, what would go in the device function is fine.\nSpeaker B: Things like on off.\nSpeaker B: Because they don't have anything to do with what you see.\nSpeaker B: Yeah.\nSpeaker B: In terms of picture and the entertainment value.\nSpeaker B: Okay.\nSpeaker B: So.\nSpeaker B: And channel.\nSpeaker B: And channel.\nSpeaker B: Because the on off also goes, you know, like on off like power, not on off sound.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Okay.\nSpeaker B: You have a choice of putting it on to others for device.\nSpeaker B: Device is basically anything which we can't categorize.\nSpeaker B: Okay.\nSpeaker A: So you're going to have audio, which is going to be like, you know, you base settings and actual volume.\nSpeaker A: Anything to do with what you hear.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: And then visual.\nSpeaker A: Anything that you can see.\nSpeaker A: Okay.\nSpeaker A: So brightness contrast, things like that.\nSpeaker A: And then just actual device things like what channel you're watching, turning it on and off.\nSpeaker A: Stuff like that.\nSpeaker A: Okay.\nSpeaker D: And I suppose quite likely what would happen is in the device category, there might be some which are just like the habitual standard and then others, which are maybe a bit more.\nSpeaker D: Like random, which we have no other place to push.\nSpeaker B: Okay.\nSpeaker B: Somewhere there.\nSpeaker B: Okay.\nSpeaker B: So I think very clearly what happens is that people at least have some in their mind.\nSpeaker B: It's easy to use.\nSpeaker B: I think that's one thing that.\nSpeaker B: And I guess from the corner point.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: So yeah, I guess just things to think about are, you know, like the fact it's got to look good because of who were, you know, targeting this up.\nSpeaker A: Something maybe kind of quirky in design, maybe make it kind of ergonomic kind of to hold, you know, things like that.\nSpeaker A: So I guess, I guess that's it.\nSpeaker A: Right.\nSpeaker A: It's the meeting over.\nSpeaker A: Oh.\nSpeaker E: Okay.\nSpeaker D: Now we get to go find out what was picked up for lunch for us.\nSpeaker D: Yeah.\nSpeaker D: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\nNone: Okay.\n"}, {"meeting_id": "TS3010b", "summary": "The team introduced their findings in turns and gradually came to some agreements. First, Industrial Designer introduced the structure of the device and suggested one battery and small size. Meanwhile, Project Manager questioned the cover design since it would not be cheap. Then, User Interface left a question whether the technical function should be basic or multi-functional. After that, Marketing concluded that less important functions should be discarded and the device should be user friendly. Lastly, Project Manager suggested discarding the teletext and asked for ideas about button design.", "dialogue": "Speaker A: I see.\nSpeaker D: Oh, it's so far, sir.\nNone: Okay.\nSpeaker B: cybersecurity Videos\nSpeaker D: Okay so welcome back what do we have to do? First I want to say I'm the secretary so I'll make the minutes you find them in your in the map in the from the group just the minutes from the first meeting you'll find the next minutes also there then I want to hear from you what you've done and after that I have some new products requirements so and after that we have to make decisions what we will do and then we're ready we're 40 minutes for this meeting after that we have lunch so first I want to ask the industrial designer to tell what he did that's my job okay\nSpeaker C: I've saved it on my computer my presentation\nSpeaker D: but where what's the name what's the name of it\nSpeaker C: it was about the working of the remote control the technical function or the functional requirements working design but safe now I don't know where it is\nSpeaker D: and this is product documents\nSpeaker C: and I'm partitioned to desktop\nSpeaker D: up up up up up my jockens f\nSpeaker C: away and and I had one presentation.\nSpeaker D: Where can you study?\nSpeaker C: Open it.\nSpeaker C: Okay, here it is.\nSpeaker D: Save as...not a cookie.\nSpeaker C: Mmm, desktop.\nSpeaker D: I'll check that.\nSpeaker D: Save.\nSpeaker C: Okay.\nSpeaker C: Very good.\nSpeaker C: A little later, but here it is.\nSpeaker D: Okay.\nNone: So...\nSpeaker C: Okay.\nSpeaker C: It's a little difficult when I'm going to tell you.\nSpeaker C: It's about the working of the remote control.\nSpeaker C: I just had an half an hour to study it and...\nSpeaker C: I don't get it.\nSpeaker C: There's 10 minutes to tell it.\nSpeaker C: 10 minutes to tell it.\nSpeaker C: Okay.\nSpeaker C: I think it will be a few minutes and...\nSpeaker C: Okay.\nSpeaker C: First, I will tell you something about the findings when I discovered about the remote control, the working of it.\nSpeaker C: Then I'll have some kind of map and it's the top of the remote control with a little bit of science.\nSpeaker C: I'll show it in a few minutes.\nSpeaker C: And then what I'll think about it.\nSpeaker C: First, the findings.\nSpeaker C: The remote control is a very difficult thing to explain to just all of you who haven't seen the remote control inside.\nSpeaker C: There's a lot of plastic on it because it's not so expensive.\nSpeaker C: And there are a lot of wires which connect the components in it, the battery and switches and things like that.\nSpeaker C: There's a lot of small electronics, so it won't be too expensive to build it.\nSpeaker C: And the 12-year-of-50, I think, will make it.\nSpeaker C: And here I have the top of the remote control.\nSpeaker C: Here's some kind of chip.\nSpeaker C: On top of this are the numbers.\nSpeaker C: You have all of your remote control and the teletext button.\nSpeaker C: Here's the battery.\nSpeaker C: When you push the button, it will be sent to the chip.\nSpeaker C: And the chip will send it to all kind of sub-components.\nSpeaker C: That's what I said.\nSpeaker C: It's very difficult.\nSpeaker C: And after that, it will be sent to infrared.\nSpeaker C: And that will send it to your television.\nSpeaker C: That's a short how it works.\nSpeaker C: I think I can make it difficult, but we all don't get it.\nSpeaker C: My preference is it won't be...we shouldn't make it too big.\nSpeaker C: Also for the cost.\nSpeaker C: We should only put one battery on it, a long lasting battery.\nSpeaker C: Also for the cost.\nSpeaker C: Use only plastic, not other materials.\nSpeaker C: Also because of the cost.\nSpeaker C: Not too much buttons on it.\nSpeaker C: You can also make a menu button.\nSpeaker C: Then you will see it on a TV.\nSpeaker C: On a TV, you can switch into the menu.\nSpeaker C: I think it's easier.\nSpeaker C: And the bleeps in you.\nSpeaker C: You told us.\nSpeaker C: But we can also use a bleep like something.\nSpeaker C: When the battery is empty, then there's a bleep.\nSpeaker C: You have to change it in a week or something.\nSpeaker C: And also the bleep, what I told you about when you lost it and you push your button, and then you were bleep bleep, and we will find it.\nSpeaker C: This is just...\nSpeaker D: Two questions. The battery.\nSpeaker D: You say one battery is cheaper.\nSpeaker D: Why?\nSpeaker C: If we use only just one small pen light, then it will be cheaper now.\nSpeaker D: When you use two, you can use it two times longer.\nSpeaker C: But then we have to make the remote control long-lasting.\nSpeaker D: So it's the size of the remote control?\nSpeaker D: Yeah.\nSpeaker D: And the buttons.\nSpeaker D: When you use it on the television, you need the television which can use it.\nSpeaker C: Yeah.\nSpeaker C: But I think our remote control is for the television.\nSpeaker C: So we sell in our company, or is it also for other television?\nSpeaker D: I think we have to use it also on other television.\nSpeaker C: Then this is an option.\nSpeaker C: So maybe just a menu button to use it on our television.\nSpeaker C: And we make it easier for our television and on the other television.\nSpeaker C: You can also use it, but then...\nSpeaker D: I think there are two different things.\nSpeaker D: We have to choose one.\nSpeaker D: That's to work on all television.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Then I think the menu button will only work on the newer televisions.\nSpeaker C: And we will look forward and don't make remote control, which is for all the televisions.\nSpeaker C: And I just have one more idea.\nSpeaker C: Maybe it's one of your tasks.\nSpeaker C: But to have a trendy remote control, we can also make something like the Nokia mobile phones to change covers.\nSpeaker C: So if you have a trendy house with all red, yellow and something, then you can put a red cover on it.\nSpeaker C: And also different things.\nSpeaker A: So we will just add to the cost.\nSpeaker C: Then it won't be... we will have just one cover on the original one.\nSpeaker C: And then you can buy the covers.\nSpeaker A: Yes.\nSpeaker A: But you have to be able to change it to make it more difficult to design.\nSpeaker C: I think it will be a little more difficult, but not too much.\nSpeaker C: Okay.\nSpeaker C: Just like with the Nokia mobile phones.\nSpeaker D: Much more Nokia telephones than these ones.\nSpeaker C: Yeah, but then we will have to just put five covers on it and see if it works, if it won't work, then we will get something else.\nSpeaker C: And we won't go further with it.\nSpeaker D: Are the profits bigger than the cost?\nSpeaker C: A cover made in China, it won't be so expensive, I think.\nSpeaker D: But also design costs. I don't think when you have remote control, do you change the coffee?\nSpeaker D: Maybe.\nSpeaker C: Would you change the coffee?\nSpeaker C: I won't.\nSpeaker C: But maybe I think trending people or children, when you're campaigned on it and the children say, Oh, this is my remote. So I made a picture on it.\nSpeaker D: I think that two less people would change it for good profit.\nSpeaker C: So, do you other people do you think about it?\nSpeaker B: Yeah, it's a good idea, but if it will make profit enough.\nSpeaker B: But it's an original idea.\nSpeaker D: Yes, it is.\nSpeaker D: But I don't think we have to do it.\nSpeaker C: You're the project manager.\nSpeaker B: Okay.\nSpeaker D: Yes, that's it.\nSpeaker D: That's clear.\nSpeaker D: Okay, thank you.\nSpeaker D: So now the user interface design.\nSpeaker C: Come on.\nSpeaker B: Yes, well, I shall give a short talk about the technical function design.\nSpeaker B: I thought the technical function design was for a remote control to have some influence on the TV set, both audio and video in a cordless way, now cords attached.\nSpeaker B: And, well, by pushing a button on the remote.\nSpeaker B: That was from my own experience and the previous meeting.\nSpeaker B: I find some interesting quotes on the web.\nSpeaker B: Well, the same idea here.\nSpeaker B: Message to the television.\nSpeaker B: And, well, basic operations like on and off and switching channels and maybe a daily text or something like that.\nSpeaker B: Well, these are two remotes.\nSpeaker B: And that's our dilemma, I think.\nSpeaker B: We just heard from the industrial designer how difficult it is, but shall we make a basic remote control with just swap in channels and volume and power button and more functions on the remote.\nSpeaker B: Maybe more devices you can influence radio or video recorder, VCR.\nSpeaker B: Well, that's our dilemma.\nSpeaker B: Any ideas about that?\nSpeaker B: Basic or multifunctional?\nSpeaker D: We'll go back on that later.\nSpeaker B: Yeah, well, that was just on my mind.\nSpeaker B: I didn't know what way we would go.\nSpeaker B: Yeah, well, that was my functional talk.\nSpeaker D: Okay.\nSpeaker D: Okay, thank you.\nSpeaker D: Then your turn, the marketing is there.\nSpeaker D: Okay.\nSpeaker A: Yeah, okay, this is a bit too far.\nSpeaker A: So I'm going to have a presentation about the market, about what people think.\nSpeaker A: We did a usability lab test with vendors' persons and we looked at several things among them behind them, the design, did they like the use of it?\nSpeaker A: What's the stage in the ad while using remote controls?\nSpeaker A: Well, what will be our market?\nSpeaker A: And we asked them if we had some new features, if that would be a good idea or not.\nSpeaker A: Well, our findings.\nSpeaker A: Our users, they dislike to look and feel of current remote controls.\nSpeaker A: They especially found them very early and they also found them hard to learn how to use it.\nSpeaker A: Well, they also shape a lot, so shaping should be very easy.\nSpeaker A: And 50% of the users only use 10% of the business, so a lot of unused business.\nSpeaker A: There is more findings on the business, which business find users very important and with which not, and how much do they use them.\nSpeaker A: Well, the most used button is the channels selection and we asked them how relevant they think the business are, the power, volume and channel selections are very relevant.\nSpeaker A: Tail attacks is less relevant but also important.\nSpeaker A: Not important, they found the audio, it's not the volume but specific, the pitch or the left or right.\nSpeaker A: The screen and the brightness and channel settings, and they also are not used very often.\nSpeaker A: Then we have a few graphs about the market.\nSpeaker A: Here we can see what the market share is of several groups.\nSpeaker A: As you can see, most users are between 36 and 45.\nSpeaker A: The younger group between 6 and 25 is not very big.\nSpeaker A: To come back on the swapping things, I don't think the younger will be most interested in it, but they are not a very big group.\nSpeaker A: In the area we asked them how would you like a new feature if you have an LCD on the remote control?\nSpeaker A: What would you think of it?\nSpeaker A: You can clearly see young users say, hey, they will be very nice and all the users think they are scared of change, they won't like it.\nSpeaker A: Another thing, how would you like to have a speech recognition on it?\nSpeaker A: Well, we see the same young users think that's an interesting idea and all users not.\nSpeaker A: We found out that there are two several markers at which we can aim.\nSpeaker A: The first are the younger, the age between 16 and 45.\nSpeaker A: They are interested in the features, as you can see here.\nSpeaker A: There are more critical on their money spending.\nSpeaker A: The second group is the older group, the age between 46 and 65.\nSpeaker A: They are less interested in new features, but they spend their money more easily.\nSpeaker A: Now, if you look back at this graph, we can see that among the first group is about 60% and the second group about 40%.\nSpeaker A: So the first group is bigger.\nSpeaker A: Then I come to my personal preferences.\nSpeaker A: The first question is also we have to ask is at which mark do we aim at?\nSpeaker A: Of course, saying we aim at a young group doesn't say that all people won't buy it, but less of them will buy it.\nSpeaker A: Well, I fact even young people say it's hard to use remote control.\nSpeaker A: So if you make a remote control that is very easy to use, especially in the test group, even the young group will also be more interested.\nSpeaker A: We can make special features, but I think it looks nice in the first time.\nSpeaker A: But when you use it, I don't know what's a good thing of speech recognition.\nSpeaker A: Well, that's my second point.\nSpeaker A: Less important functions should be described from the remote control.\nSpeaker A: This is about the discussion we had earlier.\nSpeaker A: You can find most functions on the FE set.\nSpeaker A: You don't have to have a lot of audio options or screen options to change the brightness and those things.\nSpeaker A: Well, the design is very important.\nSpeaker A: One thing I didn't say I think is that a lot of users also said that I would buy a good looking remote control if there would be one, but they found most remote controls very early.\nSpeaker A: So the design of our remote control is very important.\nSpeaker A: It should be very self-friendly as most users use it for that.\nSpeaker A: That were my findings.\nSpeaker C: Okay, thank you.\nSpeaker C: One question, if we aim for the younger people, and there will be a lot of features like LCD or the speech recognising, the cost will be a lot higher.\nSpeaker C: I think we don't have that in our budget.\nSpeaker B: Do you think?\nSpeaker B: I don't think 25 euros for a remote is really cheap or something.\nSpeaker B: It's hard to get the younger.\nSpeaker D: I think LCD is cheaper than speech recognitions.\nSpeaker D: I think that can be a good option.\nSpeaker D: LCD, just LCD.\nSpeaker D: Only the LCD.\nSpeaker D: But we'll get back on that.\nSpeaker C: Okay.\nSpeaker D: Now.\nSpeaker D: Oh, good.\nSpeaker D: What to do?\nSpeaker D: We'll go back on the decisions later.\nSpeaker D: Now, we have a few new product requirements.\nSpeaker D: First, teletext.\nSpeaker D: We have internet now, so we don't need the teletext anymore.\nSpeaker D: So, not necessary.\nSpeaker D: Next, only for the television, so we don't look at other things like the radio or something.\nSpeaker D: Only the television.\nSpeaker D: Third.\nSpeaker D: We look at the 8th group of 40 plus.\nSpeaker D: Now, younger than 40.\nSpeaker D: Big group, and like you showed, not very much people buy our stuff.\nSpeaker D: Fourth point, our corporate color and slogan must be used.\nSpeaker D: Very important for the design.\nSpeaker D: So, you can see it on our side.\nSpeaker D: Next.\nSpeaker D: We have to make our decisions what we want to do.\nSpeaker D: So, like you said, we need the 8th guy.\nSpeaker D: Maybe it's good to put it in a document.\nSpeaker D: We have to decide what controls do we need.\nSpeaker D: So, maybe you can tell us.\nSpeaker A: Maybe we can first have a discussion on the product requirements you just said.\nSpeaker A: Sorry.\nSpeaker A: The requirements you just said.\nSpeaker A: Yes, maybe you should first have a discussion about that.\nSpeaker A: Yes, I personally think teletext is a good option.\nSpeaker A: Not everyone who's looking to speak can go to the internet when they want to see the latest news.\nSpeaker D: But we don't use it.\nSpeaker D: It's not my requirements.\nSpeaker C: We'll just have to do that.\nSpeaker C: We have to do this.\nSpeaker C: Okay, no discussion about it.\nSpeaker D: So, what controls do we need?\nSpeaker D: We'll first.\nSpeaker D: Power button.\nSpeaker D: Power.\nSpeaker B: I think separate channels.\nSpeaker D: Oh, channel.\nSpeaker B: But then both the separate channels.\nSpeaker B: So, zero to nine or something.\nSpeaker D: Zero to nine.\nSpeaker D: It's easy to pick what was your one.\nSpeaker A: Okay.\nSpeaker D: We have.\nSpeaker D: Where was that example?\nSpeaker D: Hello.\nSpeaker D: Okay.\nSpeaker D: What do we need?\nSpeaker C: All of zero to nine.\nSpeaker A: To change to the next channel, just one button to move up, move down.\nSpeaker D: Yeah, that's the channel.\nSpeaker D: Do we make menu?\nSpeaker D: Menu.\nSpeaker D: Here's the new data efficient.\nSpeaker D: I think the only one or two numbers.\nSpeaker D: And that's it.\nSpeaker A: I think it will be quite useful for Avos, up down for channel selection, left, right for volume, and a menu button.\nSpeaker A: And if you press the menu button, you get into the menu and you can use the same buttons, but to tend to scroll to the menu and change options on the LCD screen.\nSpeaker A: You mean?\nSpeaker A: Well, yeah, that depends on if you have the menu on the TV, or you get the menu on the LCD screen on the remote control.\nSpeaker D: I think it's better to have it on the remote control.\nSpeaker D: Because it has to work on all television.\nSpeaker D: So, yes.\nSpeaker C: But then we'll come to the cast.\nSpeaker D: Yes.\nSpeaker D: But if we have this.\nSpeaker A: But if we aim at the younger market, as we've seen in the usability lab, they will buy a nice-looking remote control.\nSpeaker A: And also to find the easy-to-use part, very important.\nSpeaker A: So if you have LCD screen and not too many buttons, I think that will increase even when it's a bit more cost to still sell.\nSpeaker D: So now we don't have a lot of buttons.\nSpeaker D: Is this enough?\nSpeaker D: Mute.\nSpeaker D: Mute.\nSpeaker B: Maybe in the menu.\nSpeaker B: Yeah, but then it's always more than one.\nSpeaker D: I think it will.\nSpeaker D: Maybe more.\nSpeaker D: No.\nSpeaker D: Well, then that's all.\nSpeaker D: This will be the buttons.\nSpeaker D: I think it's enough for the next phase.\nSpeaker D: So you can go on to.\nSpeaker C: But now we have only the buttons.\nSpeaker C: We don't yet have to decide what the remote control would look like.\nSpeaker D: No, that's for the next phase.\nSpeaker D: If we check.\nSpeaker D: Phase two is the conceptual design.\nSpeaker D: So then we will have the concepts.\nSpeaker D: So next point.\nSpeaker D: Now we have lunch break.\nSpeaker D: After that we have 30 minutes for work.\nSpeaker D: And you can find the minutes in the product document for the inclusive buttons.\nSpeaker D: In the video action you can find them in the email.\nSpeaker D: So now it's time for lunch.\nSpeaker C: Thanks for coming.\nSpeaker C: Thank you.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\nNone: Thank you.\n"}, {"meeting_id": "IS1002b", "summary": "The group had a heated discussion over the functional design and the market positioning of their new product. First, User Interface introduced the basic components of the remote control and the cost of them. When discussing the user interface design, the group agreed that the product should be good-looking and branded with their company logo. Industrial Designer thought they should keep the functions simple and basic and aimed at the exterior design. Then, the group talked much about the teletext and whether the controller should be only for TV. When it came to the market positioning, the opinion of User Interface differed from Marketing's - the former tended to keep it simple but the latter thought there should be something special about the product to make it exclusive.", "dialogue": "None: I wasn't paying for it.\nNone: Yeah, I wasn't paying for it.\nNone: Maybe Pedro, which is here, still in presentation.\nNone: Who doesn't?\nNone: We all do. We can switch around.\nNone: Okay.\nNone: All right.\nNone: Swap these.\nNone: No.\nNone: Okay.\nNone: I see that it moves your mind.\nNone: Okay.\nSpeaker E: So, you'll click for me and stuff like that.\nNone: See, you can stay seated for years.\nSpeaker E: Yeah. Okay.\nSpeaker E: I hate these things.\nSpeaker C: Aha.\nNone: That's a great opportunity to learn how to be patient.\nNone: Not my thing.\nNone: Yeah.\nNone: It's a new thing to learn there, isn't it?\nNone: Yeah.\nSpeaker D: Sorry for holding us up, boss.\nSpeaker D: A couple of the tech guys got a little delayed there.\nSpeaker A: That's all right. I understand you guys have a lot of work there.\nSpeaker D: That's right. We certainly do.\nSpeaker D: We're both.\nNone: Me, I'm not technical enough to work a clip.\nNone: Which clearly doesn't work.\nSpeaker C: Damn.\nSpeaker C: This thing.\nNone: Yeah, well, we understand why you're a project manager and not technical director, don't we?\nSpeaker E: Yes, we do.\nNone: Oh, that's meant to be in there.\nNone: Aha.\nNone: I see.\nNone: Oh.\nNone: Oh, that's so good.\nNone: Wow.\nNone: Cheers, guys.\nSpeaker A: Okay.\nSpeaker A: So, we are here to talk about functional design.\nSpeaker A: Now, hopefully we've all got a better idea from than we did for leaving the last meeting as to what it is we're up to now.\nSpeaker A: So, here's an agenda.\nSpeaker A: A lot of them.\nSpeaker A: You should know that I'll be taking minutes in all the meetings as I was struggling to last time.\nSpeaker A: And that'll be easier for me now because I'm not actually giving the whole presentation.\nSpeaker A: The additional points are just the stuff that was sent that I forwarded on from upper management, having a few bright ideas to make our lives painful.\nSpeaker A: Now, you can all give your presentations.\nSpeaker A: We can talk about their requirements and hopefully come to some decisions.\nSpeaker A: Right.\nSpeaker A: 40 minutes for this meeting, so I bit more time than last one.\nSpeaker A: Here's the additional points.\nSpeaker A: I just wanted to put those in there to see if you guys had any comments on them.\nSpeaker A: Did you all receive that email?\nSpeaker A: Yeah.\nSpeaker A: So, does anyone have any overall?\nSpeaker C: Well, what comes up for me is that if we're going to be marketing a product that is going to be having no teletext, people are very comfortable with the idea of having teletext and using teletext.\nSpeaker C: And so, we're not, we're going to be a new product without something that people are very comfortable having right now.\nSpeaker C: So, that's from a marketing perspective.\nSpeaker C: I see a lack.\nSpeaker C: And so, we have to go, I think, in the other direction, what are we going to have that makes this thing better than...\nSpeaker A: Well, that first point could also be an opportunity because we've got some outdated, some sort of remote control that can work with the internet.\nSpeaker A: There is the opportunity that's presented, I guess.\nSpeaker C: Yeah, no, I agree with you.\nSpeaker C: So, what I'm talking about is I see that one side, we're eliminating something, so we have to come up with another side, which is what are we going to be targeting our market that identifies our product as better than because it doesn't have teletext, it has dot to dot to dot.\nSpeaker C: So, that's... that was my reaction.\nSpeaker A: Yeah, but we are sort of being dictated that this should only be for the television, so we're quite fixed.\nSpeaker A: So, we're really probably in terms of marketing and looking for something that's... that's a cost winner rather than a fantastic new featured product.\nSpeaker A: Okay.\nSpeaker E: But we're designing only the remote, we're not designing the TV.\nSpeaker E: So, we're going to be removing the teletext out of any TV.\nSpeaker E: Yeah.\nSpeaker E: People use our remote with.\nSpeaker E: That's right.\nSpeaker E: Which is kind of a stupid thing.\nSpeaker A: There's also the potential for market.\nSpeaker A: There's a market here for lost to out-text.\nSpeaker A: For example, someone that just goes to the shop that wants a replacement once it's too possible, 25 euros is a selling price.\nSpeaker A: We really have to innovate here, I guess.\nSpeaker C: That's what I'm talking about is that we have to find something that is going to be very attractive about this product because some people are going to be unhappy because they can't access their teletext.\nSpeaker C: Okay.\nSpeaker C: Because we're talking about 80% or 90% or 95% of the televisions out there are teletexts.\nNone: Yeah.\nSpeaker C: Yeah.\nSpeaker C: It's nothing of criticizing the product at all.\nSpeaker C: It's just when we eliminate that, then what do we bring?\nSpeaker C: What are we bringing to take the place of this?\nSpeaker C: In my opinion, we have to double up.\nSpeaker C: If we lose one, we need to bring two or three.\nNone: Okay.\nSpeaker A: I think that last point is probably quite straightforward.\nSpeaker A: Obviously, it has to be branded.\nSpeaker E: So the double R will be on the product.\nSpeaker A: Yeah.\nSpeaker E: I think one of the things that we should also keep in mind when we're doing this, I mean our company slogan is we put the\nSpeaker D: fashion in electronics, right? So I think our target here is to have some kind of very sleek, nice looking remote and we want it to be functional as well.\nSpeaker D: So we're going to be able to do that.\nSpeaker D: So you have this.\nSpeaker D: So we have three presentations and I think we'll go in order of participant number here.\nSpeaker D: So we can have a look at the work that's in the\nSpeaker A: website and we can have a look at the program.\nSpeaker A: So we have three presentations and I think we'll go in order of participant number here.\nSpeaker A: So we can have a look at the working design first and\nNone: participant two. That's...\nSpeaker A: Yes.\nNone: That's fine.\nNone: Okay.\nNone: Okay.\nNone: So...\nSpeaker D: I'm going to have that to click it on us.\nSpeaker D: So you all know me.\nSpeaker D: I'm the industrial designer.\nSpeaker D: And we have some basic components that our remote is going to need.\nSpeaker D: Just basically every remote will need them.\nSpeaker D: We have a power source.\nSpeaker D: We have to decide on our user interface, which is his department, but the user interface is also a major component.\nSpeaker D: We need a programmable digital signal processor to take the input from the user and translate that into electronic signal which we pass to the infrared LED which you aim at the television and it receives that signal.\nSpeaker D: You need an on-off switch.\nSpeaker D: I don't know how that got in there.\nSpeaker D: And we also need to have the...\nSpeaker D: If we want a universal remote, we need to have encryption codes for the different makes of TVs.\nSpeaker D: So we need to know all the different signals and so that will require some memory as well.\nSpeaker D: So here's just a basic layout of how the remote would work.\nSpeaker D: You have the power sources in the upper right-hand corner there and you can see that we have the user interface here which is connected to the chip which does all of our signal processing and then passes that signal into the infrared LED and that signal is then emitted and received by the television at the photo transistor.\nSpeaker D: So those are the basic components that need to go into this and everything else is pretty much open to move around.\nSpeaker C: It would just be considered just a standard...\nSpeaker C: This is not cutting edge technology we're talking here.\nSpeaker C: We're talking about existing technology.\nSpeaker C: Nothing's been modified or upgraded or new discoveries.\nSpeaker D: This is just a basic layout of all the components that are going to be absolutely necessary to have a working remote.\nSpeaker D: We can add things in like if we wanted some voice recognition.\nSpeaker D: I mean that you can kind of say would fall under the user interface and the digital signal processing chip.\nSpeaker A: Do we have an idea of costs of different components?\nSpeaker D: Well the most costly components are going to be the chip and basically the user interface and the casing are going to be expensive as well.\nSpeaker D: The LED and the transistors and everything else are pretty cheap.\nSpeaker D: So depending on what we want our functionality to be, the chip could be expensive or it could be cheap.\nSpeaker D: Depending on the amount of memory we need in there and stuff like that.\nSpeaker D: Do we have any ballpark figures for that yet?\nSpeaker D: I don't have any figures right now.\nSpeaker D: We have to wait until we get to the more specific design phase for that.\nSpeaker D: And I think a significant part of the cost could be the actual casing itself and the buttons and things like that.\nSpeaker D: Basically.\nSpeaker D: So that's all I have really.\nSpeaker A: Okay, thanks.\nSpeaker A: And we have participants through which I believe is Pedro.\nSpeaker D: Bye.\nSpeaker A: You can give you that.\nSpeaker A: Go on.\nSpeaker A: And you want to get?\nSpeaker A: We're fighting over it.\nSpeaker A: There's almost more fun.\nSpeaker A: You're a slideshow?\nSpeaker A: I think that's what you want to do here.\nSpeaker A: Just go up to view.\nSpeaker A: Click down here.\nSpeaker A: Okay.\nSpeaker E: So yeah, function design.\nSpeaker E: You guys know me, Pedro.\nSpeaker E: And what I've found is we want to do fashion and I think honestly we should keep technology low and just simple basically and try to aim for design.\nSpeaker E: It's basically a case we'll cost the same if it looks good or bad.\nSpeaker E: So we have to make it look good.\nSpeaker E: Something cute and small.\nSpeaker E: The big chunk of remotes are that in the 80s so we should just go for something that people like to actually look at.\nSpeaker E: And although most people will buy televisions and everything that have loads of little functions and everything.\nSpeaker E: And they mostly end up using simple functions and little things and most of people won't get too mad of actually having to go to the television to, for instance, tune in their stations.\nSpeaker E: There's no need to have that in the remote.\nSpeaker E: So as for what I would recommend for the interfances and I will change the colors on the logo.\nSpeaker E: But we should go for the user oriented device.\nSpeaker E: So simple controls and good ergonomics.\nSpeaker E: And although I'm still here recommending the teletext so I'll remove that I guess.\nSpeaker E: But we should go for the...\nSpeaker A: I think what the management recommendation was less that theres a worry that teletext would become outdated rather than we shouldn't have it.\nSpeaker A: So I still think if it's cheap enough functionality wise to include it really should be in there.\nSpeaker A: Because otherwise we're just going to...\nSpeaker A: I mean even if it's necessary or not, if you're given the choice between a remote west teletext or without.\nSpeaker A: When it really, if it isn't more expensive for us to make because as far as I understand it, it can be operated with the same set of buttons.\nSpeaker A: Right.\nSpeaker A: So it should be in there.\nSpeaker D: It's just the cost of an extra button. I mean software wise there's really no difference.\nSpeaker C: Okay.\nSpeaker C: I have maybe a silly question.\nSpeaker C: In the very beginning it said with the advent of computers there's going to be outmoded teletext.\nSpeaker C: I don't understand how those two things are connected.\nSpeaker C: How does computers and teletext...\nSpeaker C: Why is one eliminating the need of the other? I don't understand that.\nSpeaker A: Well maybe what we're getting into here is the idea of internet through the TV for example.\nSpeaker A: So that might play on what we can do.\nSpeaker E: They're basically aiming at saying that you would use a couple of years ago teletext would be the easiest way to check for a...\nSpeaker E: Scheduling.\nSpeaker E: The scheduling.\nSpeaker E: The next program and stuff like that.\nSpeaker D: Let's find out what you're watching even if there's commercials.\nSpeaker D: Okay.\nSpeaker E: And now you can look at it on the internet.\nSpeaker E: But I still think teletext is quite more convenient.\nSpeaker E: Until you have the same commodities.\nSpeaker C: I don't see the crossover between computers and television.\nSpeaker C: I do see the crossover in some sense.\nSpeaker C: But it's not happening.\nSpeaker C: The remote is used for television.\nSpeaker C: So if we make the crossover and we're going to view television on computers...\nSpeaker A: Then we're losing the necessity of the remote.\nSpeaker A: Well there is for example in digital TV systems you can press a button and buy things in adverts.\nSpeaker A: And you can view through a catalog for example. A family member has got a system where you just...\nSpeaker A: And there's other features for example in other systems where you can pause live TV and things like that.\nSpeaker A: It just features from the internet.\nSpeaker A: From computers that are coming into the TV under the covers.\nSpeaker A: But you still use it through a teletext.\nSpeaker A: So the things to think here are that there's going to be more functionality.\nSpeaker E: Yeah but we're not aiming to recommend for that.\nSpeaker E: That's the thing. All of those require other commands with more complexity and more software.\nSpeaker D: I think a lot of that's proprietary anyways. You're not going to be able to command a TV with our remote.\nSpeaker A: I don't think.\nSpeaker A: But still there's an opportunity if it's for example a trainable one then we're just simply having an up-down left-right.\nSpeaker A: An okay button or something like that might do as well in future proofing it.\nSpeaker E: Yeah but we would increase the price to try to make it a trainable one.\nSpeaker E: Okay.\nSpeaker E: Well I don't know. One way or the other teletext was there but I guess we can remove it or make two separate.\nSpeaker D: I think if possible you should try to talk with management about that.\nSpeaker C: I don't see the logic. I don't see the logic in elimination of teletext.\nSpeaker C: But I'm not a tech mind either. I just don't see the crossover between computers.\nSpeaker C: Because we're designing something for television.\nSpeaker A: I'll communicate that back to those guys. The message we really want to be sending to them is that although teletext may become outdated there's no logic in not having it in there anyway. It doesn't affect the price.\nSpeaker A: We're selling it to an existing market.\nSpeaker A: I think what they're pushing us towards here is in terms of thinking of ways to future proof for a system for future systems that have something else other than teletext maybe.\nSpeaker E: We're not putting anything in the place of teletext.\nSpeaker C: We're also marketing a product. What I'm seeing is a mid-range cost product.\nSpeaker C: We can't go and pump a whole bunch of technology into this thing because all of a sudden we're going to have cost overruns.\nSpeaker C: If we're going to choose to drop teletext again what are we adding to the product that makes it marketable?\nSpeaker A: So for keeping it basic, what we're looking to sell it basically is it's just being very easy to use, looking exceptionally good.\nSpeaker A: That's what I think because we really don't have anything else there, do we?\nSpeaker C: I don't see it. And to me if I'm going to market a product for beauty, for design, I'm going to try to market it at a much higher price.\nSpeaker C: I need to make it special with a high price tag. I don't want to make it economically competitive.\nSpeaker C: I want to market it as exclusive. So I would market this product at 89 euros and come up with some really beautiful exterior design.\nSpeaker C: But I don't think we have that flexibility.\nSpeaker A: If design is cheap and functionality is basic then 25 euros is probably a high price for commoner garden standard TVs.\nSpeaker A: So the place where we're going to justify that cost is through design, through making it a sleek elegant high priced basic remote.\nSpeaker A: Does that make sense?\nSpeaker C: No, I understand what you're saying.\nSpeaker C: But what I'm... Okay, we probably need to move along.\nSpeaker C: My concern is trying to find a marketing niche for this product.\nSpeaker C: And if I'm coming in with 25 euros which is mid-market price, then what am I going to give these people for this?\nSpeaker C: Yeah.\nSpeaker C: So that's just my question. But we can keep talking.\nSpeaker A: Okay. Sorry, can I cut into you there?\nSpeaker E: No, I would advocate for continuing teletext and those would be the basic commands.\nSpeaker E: As for really the case of design, maybe I'll come up with more concrete ideas right now.\nSpeaker E: It's just the idea of simplicity and slickness, cute and small.\nSpeaker D: I'm just thinking with all these universal remotes that are out there, how many people actually use every feature that are...\nSpeaker D: like these trainable remotes and things like that, where it's just so confusing to use all these functions.\nSpeaker D: I think the largest portion of the market is just going to... They lost their TV remote, they need another one that will work with their TV.\nSpeaker D: They want something that looks nice, that isn't going to break when they drop it, that maybe it's ergonomic, it feels good in your hand, something like that.\nSpeaker D: I think that's going to be where you're going to find the biggest market share.\nSpeaker A: So we're looking for something that looks good and just works rather than looking for any special features.\nSpeaker D: I think the more bells and whistles we add, it's just going to cut into our profits because I think as we add costs for things like...\nSpeaker D: if we had the voice recognition and things like that, I think what percentage of the market is actually going to use that?\nSpeaker D: Maybe 5%.\nSpeaker A: We can talk a lot more potentially in the marketing presentation about this.\nSpeaker A: Yeah, okay. I'm sorry, I didn't mean to cut. Sorry, boss. I'm not the boss.\nSpeaker C: Pedro, I just want to say quickly, I would really like it when you can come up with some more bells and whistles, huh?\nSpeaker C: Well, the thing that come to my mind is something that's voice-activated.\nSpeaker C: I know we're getting into some big money on this.\nSpeaker A: I think that's probably a question more for...\nSpeaker A: Is that for over here?\nSpeaker C: That's the idea.\nSpeaker C: And I think of voice-activated. I think of how many times I've lost my remote and I can't find it.\nSpeaker C: So some way that I can find my remote, by clapping my hands or something.\nSpeaker C: And so...\nSpeaker C: But that's just something. I'm trying to find some bells and whistles because when I talk about taking out teletext, it's like, what are we going to put in?\nSpeaker C: What makes this thing attractive? And it's only for televisions.\nSpeaker C: So our competition out there has got these multiple adapters where they can use them for their VCRs, their DVDs, their televisions.\nSpeaker C: And we're coming out with one remote for television only.\nSpeaker C: And so to me, we have to make this a really special product.\nSpeaker C: I really have to say, I don't see the market niche for this product. At this price, I don't see it yet.\nSpeaker C: I go along with this because this is what we're given to try to market. But I don't see the market niche for this product.\nSpeaker C: Without something really special to identify it as unique in the market.\nSpeaker E: That should be designed.\nSpeaker E: Yeah.\nSpeaker E: That should be the design, basically. I think technology, we're not in the price range to do it.\nSpeaker E: We don't have the money to do it towards technology, so we should have to do it in the box.\nSpeaker C: Yeah.\nSpeaker C: Okay. Well, so that's up to you then.\nSpeaker C: To make this box in something that's just absolutely extraordinary.\nSpeaker C: Okay.\nSpeaker C: We can make it in the form of a gun. We can sell it in the United States.\nSpeaker A: I have a question for you. Does any of the features that Pedro's spoken about here have any implications we might not have thought about in terms of just pure functionality?\nSpeaker A: In terms of making the worker the cost of that?\nSpeaker D: I think all these things are pretty standard. I think we'll be okay.\nSpeaker A: Okay.\nSpeaker A: Cheers.\nSpeaker A: And to participant.\nSpeaker C: Well, for marketing for me is...\nSpeaker C: And how do I go here? Okay.\nSpeaker E: You can just click.\nSpeaker E: Go.\nSpeaker D: There you go. You just get off that. You just click anywhere.\nSpeaker C: Uh-huh.\nSpeaker C: Yeah. What for me is it... I don't know what I'm marketing right now.\nSpeaker C: Okay. Until you spoke and when I wrote this, I don't know what I'm marketing.\nSpeaker C: I just know that I was identified as a... We identified ourselves as a developer, as a manufacturer, and as a distributor to other wholesalers.\nSpeaker C: And so the 25 euros that we've identified as the sale price is a wholesale price rather than a retail price.\nSpeaker C: That's what we decided here.\nSpeaker C: So what I did is I decided that this needs to be a product and a sense can market itself.\nSpeaker C: So what we've identified here as our main marketing strategy is in design.\nSpeaker C: We're making the most beautiful, attractive, whatever we decide it is.\nSpeaker C: So that means we have about seven more minutes or ten minutes left.\nSpeaker C: So I would like to have a product that we can sell easily.\nSpeaker C: So I saw inspiration. So having something beauty, something attractive, something that in a sense will sell itself.\nSpeaker C: Innovation, I think, fits into what we're talking about here with design.\nSpeaker C: Cutting-edge technology, I don't think we're going to have that.\nSpeaker C: These were ideas I was putting together unless we come up with some new age product as far as the casing is concerned.\nSpeaker C: I want to make this thing something that I can identify as special in some way.\nSpeaker C: So maybe we can have some... I talked about environmentally sensitive.\nSpeaker C: Maybe we have a product that can be identified in some way as advantageous in the home.\nSpeaker C: I don't know. These are just thoughts.\nSpeaker C: I wanted to talk about who we are as a corporation.\nSpeaker C: That we're new, we're aggressive, we're competitive, we're trying to provide a product from a new perspective rather than from an old corporate line.\nSpeaker C: So to me, it's about selling our identity, our corporate identity along with the product.\nSpeaker C: What I found is that the projected costs are competitive.\nSpeaker C: We're kind of in the middle of the market.\nSpeaker C: But what I'm seeing is that the market is ready for... I identified new technology.\nSpeaker C: But again, because what I see is budget restrictions and limitations, I don't know that we can bring any new technology into this thing.\nSpeaker C: If we could have a technological something.\nSpeaker C: I have an idea.\nSpeaker D: Please.\nSpeaker D: And it's kind of along the slides of environmentally sensitive and it may even fit into ergonomics and even kind of address the issue of losing the remote.\nSpeaker D: What we're saying is a common issue.\nSpeaker D: What we can do is, well, you know that batteries, throwing robots run through batteries like crazy.\nSpeaker D: Yeah.\nSpeaker D: And for some people, maybe not a solar remote, but instead what about if we had a power cradle?\nSpeaker D: A cradle that is there to hold the remote when you're not using it.\nSpeaker D: So you'll always know where the remote is because you have to put it back there to charge the remote.\nSpeaker D: And instead of having replaceable batteries, we'll just have a power cell that stays there for the life of the remote.\nSpeaker D: You never have to go through all these different batteries.\nSpeaker D: And also, you can, I think it's a good opportunity for the user design, not just for the user design, but also for the look and feel of the remote as a whole.\nSpeaker D: You know, you could have some kind of neat little, you know, sexy design for a cradle in the remote itself.\nSpeaker C: Well, I like it. I like the idea, but we're talking about end cost is going to probably double.\nSpeaker D: It would increase the cost.\nSpeaker C: But boy, we can sell this thing because there's no batteries. It's environmentally sensitive.\nSpeaker C: We can identify it as a safer product in some form.\nSpeaker D: And you could page the remote if you lose it. Maybe there's a button on the cradle.\nSpeaker C: Yep. That's right. I really see.\nSpeaker A: No, the question is, can we make this for less than 1250?\nSpeaker C: No, no, no, we have to change the end cost.\nSpeaker A: I don't necessarily have to change the end cost because can we do that without changing it 1250 per product if we basically can sell more based on this?\nSpeaker B: Yeah, that's going to be up to these guys.\nSpeaker B: Well, my reaction is no.\nSpeaker D: What I can do is I can look into, Pedro and I can look into how much this might increase our costs.\nSpeaker D: And you can look into what kind of effect an increased cost is going to have in our final numbers.\nSpeaker A: Yeah, they can certainly push for more expensive products.\nSpeaker C: I'm an advocate to make this an exclusive product.\nSpeaker C: Let's sell this thing wholesale for 50 euros, 60 euros.\nSpeaker C: Let's make this thing really exclusive, environmentally sensitive, high tech design, ergonomics, all of this.\nSpeaker C: Just make this thing the Rolls Royce of remote controls.\nSpeaker C: Every home's got to have, if you don't have one, what kind of remote do you have?\nSpeaker C: Oh, you've got one of those. Oh, fantastic. I want one of those too.\nSpeaker C: I just don't have enough money right now.\nSpeaker D: I just don't know about that because in order to price it at that kind of level, I think we'd need to have every functionality that all the other competitors are offering.\nSpeaker D: Being able to operate the DVD player and the stereo system and all that, which is a pretty basic function that we've opted out of.\nSpeaker D: And additionally, we're maybe not supporting teletext.\nSpeaker A: I think we can see with certainty now that we are supporting teletext.\nSpeaker D: But I do think there are some basic features that we don't have that a lot of other remotes are going to have.\nSpeaker C: One thing I don't understand is they've given us this package.\nSpeaker C: Now, in my opinion, we need to give them a package back.\nSpeaker C: They say, okay, here you go.\nSpeaker C: They give us a foundation of what they want.\nSpeaker C: And it's our place to kind of analyze and evaluate.\nSpeaker C: I don't think it's our place to create their product for them.\nSpeaker C: Maybe it is.\nSpeaker C: I don't know what kind of role we have in a corporate ladder.\nSpeaker C: But to me, it's like, okay, here's our ideas.\nSpeaker C: And then let them look at those ideas and they say, yeah, well, we can raise the price 20%.\nSpeaker C: We like this idea, this idea, no.\nSpeaker C: But to me, it's about a choice.\nSpeaker C: Do we follow their directives or do we make presentations back and then discuss?\nSpeaker C: Because I do not see the market niche for the product that they're handing us right now.\nSpeaker C: I don't see it.\nSpeaker D: I have to disagree, though.\nSpeaker D: I think our market niche is basically people who have lost their remote or have broken their remote and don't want to send back to the manufacturer.\nSpeaker D: They just want to run out to the store and get a remote.\nSpeaker D: And then they're going to look on the shelf and they're going to see ours is the nicest-looking remote.\nSpeaker D: It does what they need.\nSpeaker D: I think that-\nSpeaker A: And to get to back to another point, sorry, for cutting in, but I think it's important to remember that this remote has to work with multiple TVs, yeah, because it's selling on its own.\nSpeaker A: It's not going to be specifically for Hitash TV's or whatever.\nSpeaker A: I understand it.\nSpeaker A: So technologically, if I understand it, TV's remote's working exactly the same way as video remotes and DVD remotes.\nSpeaker A: All you need to do is train them to the individual one or all you need to do is know they're like some four-digit code.\nSpeaker A: So saying that it works with all TVs is equivalent to saying that it'll work with DVDs.\nSpeaker A: Nope.\nSpeaker C: They've identified the product as not working for anything with televisions.\nSpeaker C: They've identified as product limitations.\nSpeaker C: That's why I say I don't see the market niche for this.\nSpeaker C: If we were going to have a product that worked for DVDs, VCRs and everything, then I can see the market niche.\nSpeaker C: But we're only identified as going for televisions without teletext.\nSpeaker C: And it's like, whoa, in my opinion, we have to go with something that's extremely exclusive to make this thing marketable because who wants just a television remote?\nSpeaker C: I don't.\nSpeaker D: Well, here's my thing about that.\nSpeaker D: If you're in the market for this ultimate remote, you're not going to go for just a television remote.\nSpeaker D: If it can't control your DVD player and if you can't get that all-in-one ultimate remote that you can throw all your other remotes away, I'm not going to buy a 75-year-old remote.\nSpeaker C: I have a tendency to agree with you.\nSpeaker C: I really do.\nSpeaker C: That's why I say I question the marketability of the product.\nSpeaker C: I really question where we create the demand.\nSpeaker A: We're really looking for something basic.\nSpeaker A: One thing that you said really struck a chord with me here in that we're carrying the corporate identity with the product, but we're actually looking for a large profit at the same time where I've been inclined to go back to a product level management and say, let's just cut down our profit expected on this product because we are actually branding our company here.\nSpeaker A: We're selling more than just the product.\nSpeaker A: We're trying to get mind-share about a real reaction to the people who are going to buy consumer electronics.\nSpeaker A: We want people to eventually say, oh, that's real reaction.\nSpeaker A: That's a good make.\nSpeaker A: It's reliable.\nSpeaker A: We're going to win mind-share by either being a fantastic product or for working, just being good, reasonable priced and solid.\nSpeaker A: If we can put an emphasis here on it not breaking, I think that's in itself extremely...\nSpeaker A: That's big.\nSpeaker C: If you drop the cost, now if you make go to the other side of the cost scale and you make it less expensive, then all of a sudden now we're going into a new market area.\nSpeaker C: We can say low cost, good design, beautiful, what did that, but what I'm seeing is right now we're middle market.\nSpeaker A: It's 25 euros a mid-market price for a remote.\nSpeaker C: For multi-function remote, that's a mid-market price.\nSpeaker C: In your discount right now you can buy them on sale sometimes for about 18, 19 euro.\nSpeaker C: That's for the multi-functions, DVD, VCR, catch-alls.\nSpeaker C: They have them in a little box in the middle.\nSpeaker C: This is the consciousness that most people have about replacement remote controls.\nSpeaker C: To market only a television remote control, we have to change consciousness.\nSpeaker C: We have to have something that will change people's thinking to identify this as advantageous.\nSpeaker C: I don't know what that is.\nSpeaker C: It's got to be low cost or high cost with special design features.\nSpeaker A: I'd be inclined to say if we can make the design better than any comparable remote while reducing the cost, then that's the way we really should play it.\nSpeaker A: If we can take it down to 15 rather than 25 and make a low profit margin on it, but we're really doing that to sell the brand.\nSpeaker A: We're five minutes left.\nSpeaker A: To sell for real reactions.\nSpeaker A: I can communicate this to the more senior people within the company here to get reactions.\nSpeaker D: What about the idea that I had with Cradle?\nSpeaker D: How do you feel about that?\nSpeaker D: How does everybody feel about that?\nSpeaker C: We're talking about the other end now.\nSpeaker C: I like the idea.\nSpeaker D: We have to find out the other end.\nSpeaker A: The problem with that in my mind, I just think of and as a consumer, would I want to buy that?\nSpeaker A: I think maybe not because I've got a remote and I'll take it to different chairs.\nSpeaker A: I don't want to go over somewhere and put it down.\nSpeaker A: If it's going to be somewhere that you put it down, then it needs to be at the side of your armchair and that implies that there's only one TV user.\nSpeaker A: It's very good for some people.\nSpeaker D: It would last for several hours on its own power.\nSpeaker D: When you're done with it, you put it back in the crib.\nSpeaker A: It takes a bit of ease of use of the remote to set an extent.\nSpeaker A: You have to sort of remember, you have to shift down the side of the couch to find it and put it back.\nSpeaker A: I think in my mind, it doesn't seem like something that I really want because of that fact.\nSpeaker A: I don't know, that's just my opinion.\nSpeaker C: Again, we have a cost issue here.\nSpeaker C: If we're going to go in this direction to me, we're going to the other side of the cost range.\nSpeaker C: Are we going to make this selling the cost as a high-tech, environmentally sensitive, whatever, are we going to go to the lower side of the range?\nSpeaker C: Again, I don't see our market niche in the middle for our product yet.\nSpeaker A: I don't see it yet.\nSpeaker A: So what do we think maybe we should, Pedro, do you have any opinion on that?\nSpeaker E: We should keep it simple, medium-low cost area and make it look good.\nSpeaker C: A question I have, because I think I agree with you, we've got to make it the box.\nSpeaker C: The box is going to sell it, I think, the outside of the casing.\nSpeaker C: Can we have multiple designs?\nSpeaker C: Have a modern, have a traditional, have a, you know, and so instead of them all looking the same, people can have maybe four or five different designs they can choose from.\nSpeaker C: I don't know what that creates cost or...\nSpeaker D: Well, the problem with that is we may end up with a whole bunch of inventory of a...\nSpeaker D: One over and over design.\nSpeaker D: Yeah, okay.\nSpeaker E: I hear that.\nSpeaker D: So, it's a bit complicated.\nSpeaker D: But what we could do is some kind of...\nSpeaker D: Well, I mean, yeah, we need to, I mean, have a few designs to look at and then come up with something that I think we all agree is, you know, the best fit.\nSpeaker D: I think what's really important is it has to look good, it has to feel good in your hand, it has to be durable.\nSpeaker D: And I think it's really important that it doesn't look cheap.\nSpeaker D: I think we have to make sure that people look at it and feel like it's like a quality remote, even though the cost may be...\nSpeaker C: What about a remote that doesn't maybe look like a remote?\nSpeaker C: Just an idea.\nSpeaker C: Just...\nSpeaker C: Okay, because I mean, we all know what a remote looks like on the outside, they're selling these things everywhere.\nSpeaker C: Maybe that could be part of our design exclusivity is we make a remote that doesn't even look like a remote.\nSpeaker C: It opens like a telephone.\nSpeaker D: Or what if it looks like a pen?\nSpeaker D: It doesn't matter, yeah.\nSpeaker C: And then just talking about some...\nSpeaker C: Something to make this thing unique.\nSpeaker C: That's the thing.\nSpeaker C: Something I'll sell whatever you guys design.\nSpeaker C: I don't have a problem selling a product.\nSpeaker C: That's not the issue.\nSpeaker C: I give you ideas you guys create the product.\nSpeaker C: Once you have the finished product, I'll find a way to sell it.\nSpeaker C: Don't worry about it.\nSpeaker C: I just give you these things now because these are my thoughts and feelings.\nSpeaker D: So just to touch on my point one last time that I had brought up, I really...\nSpeaker D: Unless you guys are die-hard against the fact that it's possible, it reduces the usability of the remote in the fact that you can't lose it on your own.\nSpeaker D: I think it still might be a good idea to investigate having like a power cradle.\nSpeaker D: Because I really think in a certain sense, it almost like for me, I would want to have that just because I know at the end of when I'm done watching TV that when I'm done, oh, I better go find the remote and put it back where it belongs.\nSpeaker D: Maybe that's just some people that like that.\nSpeaker A: Oh, there's also remote controls that I remember that worked.\nSpeaker A: This hasn't been done in a while, I think, but as a watch.\nSpeaker A: As a watch.\nSpeaker A: Yeah, there was remote control watches.\nSpeaker A: But I think there are, sorry, 80s things, so it might not be easy to market in it.\nSpeaker A: But the technologies came along that it might be cheaper to make.\nSpeaker A: Now those things might have been expensive for all I know.\nSpeaker C: One thing I just thought about when you were saying that is that our target market is going to be someone who has lost or broken.\nSpeaker C: That's what I was saying.\nSpeaker C: Who's lost or broken their, their remote.\nSpeaker C: So how many remotes do they want to buy?\nSpeaker C: Can we sell them a remote that can maybe guarantee they will not need to buy another one?\nSpeaker C: And so with this kind of system or a locator or something to think about how have people been losing or breaking their remote?\nSpeaker C: How does this happen?\nSpeaker C: They'll also presumably.\nSpeaker C: And then design the product maybe some component of the product that can identify this as the last one you'll ever need to buy.\nSpeaker A: Well, if we're going down that then we can, we don't need to go, it strikes me that the locator feature might actually be quite expensive to make.\nSpeaker A: Whereas actually just putting it in the power cradle might not be too expensive at all because that's just done with mobile phones.\nSpeaker A: You can just, we have the locator.\nSpeaker A: So we have actually similar products that we can take from and potentially cut costs there.\nSpeaker A: We need to look into whether we can do that, but I think we have some mobile phone based products checking quickly our internet.\nSpeaker A: We really need to finish up here.\nSpeaker C: But we can make it in the shape of a coffee pot.\nSpeaker A: Yeah, we have the power adapter products.\nSpeaker A: We sell power adapter products.\nSpeaker A: So for example, there's nothing stopping us.\nSpeaker C: I think it's telling us to stop from meeting.\nSpeaker C: So we have the, we probably have, have online in our, in our system.\nSpeaker C: We have the ability to create this we're talking about.\nSpeaker A: Yeah, without it being too.\nSpeaker A: In fact, we can cut costs through using the manufacturing that's already going to be\nSpeaker C: in place here. If we can create this, we have probably five or six or seven really strong marketing characteristics.\nSpeaker D: I think one of the things that we can put right on the box is never buy another battery.\nSpeaker D: Yeah.\nSpeaker D: You know, something like that.\nSpeaker A: Okay.\nSpeaker A: Thanks guys.\nSpeaker C: No, that's nobody want to want.\nSpeaker C: Oh, look at here.\nNone: We have to leave.\nNone: I don't, but I don't want to leave.\nNone: Just because you learn more, do not want to go without your application and others because you can essentially provide that value over there and help out.\n"}, {"meeting_id": "TS3006b", "summary": "The whole meeting was about what the group should do for the remote control, from a good-looking appearance to a joyful user experience. At the beginning, the group had individual presentations about their previous research findings. Through those presentations, all of the four got a general idea about the target market, the functions and the layout. They talked about the functions carefully as well as the interface design of the remote control later on. Finally, under the leadership of the Project Manager, all the members got some assignments for their next detailed design.", "dialogue": "None: it's not even cheap greenhouse Editor for that reason.\nNone: No.\nNone: Ah yeaahahah.\nNone:!\nSpeaker C: I didn't see it yet I think.\nSpeaker A: Did I miss something?\nSpeaker C: Yeah, pretty much.\nSpeaker C: I thought I'd come and mail you so I thought I'd just drop it in.\nSpeaker D: Hey, what's wrong with my computer?\nNone: It's an iconita who is telling you.\nSpeaker C: Yeah, that's my presentation.\nSpeaker A: Oh, right.\nSpeaker D: Hey, everybody ready?\nSpeaker D: Sorry, I'm not a good player.\nSpeaker D: Okay, where do I find this?\nSpeaker A: I'm not a good player.\nSpeaker A: And then, I'm a paradise.\nSpeaker B: You read the new flash?\nSpeaker C: No, what was it?\nSpeaker C: Oh, no, press it.\nSpeaker D: Yeah, no, but my screen is reduced in size.\nSpeaker C: What?\nSpeaker D: Feedback?\nSpeaker A: Okay.\nSpeaker B: I'll delete.\nSpeaker B: Let's see.\nSpeaker B: So it doesn't draw the attention away.\nSpeaker A: I made my own map.\nSpeaker D: No, not this, but the task.\nSpeaker D: No, I just flipped it, close it, took it here and then this happened.\nSpeaker D: And that's not the way it's.\nSpeaker D: Where was it?\nSpeaker D: It's settings.\nSpeaker B: Okay, all right.\nSpeaker D: Thank you.\nSpeaker D: Do you guys like your task?\nSpeaker D: I get my function, you know, the information that I need.\nSpeaker B: So frustrating.\nSpeaker B: Yeah, I don't like the end and just typing it to that.\nSpeaker D: Oh, I have to do that.\nSpeaker D: So, this presentation is mainly based on my own ideas because I hadn't time to integrate the information yet.\nSpeaker C: Really annoying.\nSpeaker C: I'm not sure.\nSpeaker C: I'm not sure.\nSpeaker B: I'm not sure.\nSpeaker B: I'm not sure.\nSpeaker B: Okay.\nSpeaker B: So, I'm going to talk about functional requirements.\nSpeaker B: Well, some research has been done, has been done, observing of 100 subjects and usability lab using the remote control.\nSpeaker B: They also fill in a questionnaire.\nSpeaker B: The findings were, well, you can see them for yourself.\nSpeaker B: They dislike to look at field current remote controls.\nSpeaker B: They're ugly.\nSpeaker B: They do not match the operating behavior of the user, so they don't match what they want to have on it.\nSpeaker B: They're often lost somewhere in the room.\nSpeaker B: It takes too much time to learn how to use a new remote control.\nSpeaker B: And they're bad for error's eye.\nSpeaker B: I don't know how a user can read that, but okay.\nSpeaker B: There was also research on the most relevant and irrelevant functions.\nSpeaker B: Most irrelevant and less used were audio settings, mono stereo, pitch, bass, screen settings for brightness and color and stuff like that.\nSpeaker B: But they are used.\nSpeaker B: They do need to be under remote control.\nSpeaker B: And if you can't control the sound settings, I mean, if you dislike a very loud bass or something, you need to change that.\nSpeaker B: Yeah, we have to.\nSpeaker B: If he doesn't have an equalizer.\nSpeaker B: Yeah, I mean, we can't, my TV has, but we can leave them away.\nSpeaker B: Most relevant and most used functions, they speak for themselves, I guess.\nSpeaker B: Power button, channel volume selection, daily text, but we can skip that because I saw the new slash and daily text is so outdated that it should not be used anymore in the future.\nSpeaker B: So forget this one and channel settings.\nSpeaker B: So for programming your channels in the right order.\nSpeaker D: Where did you guys get that new slash?\nSpeaker D: Yeah, I didn't get anything.\nSpeaker B: On the project.\nSpeaker C: Not by mail.\nSpeaker C: I received mail, but you don't.\nSpeaker B: No, so it's a text file in the project folder.\nSpeaker B: So daily text can be skipped.\nSpeaker C: That's in the presentation.\nSpeaker B: There was some research on new features in the remote control about an LCD screen and speech recognition.\nSpeaker B: Well, we got an update for the audience or the targeted group.\nSpeaker B: So it's above 40, I guess.\nSpeaker B: The new product or below.\nSpeaker B: Because that's pretty relevant.\nSpeaker B: I thought our ads.\nSpeaker C: Our current customers are in the age group 40 plus.\nSpeaker C: Yeah.\nSpeaker C: And the new product should reach new markets.\nSpeaker C: Below.\nSpeaker B: Okay, well.\nSpeaker B: That's in the new.\nSpeaker B: Okay, that's good to know.\nSpeaker B: Because you see a clear distinction between the age groups concerning the features.\nSpeaker B: And above 40 people are not so interested in the screen or speech recognition.\nSpeaker B: But below that age they pretty much are.\nSpeaker B: So I think we can build that in.\nSpeaker B: Yeah, well, we can skip this part as well because I thought our ad above 40.\nSpeaker B: So we could skip the features, but we just have to build them in because they find it very interesting.\nSpeaker B: Well, we have to keep all the classic functions but make the buttons as user friendly as possible.\nSpeaker B: And also, so not only the design of the buttons but how you can push them and stuff like that.\nSpeaker B: So the physical aspect of it.\nSpeaker B: And I think certainly for the lower age groups, a nice design.\nSpeaker B: Which does not make the remote control or no to stay in your room.\nSpeaker B: It's actually a part of your interior of your design in your room.\nSpeaker B: So the people can say, whoa, what's that?\nSpeaker B: Well, that's my remote control.\nSpeaker B: So it has to look nice and feel nice and have all the functions.\nSpeaker C: But it also needs to have corporate identity.\nSpeaker B: Yeah.\nSpeaker B: So the logo has to be present and the colors as well.\nSpeaker C: We can change much of that.\nSpeaker B: Two years.\nSpeaker B: Yeah.\nSpeaker B: But I don't think that's a problem because the thing has to have a color anyway.\nSpeaker B: And most of the time there is a brand present on it.\nSpeaker B: So I think that's not going to affect it very much.\nSpeaker B: But that are the consequences of the marketing part.\nSpeaker C: It's open already so you can use all top.\nSpeaker A: Oh, you're here.\nSpeaker C: Oh, no.\nSpeaker A: You pressed all the four?\nSpeaker A: No, no, no.\nSpeaker A: I pressed the mouse button.\nSpeaker D: Oh, great.\nSpeaker D: That's a self-destruct button.\nSpeaker C: Yeah.\nSpeaker C: Maybe you can do it from your computer.\nSpeaker C: Yeah.\nSpeaker A: Yeah.\nSpeaker A: And if you go stand around here.\nSpeaker A: Yeah.\nSpeaker D: Sure.\nSpeaker C: All right.\nSpeaker C: That's nice.\nSpeaker C: Yeah.\nSpeaker C: All right.\nSpeaker A: I had two examples.\nSpeaker A: This one is the advanced one with a lot of options and functions and buttons.\nSpeaker A: This is the easy one.\nSpeaker A: I think we have to combine them.\nSpeaker A: And yeah, merge the best functions of all examples.\nSpeaker A: But yeah, the H is under 40.\nSpeaker B: Yeah.\nSpeaker B: And marketing research stated that that kind of users are not afraid of a lot of functions.\nSpeaker A: So not too much.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker D: We do have to integrate the screen and the speech recognition.\nSpeaker D: Yeah.\nSpeaker A: All right.\nSpeaker A: Yeah.\nSpeaker A: This one we can remove for kids.\nSpeaker A: It's just only for adults.\nSpeaker A: So we can use some advanced options.\nSpeaker A: Yeah.\nSpeaker A: From age of 16.\nSpeaker A: Yeah.\nSpeaker A: But I prefer we use the basic options.\nSpeaker A: Yeah.\nSpeaker A: We have to make them very easy.\nSpeaker A: So for just zipping around channels, you can just push one button.\nSpeaker A: But if you want to use your video recorder or something else, you should use an advanced option.\nSpeaker B: Yeah.\nSpeaker B: But the news flash also stated that it should control only one device.\nSpeaker B: So only your television.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: So it's very easy.\nSpeaker B: Now, yeah.\nSpeaker A: So there are not extra options in this case.\nSpeaker A: Okay.\nSpeaker A: Also, yeah.\nSpeaker A: We have to make it fashionable.\nSpeaker A: Yeah.\nNone: You want to set it before.\nSpeaker A: Yeah.\nSpeaker A: The basic functions.\nSpeaker A: Yeah.\nSpeaker A: Only use the extra functions if they are really needed.\nSpeaker C: Yeah.\nSpeaker C: So maybe you can hide them or something.\nSpeaker B: Yeah.\nSpeaker B: Well, what we can do with the screen is all the configuration options.\nSpeaker B: You can put that in the screen.\nSpeaker A: Yeah.\nSpeaker B: Screen menu to do that.\nSpeaker B: And then the basic function just under defies itself.\nSpeaker B: So it looks very simple and all the advanced features are hidden in the screen with a clear menu.\nSpeaker C: Yeah.\nSpeaker C: Uh, functionality is the screen.\nSpeaker C: What does the screen do?\nSpeaker C: All right.\nSpeaker B: All right.\nSpeaker C: It's low on power.\nSpeaker C: What are they?\nSpeaker B: So what does the screen do?\nSpeaker C: They said they needed it.\nSpeaker C: What does it do?\nSpeaker C: What do they want?\nSpeaker A: For the advanced functions, I think.\nSpeaker A: Yeah.\nSpeaker B: Well, it didn't say what they want to do with the screen.\nSpeaker B: My guess is pretty handy for advanced functions.\nSpeaker B: Right.\nSpeaker B: Searching for channels.\nSpeaker B: Yeah.\nSpeaker B: Searching for channels.\nSpeaker B: Programming them.\nSpeaker A: Nevermind.\nSpeaker A: I'm almost finished.\nSpeaker A: Okay.\nSpeaker A: We have to watch out for the, if we make it very fashionable, the functionality will go down.\nSpeaker A: So we have to make a compromise between functionality and fashion.\nSpeaker A: Complement them for them.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: That was the end.\nSpeaker A: Yeah.\nSpeaker D: Okay.\nSpeaker D: Well, my presentation is a bit sucky.\nSpeaker D: Well, you can improvise, right?\nSpeaker D: Yeah, a little bit.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker D: No.\nSpeaker C: Do you think there are requirements?\nSpeaker D: Yeah, I think that will be it.\nSpeaker D: No.\nSpeaker B: So we can go for it.\nSpeaker D: That was, you know, you know.\nSpeaker D: Let me check.\nSpeaker C: It's not really English.\nSpeaker D: I know.\nSpeaker C: The kickoff.\nSpeaker C: Our working design.\nSpeaker C: We got it.\nSpeaker B: So we can go for the look and feel of the left example and then a screen on top of it.\nSpeaker C: Here you go.\nSpeaker D: All right.\nSpeaker D: How do I skip pages?\nSpeaker C: Just press the keys.\nSpeaker D: Yeah.\nSpeaker D: All right.\nSpeaker D: Yeah.\nSpeaker D: Well, I was working on this before I got my information.\nSpeaker D: So I was just working off the top of my head using my common knowledge about remote controls and, well, the info on the website, which came to it.\nSpeaker D: So I didn't really know what kind of functions we had to put into it yet.\nSpeaker D: This is basically an overview of what we discussed in our last meeting.\nSpeaker D: Those were my starting points.\nSpeaker D: I was working on a, yeah, on a schedule and I was supposed to do it like this, but, yeah.\nSpeaker D: Then the information came and it was kind of exact with all the steps in the remote control that I had to follow.\nSpeaker D: So I was trying to organize them for myself and then make the design, the design, the actual design, but I never came around to do that.\nSpeaker D: So I'm not really sure what I'm supposed to say about it.\nSpeaker D: I mean, everything speaks for itself, I guess.\nSpeaker D: I mean, you press a button, it sends a signal to a chip, which translates it into infrared signal of certain spatial frequency.\nSpeaker D: Yeah, frequency.\nSpeaker D: And our temporal frequencies, actually, and then through a transformer, the signal gets boosted and then sent to the receiver on the TV to feel translated into the function.\nSpeaker D: Yeah, well, this was actually all I got around to do.\nSpeaker D: I mean, I don't know if I'm too slow for this stuff, but then work harder.\nSpeaker A: Should we make a list of all the functions we want to...\nSpeaker B: Yeah, we want to incorporate into it.\nSpeaker C: Okay, for those that didn't see yet, the basic new requirements of the management were no-time text, only for TV.\nSpeaker C: It should be designed for a user group below 40, but I don't think it's wrong if we can target the current customer group as well.\nSpeaker C: And the corporate identity should be clear in the design of the remote control.\nSpeaker C: And we have to decide on the functions and on the...\nSpeaker C: Let's see, what was it?\nSpeaker C: The target group.\nSpeaker C: We have to make me clear what that is.\nSpeaker B: The target group of users?\nSpeaker B: Yeah, it uses.\nSpeaker B: Because it says below 40, I mean...\nSpeaker B: Yeah, I think it's easy, but...\nSpeaker B: Yeah, or male and female.\nSpeaker B: Or in just...\nSpeaker B: No, it's below 40.\nSpeaker B: The marketing research started on the age of 16, 16 to 25, 25 to 35, 35 to 45, something like that.\nSpeaker C: So below 40 is okay, but we need an lower level in which to...\nSpeaker C: How do you miss?\nSpeaker C: So is it from 16 to 40?\nSpeaker C: Is it from 20 to 40?\nSpeaker B: Yeah, it's from 16 to 40.\nSpeaker B: I guess people of 16 are quite used to technology.\nSpeaker B: So they adapted pretty soon, I guess.\nSpeaker A: If we have a larger public, we have more options to sell our product.\nSpeaker D: Yeah, so what we might want to do is...\nSpeaker D: Yeah, have customized the screen functions, if you know what I mean.\nSpeaker D: Younger users are more acquainted with these kind of machines, they can use more advanced functions, but maybe elderly people don't understand it so well.\nSpeaker D: So they need a simpler remote.\nSpeaker D: And that you can choose what the design displays or whatever.\nSpeaker A: That's right, I wanted to make two different groups of functions.\nSpeaker A: The simple functions for the whole public and the advanced options for the younger people who are more experienced.\nSpeaker B: But they all incorporate it in the screen or just on the remote itself.\nSpeaker A: I should put the advanced options onto the screen display.\nSpeaker B: Yeah, like you have a visual of the brightness with a sliding bar and a bus and stuff like that for sound and visuals.\nSpeaker C: You should have a menu for all the issues.\nSpeaker B: Make a drawing.\nSpeaker B: Select.\nSpeaker B: That's okay.\nSpeaker C: That's very big.\nSpeaker C: Yeah, we get there.\nSpeaker B: Yeah, well, this is basically a remote.\nSpeaker B: Usually the power button is on top, I guess.\nSpeaker C: It should fit right into your hand.\nSpeaker B: Right.\nSpeaker B: Most people are right-handed, so maybe less than a special decision, but okay.\nSpeaker B: If you put it like here or something, I don't know.\nSpeaker B: Then you can put a screen on a mobile phone, also on top, I guess.\nSpeaker A: Yeah, but if you are using the normal functions, the basic functions, you normally do this.\nSpeaker C: So it should be a little bit longer so that you can put your thumb, also reach the middle, or you should screen at the bottom.\nSpeaker C: Then you can see it very clearly.\nSpeaker B: Okay, so you have your hand and the screen is below and the buttons are in the middle.\nSpeaker B: For example, if you put the screen here, it's more about the functions now than the layouts.\nSpeaker B: It doesn't work too well.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker C: Right.\nSpeaker B: Okay, you get it.\nSpeaker B: For example, if you put all the...\nSpeaker B: Right.\nSpeaker C: You want the normal piece of paper and there's a 10?\nSpeaker B: Yeah.\nSpeaker B: Yeah, my piece is...\nSpeaker B: Kind of works.\nSpeaker B: Like if you have 10 buttons for all the channels, an ear button for flipping.\nSpeaker B: Yeah.\nSpeaker A: And the following?\nSpeaker B: Yeah.\nSpeaker B: That's usually like here, here, here, here.\nSpeaker B: Yeah.\nSpeaker B: So we have up and down for the channels.\nSpeaker B: Left and right.\nSpeaker B: Left and right for the volume.\nSpeaker C: Those can also be used for the menu.\nSpeaker D: Yeah, exactly.\nSpeaker D: Yeah.\nSpeaker D: But this is really your department that we need just functional displays.\nSpeaker B: Yeah.\nSpeaker B: And for a cursor, this is function.\nSpeaker B: And you...\nSpeaker B: Yeah.\nSpeaker B: If you can use the...\nSpeaker B: Most of the time you have one button in the middle.\nSpeaker B: It says menu.\nSpeaker B: Yeah.\nSpeaker B: And then if you press it, the screen gets activated and then you can use these buttons to scroll up and down.\nSpeaker B: And left and right to go into a function.\nSpeaker B: Yeah.\nSpeaker B: And then just...\nSpeaker B: Most of the time this menu button is also like, okay, to confirm a kind of action.\nSpeaker B: So you scroll into it, okay, you select a function like base.\nSpeaker B: You just adjust it with these two buttons.\nSpeaker B: Yeah.\nSpeaker B: Then okay to confirm.\nSpeaker B: And then on each screen, this should be an option to go back to an upper level.\nSpeaker B: And then finally say, okay, exit or one button to exit it.\nSpeaker B: Yeah.\nSpeaker B: In one time, I don't know that's not really my department.\nSpeaker B: That's more your department too.\nSpeaker A: It's a logo on the remote control.\nSpeaker B: Yeah.\nSpeaker B: But it should be if the screen is here, then the logo should be on top.\nSpeaker B: Yeah.\nSpeaker B: I mean, it's...\nSpeaker C: Yeah, but that depends on where you put the screen.\nSpeaker C: Yeah.\nSpeaker C: It's essential that there is a screen.\nSpeaker B: Yeah.\nSpeaker B: I mean, the position of the screen is also more essential than...\nSpeaker B: I mean, we look where we have space left and then put the logo over there.\nSpeaker C: And for the speech recognition part, if we want to incorporate that, we need the microphone.\nSpeaker C: Yeah.\nSpeaker B: So it should be...\nSpeaker B: I mean, if you have it in your hand here, it should be on top somewhere.\nSpeaker B: Yeah.\nSpeaker D: Why do we want to put the display in the bottom?\nSpeaker C: No, that's not sure.\nSpeaker C: That's not sure, but...\nSpeaker B: Maybe.\nSpeaker B: Maybe.\nSpeaker D: Because if you use the functions, your hand will block the display.\nSpeaker A: Yeah, okay.\nSpeaker A: But that's only for the basic function.\nSpeaker A: If you are going to use the iPhone option, you are going to press the menu button and then...\nSpeaker B: Yeah.\nSpeaker B: I mean, the screen on top looks more...\nSpeaker B: Normal.\nSpeaker B: Logical to me also because people use mobile phones and they also have a screen on top.\nSpeaker A: Okay.\nSpeaker B: Yeah.\nSpeaker B: So you just have to reach a little bit for the power button.\nSpeaker B: Yeah.\nSpeaker B: I mean, if you grab it...\nSpeaker B: Once it's on, it's on.\nSpeaker B: But most of the times, if a TV is on standby, people just press the check.\nSpeaker B: Yeah, we put it on top.\nSpeaker B: So we put this on top and then make the corporate logo like over here, R-R.\nSpeaker B: And the microphone, I mean, it can be very small if you look at your mobile phone.\nSpeaker B: So these are some stripes, little holes.\nSpeaker D: Maybe on the top or even on the side.\nSpeaker B: Yeah.\nSpeaker B: Maybe on the side, I mean, if the microphone is good...\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker B: So on the top is better.\nSpeaker D: But if you hold the remote like this, if you put it on the top on the side, I don't know.\nSpeaker D: Yeah.\nSpeaker D: It should be able to...\nSpeaker C: Yeah.\nSpeaker C: It doesn't...\nSpeaker D: It doesn't...\nSpeaker D: Okay, well, leave this up to you.\nSpeaker B: Yeah, it doesn't matter that much.\nSpeaker B: So...\nSpeaker B: But the screen is on top.\nSpeaker B: Which functions did we have left?\nSpeaker B: I mean, this is basically numbers for you.\nSpeaker B: For you.\nSpeaker B: For you.\nSpeaker B: Channel up and down.\nSpeaker B: Channel up and down.\nSpeaker C: Up there.\nSpeaker C: Outphones options.\nSpeaker C: Yeah.\nSpeaker C: Maybe we should decide what outphones options we want to put in the LCD screen.\nSpeaker B: Yeah.\nSpeaker B: That's a good one.\nSpeaker D: Yeah.\nSpeaker D: So we needed to integrate those sound and image options, right?\nSpeaker D: Yeah.\nSpeaker C: Like...\nSpeaker D: Sound.\nSpeaker D: Sound.\nSpeaker D: So we need kind of an equalizer.\nSpeaker B: Yeah.\nSpeaker B: Equalizer.\nSpeaker B: So if you have sound.\nSpeaker B: But not too advanced.\nSpeaker B: Most TVs use only treble and bass.\nSpeaker B: Yeah.\nSpeaker B: Treble and middle bass or something.\nSpeaker B: Yeah.\nSpeaker B: It's pretty hard to write.\nSpeaker B: And...\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: But you have sound?\nSpeaker C: Yeah.\nSpeaker B: Sound.\nSpeaker B: Yeah.\nSpeaker B: You have a digital...\nSpeaker B: Better write it down over there.\nSpeaker B: So we have sound.\nSpeaker B: Just a...\nSpeaker B: Secret.\nSpeaker B: Yeah.\nSpeaker B: Yeah.\nSpeaker B: Sound.\nSpeaker B: And then...\nSpeaker B: Within sound, I guess, treble and bass.\nSpeaker B: Yeah.\nSpeaker D: Yeah.\nSpeaker D: The mono stereo option.\nSpeaker B: Yeah.\nSpeaker D: And also...\nSpeaker D: There was something else also.\nSpeaker D: And then...\nSpeaker D: Pitch.\nSpeaker B: Pitch.\nSpeaker B: Yeah.\nSpeaker B: But pitch isn't that...\nSpeaker B: Yeah, that's the height.\nSpeaker B: Yeah.\nSpeaker B: The frequency is down.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: Why would you use that?\nSpeaker B: Yeah.\nSpeaker B: Isn't this...\nSpeaker D: It depends on the signal of the program you're watching.\nSpeaker C: Yeah.\nSpeaker C: And also the tuning part?\nSpeaker B: Programming part.\nSpeaker B: Yeah.\nSpeaker B: Channel programming.\nSpeaker B: So we have sound, yeah.\nSpeaker B: Channel programming.\nSpeaker B: And...\nSpeaker D: Yeah.\nSpeaker D: In the functionality of the...\nSpeaker D: Telefician.\nSpeaker D: No, no.\nSpeaker D: Of the remote.\nSpeaker D: Do we want the buttons to make sound when you press them?\nSpeaker D: As a confirmation?\nSpeaker D: Or whatever, you know?\nSpeaker D: I don't know.\nSpeaker B: I think it gets annoying.\nSpeaker B: I mean, most mobile phones use that in the beginning.\nSpeaker A: But...\nSpeaker A: You can disable...\nSpeaker D: Under the advanced option.\nSpeaker D: You can put those in.\nSpeaker A: The...\nSpeaker A: The telefician itself has also the options with brightness and screen color, etc.\nSpeaker A: So...\nSpeaker A: Contrast.\nSpeaker B: Yeah.\nSpeaker B: But people don't want to sit underneath...\nSpeaker B: Underneath in front of the telefician which is usually only three buttons and it's very hard to.\nSpeaker B: Contrast.\nSpeaker B: Yeah.\nSpeaker B: Contrast and brightness.\nNone: Point.\nSpeaker B: Yeah.\nSpeaker B: That's the most used, I guess, if you don't get your money turned.\nSpeaker B: Yeah, well, I guess this were the only ones, I guess.\nSpeaker B: Okay.\nSpeaker B: It's easy.\nSpeaker B: Yeah.\nSpeaker D: But so we have...\nSpeaker D: We have TV options which is all this.\nSpeaker D: The sound...\nSpeaker C: The options.\nSpeaker D: The sound image.\nSpeaker D: And you have in the...\nSpeaker D: The...\nSpeaker D: Indeed.\nSpeaker D: The remote control options.\nSpeaker D: Yeah.\nSpeaker D: So we need two menus kind of thing.\nSpeaker C: Yeah.\nSpeaker C: You have basically a button menu which you can use directly.\nSpeaker C: Uh-huh.\nSpeaker C: According to the opris port and the LCD options.\nSpeaker C: Are activated based on some software options that's communicated with the infrared.\nSpeaker C: Yeah.\nSpeaker B: With the chip and then trans...\nSpeaker B: I mean...\nSpeaker C: So you have an additional processor in.\nSpeaker B: Yeah.\nSpeaker B: Well...\nSpeaker B: Yeah.\nSpeaker B: We have power button, whether that's present.\nSpeaker B: Channel volume selection present.\nSpeaker B: Numbers present.\nSpeaker B: Yeah.\nSpeaker B: Audio settings, monosterior pitch, bass, treble, screen settings, brightness and color.\nSpeaker C: Yeah.\nSpeaker C: Yeah.\nSpeaker B: Contrast this.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker C: Color and brightness.\nSpeaker B: Yeah.\nSpeaker B: Um...\nSpeaker B: And which is a channel setting.\nSpeaker B: So channel programming.\nSpeaker B: So you have an option to start scanning all the frequencies.\nSpeaker B: Yeah.\nSpeaker B: And when it encounters one.\nSpeaker B: Well, it shows on your TV.\nSpeaker B: And then you can...\nSpeaker B: Um...\nSpeaker B: Uh...\nSpeaker B: And then you can select a number in your remote.\nSpeaker C: So the channel program is auto-seek.\nSpeaker C: Yeah.\nSpeaker B: Auto-seek.\nSpeaker B: Uh...\nSpeaker C: Name a channel.\nSpeaker C: Or...\nSpeaker B: Well, most TVs automatically display the name which they get through the cable.\nSpeaker C: Oh, they get the ultimate name.\nSpeaker C: Yeah.\nSpeaker B: So you only have to choose the position on your...\nSpeaker B: It only has to match the channel frequency on your TV.\nSpeaker B: Yeah.\nSpeaker B: With the position on your TV.\nSpeaker B: And so your remote.\nSpeaker C: Yeah.\nSpeaker C: But can you also say I want Veronica on the channel number 5?\nSpeaker B: If you already programmed it.\nSpeaker B: Yeah.\nSpeaker B: If you want to move it.\nSpeaker B: Yeah.\nSpeaker B: That should be possible too.\nSpeaker B: How do you call that?\nSpeaker B: Yeah.\nSpeaker B: How do you call that?\nNone: Mmm.\nSpeaker B: Well, you should be able to browse through a list or something which displays all the values, all the channels which are possible.\nSpeaker B: I mean like 1 to 30 or 4.9 here or whatever.\nSpeaker B: Just browse through it and then...\nSpeaker B: Um...\nSpeaker B: Some kind of way.\nSpeaker B: See if it is programmed.\nSpeaker C: Swap channels?\nSpeaker C: Yeah.\nSpeaker B: Swap channels.\nSpeaker C: Swap is good option.\nSpeaker C: Okay.\nSpeaker B: So most of the time, if you swap it.\nSpeaker B: Let's say for example you have RTL5 on channel 5.\nSpeaker B: And Veronica on channel 8 and you want to move Veronica to channel 5.\nSpeaker B: Most of the time you overwrite the previous one.\nSpeaker C: Well, that's sort of...\nSpeaker C: Yeah.\nSpeaker B: It's pretty...\nSpeaker B: Yeah.\nSpeaker B: I mean you have to look on the menu.\nSpeaker D: Those are working.\nSpeaker D: On the TV.\nSpeaker C: Also.\nSpeaker D: He only has to figure out how it has to look.\nSpeaker C: Yeah, but also which buttons you have to press to get a certain result.\nSpeaker C: Yeah, okay.\nSpeaker C: The working design was to specify how the physical interaction between the components was.\nSpeaker C: I...\nSpeaker C: Yeah.\nSpeaker D: It treated your homework.\nSpeaker D: But um...\nSpeaker D: Yeah.\nSpeaker B: Or is it too hard to just do it on your remote to program the channels?\nSpeaker D: No.\nSpeaker C: It should be able to do a thing with the television is difficult.\nSpeaker C: But that's not our part.\nSpeaker C: No.\nSpeaker C: We don't have to design protocol.\nSpeaker B: That's true.\nSpeaker B: That's true.\nSpeaker B: Thank God.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: So we have to figure out a way to do that easily.\nSpeaker B: I mean the auto-secret is not a problem.\nSpeaker B: You are a counter-channel and then it just says on which number do you want to save this?\nSpeaker B: And you just press the number on your remote and then say confirm.\nSpeaker B: And it's safe.\nSpeaker B: It's easier.\nSpeaker B: It's harder to, if you have already programmed it, to swap.\nSpeaker C: So we have to think of something for that.\nSpeaker C: The user interface design can design a menu for all the use functions.\nSpeaker B: Kind of structure into layers.\nSpeaker A: I mean you make a map with all the...\nSpeaker A: Yeah, because it's now...\nSpeaker A: There are also documents.\nSpeaker C: Yeah.\nSpeaker C: Yeah, I can.\nSpeaker C: I'll just try to reorganize things.\nSpeaker C: So you design the basic function menu for the LCD screen.\nSpeaker B: And the layout of the thing itself.\nSpeaker B: I think the layout of the screen.\nSpeaker C: I think you can concentrate more on the bit and placement and the placement of the screen itself on the...\nSpeaker A: Isn't that more than the...\nSpeaker A: This is the interface.\nSpeaker C: No, I think there's more in...\nSpeaker D: All the functional aspects of the...\nSpeaker D: Maybe more than...\nSpeaker D: I think I am in my department.\nSpeaker D: I have to know what it has to do.\nSpeaker D: So if you want to...\nSpeaker D: You know, the changing the channels kind of thing is...\nSpeaker D: I have to integrate that in the design.\nSpeaker C: So you lay out your function.\nSpeaker C: Yeah.\nSpeaker C: One function, okay.\nSpeaker D: I think that's a good example.\nSpeaker A: I have to make the menu layout also.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker B: I guess so.\nSpeaker C: I think that's your department.\nSpeaker C: Yeah.\nSpeaker C: We have to kind of work together.\nSpeaker D: If I make the...\nSpeaker D: The menu, like...\nSpeaker D: I have to state which function has to be in the menu and then you have to decide it in the way that is your friend.\nSpeaker A: You are going to make a list of what functions are going to be on the menu screen.\nSpeaker A: And I'm going to make a nice menu.\nSpeaker A: Yeah.\nSpeaker B: With some layers in it.\nSpeaker C: Yeah.\nSpeaker C: And also...\nSpeaker C: Make clear which buttons to press to get certain results.\nSpeaker C: Yeah.\nSpeaker C: Because that's always difficult to every device that's on...\nSpeaker B: Yeah.\nSpeaker B: I guess this button, the okay menu, okay.\nSpeaker B: Or you can incorporate two different buttons like you have on your mobile phone.\nSpeaker B: So this is only to get in the menu or to exit it.\nSpeaker B: And then I want to confirm.\nSpeaker B: I want to go one step back.\nSpeaker B: So it's like the the say button on your mobile phone if you have a Nokia or like that or the no button.\nSpeaker B: To go one step back, it's only two extra buttons.\nSpeaker B: If it's very clear that they are for the screen.\nSpeaker A: I think we have to make two groups.\nSpeaker A: One group for the display and one group for the basic functions.\nSpeaker C: Yeah, but they're incorporated.\nSpeaker A: Yeah, because this is used for both.\nSpeaker A: Yeah.\nSpeaker A: But maybe that's not...\nSpeaker A: If you're zapping with your remote control, you can press the menu button and then you are suddenly into the...\nSpeaker B: To your screen.\nSpeaker B: Okay, so you should leave the menu button out of here.\nSpeaker B: And just put it under the screen.\nSpeaker A: Just group.\nSpeaker A: Yeah, we make these advanced functions and these basic functions.\nSpeaker A: We make a line between them.\nSpeaker B: But we should place the screen on top, right?\nSpeaker A: Well, oh yeah, okay.\nSpeaker A: We swap the...\nSpeaker C: You use the port.\nSpeaker C: Yeah, you just find out and then the headlady is out.\nSpeaker C: Do we do the extra two buttons or not?\nSpeaker D: I think you should.\nSpeaker D: It's easier if you put too many functions in one button, it gets confusing.\nSpeaker A: Yeah, okay, that's true.\nSpeaker A: So we have a menu button and a...\nSpeaker C: Okay, in Beck.\nSpeaker C: Yeah.\nSpeaker B: Okay, in Beck.\nSpeaker B: Yeah, our confirm and back.\nSpeaker C: The four arrows.\nSpeaker B: Yeah.\nSpeaker C: But those are still doubly used.\nSpeaker B: Should we save this picture or, you know, what it looks like?\nSpeaker A: Yeah, I'm not clear about the extra two buttons.\nSpeaker A: We have a menu button.\nSpeaker B: Yeah, yeah.\nSpeaker B: Just to activate the screen.\nSpeaker C: Yeah, okay.\nSpeaker C: So...\nSpeaker C: So you button access the menu in the LCD screen.\nSpeaker B: And then with these buttons, you can navigate.\nSpeaker B: You navigate.\nSpeaker C: Okay.\nSpeaker C: But you can also navigate the channels and the volume.\nSpeaker A: Okay, so that's not...\nSpeaker A: Yeah, those are both... both.\nSpeaker D: Hey, is it interesting for users if we put lead lights under the buttons so that if you...\nSpeaker D: Press on the menu function, that only the buttons that are used for the menu are...\nSpeaker D: Liden up, yeah.\nSpeaker B: Yeah.\nSpeaker B: That's very good ID.\nSpeaker C: Five minutes.\nSpeaker B: All right.\nSpeaker B: Yeah, that's a good ID because then it becomes clear which buttons are active now which you can use.\nSpeaker D: Yeah.\nSpeaker D: And then it's also easier to integrate several functions in one button.\nSpeaker D: Yeah.\nSpeaker D: Sure.\nSpeaker D: Okay.\nSpeaker B: Yeah.\nSpeaker C: So...\nSpeaker B: Anything else?\nSpeaker B: Those buttons are lit up.\nSpeaker B: But just one thing.\nSpeaker B: Should we use those two then?\nSpeaker B: Or only this two to scroll?\nSpeaker B: And then use the two functional buttons to confirm to go into something.\nSpeaker B: Oh no, we have to use this to adjust...\nSpeaker D: Or some some bars.\nSpeaker D: Or keep it optional because maybe you can... you know, go left, right and up and down.\nSpeaker B: Yeah.\nSpeaker B: Maybe we should use this also as an okay button still.\nSpeaker B: And then just only a back button.\nSpeaker C: Well, we have those buttons.\nSpeaker D: The problem with the okay button in the middle is sorry.\nSpeaker D: Yeah.\nSpeaker D: Sorry.\nSpeaker D: It's...\nSpeaker D: If you're pressing up and down, you can easily press it okay once you're not already at your...\nSpeaker A: Or maybe make one okay button...\nSpeaker A: And one back.\nSpeaker A: And one navigation button.\nSpeaker A: Yeah.\nSpeaker A: With the channel and the volume.\nSpeaker A: Yeah.\nSpeaker A: A multifunctional navigation button.\nSpeaker C: Okay.\nSpeaker C: That's what we decided.\nSpeaker C: Yeah.\nSpeaker C: Okay.\nSpeaker D: Okay.\nSpeaker D: So...\nSpeaker C: You want to close down?\nSpeaker C: Yeah.\nSpeaker C: I want to close down.\nSpeaker C: Sorry.\nSpeaker C: Sorry.\nSpeaker C: Not because it doesn't like you.\nSpeaker C: But we have lunch break already.\nSpeaker C: And then we can work for 30 minutes.\nSpeaker C: We'll have to write very fast to make the minute.\nSpeaker C: Yeah.\nSpeaker C: And then we'll see.\nSpeaker D: Alright.\nSpeaker D: How long is the lunch break?\nSpeaker C: I don't know.\nSpeaker C: We have to ask all of me.\nSpeaker A: Do we have to write down our stuff now?\nSpeaker A: First lunch break.\nSpeaker A: I believe that's first lunch break.\nSpeaker A: Everything in my head now.\nSpeaker A: Yeah.\nSpeaker A: There you go.\nSpeaker A: Okay.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker A: Yes.\nSpeaker B: I think you can put a laptop back in the room.\nSpeaker C: Yeah.\nSpeaker C: Yes, sir.\nSpeaker C: Time pressure.\nSpeaker C: Okay.\nSpeaker B: Yeah, it's a lot of pressure.\nSpeaker C: Yeah.\nSpeaker D: Sorry for my not finished presentation.\nSpeaker D: Sorry.\nSpeaker D: Yeah.\nSpeaker D: We'll kick your ass later.\nSpeaker B: Bring it on.\nSpeaker B: I don't know if it works.\nSpeaker B: You saved it.\nNone: That's safe.\nNone: Out of my head.\nSpeaker B: Yeah.\nSpeaker B: Okay.\nSpeaker B: We'll see.\nSpeaker B: Just shoot back my laptop.\nSpeaker B: Smart board.\nSpeaker B: Alright.\nSpeaker B: You can use it.\nSpeaker B: Yeah.\nSpeaker B: You can open it with picture preview or stuff like that.\nSpeaker A: Some maps with all the data we have.\nSpeaker C: Yeah.\nSpeaker C: I tried to organize it by these three.\nSpeaker A: Yeah.\nSpeaker A: Yeah.\nSpeaker A: It's just my own map.\nSpeaker A: Yeah.\nSpeaker C: I don't really mind.\nSpeaker C: I just put the minutes here.\nSpeaker C: We'll see.\nSpeaker A: But you've got some actual information.\nSpeaker C: Yeah.\nSpeaker C: That's in the functional design.\nSpeaker C: Okay.\nSpeaker C: Yeah.\nSpeaker C: It's just basically what I just showed.\nSpeaker A: Where do you get?\nSpeaker A: Yeah.\nSpeaker A: I got it by the only one.\nSpeaker A: Yeah.\nSpeaker C: Okay.\nSpeaker B: I'm glad you're here.\nSpeaker C: I don't do it.\nSpeaker B: Okay.\nSpeaker B: Alright.\nNone: Yeah.\nSpeaker C: Make me proud.\nSpeaker D: I'll try to.\nSpeaker D: So first we have a lunch break now.\nSpeaker B: Yeah.\nSpeaker C: Alright.\nSpeaker C: I believe so.\nSpeaker C: Oh, just ask.\nSpeaker D: Mm-hmm.\nSpeaker D: I don't know where she's at though.\nNone: Okay.\n"}, {"meeting_id": "ES2014b", "summary": "The Project Manager gave a brief review of the team's last meeting, then Industrial Designer, User Interface and Marketing gave their presentation each. Industrial Designer's presentation is about the components and system of the remote control. User Interface gave the presentation about the design of the remote. The team agreed that there should be 17 buttons on the front of the remote and two on the back. Marketing's presentation is about the research into the functional requirements that people want out of their remote control. The team agreed that they could add a clap-back function to the remote.", "dialogue": "None: and Right.\nSpeaker A: So, so, where's the PowerPoint presentation?\nSpeaker A: So, I have PowerPoint.\nSpeaker A: Right.\nSpeaker A: So, right.\nSpeaker A: Okay, so we've got So, we've got new project requirements.\nSpeaker A: So, basically, we've got three things.\nSpeaker A: We've got 40 minutes in which to fill this meeting to discuss the various options.\nSpeaker A: Three presentations.\nSpeaker C: We have a, I guess, we have a presentation.\nSpeaker A: I see. Right. That's nice. No one from each of you.\nSpeaker A: New project requirements. So, do we want to do the presentation first or do we want to?\nSpeaker A: I got three things, basically relating to the amount being only for TV.\nSpeaker A: We discussed that last time.\nSpeaker A: And in actual fact, I was pretty well put with any of these.\nSpeaker A: In fact, actually, we're going to be forced over in a sense.\nSpeaker A: We've got teletext outdated.\nSpeaker A: Did you get any information on that?\nSpeaker A: We didn't. Right.\nSpeaker A: And the corporate image was the final thing.\nSpeaker A: So, I got that in email form.\nSpeaker A: Right. Okay. So, I guess if we go ahead with the, with the three presentations.\nSpeaker A: So, we'll start with yourself on the basis that.\nSpeaker A: Sorry. Yep.\nNone: Oh.\nSpeaker D: I'm a question.\nSpeaker C: Yeah.\nSpeaker B: Usually a little thing at the top, right?\nSpeaker B: Oh, hang on.\nSpeaker C: This is good.\nSpeaker C: Oh, I got it.\nSpeaker C: I passed it all five times now.\nSpeaker C: Okay, let's be one.\nSpeaker C: Okay, let's go again.\nSpeaker A: Where's it going?\nSpeaker C: I should be at this point.\nSpeaker C: Okay, let's go again.\nSpeaker C: Okay, I'm going to be looking at the working design of the remote control.\nSpeaker C: I've just got three sections for this.\nSpeaker C: The research I made on the remote control itself.\nSpeaker C: And then that involves the components required in the systems design of the actual device.\nSpeaker C: So having researched the existing models within the market, my primary research of the internet, I've established what the components required for the remote control to function at HDR.\nSpeaker C: And then also the methods in which these components interact together for the remote to actually do what you wanted to do and how it connects with the television.\nSpeaker C: The basic components are an energy source, which I guess, in most of these models will be a battery supply, where for that will be two batteries, four batteries, and it may vary.\nSpeaker C: We then have a user interface, which is basically the buttons on the actual remote.\nSpeaker C: The various functions you saw, changing channel, channel left and down one, is that that.\nSpeaker C: There's also a chip inside the remote, which does all the computer type things.\nSpeaker C: And then a sender, which is usually found in infrared device, which sends a signal to the actual television.\nSpeaker C: And the nice part is the receiver, which is important in the system, but it's not actually part of the remote itself, because that's obviously found in the television.\nSpeaker C: And I've actually drawn the board because it was a little tricky on PowerPoint to get into working, so let's go see if it.\nSpeaker C: Do we have a cloth to wipe this down with?\nSpeaker A: There's the rubber on the right, I think.\nSpeaker A: Get rid of the magic.\nSpeaker C: So we start off with a power supply, which is probably going to be the battery.\nSpeaker C: This is how the basic system works.\nSpeaker C: So we have a new brush, and then sends the appropriate message to the sender.\nSpeaker C: So that's the emotion that self-class composes the mode and how to work together. This is the user interface.\nSpeaker C: This is the chip itself, which then sends the sender to the infrared.\nSpeaker C: And then on the set, the thing we have on the television, we have a receiver.\nSpeaker C: And the sender sends a message to the receiver.\nSpeaker C: So the top is the power source, yes?\nSpeaker C: So moving on to personal preferences, as I've said, the battery seems the best option for the actual remote, just because of the size you don't know, a cable attached to the remote, it's not really remote.\nSpeaker C: And then the sender in the infrared has been used quite successfully.\nSpeaker C: If the battery is on reasonable power, they will seem to work for any way to have to be pointed directly at the television itself.\nSpeaker C: The battery is the in the sender.\nSpeaker A: OK, so now more design.\nSpeaker D: Mine's not quite as complicated as all that.\nSpeaker A: That's what we like to hear.\nSpeaker D: OK, so I'm going to talk a bit about the technical functions design.\nSpeaker D: So the basic method of this is to send a signal from the remote to the television set so that the desired function is performed.\nSpeaker D: An example of the function could be to change the volume up or down, so obviously you need two different buttons for that.\nSpeaker D: To change the channel, either by pressing the number that you want or by channel up or down, to switch the television on or off, maybe a standby button.\nSpeaker D: Here are two example results.\nSpeaker D: By the look of it, the both have kind of play and fast forward rewind functions, so I think the incorporate a kind of video function, which we won't have to worry about.\nSpeaker D: But as you can see, the left remote is quite busy looking, quite complicated, whereas the right remote is much simpler.\nSpeaker D: It looks much more user friendly.\nSpeaker D: So my personal preference would be the right remote. So it's got nice big buttons, it's got a very limited number of buttons.\nSpeaker D: They're nice, kind of clearly labeled.\nSpeaker D: I like the use of the kind of symbols, like the triangles and squares and arrows, as well as the words on the play functions and all that.\nSpeaker D: So it's very user friendly. It's got a little splash of color. Could maybe do a small color.\nSpeaker A: Well, there's a couple of things there.\nSpeaker A: We have to remember that we have our own logo and color scheme.\nSpeaker A: So basically we'd have to be putting that on the product.\nSpeaker D: Do we get to see that?\nSpeaker A: I haven't as yet.\nSpeaker A: But I got an email that basically said to make sure that whatever device we come up with at the end of the day had to incorporate the corporate color and slogan.\nSpeaker A: So I'm guessing that I noticed on the bottom there it's got what's that EP-O-G-E-E that might be the corporate color scheme.\nSpeaker A: Although the only color I could see in that is the red.\nSpeaker C: Would you be able to get rid of the X-Men's here, the sort of circular section, setting three for a video as well?\nSpeaker C: We could dispense with that a little bit as well and just get it down to just the numbers in the volume.\nSpeaker C: What we might have seen in section like all the examples in this one.\nSpeaker C: Yeah, just a little bit is that I think that's still in my video.\nSpeaker C: Yeah.\nSpeaker D: All my parts were made with the video as well.\nSpeaker C: Yeah.\nSpeaker D: And I don't really think that I need nine numbers.\nSpeaker D: I mean, how often do you send me nine? I think just one six and then channel up and down.\nSpeaker A: Well, for general television purposes, obviously you have channels one to five at this point in time and we'd have to have some room for future such channels.\nSpeaker A: As people used to seeing us, we didn't have them.\nSpeaker A: Well, possibly.\nSpeaker A: But the other thing is that with the current expansion of channels and the process of taking place.\nSpeaker A: Certainly the button up and down.\nSpeaker A: But I mean, how many channels do we have to actual television channels do we have to prepare for?\nSpeaker A: I would have thought that the fact of it is forever expanding.\nSpeaker A: And the moment we've got, although you've got the five standard, you've got the BBC have come up with a further six.\nSpeaker A: And there's, I don't know exactly how many channels there are on when you take into account.\nSpeaker A: Sky and various other.\nSpeaker A: And various others.\nSpeaker A: So I would have thought that we wouldn't, you know, rather than.\nSpeaker A: Okay, if the time of flicking from one to other, but presumably it'll take a second because you have to be able to stop it.\nSpeaker A: Maybe you could have a fast forward on the on the channels that and then you could dispense with more otherwise.\nSpeaker A: You want, you want to get fairly quickly to the channel that you wanted.\nSpeaker D: Some remote type kind of favorite options where if you always pick from channel one to channel six.\nSpeaker D: If that's your favorite, you just bypass two to five.\nSpeaker A: I suppose in a sense you could have them if you've got a hundred channels.\nSpeaker A: Then if you had sort of an easy way of getting rather than having one to a hundred, you could go one to ten, ten to twenty and then have a second button to get you to the actual channel you want.\nSpeaker A: Then that would cut down your time.\nSpeaker D: I think a lot of them, like, keep on sky stuff, that would be true to one channel and then you'd have another remote for all of those channels.\nSpeaker D: I'd get a 55.\nSpeaker D: Whatever.\nSpeaker A: Okay.\nSpeaker A: Okay.\nSpeaker A: Number of buttons.\nSpeaker A: User friendly.\nSpeaker D: That's called nine to not really accessible.\nSpeaker C: I suppose you've got the last one which makes the talent to me.\nSpeaker C: It's like multiple, so you can make any number.\nSpeaker C: So with that we can buy from any problems with.\nSpeaker A: Well, that's true.\nSpeaker A: Yeah, you could get 50 by five and a zero or whatever.\nSpeaker A: That makes sense.\nSpeaker C: Does that facilitate having all the numbers you could have in you?\nSpeaker A: Yeah.\nSpeaker A: That's...\nSpeaker A: So what was the circular thing that you were?\nSpeaker D: I think that's just for a video.\nSpeaker D: So what wouldn't need any of that at all?\nSpeaker D: So we get...\nSpeaker A: If it's just for TV, which is what it is at the moment.\nSpeaker C: Whoa.\nSpeaker C: Should we get to 100 buttons?\nSpeaker C: We've got ten, eleven, twelve.\nSpeaker C: We've got fourteen that we need, I guess.\nSpeaker C: Which isn't really the team, I'd be quite used to making these guys for the whole thing.\nSpeaker A: Well, we've got this remote for TV only, otherwise Project would become too complex with which we'd endanger the time to market was one of the considerations.\nSpeaker A: I don't know.\nSpeaker A: Did you have that information being the marketing?\nSpeaker A: Or was I meant to give you that information?\nSpeaker B: I'm not sure.\nSpeaker B: I had some more information.\nSpeaker B: I had some more information.\nSpeaker B: Right, no, from the company.\nSpeaker A: Right, okay.\nSpeaker A: So basically, time to market seems to be important.\nSpeaker A: Therefore, speed of delivery.\nSpeaker A: We've only got about another four hours left.\nSpeaker D: Okay, so is everyone happy with that?\nSpeaker C: Yeah, okay.\nSpeaker D: Right, well, I asked the end of my presentation.\nSpeaker E: Please.\nSpeaker C: I think you'll just give us a second to...\nSpeaker A: Yeah, I think she's at twenty seconds to...\nSpeaker C: I'm sure we'll have her hands to play.\nSpeaker B: Okay, there we go.\nNone: Yeah, there we go.\nSpeaker B: Right, we've done some research into the functional requirements that people want out of their remote control.\nSpeaker B: And first off, we should state that the remote control is for controlling the TV.\nSpeaker B: And how do people use it?\nSpeaker B: We ask them which buttons were useful for them.\nSpeaker B: How does a remote control look and feel for them?\nSpeaker B: And what improvements would they like to remote control?\nSpeaker B: And we did that by giving them a questionnaire that we prepared and asking them to fill in the answers.\nSpeaker B: And three-quarters of them found that remote controls are ugly and that even high proportion would spend more for a fancier remote control.\nSpeaker B: And that, of all the buttons on the remote control, the setting buttons for the picture, picture and brightness and the audio settings, they weren't used very often at all.\nSpeaker B: People concentrating on the channel buttons and the volume buttons and the power buttons.\nSpeaker B: We also asked them about speech recognition for remote control.\nSpeaker B: And young people were quite receptive to this.\nSpeaker B: But as soon as we got sort of over about into 35 to 48, 45 age movement and older, people were quite so keen on speech recognition.\nSpeaker B: There was a lot more older people who didn't know whether they wanted it or not as well.\nSpeaker B: We also asked what frustrated people about remote controls.\nSpeaker B: And the number one frustration was that the remote was lost somewhere else in the room.\nSpeaker B: But they couldn't find it.\nSpeaker B: And the second biggest frustration was that if they got a new remote control, it was difficult to learn all the buttons and all the functions and to find your way around it.\nSpeaker B: Okay, so, what this is not a preference is from the marketing is that we need to come up with some sort of sleek, sort of good looking, high tech, the design which looks high tech basically.\nSpeaker B: And that we should come up with fewer buttons than most of the controls on the market.\nSpeaker B: And we should sort of concentrate on the channels and power and also volume.\nSpeaker B: So I think as Louise said, we could maybe come up with a menu, sort of an LCD menu for other functions on the remote control.\nSpeaker B: And that's worth thinking about.\nSpeaker B: And maybe we could think about speech recognition as well because some young people are perhaps the ones that are going to buy our new product if we aim it.\nSpeaker B: So, you know, sort of a high tech design.\nSpeaker B: That might be the market that we're looking for.\nSpeaker B: So, let me think about using speech recognition as well as a sign to control if it's lost in a room rather than sort of having it to speech recognition to change the channel.\nSpeaker B: Because there's a problem with that in that the television makes noise so it could end up talking to itself and changing its channels.\nSpeaker B: Okay.\nSpeaker B: And at the end of this slide, that's it.\nSpeaker A: So, what was that last bit there? About speech recognition.\nSpeaker A: Yeah, okay. But that was only for young people that preferred it.\nSpeaker B: Older people did.\nSpeaker B: Young people.\nSpeaker B: Yeah.\nSpeaker B: They said that they'd be interested in remote control, which offered that possibility.\nSpeaker B: And as you go through the age groups, people have got less and less interested in sort of a remote control that you can talk to.\nSpeaker C: I think it seems the technology would be quite advanced and it might end up casting more than a 1250 budget.\nSpeaker C: And possibly thinking about the weather mode being lost, we could have a nearby phone and you can ring it, maybe we can have some kind of sense, which just kept them away.\nSpeaker C: You can basically just go to the bus assistant between the two.\nSpeaker C: So, you can press the button, which is always kept in one place and then it goes to the other bus.\nSpeaker B: Wherever the map actually is. Yeah, we'd have to put two products together as well, which is, again, we'll do a bit of it.\nSpeaker D: There's key rings, but you can whistle out a clap-back, I can't remember.\nSpeaker D: Yeah.\nSpeaker D: And then the whistle back or something like that.\nSpeaker D: It's not reasonable.\nSpeaker D: It's not reasonable.\nSpeaker E: Mm-hmm.\nSpeaker C: So, I guess it's a little tough to say, or like the same technology could be put inside there.\nSpeaker A: Well, you're trying to avoid having a second product because obviously you could have a second product that gave you the only pitch, which would be, which would set the remote off to say, here I am sort of like, you know, without sound recognition.\nSpeaker A: But if you, I know, it's going to say a sharp noise, you know, a clapping of hand or whatever.\nSpeaker A: You want to try and just have the one product that, if...\nSpeaker C: Yeah, if we can have it in the actual amount, everything in one.\nSpeaker C: Yeah.\nSpeaker A: I don't know.\nSpeaker A: Talk about, I mean, obviously, if you've got voice recognition, then you can do it in that way because it will recognize the voice and you can give it a command, a sick command, whatever that happened to be.\nSpeaker A: Yeah.\nSpeaker A: But you can then go to the point, if you're not going with voice recognition then...\nSpeaker C: You can have an option to turn it off or...\nSpeaker A: Yeah.\nSpeaker A: So, any suggestion?\nSpeaker A: Well, any conclusions?\nSpeaker B: Would it take quite a while to sort of develop this feature recognition software in the remote control?\nSpeaker A: Well, if it does, then we can't.\nSpeaker A: It's that simple because we've got three primary requisites from a new meal that was sent to me.\nSpeaker A: Whereby we had the design logo was one, which we've already mentioned.\nSpeaker A: We've got...\nSpeaker A: The remote was only for the television and not for...\nSpeaker A: Because that would make it too complex and we have to get out market quickly.\nSpeaker A: And the third thing was that teletext as far as the management is concerned has become outdated due to the popularity of the internet.\nSpeaker A: So, that means that...\nSpeaker A: So, these are the sort of three extra parameters that have been put on this project.\nSpeaker A: So, we're being focused effectively directly at a television and it seems to me that the management is wanting us to go down a narrow path and not opening out.\nSpeaker A: So, anything that is to be added such as voice recognition, etc.\nSpeaker A: Has to be very simple and has to be very quick because time to market is critical.\nSpeaker C: As soon as we could get something in which was quite quick and simple, I would give us an advantage of it a little more.\nSpeaker C: It would.\nSpeaker A: But probably quick and simple is primarily rather than added extras.\nSpeaker A: Added extras would be nice.\nSpeaker A: But the primary considerations to get the project finished within this short time window, which effect when now is four hours.\nSpeaker A: So, and we've got to get to the end.\nSpeaker A: I think first and foremost, we've got to get to the end and get to the end with some added extras if possible.\nSpeaker A: Okay.\nSpeaker A: One screw is this.\nSpeaker A: So, then I'm coming.\nSpeaker A: Okay.\nSpeaker A: So, I need to...\nSpeaker A: Right.\nSpeaker A: So, I don't know how long we have left over time.\nSpeaker A: But we have to make the decisions on the remote control functions and how we're planning to proceed so that at the next meeting, each person that's got a task to do is clear from this meeting what that task is.\nSpeaker A: And also, when the next meeting is, I...\nSpeaker A: So, we'll know how long we've got to complete that task.\nSpeaker A: And then we can report back at the next meeting and say, right, okay, yes, we've achieved this or we haven't achieved this.\nSpeaker A: This is how far we've progressed.\nSpeaker A: Does that make reasonable sense?\nSpeaker A: Yeah.\nSpeaker A: So, we have to come effectively to the decision on the remote functions so that you can decide what you're going to be doing.\nSpeaker A: And if between the time of this meeting finishing and the next meeting starting, if you get any additional information that only you have at that point in time, you think would be relevant to other people in terms of their decision making process, then we should communicate that as quickly as possible and not wait until the next meeting, do it via the email so that rather than coming, if you get the information just before the next meeting, that's fine.\nSpeaker A: Come along with the next meeting, we can discuss it then and take whatever actions appropriate.\nSpeaker A: But if you get it well before the next meeting, let everybody else know, because that might have an impact on their...\nSpeaker A: What they come up with effectively at that next meeting.\nSpeaker A: Okay.\nSpeaker A: All right.\nSpeaker B: So, do we need to decide on the functions now?\nSpeaker A: I would guess so.\nSpeaker D: I think it would be really easy, if we did have some sort of whistle-bath kind of function.\nSpeaker D: Because that also kind of the frustration of losing it.\nSpeaker B: Yeah, that was the number one frustration people said.\nSpeaker B: I don't think there's anything else on the market that does that.\nSpeaker D: Yeah.\nSpeaker D: I don't really know about the voice recognition thing.\nSpeaker C: Well, maybe we should concentrate just on the whistle-bath.\nSpeaker A: Something simple.\nSpeaker A: The primary consideration is to get it there in time.\nSpeaker A: Time short, you want something to meet the major concerns of the consumer so that we can have it as a selling point for the product.\nSpeaker A: Something that's quick and simple.\nSpeaker A: Yeah.\nSpeaker D: And that one would put off the kind of older generation either, because everyone would whistle a clap.\nSpeaker D: And they wouldn't have to be kind of scared of this retent\nSpeaker A: other than a whistle would be on the basis that if we've got, if we're catering to the whole age range, you want something that's easy to do. Something that, well, I don't know, if you think that more people can whistle than clap, then that's fine.\nSpeaker A: They can go for that option.\nSpeaker A: But if I would have thought that more people could clap rather than whistle, so clap option.\nSpeaker A: Yeah.\nSpeaker A: Okay.\nSpeaker D: We've already decided that we don't need a teletext, but we haven't yet.\nSpeaker A: Effectively, that's what they're saying.\nSpeaker A: If people are now using the internet, then you don't need teletext.\nSpeaker A: So take out teletext.\nSpeaker C: Do we decide on having the 10 numbers and then the little good gen X to it, which kind of enables you to put into account?\nSpeaker B: I think nowadays you can just get ones where it gives you a second or two to press another number.\nSpeaker B: Yeah.\nSpeaker B: So you can press any two and it will sort of put them together.\nSpeaker C: Okay.\nSpeaker C: 10 numbers and then some kind of device to allow a, yeah.\nSpeaker C: I'll put delay to a multiple number.\nSpeaker C: A multiple number.\nNone: Yeah.\nSpeaker C: Do we decide on the other functions as in setting the audio and tuning it and stuff like that?\nSpeaker C: You have an idea about the menu?\nSpeaker B: We could possibly put an out there, sort of the LCD menu in, but that again is probably an expense.\nSpeaker C: But just thinking, um, people probably, I mean, you don't have to have a opportunity to use one, but you have to be able to tune it that one.\nSpeaker C: Yeah.\nSpeaker C: So in a family TV break, you're going to have to be able to tune it.\nSpeaker C: You can't really avoid that.\nSpeaker C: No.\nSpeaker C: You decide the new dish the market is pushed to it by themselves.\nSpeaker C: Yeah.\nSpeaker A: But that's relying on the television market changing to an automatic, and if it is at the moment, that's fine.\nSpeaker A: But at the moment, it's not. So it seems to me that you have to have a device that caters, because otherwise it would make it your device would become, you know, probably or only operable in certain circumstances and the idea is to have an international market, which is.\nSpeaker A: Yeah.\nSpeaker A: And it's also, it's the product we've got is something that's the, I would have said the lower end of the cost scale. So we're not really going for something that's.\nSpeaker A: I suppose.\nSpeaker B: If people are buying remotes, then they're probably buying it to replace another remote.\nSpeaker B: Possibly.\nSpeaker B: Most of the tellies come with remote. So that's right.\nSpeaker B: And we're maybe talking about the placing remotes for slightly older television. So we maybe need to keep the tuning functioning.\nSpeaker C: Yeah.\nSpeaker C: So how does manual function with you? Maybe have like one menu button and you use the other buttons and then the other buttons to actually use other functions.\nSpeaker C: Yeah.\nSpeaker D: And I'm going to use it.\nSpeaker D: Yeah.\nSpeaker D: Yeah.\nSpeaker D: Because you do need a kind of brightness and contrast mode as well. My dad was watching the film the other week when it was too dark.\nSpeaker D: So I had to go through it and tell the brightnesser.\nSpeaker C: So we're going to have the individual numbers and then the menu function and maybe sort of slightly more advanced instruction throughput.\nSpeaker C: And with it, guys, I think it would be quite hard just for people to grasp just stuff like the menu that you get on different buttons.\nSpeaker C: So we have to have some better instructions about how that actually works.\nSpeaker B: I'm not sure whether sort of having people have a booklet because the second most annoying thing that people found was to learn the new one.\nSpeaker B: Right.\nSpeaker B: So maybe next to each of the buttons, each of them could have a number and then also a function right next to it.\nSpeaker B: So you're basically pressing.\nSpeaker C: But that also goes back to the original design also those two and there was all in the left hand side which had a little double function instead which can look too busy and run a shot.\nSpeaker C: Okay.\nSpeaker D: Well.\nSpeaker D: Well.\nSpeaker D: It's a very good idea.\nSpeaker A: So you keep.\nSpeaker A: We've got five minutes before we wind up those meeting.\nSpeaker A: So I've been told I don't know if you've got the same.\nSpeaker A: Okay.\nSpeaker A: So so keep.\nSpeaker A: Keep detailed functions.\nSpeaker A: Hidden at the back.\nSpeaker D: So we're going to have to work out what's going to be on these other functions as well.\nSpeaker C: So we're going to have like two set.\nSpeaker C: That's right.\nSpeaker C: Two set the list.\nSpeaker A: Yeah.\nSpeaker A: You do two functions hidden at back.\nSpeaker A: Can bring out when needed.\nSpeaker B: The detailed ones will be sort of.\nSpeaker B: Right.\nSpeaker A: That's right.\nSpeaker A: So we're just so you've got which ones are going to be on the front which ones are going to be at the back.\nSpeaker C: We have to decide.\nSpeaker C: Okay.\nSpeaker A: So.\nSpeaker C: So on the.\nSpeaker A: So on the front.\nSpeaker C: Numbers.\nSpeaker C: Do you have a mute button as well?\nSpeaker C: Sorry.\nSpeaker C: A mute button as well.\nSpeaker E: Yeah.\nSpeaker B: And probably power one as well.\nSpeaker D: I'm not a totally nice.\nSpeaker D: Not in a shooting.\nSpeaker D: Yeah.\nSpeaker D: But I'm not using standby.\nSpeaker C: Because of the.\nSpeaker C: Like we have to have a similar address.\nSpeaker D: Like if you boil the kettle last four.\nSpeaker D: Yeah.\nSpeaker A: So are you having the standby on the front then?\nSpeaker D: I think you probably should.\nSpeaker A: Well.\nSpeaker C: We're going to have a channel up and down as well as a number of buttons.\nSpeaker B: Because yeah the market research said there was quite a lot of people to just zap around and flick.\nSpeaker C: So we've got ten.\nSpeaker C: Channel down.\nSpeaker A: What else we got.\nSpeaker B: What was that six numbers is ten volume is.\nSpeaker B: Volume but any volumes.\nSpeaker B: One up one down.\nSpeaker A: Okay.\nSpeaker D: Only.\nSpeaker B: That's sixteen.\nSpeaker C: So anything else.\nSpeaker B: I don't think so.\nSpeaker A: I'm going to put a button standby channel up and down.\nSpeaker A: So is that it?\nSpeaker A: Yes.\nSpeaker A: Okay.\nSpeaker A: That's sixteen buttons you reckon.\nSpeaker A: And then at the back.\nSpeaker B: You've got brightness and contrast.\nSpeaker C: Maybe we're going to run out of time one of the list of these and then get back and then like to make them just at the start.\nSpeaker C: Okay.\nSpeaker A: So on the back.\nSpeaker A: You're also going to have the channel tuner in there as it were.\nSpeaker A: Tuner would that have a button down?\nSpeaker C: Okay.\nSpeaker A: Okay.\nSpeaker B: And then maybe sort of an enter button for sort of saying that you want that particular thing tuned in.\nSpeaker B: So you go up and down and then it finds that there's only one of this way.\nSpeaker B: Enter.\nSpeaker B: Select.\nSpeaker C: I guess we're keeping simple.\nSpeaker C: I don't mean any other options because it's just volume up.\nSpeaker C: I would have asked.\nSpeaker D: So I think there's quite a lot of like doggies around, swedling, surveillance, sort of things.\nSpeaker C: Yeah.\nSpeaker C: So they have their own controls on their actual products.\nSpeaker C: What do you have to do via the remote?\nSpeaker D: I think they've got their own controls in this kind of like hidden panel.\nSpeaker C: As soon as we got there they've got their own controls and we can avoid it.\nSpeaker C: Yeah.\nSpeaker D: Oh no, I mean, like this kind of individual.\nSpeaker D: Oh, okay.\nSpeaker D: Like on the TV.\nSpeaker C: But I don't know what I've never used them.\nSpeaker A: I don't know there's something to do with.\nSpeaker A: Maybe less.\nSpeaker A: It comes up.\nSpeaker C: Well, you might get some research on.\nSpeaker A: What should we look into that and then just get back to the.\nSpeaker A: Okay.\nSpeaker A: Right.\nSpeaker A: So I'll do the minutes of this meeting.\nSpeaker A: I'm not sure.\nSpeaker A: 40 minutes, I believe is the time.\nSpeaker B: Okay.\nNone: Okay.\nSpeaker A: Come on.\nNone: Okay.\n"}]
\ No newline at end of file